63378 THE WORLD BANK Research Observer EDITOR Emmanuel Jimenez. World Bank CO-EDITOR Luis Serven. World Bank EDITORIAL BOARD Harold Alderman. World Bank Barry Eichengreen. University of California-Berkeley Marianne Fay. World Bank Jeffrey S. Hammer. Princeton University Ravi Kanbur. Cornell University Howard Pack. University of Pennsylvania Ana L. Revenga, World Bank Sudhir Shetty, World Bank The World Bank Research Observer is intended for anyone who has a professional interest in development. Observer articles are written to be accessible to nonspecialist readers; con­ tributors examine key issues in development economics, survey the literature and the lat­ est World Bank research. and debate issues of development policy. Articles are reviewed by an editorial board drawn from across the Bank and the international community of econo­ mists. Inconsistency with Bank policy is not grounds for rejection. The journal welcomes editorial comments and responses, which will be considered for pub­ lication to the extent that space permits. On occasion the Observer considers unsolicited contributions. Any reader interested in preparing such an article is invited to submit a proposal of not more than two pages to the Editor. Please direct all editorial correspon­ dence to the Editor. The World Bank Research Observer, 1818 H Street, NW. Washington, DC 20433, USA. The views and interpretations expressed in this journal are those of the authors and do not necessarily represent the views and policies of the World Bank or of its Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of data included in this publication and accepts no responsibility whatsoever for any consequences of their use. When maps are used. the boundaries. denominations, and other information do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. For more information, please visit the Web sites of the Research Observer at www.wbro.oxfordjournaLs.org, the World Bank at www.worldbank.org, and Oxford University Press at www.oxfordjournaLs.org. THE WORLD BANK Research Observer Volume 25 • Number 1 • February 2010 Symposium on Agriculture and Development Agricultural Growth and Poverty Reduction: Additional Evidence Alain de Janvry and Elisabeth Sadoulet 1 Policy Reforms Affecting Agricultural Incentives: Much Achieved, Much Still Needed Kym Anderson 21 Agricultural Employment Trends in Asia and Africa: Too Fast or Too Slow? Derek Headey, Dirk Bezemer. and Peter B. Hazell 57 Symposium on Implications of Economic Crises Are All the Sacred Cows Dead? Implications of the Financial Crisis for Macro- and Financial Policies Ash Demirgiic;-Kunt and Luis Serven 91 The 2007 Meltdown in Structured Securitization: Searching for Lessons, not Scapegoats Gerard Caprio. Jr.. Ash Demirgiic;-Kunt. and Edward J. Kane 125 Higher Food Prices in Sub-Saharan Africa: Poverty Impact and Policy Responses Quentin Wodon and Hassan Zaman 157 Subscriptions A subscription to The World Bank Research Observer (ISSN 0257-3032) comprises 2 issues. Prices include postage; for subscribers outside the Americas, issues are sent air freight. Annual Subscription Rate (Volume 25, 2 issues, 2010) Academic libraries Print edition and site-wide online access: US$180/£120/€180 Print edition only: CS$165/£llO/€165 Site-wide online access only: US$1501L1l0/€150 Corporate Print edition and site-wide online access: US$272/£181/€272 Print edition only: US$250/£167/€2S0 Site-wide online access only: US$227/£1S1/€227 Personal Print edition and individual online access: US$55/£37/€55 Please note: US$ rate applies to US & Canada. Euros€ applies to Europe, UK£ applies to UK and Rest of World. Readers with mailing addresses in non-OECD countries and in sOdl1list economies ill urmsitioll are eJigible to rueive complimentary subscriptions on request by writing to the UK address below. There may be other subscription rates available; for a complete listing. please visit www.wbro.oxfordjournals.org/subscriptions. Full pre-payment in the correct currency is required for all orders. Payment should be in l'S doliars for orders being delivered to the USA or Canada: Euros for orders being delivered within Europe (excluding the EK); GBP sterling lor orders being delivered elsewhere (i.e .. not being delivered to USA. Canada. or Europe). All orders should be accompanied by full payment and sent to your nearest Oxford Journals office. Subscriptions are accepted for complete volumes only. Orders are regarded as firm. and payments are not refundable. Our prices include Standard Air as postage outside of the l'K. Claims must be notified within four months of despatch/order date (whichever is later). Subscriptions in the EEC may be subject to European \\>\T. If registered. please supply details to avoid unnecessary charges. Por subscriptions that include online versions. a proportion of the subscription price may be subject to UK \i\T. Subscribers in Canada. please add GST to the prices quoted. Personal rate subscriptions are only available if payment is made by personal cheque or credit card. delivery is to a private address. and is for personal use only Back issues: The current year and two previous years' issues are available from Oxford LTniversity Press. Previous volumes can be obtained from the Periodicals Service Company. 11 Main Street. Germantown. NY 12526. USA. F.-mail: psc@periodicals.com. Tel: (518) 537-4700. Fax: (518) 537-5899. Contact information: Journals Customer Service Department. Oxford l'niversity Press. Great Clarendon Street. Oxford OX2 6DP, UK. E-mail: jnls.cust.serv@oxfordjournals.org. Tel: +44 (O)l865 353907. Fax: +44 (0)1865353485. In the Americas, please contact: Journals Customer Service Department. Oxford Cniversity Press. 2001 Evans Road. Cary. NC 27513. USA. F.-mail: jnlorders@oxfordjournals.org. Tel; (800) 852-7323 (toll-free in USA/Canada) or (919) 677-0977. Fax: (919) 677-1714. In Japan, please contact: Journals Customer Service Department, Oxford University Press. 4-5-1 0-8F Shiba. Minato-ku. Ibkyo. 108-8386. Japan. E-mail: custserv.jp@ox[ordjournals.org. Tel: +81 3 5444 5858. Fax: +81 3 34542929. Postal information: The World Bank Research Observer (lSSN 0257-3032) is published twice a year. in February and August. by Oxford University Press for the International Bank for Reconstruction and Development/THE ImBLD RA'iK. Postmaster: send address changes to Tile World Bank Research Observer. Journals Custnmer Service Department. Oxford University Press. 200] Evans Road, Cary. "IC 27513-2009. Periodicals postage paid at Cary. NC and at additional mailing offices. Communications regarding original articles and editorial management should be addrcsstxl to The Editor. The World Bank Research Observer. The World Bank. 1818 H Street. N\-V, Washington, D.C. 20433, USA. Oxford Journals Environmental and Ethical Policies: Oxford Journals is committed to working with the global community to bring the highest quality research to the widest possible audience. Oxford Journals will protect the environment by implementing environmentally friendly policies and practices wherever possible. Please see http://www.oxfordjournaL,.org/ ethicalpolieies.html for further information on Oxford Journals' environmental and ethical poliCies. Digital Object Identifiers: For information on dois and to resolve them, please visit www.doi-org. Permissions: For information on how to request permissions to reproduce articles or infonnation from this journal. please visit www.oxfordjournals.org/jnls/permissions. Advertising: Advertising. inserts. and artwork enquiries should be addressed to Advertising and Special Sales. Oxford Journals. Oxford University Press. Great Clarendon Street. Oxford. OX2 6DR UK. Tel: +44 (0)lH65 354767; Fax: +44(0)1865 353774; E-mail: jnlsadvertising@oxfordjournals.org. Disclaimer: Statement, of [act and opinion in the articles in The Worlli Bank Researdl Observer are those of the respective authors and contributors and not of the International Bank for Reconstruction and Development/TIlE WORLD BANK or Oxford University Press. Neither Oxford Cniversity Press nor the International Bank for Reconstruction and Development/THE WORLD SA'>K make any representation. express or implied. in respect of the accuracy of the material in this journal and cannot accept any legal responsibility or liability for any errors or omissions that may be made. The reader should make her or his own evaluation as to the appropriateness or otherwise of any experimental technique described. Paper used: The World Bank Research Observer is printed on acid-free paper that meets the minimum requirements of ANSI Standard Z39.48-1984 (Permanence of Paper). Indexing and abstracting: The World Bank Research Observer is indexed and/or abstracted by ABilINI'ORM. [,AB Abstracts. Current Con!ents/Social and Behavioral SCiences. Journal of Economic UteraturelEconLit. PiUS International. RePEc (Research in Economic Papers), Social Services Citation Index. and Wilson Business Abstracts. Copyright 2010 The International Bank for Reconstruction and DevelopmentlTHE WORLD BANK All rights reserved: no part of this publication may be reproduced. stored in a retrieval ''Ystem, or transmitted in any form or by any means. electronic, mechanical, photocopying. recording. or otherwise without prior written permission of the publisher or a license permitting restricted copying issued in the UK by the Copyright Licensing Agency Ltd. 90 Toltcnham Court Road. tondon WIP 9HE. or in the CSA by the Copyright Clearance Center. 222 Rosewood Drive, Danvers. MA 01923. Typeset by Techset Composition Limited. Chennai. India: Printed by Edwards Brothers Incorporated. USA Agricultural Growth and Poverty Reduction: Additional Evidence Alain de Janvry • Elisabeth Sadoulet Agricultural growth has long been recognized as an important instrument for poverty reduction. ret, measurements of this relationship are still scarce and not always reliable. The authors present additional evidence at both the sectoral and household levels based on recent data. Results show that rural poverty reduction has been associated with growth in yields and in agricultural labor productivity, but that this relation varies sharply across regional contexts. GDP growth originating in agriculture induces income growth among the 40 percent poorest, which is on the order of three times larger than growth originating in the rest of the economy. The power of agriculture comes not only from its direct poverty reduction effect but also from its potentially strong growth linkage effects on the rest of the economy. Decomposing the aggregate decline in poverty into a rural contribution, an urban contribution, and a popUlation shift component shows that rural areas contributed more than half the observed aggregate decline in poverty. Finally, using the example of Vietnam, the authors show that rapid growth in agriculture has opened pathways out of poverty for farming households. While the effectiveness of agricultural growth in reducing poverty is well established, the effectiveness of public investment in indUcing agricultural growth is still incomplete and conditional on context. JEL codes: 013, 131 Poverty reduction can be achieved through two instruments: transfers and pro­ poor growth. Transfers require foreign aid or taxation of the incomes accruing to the non-poor and distribution through a variety of social programs. It has been widely used. especially in dealing with emergencies or to achieve quick poverty reduction results, and to address poverty among categories of the population that could not generate autonomous incomes even under the best of circumstances. It has the appeal of relative ease of implementation compared to promoting auton­ omous income growth of the poor. Sustained poverty reduction through The World Bank Research Observer © The Author 2009. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development I THE WORIll MNK. All rights reserved. For permissions. please e-mail: journals.permissions@oxfordjournals.org doi; 10. 109 3/wbro/lkp015 Advance Access publication November 9, 2009 25: J-20 redistribution is. however. politically difficult to implement. particularly when poverty is a mass phenomenon. as it is in most developing countries. It can be hugely expensive if it has to be sustained year after year to reduce poverty signifi­ cantly. And it is not a dignified way of dealing with poverty when the poor have the capacity to generate autonomous incomes. if they are given the chance. For this reason, pro-poor growth, that is growth that benefits the poor,l is the better alternative to poverty reduction for those who can work. This. however, raises the question of identifying the pathways through which growth helps reduce poverty, not surprisingly one of the most fundamental topics in development economics. For producers. how growth helps reduce poverty depends on access to assets and on how they are able to use these assets for income generation. For rural workers, it depends on the ability to link to expanding employment opportunities in good jobs in agriculture and the rural non-farm economy. As shown by Lipton (1991), the Green Revolution in Asia increased land productivity faster than labor productivity, with the result that agriculture was able to absorb more labor and help reduce poverty. For consumers, if agriculture is incompletely tradable, growth in food production can help lower the domestic prices of consumption goods and raise real incomes. This will benefit the urban poor. landless rural workers. and the many poor net-buyers among smallholders. Recent estimates show that a majority of smallholders are in fact net buyers, benefiting more from a decline than from a rise in the price of food. The main long-run effect of growth in cereal yields on poverty reduction in India. in a context of non-tradability. was through a decline in the price of food (Datt and Ravallion 1998). Conditions are changing. however. With increasing tradability of agriculture. productivity gains in agriculture will be transmitted increasingly less via lower food prices, and increasingly more through higher employment and wages (Valdes and Foster 2007). Growth can thus offer a multiplicity of pathways out of poverty. These pathways depend on the sector where growth occurs. broadly agri­ culture. industry. or services. And they depend on the structure of production, in particular asset distribution among producers (farm or firm size) and the labor intensity of production. In this paper, we present new evidence on the capacity of agricultural growth to serve as an effective instrument for poverty reduction. We look at: the poverty reduction value of land and labor productivity growth and of GDP growth orig­ inating in agriculture versus the rest of the economy; the comparative linkage value of a quantum of sectoral growth for aggregate growth and poverty reduction; the contribution made by rural areas to aggregate poverty reduction under alternative migration scenarios; and the household pathways out of poverty in the context of aggregate growth, in particular via market-oriented smallholder farming. The key relation between public investment and sectoral growth response needs to be determined in order to decide when to use an 2 The IVorld Bank Research Observer. vol. 25, 110, 1 (February 2010) agriculture-first poverty reduction strategy. While this relation is central to deciding on use of agriculture for development. it remains difficult to establish and dearly conditional on the circumstances where it applies. We review evidence indicating that there are many situations where investing in agriculture for poverty reduction may be the preferred strategy. Productivity Growth and Rural Poverty: Regularities Productivity gains are the main mechanism whereby growth is achieved. In agri­ culture, most important are land and labor productivity. Regularities in the pro­ ductivity-poverty relation for these two types of productivity gains are suggestive of what agriculture can do for poverty reduction. Land Productivity and Poverty In agriculture, yield increases are the main source of output growth once the agricultural frontier has been exhausted. This is the case in East Asia where new land for area expansion is hardly available. This is also increasingly the case in Africa, where population pressure on the land and the increasing speed of rotations between cultivation and fallow periods needs to be compensated by rising yields to maintain output. Rising yields thus support output gains which in turn can increase incomes in self-employment and employment opportunities for those on the labor market. In figure 1, observations on cereal yields measuring Figure 1. Cereal Yields and Rural Poverty 140 y-----------------, Eas[era Eumpe and Celliral A!:rill ~ ! ~ IOD...-__."""'"--------------j ~ 121l L - - - - f - - - - ' ' ' ' " ' - - - - - - - - - - - j " g; 8i 90 +--+-----"'<--~;:__-------4 ,; 100 f--~~~~~=5~:::::::O'~~~=-1 !j Brazil ~ :¥ 8u t-----'r--------;::::,..-=:::::---' ~ ~ 15 80 ~ ~ ~ e ~ 60~----~~---------_1 ~L----~-----_- __ ---~ 95 IOU 105 110 J 15 120 125 130 95 100 105 110 115 120 125 130 US l40 Cereal Yields index, 1993=100 Cereal yields index, 1993=100 Note: Observations are for 1993, 1996, 1999, and 2002. Sources: Poverty data from Ravallion, Chen. and Sangraula (200;) using a Sl.08/day poverty line in 1993 PPP: yield data from FAO (2006), de Janvry alld Sadoulet 3 average land productivity and rural poverty indices are reported for 1993, 1996, 1999. and 2002, with a base of 100 in 1993 for each of five regions and three major countries. The expected inverse relationship between rising yields and falling rural poverty is visible. It does. however. vary widely across regions. In East Asia. a 10 percent growth in cereal yields is accompanied by a decline in rural poverty of more than 53 percent. In Eastern Europe and Central Asia. after a tran­ sition over which yields were stagnant and poverty rose. subsequent yield gains were associated with a rapid decline in rural poverty. In Latin America and the Caribbean. gains in cereal yields were very large. growing at an average annual rate of 2.5 percent. yet rural poverty hardly declined. Clearly, the way yield gains were achieved did not help reduce poverty. In Sub-Saharan Africa. yields were largely stagnant and the rural poverty rate remained unchanged. Similar patterns are observed at the country level. with elasticities of rural poverty reduction with respect to cereal yield growth equal to - 5.1 in China. -1.2 in India. and -0.6 in Brazil. These are simple correlates. yet the contrast is telling of how land pro­ ductivity gains can matter for rural poverty reduction. but differentially across the contexts in which they occur. The China - Brazil contrast is revealing of the impor­ tance of a more egalitarian land tenure system in transmitting land productivity gains into poverty reduction. Labor Productivity and Poverty Labor productivity in agriculture is also a major determinant of agricultural incomes. It can increase as a consequence of technological change in agriculture or of out-migration from agriculture. The labor productivity-poverty relation can, however. be quite different across countries according to the production structure: strong if smallholders participate in the gains in labor productivity and if agricul­ ture is labor intensive; weak if otherwise. This is exemplified in figure 2 by the contrasts in the labor productivity-rural poverty relation over the period 1993 2002 across major regions and countries. Labor productivity is measured by the average value added in agriculture per worker in the sector. The two extreme cases are East Asia and Latin America and the Caribbean. In East Asia. labor pro­ ductivity gains were large and the rural poverty rate fell sharply. Agriculture is practiced by smallholders and it is labor intensive. In Latin America and the Caribbean. labor productivity gains were very large as well. but rural poverty hardly fell. Agricultural growth in countries such as Brazil occurred mainly in mechanized large farms with little employment creation. Labor productivity was further enhanced by rapid rural-urban migration. leading to an absolute decline in agricultural labor, yet without decline in rural poverty. Other regions span the range between these two extremes. In Sub-Saharan Africa, with high population growth and limited employment opportunities. labor productivity gains in 4 The World Bank Research Observer, vol. 25. no. 1 (February 2010) Figure 2. Agricultural Labor Productivity and Rural Poverty 1I0,..--_ _ _ _ _ _ _ _ _ _ _- - , J40,..----------------, l()+--_~--~·--_--.....,-----1 95 100 105 110 115 120 125 130 9f) 110 U() 150 170 Agricultural labor productivity index. 1993=100 Agricultural tabm productivity index, 1993=100 Note: Observations are for 1993, 1996, 1999, and 2002. Sources: Poverty data from Ravallion. Chen, and Sangraula (2007) using a $1.08/day poverty line in 1993 PPP; agricultural labor productivity data from World Bank (2006). agriculture were low, and poverty reduction was equally low. In South Asia, India most particularly, low rural-urban migration rates and low growth in agricul­ tural production during the period that followed the Green Revolution also atro­ phied productivity gains. Permanence of rural poverty and rising disparities between rural and urban incomes as growth accelerates in other sectors of the economy are a major politi­ cal issue. In Eastern Europe and Central Asia, labor productivity fell during the transition out of collective farming into a market economy, but it was sub­ sequently followed by rapid labor productivity gains and sharp rural poverty reduction. The elasticities of poverty reduction with respect to agricultural labor productivity growth are -1.2 in China, -1.2 in India, and -0.3 in Brazil. Differences in these correlates show that labor productivity gains in agriculture can be quite effective for poverty reduction, but that the structural conditions under which agricultural growth occurs matter for the poverty reduction effect it can have. This in turns tells us that policy instruments can be used to enhance the poverty reduction value of agricultural growth. The Power of Growth Originating in Agriculture for Poverty Reduction: Causalities Reliable estimates of the growth-poverty relation are few as identification of caus­ ality in this relation is difficult to establish. Different studies used different indi­ cators of outcome (income of the poor, poverty rates) and different concepts of de Janvry and Sadoulet 5 growth (agricultural labor productivity. sectoral value added). Results in general support the high poverty reduction capacity of agricultural growth. However. because concepts used are different. results are not directly comparable. Bravo-Ortega and Lederman (2005) estimated the effect of an increase in sec­ toral labor productivity on GDP growth and the income of the poor. They found that overall GDP growth originating in an increase in agricultural labor pro­ ductivity is on average 2.9 times more effective in raising the income of the poorest quintiles in developing countries than an equivalent increase in GDP growth originating in non-agricultural labor productivity. Christiaensen and Demery (2007) estimated the effect of sectoral growth on the headcount poverty rate rather than on the income of the poorest. They found for Africa that overall GDP growth coming from agriculture is 2.7 times more effective in reducing l$/day poverty in the poorest quarter of countries in their sample. and 2 times more effective in the richest quarter of countries, than growth coming from non­ agriculture. Ravallion and Chen (2007) estimated the effect of sectoral growth on the headcount poverty rate in China using annual poverty data over 21 years. They find that the primary sector has a 3.5 times larger impact on poverty reduction that either the secondary or tertiary sectors. Using cross-country data for 55 countries with spells of observations, Loayza and Raddatz (forthcoming) show that what matters for the poverty reduction capacity of growth is the unskilled labor intensity of a sector. In that perspective. agriculture comes ahead of industry and services. Growth originating in agriculture is 2.9 times more poverty reducing than growth originating in manufacturing and 1.8 times that of growth originating in construction. For the World Development Report 2008 on Agriculture for Development, Ligon and Sadoulet (2007) estimated the expenditure growth effect for each household decile in the distribution of expenditures due to GDP growth originating in the agricultural sector and to GDP growth originating in the non­ agricultural sectors, respectively. These estimations are obtained from the information available in the World Bank's PovCal database (World Bank 20(8) for 42 countries that have at least three expenditure surveys over the period 1978 to 2003. Estimations are done with rigorous econometric methods that ensure that the results can be interpreted as a causal effect of sectoral growth on house­ hold expenditures. and are shown to be robust to a variety of specification checks? Results indicate that GDP growth originating in agriculture has a much larger positive effect on expenditure gains for the poorest households than growth orig­ inating in the rest of the economy. Figure 3 shows the relative strength of these effects measured as the ratio of the estimated coefficients of agricultural and non­ agricultural growth on household expenditures. Overall growth originating in agriculture is estimated to be at least three times as effective in reducing poverty 6 The Rhrld Bank Resfarc/l Observer. vol. 25. 110. 1. (February 2010) Figure 3. Expenditure Effects of GDP Growth Originating in Agriculture Relative to Non­ agriculture across Expenditure Deciles, from Poorest to Richest 8 7 o +---~--~--~--~---4---'----r---~------~ o 2 3 4 5 6 7 8 9 to Expenditure deciles Source: Ligon and Sadoulet (2007). as overall growth originating in the rest of the economy. This statement is based on the relative impacts of growth from agriculture and non-agriculture on the expenditures of the poorest four deciles which have a median value of 3.1. The relative impact is significantly different from 1 for the poorest 50 percent of the population. Further sectoral disaggregation of non-agriculture shows that other sectors can also have high poverty reduction value, and that this varies across regions. Thus, Hasan and Quibria (2004) found that. while growth in agriculture is most effec­ tive for poverty reduction in Sub-Saharan Africa and South Asia. growth in industry is most effective for East Asia and in services for Latin America. Ravallion and Datt (1996) and Foster and Rosenzweig (2005) for India, and Suryahadi, Suryadarma, and Sumarto (2008) for Indonesia. all find that agricul­ tural growth is key to reducing poverty in rural areas. But they also find that informal services. rural factory employment. and both urban and rural services, respectively, have important impacts on rural poverty reduction, complementing the role of agriculture. Loayza and Raddatz (forthcoming) singled out growth in construction as the most poverty reducing sector in non-agriculture. expectedly because it is the next most intensive sector in unskilled labor after agriculture. So, while growth orig­ inating in agriculture has strong powers for poverty reduction. there are other sectors that can be quite effective as well, especially if they are intensive in de Janvry and Sadolilet 7 unskilled labor and are located in the rural non-farm economy. This suggests that a growth strategy for poverty reduction must focus not only on agriculture growth but on the growth of these other strategic sectors as well. Opening the Growth - Poverty BLack Box: The RoLe of Linkages Agricultural growth contributes to both aggregate growth and overall poverty reduction through two effects: directly as a sector of economic activity. and indirectly through growth linkages with non-agriculture. What are the relative contributions to growth and poverty reduction of each of these two effects? In this section, we compare the aggregate growth and poverty effects of a one percent growth in both agriculture and non-agriculture. The absolute levels of these effects are obviously affected by the sizes of these sectors. However. the focus of this section is on the relative importance of the direct and linkage effects, which is not. The role of linkages is illustrated in figure 4 with results for China over the 1980-2001 period. This was a time of rapid growth, not only for the non-agri­ cultural sector (growing at an average 9.3 percent annual rate) but also for the agricultural sector (growing at an average 4.6 percent annual rate). where Figure 4. Estimates of the Agricultural Growth-Non-agricultural growth poverty linkages for China. 1980-2001 Non-ag. Indirect contributions growth 0.29% To growth: 0.29*0.78=0.23% To poverty reduction: -2.25*0.23=-{l.52% Direct contributions To growth: 0.22% To poverty reduction: -7.85*0.22~1.73% Direct contributions To growth: 0.78% To poverty reduction: -2.25*0.78=-1.76% Indirect contributions To growth: 0.64*0.22=0.14% To poverty reduction: -7.85*0.14=-1. \0% 8 The World Bank Research Observer. vol. 25. no. 1 (February 2010) Table 1. Direct and Indirect Contributions of Sectoral Growth to Aggregate Growth and Poverty Reduction in China. 1980-2001 ~~~~~----------------- Contributions to Contributions to growth (%) poverty (%) - -.... --~ Sectoral growth Aggregate growth Direct Indirect Poverty reduction Direct Indirect ...- -..- ...- -.. ---~ Agriculture 1 % 0.45 49 51 2.24 77 23 ;-';on-agriculture 1 % 0.92 85 15 -2.85 62 38 growth was driven by improved incentives (the household responsibility system replacing collective farms. and domestic market liberalization replacing regional food self-sufficiency). During these years, the sectoral shares of GDP were on average 22 percent for agriculture and 78 percent for non-agriculture. A one percent growth of the smaller agricultural sector induces a 0.29 percent growth in the much larger non-agricultural sector. 3 This growth in the non-agricultural sector amounts to 0.2.3 percentage points of aggregate economic growth. Conversely, a 1 percent growth of the non-agricultural sector induced a 0.64 percent agricultural growth. This added 0.14 percent points to aggregate econ­ omic growth, a smaller indirect contribution due to the lower share of the agri­ cultural sector in GDP. Given the relative sizes of the two sectors. these multipliers can also be read as $1 growth in agriculture inducing $1 growth in non-agricul­ ture, while $1 growth in non-agriculture induced $0.18 in agriculture, showing the very strong growth linkages arising from agriculture at that particular time in China. Combining the direct and linkage effects shows that a 1 percent growth in agriculture has an aggregate growth effect of 0.45 percent. lower than the 0.92 percent aggregate growth induced by a 1 percent growth in the .3.5 times larger non-agricultural sector. In terms of poverty reduction. the growth elasticities were estimated by Ravallion and Chen (2007) to be -7.S5 for agriculture and - 2.25 for non-agri­ culture. As a result. a 1 percent growth in agriculture would induce a direct reduction in the poverty rate of 1.7.3 percent, about the same as the 1.76 percent direct contribution induced by a 1 percent increase in non-agriculture. This is despite the fact that the share of agriculture in GDP is only 22 percent. Combining the direct and indirect effects gives an overall poverty reduction of 2.24 percent following a 1 percent growth in agriculture, and 2.85 percent fol­ lowing a 1 percent growth in non-agriculture. The structure of direct and indirect contributions to aggregate growth and poverty reduction coming from a 1 percent sectoral growth is presented in table 1. The remarkable feature is the large indirect contribution of agriculture to growth (51 percent of the total effect). while for non-agriculture the largest con­ tribution is direct (85 percent). The effect is the opposite for poverty: agriculture de /urlVry and Sadoulet 9 has a large direct contribution to overall poverty reduction (77 percent), while it is non-agriculture that has the relatively larger indirect effect (38 percent). Linkage effects of agriculture on the rest of the economy are thus important for growth; direct effects are important for poverty reduction. Finally, if we return to a comparison of the poverty reduction value of a 1 percent GDP growth coming from agriculture versus non-agriculture, we see that the first contributes a 10.2 percent reduction in poverty while the latter contrib­ utes 3.7 percent. We thus rediscover for China during the 1980~2001 period the result obtained by Ligon and Sadoulet (2007) using cross-country data: GDP growth originating in agriculture is about three times (2.8 times for China) more effective for poverty reduction than growth originating in non-agriculture. These particular results are specific to China in the 1980~2001 period. They show that agriculture poverty reduction effects are relatively more direct than its growth effects. The importance of the linkage effects on non-agriculture as opposed to the direct effect is largely related to the mere size of the agricultural sector that implies that most of its linkages are externalized to the other sectors. The fact that this is not so for the poverty effect reveals the fundamental poverty reduction value of agricultural growth. However, simply because of its relatively small share in aggregate GDP. a percentage point growth in agriculture can have less aggregate growth and even less poverty reduction effect than a percentage point growth in the large non-agricultural sector. An Agriculture-first Strategy for Poverty Reduction: Piecing together the Evidence Is it justified to invest public resources in agriculture as the most cost effective option in using growth to reduce poverty? In asking this question, we are not trying to compare the cost effectiveness of transfers versus investments in growth to reduce poverty, only of the latter across sectors of economic activity. If agricul­ ture were the most cost effective investment. this would be the argument in support of an "agriculture-first" strategy for poverty reduction (Suryahadi, Suryadarma, and Sumarto 2008). Not surprisingly. the answer is that it depends on country context. though there are many cases where focusing on agriculture as the preferred strategy is plausible. The empirical evidence presented above allows us to make two strong state­ ments on the role of agricultural growth for poverty reduction. The first is that GDP growth that originates in agriculture (that is for an equal 1 percent of GDP growth) tends to be more effective for poverty reduction than growth that orig­ inates in other sectors of the economy. with unskilled labor intensive activities 10 The World Bank Research Observer, vol. 25, no. 1 (February 2010) located in the rural non-farm economy as strong complementary instruments. The second is that the growth of agriculture makes relatively large indirect contri­ butions to aggregate growth, while its contributions to poverty reduction are larger via direct than indirect effects. Agriculture grm<\<1;h is thus a good servant of aggregate growth and a direct actor for aggregate poverty reduction. But comparison of the poverty reduction value. neither of a given aggregate growth originating in either sector, nor of a 1 percent growth in sectors of differ­ ent sizes, can answer the policy question of whether to invest in agricultural growth to maximize poverty reduction. The key question is how much growth do we get from public investment in agriculture versus investment in other sectors of the economy? This is where the information is still incomplete. due to both con­ ceptual and econometric reasons. Conceptually; it is difficult to define sectoral investment. Most investments. such as infrastructure, health. and education, have strong intersectoral spillovers. Econometrically; it is difficult to consider investment exogenous to growth outcomes, as investments are targeted where growth poten­ tial is the highest. More effective is to go to detailed case studies. Investment in research and development (R&D) for agriculture tends to have large geographical spillovers, creating high rates of returns for such investments. While there is undoubtedly selection of successful cases in measuring rates of return from specific agricultural R&D investments, a large number of success stories shows that high returns are at least broadly possible. with an average 43 percent rate of return across 700 studies, well above the opportunity cost of public investment (Alston and others 2(00).4 Brazil has made significant investments in a premier public agricultural research institution, EMBRAPA, and reaped spectacular land productivity gains in huge areas of the country. fueling its agro-export boom. Investment in rural roads can similarly have high rates of return, but the level of this return and the incidence of gains and losses across households depend importantly on complementary investments and on households' distance from market and asset endowments (van de Walle and Mu 20(7). As should be expected, rates of return to public investment in agriculture thus vary depending on context. and the incidence of gains and losses can be quite unequally distributed. The causal chain running from public investment to agricultural growth. overall grovvth. and poverty reduction thus critically depends on the investment­ growth linkage which remains weakly established and conditional on circum­ stances that are varied and complex. There are, however, sufficient case studies of competitive returns to make the case for investing in fostering agriculture growth as an effective strategy for poverty reduction under the right conditions. This is more likely to be the case where agriculture is a high share of GDP. competitive advantage is located in agriculture. and the majority of the poor are in the rural sector. These are the defining characteristics of the "agriculture-based countries," de Janvry and Sadoulet 11 mainly poor countries located in Sub-Saharan Africa and also in Central America and the Caribbean (World Bank 2007). "Agriculture-based" conditions are also found in many regions internal to large countries. making the growth-poverty role of agriculture important outside of the poor countries as well. The Contribution of Rural Areas to the Decline in Overall Poverty While most countries and regions of the world have experienced a decline in the rural poverty rate over the period 1993-2002. often larger in percentage points than the decline in urban poverty. this does not necessarily mean that most pro­ gress in poverty reduction came from the rural areas. Indeed. higher urban incomes have induced important rural-urban migration flows. raising the possi­ bility that reductions in rural poverty were due to migration of the poor as opposed to a genuine decline in poverty among the non-migrants that stay in rural areas. Of interest is thus to estimate what has been the contribution of rising incomes in rural areas to overall poverty reduction, net of the role of migration. This is done using a standard decomposition of aggregate poverty reduction into sectoral changes (urban and rural) and a population shift com­ ponent based on the transition of migrants from rural to urban areas. In this decomposition. the "rural contribution" is the decline in aggregate poverty that is due to the decline in poverty of the population of non-migrants that remain rural. Performing this decomposition thus requires specifying who migrates out of rural areas among the poor and the non-poor. Not having this information on the composition of migrants. we simulate three cases that establish a range of possible values for the rural contribution. The first case consists in assuming that migration is poverty neutral, that is that the poor and non-poor migrate at the same rate out of rural areas. In this case. the decline in the poverty rate of non­ migrants is equal to the observed decline in the rural poverty rate. The second case considers the extreme condition where all migrants are poor. If the poor are more likely to migrate, migration in itself contributes to the reduction of poverty in rural areas by its selection process. The "rural contribution" is due to the reduction of poverty in rural areas beyond the poverty reduction effect of migration of the poor. This case thus gives a lower bound to the genuine reduction of aggregate poverty achieved in rural areas. The third case considers the other extreme condition where it is the non-poor who are more likely to migrate, as documented for many countries. In this case, selection into migration contributes to an increase in the poverty rate in rural areas. The reduction in rural poverty among the non-migrants is even higher than the observed decline 12 The World Bank Research Observer. vol. 25. no. 1 (February 2010) in poverty. The extreme case where all migrants are non-poor gives an upper bound for the rural contribution. Table 2 reports the rural contribution to poverty change under the three scen­ arios, which give a range for the rural contribution to poverty reduction. 5 For the world, the aggregate poverty rate over the 1993-2002 period declined by 8.8 percentage points. Of these at least 45 percent and up to 93 percent can be attrib­ uted to the decline in poverty among the rural population, with the intermediate value of 56 percent under poverty neutral migration. These aggregate numbers are however dominated by China's extraordinary success. For the rest of the world, the decline in poverty was a modest 1.8 percentage points, of which 79 percent (using neutral migration) was due to rural areas. Heterogeneity is seen across regions. In China, 72 percent of the population and 86.5 percent of the poor resided in rural areas in 1993. Over these ten years, China experienced a huge decline of 30 percentage points in poverty rates, with poverty declining both in the urban (25 percentage points) and rural (24 percen­ tage points) areas. So, without migration, both sectors would have contributed to the decline in the poverty rate proportionately to their share in poverty. But a large migration (8 percent of the rural population) moved people from the high rural poverty rate (88.6 percent in 1993) to the much lower urban poverty rate (35.6 percent in 1993). This resulted in a rural contribution to the decline in poverty equal to about half of the aggregate. Since the poverty rate was so high, there is not much difference between the "neutral" migration (which assumes that 88.6 percent of migrants are poor) and the migration of poor (which assumes that 100 percent of migrants are poor) scenarios. Aggregate results for East Asia are dominated by the Chinese experience. The situation in India is quite different. Urban and rural poverty rates are not very different (91.5 percent vs 82.3 percent in 1993), and neither one changed much over the ten-year period. In addition there was almost no migration. So aggregate poverty only declined by 3.5 percent. With a somewhat lower decline in poverty in rural areas and a small migration. the rural contribution ranges from 56 to III percent, and is equal to 61 percent under the assumption of neutral migration. Hence. the contributions of the rural sector is in percentage terms similar to what happened in China, but with low reductions in both sectors and low migration effects. rather than high reductions in both sectors and high migration effects. South Asia as a whole is not very different from India. with slightly lower poverty rates. and even less poverty reduction in rural areas. So aggregate poverty has declined by a very small 1.7 percentage point, and only 33 percent came from rural progress, if migration is assumed to have been neutral. Latin America and the Caribbean is also different. It started with a large differ­ ence between rural and urban poverty rates (47 percent vs 23 percent) but with a small share of the population residing in rural areas (28 percent). Of the poor. de Jrmvry and Sadoulet 13 t-> "'" Table 2. Contributions of the Rural Sector to Aggregate Poverty Change, 1993-2002 Share of rural Rural poverty Urban povertH in total Comribution of rural sector to aggregate rate rate population Aggregate poverty rate poverty change (%) Change ,\;1igratian of Neutral Migration Region 1993 2002 1993 2002 1993 2002 1993 2002 1993-2002 non-poor migration a/poor ~ East Asia Pacific 85.1 63.2 38.6 ] 7.8 68.9 61.2 70.6 45.6 25.0 80 53 49 ~ China 88.6 65.1 35.6 10.7 72.0 60.9 73.8 43.8 30.0 81 48 44 is:: Latin America and Caribbean 47.3 46.4 22.8 27.] 27.7 23.8 29.6 31.6 2.1 -99 -lO 88 IJ:j !'l ;;;. South Asia 87.6 86.8 78.0 74.6 74.3 72.2 85.] 83.4 -1.7 141 33 17 India 91.5 88.6 82.3 78.1 73.8 71.9 89.1 85.6 -3.5 111 61 56 f .... B- Sub-Saharan Africa Total 85.2 78.2 82.4 69.7 66.9 39.1 68.5 33.7 70.2 61.9 64.8 57.7 79.8 63.3 77.5 54.4 -2.2 -8.8 292 93 81 56 45 45 ~ .., Less China 73.7 71. 3 40.0 40.5 58.4 56.6 59.6 57.9 1.8 153 79 52 '" til Sources: Poverty data from Ravallion. Chen. and Sangraula (2007) using a PPP$2.15/day poverty line; contribution of rural sector from authors' calculations. '"" ~ t" :,n ~ '" ;(;I "" § ~ tv a ,.... -3 44.3 percent resided in rural areas in 1993. Over the ten-year period, there was some migration. almost no decline in the rural poverty rate, and an increase in urban poverty. Aggregate poverty increased from 29.6 to 31.6 percent. Depending on the migrant composition, the contribution of the rural sector to changes in aggregate poverty is estimated to be an increase in poverty when only the poor migrate, giving maximal contribution to migration, or a decrease in poverty when only the non-poor migrate. With poverty-neutral migration, the rural con­ tribution to poverty was a negative 10 percent. indicating again the Latin American exception in helping the rural poor benefit from agricultural growth. In Sub-Saharan Africa. we observe a small decline in rural poverty but an increase in the urban poverty rate. adding up to an overall decline in poverty because the population is still predominantly rural. Because poverty rates are much higher in rural (85.2 percent in 1993) than in urban areas (66.9 percent). even under the neutral migration scenario, the urban increase in poverty rate was largely due to the in-migration of rural poor. With failing agriculture and failing aggregate growth, rural-urban migration contributed to an increase in the urban poverty rate. Rural poverty decline contributed 81 percent of aggregate poverty reduction under the neutral migration scenario. We can thus conclude that the rural sector's contribution to aggregate poverty reduction was large overall. It accounted for more than half of aggregate poverty reduction worldwide and up to three-quarters for the rest of the world when China is excluded. Note that this is an accounting decomposition, measuring the share of the decline in poverty among the rural population in the aggregate decline in poverty, not a causal relationship between rural growth and poverty reduction. Most of this effect expectedly came from incomes generated in the rural sector. mainly from agricultural growth. Other factors that have contributed to rural income growth include public transfers and remittances received by rural households which are derived from urban and foreign income growth. Migration also contributes to rural income growth through rising wages by reducing the rural labor supply. Household-level Analysis: Agricultural Growth Offering Pathways out of Poverty How do different categories of farming households benefit from agricultural growth? Agriculture offers rural households a number of pathways out of poverty: they can increase their incomes by selling agricultural products on markets (market-oriented farming households), they can leave the subsistence economy and become market participants (market entrants). and they can de la/wry arid Sadoulel 15 Table 3. Pathways out of Poverty in Vietnam. 1992/3-1997/8 Categories of farming households Market-oriented Market entrants Subsistence-oriented Base Base Base period % change period % change period % change 199213­ 199213­ 199213­ 199213 199718 199213 199718 199213 199718 Share of farming households (%) 28 13 6 Poverty outcomes Share of households below the poverty 64 -42 73 35 86 28 line (%) Income structure Share of agricultural income in total 83 12 83 20 80 -23 income (%) Share of high value and industrial 29 34 21 49 14 1 crops in gross agricultural income (%) Source: Data from Vietnam Living Standard Surveys (World Bank 2009). improve their well-being in the subsistence economy either through farming or through other sources of income (subsistence-oriented households). Vietnam offers a good case study, both because agricultural growth was rapid (an average annual growth in real agricultural value added of 4.1 percent in the period 1992-98) and because we have rare household panel data over that period that allow us to track what happened to different categories of households. In table 3, we look at rural farming households defined as those with more than 50 percent of their income. including home consumption. derived from agriculture, and recognize three path'ways: market-oriented farming households selling more than 25 percent of their agricultural production in both the initial (1992/3) and term­ inal (1997/8) years; market entrants who were selling less than 10 percent in the base period and more than 25 percent in the terminal year: and subsistence­ oriented farming households who were selling less than 10 percent in both initial and terminal years. These three groups constitute 47 percent of the farming households, the remaining having more mixed income strategies. Market-oriented households benefited most from this period of rapid agricul­ tural growth, with a 42 percent reduction in their poverty rate. While they diver­ sified away from agriculture as a source of income, they also diversified away from staple crops (rice) toward high value and industrial crops. Among market entrants, poverty fell by 35 percent and they also importantly diversified both away from agriculture and toward high value and industrial crops in agriculture. Finally, for the subsistence farmers in the base period that remained in that 16 The World Bank Research Observer. vol. 25, no. 1 (February 2010) category through the period, poverty was reduced by 28 percent. They continued to produce staple crops for home consumption. Their income gains were mainly derived from diversifying away from agriculture, benefiting from employment cre­ ation in agriculture and in the rural non-farm economy driven by overall agricul­ tural growth. Agricultural growth can thus pull farming households out of poverty along a multiplicity of pathways. The implication is that making these pathways more effective for poverty reduction will require specific policies for specific categories of households: supporting competitiveness for market-oriented farming households; enhancing access to assets and to markets to favor market entry for subsistence households; and improving production for home consumption and entry into rural labor markets for subsistence-oriented farming households. Designing specific policies for specific categories of households is thus a very important prin­ ciple in making policy in support of agriculture for development. It requires access to information about the opportunities and constraints that apply to each of these categories of households that can only be obtained through their active participation in policy design. Policy Implications: Agriculture for Development With two competing approaches to poverty reduction-transfers and pro-poor growth-a key policy issue is to find the right balance in allocating public resources and foreign aid budgets between the two approaches for optimum com­ plementarity. The dilemma is particularly stark when poverty and hunger are high and time to achieve relief is short. tilting public expenditure priorities toward transfers as opposed to promoting the growth of autonomous incomes. In recent years, increased emphasis has been given to poverty reduction via transfers. some­ times conditional on behavior toward child education and health. This has con­ tributed to the neglect of agricultural growth as an instrument for poverty reduction. This shift in emphasis has been the product of a complex set of circum­ stances including the urgency of addressing the poverty effects of the debt crisis and stabilization policies. low profitability of investments in agriculture due to declining commodity prices on international markets, low effectiveness of poverty reduction projects based on agriculture (using approaches such as state-led inte­ grated rural development projects, parastatal agencies for marketing, subsidized credit. and the training-and-visit approach to extension). and perceived inevitable conflicts between agriculture and the environment. Yet. we know that autonomous income growth is, in the long run. the better instrument for poverty reduction among those who have the potential to work. And we have seen that agricultural growth-along with the growth of unskilled de JaTl'lry and Sadoulet 17 labor intensive activities in the rural non-farm economy-can be particularly effective for poverty reduction via autonomous income growth. Growth originat­ ing in agriculture can be three times more effective in reducing poverty than growth originating in the rest of the economy. Yet. not all agricultural growth is equally effective. Growth in cereal yields and in agricultural labor productivity have been associated with greater poverty reduction in East Asia than in Latin America. Agricultural grmvth can have not only strong direct poverty reduction effects on but also strong growth linkages to the rest of the economy. The contri­ bution of the rural sector to aggregate poverty reduction. largely driven by agri­ cultural growth. has been about half of total poverty reduction. even under conservative assumptions regarding the contribution of migration. and it has been particularly high in Sub-Saharan Africa, precisely where it matters the most. Heterogeneity of rural populations suggests that there exist multiple path­ ways for using agriculture to help rural households move out of poverty. with market-oriented smallholder farming the most effective one. Existence of multiple pathways amplifies the array of policy instruments that can be used. calling on the design of specific interventions for specific categories of rural households. This suggests that a return to using growth in agriculture as an instrument for poverty reduction may be warranted under many circumstances. While much is still left to be researched. in particular to determine how to achieve growth in agriculture more cost effectively and how to make it more pro-poor. we now have a better understanding of how agriculture contributes to poverty reduction. and what features of the structural context can enhance this effect. Conditions to invest profitably in agriculture are currently more favorable than they have been for the last 35 years: markets are significantly less distorted. commodity prices are higher. markets for high value crops and animal products are expanding. and there are numerous technological and institutional innovations available to enhance supply response. New options to design investments in agriculture so they are more pro-poor are also emerging. This includes projects that are more decentralized; more participatory; give greater attention to not only access to assets for the rural poor but also to the role of the market. public goods. and insti­ tutional conditions for effective use of these assets; and seek to make growth more compatible with environmental protection and more resilient to climate shocks. Commitments by governments and international development agencies to place more resources in support of agriculture-based development projects have been made, in particular in response to the stress of the food crisis. Current conditions are thus generally favorable to use again agricultural growth-along with other linked sectors-as an effective instrument for poverty reduction. Realizing this potential requires careful design of investment in agriculture to achieve growth, political commitments by governments and donors. and a voice for the private sector and civil society to ensure that these commitments are implemented. 18 The World Bank Research Observer. vol. 25. no. 1 (1:ebnlary 2010) Notes Alain de Janvry (corresponding author) and Elisabeth Sadoulet are at the University of California at Berkeley; email addresses: alain@berkeley.edu, esadoulet@berkeley.edu. The authors are indebted to Derek Byerlee and the World Development Report 2008 team for their contributions to thL~ research. 1. Though this does not necessarily benefit the poor proportionately more than the non-poor, as in the UNDP's definition of pro-poor growth; see Ravallion (2004), 2. Details are given in Ligon and Sadoulet (2007). The expenditure equation they estimate for deciles. countries. and years uses year and country-decile fixed effects, as well as instrumentation of sectoral income growth using the average of neighboring countries' growth rates of agriculture value added as an instrument for own-country agriculture value added growth. Estimates are shown to be robust to a range of alternative specifications designed to challenge the result obtained. 3. Data from World Bank (2006). These estimates of intersectorallinkages are obtained from a vector autoregressive model with the optimal lag order, Details on the e~1imation are given in the Appendix to de Janvry and Sadoulet (2008). 4. This finding justifies investing in agricultural R&D even if rates of return to investment in non-agricultural R&D are high or even higher for as long as a capital market exists. 5. Details on the equations used to calculate the rural contribution are given in the Appendix to de Janvry and Sadoulet (2008). References The word processed describes informally reproduced works that may not be commonly available through libraries. Alston. Julian. Connie Chan-Kang. Michele Marra. Philip Pardey. and T.]. Wyatt. 2000. A Meta­ Analysis of Rates of Return to Agricultural RE7D: Ex Pede Herculem? Washington, DC: International Food Policy Research Institute. Bravo-Ortega. Claudio, and Daniel Lederman. 2005. ':Agriculture and National Welfare around the World: Causality and International Heterogeneity since 1960." Washington. DC: World Bank Policy Research Working Paper Series 3499, Christiaensen, Luc, and Lionel Demery, 2007. Down to Earth: Agriculture and Poverty Reduction in Africa, Directions in Development. Washington. DC: World Bank. Datl. Gaurav. and Martin Ravallion. 1998. "Farm Productivity and Rural Poverty in India." Journal of Development Studies 34(4): 62-85. FAO (Food and Agriculture Organization). 2006. "FAOSTAT.·· Rome: Food and Agriculture Organization. Foster. Andrew, and Mark Rosenzweig. 2005. ':Agricultural Development, Industrialization. and Rural Inequality." Processed. Economics Department. Brown University, Hasan, Rana. and M.G. Quibria, 2004. "Industry Matters for Poverty: A Critique of Agricultural Fundamentalism," Kyklos 57(2):253-64. de ]anvry. Alain. and Elisabeth Sadoulet. 2008. ':Agricultural Growth and Poverty Reduction: Additional Evidence. Appendix," Paper prepared for the WDR. available at http://are.berkeley ,edul -sadouletl Ligon. Ethan. and Elisabeth Sadoulet. 2007, "Estimating the Effects of Aggregate Agricultural Growth on the Distribution of Expenditures," Background paper for the WDR 2008. Lipton. Michael. 1991. New Seeds and Poor People, With Richard Longhurst, London: Unwin Hyman. de JanlJry and Sadoulet 19 Loayza, Norman, and Claudio Raddatz. Forthcoming. "The composition of Growth Matters for Poverty Alleviation." Journal of Development Economics. Ravallion, Martin. 2004. "Defining Pro-poor Growth: A Response to Kakwani." One Pager. Brasilia: International Poverty Center, UNDP. Ravallion, Martin, and Gaurav Datt. ] 996. "How Important to India's Poor is the Sectoral Composition of Economic Growth?" World Bank Economic Review 10(1):1-26. Ravallion, Martin, and Shaohua Chen. 2007. "China's (Uneven) Progress Against Poverty." Journal of Development Economics 82( I): 1-42. Ravallion, Martin, Shaohua Chen, and Prem Sangraula. 2007. "New Evidence on the Urbanization of Global Poverty." Background paper for the World Development Report 2008, The World Bank. Suryahadi, Asep, Daniel Suryadarma, and Sudarno Sumarto. 2008. "The Effects of Location and Sectoral Components of Economic Growth on Poverty: Evidence from Indonesia." Journal of Development Economics 89(1): 109-17. Valdes, Alberto, and William Foster. 2007. "Making the Labor Market a Way out of Rural Poverty. Rural and Agricultural Labor Markets in Latin America and the Caribbean." Background paper for the World Development Report 2008, The World Bank. van de Walle, Dominique, and Ren Mu. 2007. "Fungibility and the Flypaper Effect of Project Aid: Micro-evidence for Vietnam." Journal of Development Economics 84(2): 667- 8 5. World Bank. 2006. World Development Indicators. Washington, DC: The World Bank. ___. 2007. Agriculture for Development. World Development Report 2008. Washington, DC: The World Bank. 2008. PovcalNet. http://iresearch.worldbank.org/PovcaINetl ___.2009. Living Standards Measurement Study: LSMS Data Table. Washington, DC: The World Bank. 20 The World Bank Research Observer. vol. 25. no. 1 (February 2(10) Policy Reforms Affecting Agricultural Incentives: Much Achieved, Much Still Needed Kym Anderson For decades, earnings from farming in many developing countries have been depressed by a pro-urban bias in own-country policies. as well as by governments of richer countries favoring their farmers with import barriers and subsidies. Both sets of policies reduce national and global economic welfare and inhibit agricultural trade and economic growth. They almost certainly add to inequality and poverty in developing countries, since three-quarters of the world's billion poorest people depend on farming for their live­ lihood. During the past two decades, however, numerous developing country governments have reduced their sectoral and trade policy distortions. while some high-income countries also have begun reducing market-distorting aspects of their farm policies. The author surveys the changing extent of policy distortions to prices faced by developing­ country farmers over the past half century, and provides a summary of new empirical estimates from a global economy-wide model that yield estimates of how much could be gained by removing the interventions remaining as of 2004. The author concludes by pOinting to the scope and prospects for further pro-poor policy reform in both developing and high-income countries. JEL codes: F13, F14, Q17, Q18 For many decades agricultural protection and subsidies in high-income (and some middle-income) countries have been depressing international prices of farm products, which lowers the earnings of farmers and associated rural businesses in developing countries (Johnson 1991; Tyers and Anderson 1992). Those policies almost certainly add to inequality and poverty, since three-quarters of the world's poorest people depend directly or indirectly on agriculture for their main income (World Bank 2007). Currently less than 15 million relatively wealthy farmers in developed countries. with an average of almost 80 hectares per worker. are being The World Bank Research Observer 1[:, The Author 2009. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development f THE WORlD BAC','K. All rights reserved. For permissions, please e-mail: journals,permissions@oxfordjournals.org doi: 10. 1093fwbrollkpO 14 Advance Access publication November 9, 2009 25:21-55 helped at the expense of not only consumers, taxpayers. and producers of other tradables in those rich countries but also the majority of the 1.3 billion relatively impoverished farmers and their large families in developing countries who, on average, have to earn a living from just 2.5 hectares per worker. But in addition to this external policy influence on rural poverty, the govern­ ments of many developing countries have directly taxed their farmers over the past half-century. A well-known and often-cited example is the taxing of exports of plantation crops in post-colonial Africa (Bates 1981). Furthermore. many developing countries in the 1960s and 1970s chose also to pursue an import­ substituting industrialization strategy. predominantly by restricting imports of manufactures. As Krueger. Schiff, and Valdes (1988. 1991) showed in their seminal multi-country study, this indirectly taxed other tradable sectors in those developing economies. including agriculture. Thus the price incentives facing farmers in many developing countries have been depressed by both own-country and other countries' farm, food, and trade policies. I will survey the extent to which government policies at home and abroad have distorted prices faced by developing-country farmers over the past half-century. I begin with a brief examination of the methodology required to measure the extent of own-country distortions to farmer incentives. I then survey analyses of the effects of those agricultural and trade policies on incentives over time, focusing on the worsening of that situation between the 1950s and early 1980s and on the progress that has been made over the 25 years since then. Notwithstanding recent reforms, many price distortions remain in the agricul­ tural sector of both developing and high-income countries. I provide a summary of new empirical estimates from a global economy-wide model that indicate how much could be gained by removing the interventions remaining as of 2004. I conclude by pointing to the scope and prospects for further pro-poor policy reform in both developing and high-income countries. National Distortions to Incentives: Basic Theoryl Bhagwati (1971) and Corden (1997) define the concept of a market policy distor­ tion as something that governments impose to create a gap between the marginal social return to a seller and the marginal social cost to a buyer in a transaction. Such a distortion creates an economic cost to society which can be estimated using welfare-measuring techniques such as those pioneered by Harberger (1971). As Harberger notes, this focus allows a great simplification in evaluating the marginal costs of a set of distortions: changes in economic costs can be evalu­ ated by taking into account the changes in volumes directly affected by such dis­ tortions. ignoring all other changes in prices. In the absence of divergences such 22 The World Bank Research Observer. vol. 25, no. 1 (February 2(10) as externalities, the measure of a distortion is the gap between the price paid and the price received, irrespective of whether the level of these prices is affected by the distortion. 2 Importantly. the total effect of distortions on the agricultural sector will depend not just on the size of the direct agricultural policy measures, but also on the magnitude of distortions generated by direct policy measures altering incentives in non-agricultural sectors. It is relative prices, and hence relative rates of government assistance, that affect producers' incentives. In a two-sector model an import tax has the same effect on the export sector as an export tax: the Lerner (1936) Symmetry Theorem. This carries over to a model that has many sectors, and is unaffected if there is imperfect competition domestically or internationally or if some of those sectors produce only non-tradables (Vousden 1990. pp. 46-7). The Symmetry Theorem is therefore also relevant for considering distortions within the agricultural sector. In particular. if import­ competing farm industries are protected. for example via import tariffs. this has similar effects on incentives to produce exportables as does an explicit tax on agricultural exports; and if both measures are in place. this is a double imposition on farm exporters. Direct Agricultural Distortions Consider a small, open, perfectly competitive national economy with many firms producing a homogeneous farm product with just primary factors. In the absence of externalities. processing, producer-to-consumer wholesale plus retail marketing margins, exchange rate distortions, and domestic and international trading costs, that country would maximize national economic welfare by allow­ ing the domestic producer and consumer prices of that product to both equal E times P. where E is the domestic currency price of foreign exchange and P is the foreign currency price of this identical product in the international market. That is. any government-imposed diversion from those two equalities. in the absence of any market failures or externalities. would be welfare-reducing for that small economy. Price-distorting Trade Measures at the National Border. The most common distortion is an ad valorem tax on competing imports (usually called a tariff). t m . Such a tariff on imports is the equivalent of a production subsidy and a consumption tax. both at rate trll' If that tariff on the imported primary agricultural product is the only distortion, its effect on producer incentives can be measured as the nominal rate oj assistance (NRA) to farm output conferred by border price support (NRA BS )' which is the unit value of production at the distorted price, less its value at the ,11ldersoll 23 undistorted free market price. expressed as a fraction of the undistorted price: 3 Ex ExP NRA Bs = ---'----'---'----- = tm (1) ExP The effect of that import tariff on consumer incentives in this simple economy is to generate a consumer tax equivalent (CTE) on the agricultural product for final consumers: CTE tm (2) The effects of an import subsidy are identical to those in equations (1) and (2) for an import tax. but tm in that case would have a negative value. Governments sometimes also intervene with an export subsidy Sx (or an export tax. in which case Sx would be negative). If that were the only intervention. then: NRA Bs = CTE Sx (3) Domestic Producer and Consumer Price-distorting Measures. Some governments provide a direct production subsidy for farmers, Sf (or production tax. in which case Sf is negative. including via informal taxes in kind by local and provincial governments). In that case. if only this distortion is present, the effect on producer incentives can be measured as the nominal rate of assistance to farm output con­ ferred by domestic price support (NRA Ds ). which is as above except Sf replaces tm or SX' but the CTE in that case is zero, Similarly. if the government just imposes a consumption tax Cc on this product (or consumption subsidy, in which case Cc is negative), the CTE is as above except Cc replaces tm or SX' but the NRA Ds in that case is zero. The combination of domestic measures and border price support provides the following total rate of assistance to output, NRA o • and total consumer tax equival­ ent, CTE: NRAo = NRA Bs + NRlil)s (4) CTE = NRA Bs + Ct (5) What If the Exchange Rate System Is also Distorting Prices? Should a multi-tier foreign exchange rate regime be in place. then another policy-induced price wedge exists. A simple two-tier exchange rate system creates a gap between the price received by all exporters and the price paid by all importers for foreign cur­ rency, changing both the exchange rate received by exporters and that paid by importers from the equilibrium rate E that would prevail without this distortion in the domestic market for foreign currency (Bhagwati 1978). This requires 24 Tile World Bank Research Observer, vol. 25, no. I (February 20ID) controls by the government on current account transfers. In the past it was common for exporters to be required to surrender their foreign currency earnings to the central bank for exchange to local currency at a low official rate, which is equivalent to a tax on exports to the extent that the official rate is below what the exchange rate would be in a market without government intervention. That implicit tax on exporters reduces their incentive to export and hence the supply of foreign currency flowing into the country. With less foreign currency. demanders are willing to bid up its purchase price, providing a potential rent for the govern­ ment which can be realized by auctioning off the limited supply of foreign cur­ rency extracted from exporters, or by creating a legal secondary market. Either mechanism will create a gap between the official and parallel rates (Dervis. de Melo and Robinson 1981). If the government chooses to allocate the limited foreign currency to different groups of importers at different rates, that is called a multiple exchange rate system. Some lucky importers may even be able to purchase it at the low official rate. The more that is allocated and sold to demanders whose marginal valuation is below the equilibrium rate, the greater the unsatisfied excess demand and hence the stronger the incentive for an illegal or "black" market to form, and for less-unscrupulous exporters to lobby the government to legalize the secondary market for foreign exchange and to allow exporters to retain some fraction of their exchange rate earnings for sale in the secondary market. Such a right given to exporters to retain and sell a portion of foreign exchange receipts would increase their incentives to export, and thereby reduce the shortage of foreign exchange and hence the secondary market exchange rate (Tarr 1990; Martin 1993). For present purposes. what matters is that. where a country has distortions in its domestic market for foreign currency. the exchange rate relevant for calculat­ ing the NRAo or the CTE for a particular tradable product depends. in the case of a dual exchange rate system. on whether the product is an importable or an exportable one, while in the case of multiple exchange rates it depends on the specific rate applying to that product each year. The precise way in which that can be handled is detailed in Anderson and others (2008). What if Farm Production Involves not just Primary Factors but also Intermediate Inputs? Where intermediate inputs are used in farm production. any taxes or sub­ sidies on their production. consumption. or trade would alter farm value added and thereby also affect farmer incentives. Sometimes a government will have directly offsetting measures in place, such as a domestic subsidy for fertilizer use by farmers but also a tariff on fertilizer imports. In other situations there will be farm input subsidies but an export tax on the final product. In principle all these items could be brought together to calculate an effective rate of direct assistance Anderson 25 to farm value added. The nominal rate of direct assistance to farm output, NRA o' is a component of that. as is the sum of the nominal rates of direct assistance to all farm inputs, call it NRA i • Where there are significant distortions to input costs, their ad valorem equivalent can be accounted for by summing each input's NRA times its input-output coefficient to obtain the combined NRA j , and adding that to the farm industry's nominal rate of direct assistance to farm output. NRA o' to get the total nominal rate of assistance to farm production, NRA. What about Post-farmgate Costs? If a state trading corporation is charging exces­ sively for its marketing services and thereby lowering the farm-gate price of a product. for example as a way of raising government revenue in place of an expli­ cit tax, the extent of that excess is treated as if it were an explicit tax. Some farm products, including some that are not internationally traded, are inputs into a processing industry that may also be subject to government inter­ ventions. In that case the effect of those interventions on the price received by farmers for the primary product also needs to be taken into account. The Mean and Variance of Agricultural NRAs. When it comes to averaging across countries, each polity is an observation of interest. so a simple average is mean­ ingful for the purpose of political economy analysis. But if one wants a sense of how distorted agriculture is in a group of countries, a weighted average is needed. The weighted average NRA for covered primary agriculture can be generated by multiplying each primary industry's share of production (valued at the farm-gate equivalent undistorted prices) by its corresponding NRA and adding across indus­ tries. The overall sectoral rate, NRAag. also could include actual or assumed infor­ mation for the non-covered commodities and, where it exists, the aggregate value of non-product-specific assistance to agriculture. A weighted average can be gen­ erated also for just the tradables part of agriculture-including those industries producing products such as milk and sugar that require only light processing before they can be traded-by assuming that its share of non-product-specific assistance equals its weight in the total. Call that NRAag t • In addition to the mean, it is important to provide also a measure of the dis­ persion or variability of the NRA estimates across the covered products. The cost of government policy distortions to incentives in terms of resource misallocation tends to increase as the degree of substitution in production increases (Lloyd 1974). In the case of agriculture which involves the use of farm land that is sector-specific but transferable among farm activities, the greater the variation of NRAs across industries within the sector, the higher will be the welfare cost of those market interventions. A simple indicator of dispersion is the standard devi­ ation around the weighted mean of industry NRAs within the agricultural sector. 26 The Vybrld Bank Researcll O/Jserver. I'OJ. 25. no. 1 (Fcbrllory 20IO) Trade Bias in Agricultural Assistance. A trade bias index is also needed to indicate the changing extent to which a country's policy regime has an anti-trade bias vvithin the agricultural sector. This is important because, as mentioned above, the Lerner (1936) Symmetry Theorem demonstrates that a tariff assisting import­ competing farm industries has the same effect on farmers' incentives as if there were a tax on agricultural exports; and, if both measures are in place, this is a double imposition on farm exporters. The higher is the nominal rate of assistance to import-competing agricultural production (NRAag m ) relative to that for expor­ table farm activities (NRAag x ) , the more incentive producers in that subsector will have to bid for mobile resources that would otherwise have been employed in export agriculture. other things being equal. Indirect Agricultural Assistance or Taxation via Non-agricultural Distortions In addition to direct assistance to, or taxation of, farmers, the Lerner (1936) Symmetry Theorem further demonstrates that their incentives are also affected indirectly by government assistance to non-agricultural production in the national economy. The higher is the nominal rate of assistance to non-agricul­ tural tradables production (NRAnonag t ), the more incentive producers in other tradable sectors will have to bid up the value of mobile resources that would otherwise have been employed in agriculture, other things being equal. If NRAal is below NRAnonag t , one might expect there to be fewer resources in agriculture than there would be under free market conditions in the country, notwithstand­ ing any positive direct assistance to farmers, and conversely: One way to capture this is to calculate a Relative Rate of Assistance, RRA, defined as: 1 + NRAag t RRA= (6) + NRArlOnag t Since an NRA cannot be less than -1 if producers are to earn anything, neither can an RRA. This measure is a useful indicator for providing international comparisons over time of the extent to which a country's policy regime has an anti- or pro-agricultural bias. National Distortions to Farmer Incentives: The Evolution of Policies Before turning to the contemporary (post-World War II) situation, it would be insightful to examine briefly the long history of government intervention in ilnderson 27 international markets for farm products by today's advanced economies. since similar political economy forces may influence policy choices in later-developing countries. Attention then turns to the price-distorting policies of developing countries since the 1950s as they became independent from their colonial masters. The Long History in High-income Countries, Briefly Britain was the first country to have an industrial revolution. Prior to that revolu­ tion-from the late 1100s to the 1660s-Britain used export taxes and licenses to prevent domestic food prices from rising excessively. But during 1660-90 a series of Acts gradually raised food import duties (making imports prohibitive under most circumstances) and reduced export restrictions on grain. These pro­ visions were made even more protective of British farmers by the Corn Laws of 1815. True, the famous repeal of the Corn Laws in the mid-1840s heralded a period of relatively unrestricted food trade for Britain, 4 but then agricultural pro­ tection returned in the 1930s and steadily increased over the next five decades. Similar tendencies have been observed in many other West European countries, although on the Continent the period of free trade in the 19th century was con­ siderably shorter. and agricultural protection levels during the past 150 years have been somewhat higher on average than in Britain. Kindleberger (1975) describes how the 19th-century free-trade movements in Europe reflected the national economic, political, and sociological conditions of the time. Agricultural trade reform was less difficult for countries such as Britain with overseas terri­ tories that could provide the metropolis with a ready supply of farm products. The fall in the price of grain imports from America in the 1870s and 1880s provided a challenge for all, however. Denmark coped well by moving more into livestock production to take advantage of cheaper grain. Italians coped by sending many of their relatives to the New World. Farmers in France and Germany successfully sought protection from imports. however. and so began the post-Industrial Revolution growth of agricultural protectionism in densely populated countries. Meanwhile, tariffs on West European imports of manufactures were progressively reduced after the General Agreement on Tariffs and Trade (GATT) came into force in the late 1940s, thereby adding to the encouragement of agricultural relative to manufacturing production (Lindert 1991; Findlay and O'Rourke 2007). Japan provides an even more striking example of the tendency to switch from taxing to increasingly assisting agriculture relative to other industries. Its industri­ alization began later than in Europe, after the opening up of the economy follow­ ing the Meiji Restoration in 1868. By 1900 Japan had switched from being a small net exporter of food to becoming increasingly dependent on imports of rice (its main staple food and responsible for more than half the value of domestic food production). This was followed by calls from farmers and their supporters for 28 2'he World Bank Research Observer. vol. 25, no. 1 (February 2010) rice import controls. Their calls were matched by equally vigorous calls from manufacturing and commercial groups for unrestricted food trade, since the price of rice at that time was a major determinant of real wages in the non-farm sector. The heated debates were not unlike those that led to the repeal of the Corn Laws in Britain six decades earlier. In Japan, however, the forces of protection tri­ umphed, and a tariff was imposed on rice imports from 1904. That tariff then gradually rose over time, raising the domestic price of rice to more than 30 percent above the import price during World War I. Even when there were food riots because of shortages and high rice prices just after that war, the Japanese government's response was not to reduce protection but instead to extend it to its colonies and to shift from a national to an imperial policy of rice self-sufficiency. That involved accelerated investments in agricultural development in the colonies of Korea and Taiwan behind an ever-higher external tariff wall that by the latter 1930s had driven imperial rice prices to more than 60 percent above those in international markets (Anderson and Tyers 1992). After the Pacific War ended and Japan lost its colonies, its agricultural protection growth resumed and spread from rice to an ever-wider range of farm products. The other high-income countries were settled by Europeans relatively recently and are far less-densely popUlated. They therefore have had a strong comparative advantage in farm products for most of their history following Caucasian settle­ ment, and so have felt less need to protect their farmers than Europe or north­ east Asia. Indeed Australia and New Zealand until the late 20th century tended-like developing countries-to adopt policies that discriminated against their farmers (Anderson. Lloyd and MacLaren 2007). Developing Countries since the 1950s In the Republic of Korea and Taiwan, China in the 1950s, as in many newly independent developing countries. initially adopted an import-substituting indus­ trialization strategy which harmed agriculture. But in those two economies that policy was replaced in the early 1960s with a more neutral trade policy that resulted in very rapid export-oriented industrialization. That development strategy in those densely populated economies imposed competitive pressure on the farm sector which, just as in Japan in earlier decades, prompted farmers to lobby (suc­ cessfully, as it happened) for ever-higher levels of protection from import protec­ tion (Anderson and Hayami 1986. ch. 2). Many less-advanced and less-rapidly growing developing countries not only adopted import-substituting industrialization strategies in the late 1950s or early 1960s (Little, Scitovsky. and Scott 1970; Balassa and Associates 1971) but also imposed direct taxes on their exports of farm products. The latter practice was especially rife in Africa (Bates 1981). It was common in the 1950s and 1960s, Anderson 29 and in some cases even in the 1970s and 1980s. also to use dual or multiple exchange rates so as to tax indirectly both exporters and importers (Bhagwati 1978; Krueger 1978). This added to the anti-trade bias of developing countries' trade policies. That policy history is now well known. and has been documented extensively in previous surveys (for example Krueger 1984). Less well-known is the extent to which many emerging economies have belatedly followed the examples of Korea and Taiwan in abandoning import­ substitution and opening their economies. Some (for example Chile) started in the 1970s. while others (for example India) did not do so in a sustained way until the 1990s. Some have adopted a very gradual pace of reform. with occasional rever­ sals. while others have moved rapidly to open markets. Some have reduced export taxes but simultaneously raised import barriers. And some have adopted the rheto­ ric of reform but in practice have done little to free up their economies. To get a clear sense of the overall impact of these reform attempts. there is no substitute for empirical analysis that quantifies over time the types of indicators raised in the theory section above. Building on recent work by the International Food Policy Research Institute (IFPRI) and the OECD (Orden and others 2007; OECD 2007). a World Bank project recently undertook such analysis. to which we now turn. National Distortions to Farmer Incentives: Empirical Estimates post-World War II After post-World War II reconstruction. Japan continued to raise its agricultural pro­ tection. just as had been happening in Western Europe. but to even higher levels. Domestic prices exceeded international market prices for grains and livestock pro­ ducts in both Japan and the European Community in the 19 50s. but by less than 40 percent. By the early 1980s the difference was more than 80 percent for Japan but was still around 40 percent for the EC-and was still close to zero for the agricul­ tural-exporting rich countries of Australasia and North America (Anderson and Hayami 1986. table 2.5). Virtually all of that assistance to Japanese and European farmers in that period was due to restrictions on imports of farm products. Since 1986 the OECD (2008) has been computing annual producer and consu­ mer support estimates by member countries. For the OECD member countries as a whole. producer support rose between 1986-88 and 2005-07 in US dollar terms (from $239 to $263 billion) but has come down when expressed as a share of support-inclusive returns to farmers (from 37 to 26 percent). Because of some switching of support instruments. including switching to measures that are based on non-current production or on long-term resource retirement. the share of that assistance provided via market price support measures has fallen from three-quar­ ters to one-half. When the PSE payment is expressed as a percentage of undistorted 30 The World Bank Research Observer, vol. 25, no. I (Febmary 2010) prices to make it an NRA. the NRA fall is from 59 to 35 percent between 1986­ 88 and 2005-07 (DEeD 2008). This indicator suggests high-income country pol­ icies have become considerably less trade-distorting, at least in proportional terms. even though farmer support in high-income countries has continued to grow in dollar terms because of growth in the value of their farm output. As for developing countries outside north-east Asia. the main comprehensive set of pertinent estimates over time is for the period just prior to when reforms became \\1despread. They were generated as part of a major study of 18 developing countries from the 1960s to the mid-1980s by Krueger. Schiff, and Valdes (1988. 1991). That study by the World Bank. whose estimates are summarized in Schiff and Valdes (1992), shows that the depression of incentives facing farmers has been due only partly to various forms of agricultural price and trade policies. including subsidies to food imports. Much more important in many cases have been those developing countries' non-agricultural policies that hurt their farmers indirectly. The two key ones have been manufacturing protectionism (which attracts resources from agriculture to the industrial sector) and overvalued exchange rates (which attract resources to sectors producing non-tradables, such as services). That indirect impact was negative for all four groups of countries shown in table 1, whereas the impact of direct agricultural policies was negative only for the two lowest-income country groups. In addition to the total assistance being more negative the poorer the country group. table 1 also reveals that it is lower for producers of exportables than for the subsector focused on import-com­ peting farm products. suggesting a strong anti-trade bias for the sector as a whole. Since there were no comprehensive multiregion studies of the Krueger, Schiff. and Valdes type for developing countries that monitored progress over the sub­ sequent reform period. s a new study was recently launched by the World Bank aimed at filling this lacuna. The new study covers not only 41 developing countries but also 14 European transition economies as well as 20 high-income countries. The results from that study 6 do indeed reveal that there has been a substantial reduction in distortions to agricultural incentives in developing countries over the past two to three decades. They also reveal that progress has not been uniform across countries and regions, and that-contrary to some earlier claims (for example from Jensen. Robinson. and Tarp 2002)-the reform process is far from complete. In particular. many countries still have a wide dis­ persion in NRAs for different farm industries and in particular have a strong anti­ trade bias in the structure of assistance within their agricultural sector; and some countries have "overshot" in the sense that they have moved from having an average rate of assistance to farmers that was negative to one that is positive, rather than stopping at the welfare-maximizing rate of zero. Moreover, the var­ iance in rates of assistance across commodities within each country, and in aggre­ gate rates across countries, remains substantial; and the beggar-thy-neighbor Anderson 31 Table 1. Direct and Indirect Nominal Rates of Assistance to Farmers in 18 Developing Countries, 1960 to mid-1980s (percent) Assistance to agric. Direct Indirect Total Assistance to agric. import-competing Country group assistance assistance assistance 8 export subsector" subsector" Very low income 23 29 -52 -49 -11 Low income 12 24 -36 -40 -13 Lower middle 0 -16 ]6 -]4 -2 income Upper middle 24 14 10 1 15 income Unweighted -8 -22 -30 -35 -9 sample a Total assistance is the weighted average of assistance to the agricultural subsectors producing exportables. importables. and non-tradables (the latter not shown above). Soun:e: Anderson (forthcoming a). summarized from estimates reported in Schiff and Valdes (1992. tables 2.1 and 2.2). practice of insulating domestic markets from international food price fluctuations continues, thereby exacerbating that volatility. The global summary of those new results is provided in figure 1. It reveals that the nominal rate of assistance (NRA) to farmers in high-income countries rose steadily over the post-World War II period through to the end of the 1980s, apart from a small dip when international food prices spiked around 1973 74. After peaking at more than 50 percent in the mid-l 980s, the average NRA for high­ income countries has fallen a little, depending on the extent to which one believes some new farm programs are "decoupled" in the sense of no longer influencing production decisions. For developing countries, too, the average NRA for agricul­ ture has been rising, but from a level of around 25 percent during the period from the mid-1950s to the early 1980s to a level of nearly 10 percent in the first half of the present decade. Thus the global gross subsidy equivalent of those rates of assistance have risen very substantially in constant (2000) US dollar terms, from close to zero up to the mid-19 70s to more than $200 billion per year at the farm gate since the mid-1990s (figure 2). When expressed on a per farmer basis, the gross subsidy equivalent (GSE) varies enormously between high-income and developing countries. In 1980-84 the GSE in high-income countries was already around $8,000 and by 2000-04 it had risen to $10,000 on average (and $25,000 in Norway, Switzerland, and Japan), or $13,500 when "decoupled" payments are included. By contrast, the GSE in developing economies was - $140 per farmer in the first half of the 1980s, which is a non-trivial tax when one recalls that at that time the majority 32 Tile World Bank Resean:h Observer. vol. 25. no. 1 (February 2010) Figure 1. Nominal Rates of Assistance to Agriculture in High-income and European Transition Economies and in Developing Countries, 1955 to 2004 (percent, weighted averages, with "decoupled" payments included in the dashed higher income countries line) 70 60 50 40 30 ~ ~ 20 10 0 1955-59 1960...(H 1965~69 1970··,i4 1915-7919&\~84 -10 -20 -30 - - Higher income countrie~ and European transition economies Higher income countries and European transition economies jncl. decoupJed payments Developing countries Source: Anderson (2009), Figure 2. Gross Subsidy Equivalent of NRAs in High-income and European Transition Economies and in Developing Countries, 1960 to 2007 (constant 2000 US$ billion) ------ 300 200 --- 100 o -100 -200 1955-59 1960--64 1965--69 197~74 1975--79 1980-84 1985-ll9 1990-84 1995--99 2000-04 2005--07 c::::J Developing countries (no averages for periods 1955--59 and 2005--07) 11IIIIIIII High-income countries and Europe's transition economies - - Net, global (decoupled payments are in eluded in the higher, dashed line) Source: Anderson (2009). Anderson 33 of these people's households were surviving on less than $1 a day per capita. By 2000-04 they received on average around $50 per farmer (Anderson 2009, ch. 1). While this represents a major improvement. it is less than 1 percent of the support received by the average farmer in high-income countries. The developing economies of Asia-including Korea and Taiwan, which were both very poor at the start of the period-have experienced the fastest transition from negative to positive agricultural NRAs. Latin American economies first increased their taxation of farmers but gradually moved during the mid-1970s to the mid-2000s from around 20 percent to 5 percent. Africa's NRAs were similar though slightly less negative than those of Latin America until the latter 1980s, before they fell back to - 7 percent (implying a gross tax equivalent per farmer of $6). In Europe's transition economies farmer assistance fell to almost zero at the start of their transition from socialism in the early 1990s: but since then, in preparation for EU accession or because of booms in exports of raw materials for energy production. assistance has gradually increased to nearly 20 percent, or $550 per farmer (Anderson 2009. ch. 1). The developing country average NRA also conceals the fact that the exporting and import-competing subsectors of agriculture have very different NRAs. Figure 3 reveals that while the average NRA for exporters in developing countries has been negative throughout (coming back from - SO percent to almost zero in 2000-04), the NRA for import-competing farmers in developing countries has fluctuated around a trend rise from 10 and 30 percent (and it even reached 40 percent in the years of low international prices in the mid-1980s). Having increased in the 1960s and 1970s. the anti-trade bias within agriculture for developing countries has diminished considerably since the mid-1980s. but the NRA gap between the two subsectors still averages around 20 percentage points. That anti-trade bias means that the rates of assistance are not uniform across commodities. which indicates that the resources that are being used within the farm sector are not being put to their best use. The extent of that extra inefficiency. over and above that due to too many or too few resources in aggregate in the sector, is indicated by the standard deviation of NRAs among covered products in each focus country. This dispersion index has fluctuated between 43 and 60 percent throughout the past five decades, with no discernible trend (Anderson 2009, table 1.6). Figure 4 shows that rice, sugar, and milk (the rice pudding ingredients) are by far the most assisted farm industries in both high-income and developing countries. Beef and poultry meat have the next highest NRAs in high-income countries followed by cotton-while in developing countries cotton has the lowest (most negative) NRA. A further decomposition of the developing countries' NRAs worth commenting on is the contribution to them from trade policy measures at each country's border as distinct from domestic output or input subsidies or taxes. Often political attention is focused much more on direct domestic subsidies or taxes than on 34 The l'orld HWlk Research Oh~erver. Fill. 2~. /lO. 1 (l:Hmwry 2(10) Figure 3. Nominal Rates of Assistance to Exportable. Import-competing. and All Covered Agricultural Products,a High-income and Developing Countries, 1955 to 2007 (percent) (a) Developing Countries 90 70 50 30 10 . ........ - . ~ -10 1965-69 1970-74 1975-79 1980-84 .' 9,!S5.89 • i99~2;'~~~f\l~5~::99~~~"20()O-O~ .. ... ... .. _ flO .. ... .. .. ... ... .. "" .. .. -. ",0~~-/""'~ -30 -50 - - Import-competing Exportables •••• Total (b) High-income Countries plus Europe's Transition Economies 90 70 50 30 .. .......... . ....... 10 -10 1955~59 1960~64 1965-69 1970-74 1975-·79 1980-84 1985-89 1990-94 1995-99 200J-D4 -30 -50 --Import-competing Exportable, •••• Total "Covered products only, The total also includes non-tradable goods. Note: The sloped straight line is an ordinary least squares regression trend line over the period shown, SOLln'e: Anderson (2009), trade measures, because those fiscal measures are made so transparent through the annual budgetary scrutiny process, whereas trade measures are reviewed only infrequently and are far less transparent. especially if they are not in the simple form of ad valorem tariffs. That attention would appear to be misplaced, however. because between 80 and 90 percent of the NRA for developing country agriculture (not including non-product-specific support which is very minor) comes from border measures such as import tariffs or export taxes (Anderson 2009. ch. 1). The improvement in farmers' incentives in developing countries is understated by the above NRAag estimates. because those countries have also reduced their /1IJdefSO/l 35 Figure 4. Nominal Rates of Assistance, Key Covered Products. High-income and Developing Countries. 1980-84 and 2000-04 (percent) (a) Developing Countries (b) High-income Countries Sugar Milk Rice e-, Rice Sugar Milk ...,387 Poultry Beef Wheat Poultry Maize Colton Pigmeat Pigmeat Coffee Soybean Soybean Maize Beef Wheat Coconut Barley Cotton Rapeseed -150 -SO 50 -150 -50 SO 150 250 Source: Anderson and Valenzuela (2008). based on estimates reported in the project's national country studies. assistance to producers of non-agricultural tradable goods, most notably manufac­ turers. The decline in the weighted average NRA for the latter, depicted in figure 5, was clearly much greater than the increase in the average NRA for trad­ able agricultural sectors for the period to the mid-1980s, consistent with the finding of Krueger, Schiff, and Valdes (1988. 1991). For the period since the mid­ 1980s, changes in both sectors' NRAs have contributed almost equally to the improvement in farmer incentives. The Relative Rate of Assistance. captured in equation (6) above, provides a useful indicator of relative price change: the RRA for developing countries as a group went from 46 percent in the second half of the 1970s to 1 percent in the first half of the present decade. This increase (from a coefficient of 0.54 to 1.01) is equivalent to an almost doubling in the relative price of farm products, which is a huge change in the fortunes of developing country farmers in just a generation. This is mostly because of the changes in Asia. but even for Latin America that relative price hike is one-half. while for Africa that indicator improves by only one-eighth (figure 6). One of the main contributors to the Asian changes is China, and a non-trivial part of its reform came through reducing its overvalued official exchange rate. There, as in some other developing countries, the distortion in the domestic market for foreign currencies was gradually reduced in an indirect way, namely, by allowing exporters to sell an increasing share of their foreign currency earn­ ings on a higher-priced secondary market. This lowered the trade tax equivalent of that distortion over time, and hence its impact on the NRA for farm and non- 36 The World Bank Research Observer. vol. 25. no. 1 (February 2010) Figure 5. Nominal Rates of Assistance to Agricultural and Non-agricultural Sectors and Relative Rate of Assistance,a Developing Countries, 1965 b to 2004 (percent. weighted averages) 80 60 40 E 20 " u ~ 0 I965--th has been employment friendly overall. Philippines Estimates of employment intensity in the case study are inadequate to arrive at a firm judgment although it appears, from findings of other studies, that there were institutional obstacles to labor absorption in agriculture and manufacturing. Thailand OEEs were higher for the nonagricultural sectors in the 1990 s (until 1996) than in the 1980 s but the overall OEE was lower due to the fact that agriculture's OEE turned from a positive value in the 1980 s to a highly negative value in the 1990 s. In the recovery period the OEE for manufacturing fell somewhat but the same for construction and services rose. The Lewis transition in agriculture of the 1990 s was reversed. India In the post-reform period the OEEs fell and employment growth fell as compared to pre- reform period. But real wages rose presumably due to a supply-induced tightening of the labor market. Sri Lanka With the exception of a few subsectors of industries and services OEEs were reasonably high and growth was employment intensive. Source: Adapted from Khan (2007). presented in table 3. These studies confirm a tendency toward declining employ­ ment intensity in agriculture, but they also suggest declining employment intensi­ ties in manufacturing. In India. capital-biased industrial policies and prohibitive labor regulations have led to an economic structure ill-suited to India's labor abundance (Besley and Burgess 2004; Kochhar and others 2006). China's story is more complex as it partly relates to the varying fortunes of China's Town and Village Enterprises (which were a source of significant labor absorption in the 1980 s before credit constraint slowed their growth). and partly to the gradual shedding of surplus labor from state-owned enterprises (Khan 2(07). Prior to the recent financial crisis, rural-urban migration in China looked set to add further pressure on nonagricultural labor markets, but this trend will mostly like resume once the short-term effects of the crisis abate. Between 1999 and 2003, for example. the number of internal migrants in China roughly doubled. from 52 to 98 million. and China's 2000 census indicated annual migration rates of 8.5 percent of the workforce, with roughly 30 percent heading to local townships, 30 percent to other counties in the same province, and 40 percent representing movement across provinces (Du. Park. and Wang 2(05). Headey, Bezemel; and Hazell 73 To summarize, we know that most of developing Asia has achieved remarkable feats of growth and poverty reduction vvith relatively little urbanization, largely through a combination of rapid agricultural growth and spatially dispersed indus­ trial growth. However. the potential of Asian agriculture to keep people on the land has diminished markedly in recent decades, and there are justified concerns that the sheer number of jobs that need to be created outside of Asia's agricul­ tural sector will impose a daunting challenge on the region. Such challenges would be large for any economy, but they are now magnified because the same problem is emerging Simultaneously for a number of very populous Asian countries. Previous research has attempted to assess what the implications arc of cheap and abundant Asian labor on labor markets in other regions of the world, such as Latin America (Wood 1997). But one might also ask whether large Asian countries will also experience lower employment growth as a result of increased competition from each other (that is in export markets).6 Agricultural Employment and Agricultural Potential in Sub-Saharan Africa The exodus of so many African people from agricultural activities, in spite of minimal or even negative economic growth. potentially poses a serious challenge to the conventional thinking that structural transformation is part and parcel of economic development. The 2008 World Development Report, for example, dis­ tinguishes between agriculture-based, transforming, and urbanized economies. but the extreme case of Nigeria seems to be one of transformation without devel­ opment. Moreover, African agriculture's poor performance has induced pessimism in some quarters with regard to the sector's capacity to achieve rapid growth and poverty reduction. Hence in this section we revisit the ongoing debate as to whether African agriculture really has the potential to achieve the kind of job cre­ ation, poverty reduction, and structural transformation that the Green Revolution achieved for Asian economics. The Case for Agriculture-based Development Strategies in Sub-Saharan Africa One of the reasons agricultural growth is thought to be so important at initial stages of development--especially for poverty reduction-is simply one of arith­ metic. Consider a stylized economy in an early stage of development: the majority of the population is rural and mostly engaged in the production of a few staple crops; most rural people are poor and typically more so than urban people (Sahn and Stifel 2003; World Bank 2003); and food consumption is a large share of a 74 The lAO rId Bank Research Observer, vol. 25, riO. I (Pebruary 2010) typical household's budget (50 percent or more). In this kind of economy the potential for a given growth rate in food production to raise the incomes and nutrition levels of most of the population, including the poorest, is tremendous (Diao, Headey, and Johnson 2008). The single most important commodity group for poverty reduction and nutritional improvements- (staple foods) -becomes cheaper for both the rural and urban populations, and the two most important assets that the poor own-their labor and their land-are suddenly in much greater demand (Lipton 2009). There are some qualifications to this stylized model, of course. First, the more food prices fall. the weaker is the incentive farmers have to increase agricultural production (Diao, Headey, and Johnson, 2008). Second, if the addition to farmers' incomes is also spent on food, demand for nonfarm labor may be limited (Dercon and Zeitlin 2009). Third, if unskilled wages do go up (or if food prices don't fall), other sectors may become less competitive, which could inhibit rather than promote structural transformation (Lewis 1954). Because of these tradeoffs there is potentially a role for strategic trade and pricing policies aimed at redistributing the effects of agricultural production growth across consumers, producers, unskilled workers, and the domestic and world economy. However, the basic reasoning of this simple model is sound, and there is also ample historical, econo­ metric, and household modeling evidence to confirm these basic intuitions (Ravallion and Datt 1996; Djurfeldt and others 2005; Chen and Ravallion 2007; Ravallion, Chen, and Sangraula 2007; Christiaensen, Demery, and Kiihl 2006; Ivanic and Martin 2008; World Bank 2008). Perhaps the most important aspect of this model. especially in the current context, is the extent to which a given agricultural growth rate raises the demand for labor. This channel is especially important because Africa is undergoing the early stages of the so called demographic transition in which the working age population becomes large relative to the dependent population (the very young and very old). Figure 4 shows that since 1985 Africa's age dependency ratio has started declining, and that it is expected to decline at a faster rate until at least 2050. The "demographic window" presented by this decline in age dependency ratios presents Africa with both an opportunity and a threat. The opportunity arises because the increasingly larger share of productively employed adults in a population raises incomes per capita, which in turn creates opportunities for increased investments in human and physical capital for the next generation. As figure 4 shows, age dependency ratios ~1:arted declining in Asia in the early 1970 s, and previous research has shown that the region's unusually fast demo­ graphic transition accounted for one-third to one-half of East Asia's dynamic growth rates during the period 1965-1990 (Bloom and Williamson 1998). However. that research also suggested that the benefits of the transition are not inevitable. If African economies cannot create sufficiently productive Headey, Bezemer. and Hazell 75 Figure 4. Trends in Age Dependency Ratios: Africa's Demographic Window Is Opening Up 40 0 1.1) 0 1.1) 0 1.1) 0 1.1) 0 1.1) 0 1.1) 0 1.1) 0 1.1) 0 1.1) 0 1.1) 0 \0 \0 IX) IX) 0) 0 0 .-t .-t m m <:t <:t " " 1.1) 1.1) 0) N N 1.1) 0) 0) 0) 0) 0) 0) 0) 0) 0'1 0) 0 0 0 0 0 0 0 0 0 0 0 ..... ..... ..... ..... ..... .-t ..... ..... ..... .-t N N N N N N N N N N N Jouree: Authors' calculations from UN (2009) data. employment-as East Asia did-these countries will end up bypassing this unique growth opportunity. Worse still a large population of underemployed young men greatly raises the threat of conflict: one study estimates that the risk of civil war is increased from 4.7 to 31 percent if the share of young men in the population doubles (Collier, Hoeffler, and Rohner 2007). This reasoning suggests that the goal of creating productive employment should be much higher on Africa's development agenda. Existing evidence also suggests that smallholder-based agricultural growth could be an engine of job cre­ ation in predominantly rural economies, and that it should generally be preferred to largeholder-based growth because of small farmers' greater use of both family labor and hired labor (Hazell and Diao 2005; Lipton 2009). As for nonagricul­ tural sectors, data on labor intensity are far too scarce and labor intensities vary so much that we are reluctant to generalize. However, we do know that many leading nonagricultural sectors in Africa. such as mining. are much less labor intensive than agriculture. This means that in a typical agriculture-based economy a purely nonagricultural growth strategy is overburdened from an employment perspective. Turning back to the stylized agriculture-based economy we considered above, let us assume that the agricultural population share is around 70 percent, the growth in the labor force is around 3 percent per annum (an average for developing countries in the 1990 s). and that agricultural growth is at least somewhat more labor intensive than nonagricultural growth 76 The World Bllnk ResearciJ Obsen'er, \'01. 25, no. 1 (February 2010) (for example a growth elasticity of 0.5 versus one of 0.3 for nonagriculture). If an industry-first strategy implies that agricultural production only keeps up with population growth, then a simple back-of-the-envelope calculation suggests that non agriculture would have to grow by over 20 percent per year to absorb surplus labor (Headey, Bezemer, and Hazell 2008).7 It is also significant that while East Asian countries like China and Vietnam have achieved remarkable rates of indus­ trial growth in the last few decades (typically in excess of 10 percent per annum), they still would have incurred significant unemployment problems had agriculture also not grown by 4-6 percent per annum. So from a job creation perspective, smallholder agriculture-based growth looks the best bet. Limitations of the Agriculture-first Model in Sub-Saharan Africa However, most critics of agriculture-first models do not contest the labor-intensive nature of smallholder-based growth. Instead they argue, for various reasons, that agriculture-based growth prospects in Africa are simply very limited. One poten­ tially very important criticism is that Asian experiences cannot simply be trans­ planted to Africa. And indeed. as we noted above. Africa's agronomic endowments are very different to those of Asia. especially in terms of the diversity of crops produced. the extent of irrigation, the diversity and quality of soil and climate types. and the generally lower population density. This implies that more locally adapted high-yielding varieties (HYVs) must be produced. while greater land availability in some parts of Africa suggests that the demand for HYVs and other land-intensive technologies may be more limited than was the case in popu­ lation-dense Asia (Binswanger and McIntyre 1987). Yet despite these limitations Africa does have a number of both experimental and localized R&D success stories (Haggblade and Gabre-Mahdin 2004). such that the real question is argu­ ably why these success stories are so rarely scaled up. Agroclimatic diversity could partly explain the poor uptake. but low R&D expenditures. pricing biases. insuffi­ cient investments in complementary investments (for example infrastructure). and unfavorable political economy factors also seem explain to the divergence between African and Asian agriculture (Djurfeldt and others 2005; Bezemer and Headey 20(8). A second problem with an agriculture-based strategy for Africa is that. relative to Asia. many African economies are abundant in oil or other mineral resources (Nigeria. Angola. Cameroon). and the group seems to be getting larger as countries like Chad, Ghana, and Uganda have recently started to develop large oil deposits. This naturally prompts some economists to argue that the comparative advantage endowed by resource abundance implies that agriculture has dimin­ ished in importance in Africa (Collier 2007; Dercon 2009). However, this argu­ ment seems to rely on some implicit assumptions that are questionable at best. Headey. Bezcmer. and Hazell 77 First, even if growth is regarded as the sale economic objective, the general equili­ brium implications of oil revenues are not very favorable because oil sectors have weak and potentially negative linkages to other sectors (because of the inflation­ ary effects of "Dutch Disease"). Second, if governments are trying to reduce poverty, then job creation is of the utmost importance. However. by themselves the oil and minerals sectors have very limited employment potential because they are capital intensive sectors. In other words. oil might have a comparative advan­ tage in growth. but it certainly does not have a comparative advantage in poverty reduction. A second weakness with the conventional comparative advantage viewpoint is that much of agriculture's economic potential lies in the African market where there is ample scope to substitute food imports with domestic food production (Diao, Headey. and Johnson 20(8). Even oil-rich Nigeria's recent economic success has been predominantly driven by agriculture. From 2000-07 agricul­ tural GDP growth in Nigeria accounted for almost half of the impressive doubling of national GDP. while the industrial sector accounted for just one-quarter. Hence over 2000-07, a period of rising oil prices. it looks like agriculture was the leading growth sector, not oil. s Moreover. it so happens that many of Africa's most mineral-rich countries are also those with tremendous biophysical potential for agricultural production. Nigeria has very good rainfall, relatively good soils. attractively large urban markets. good access to export markets via its coastal ports and its northern borders into the Sahel, and-despite high levels of popu­ lation density in some parts of the country-large tracts of unexploited but fertile land. The Democratic Republic of Congo has similar if not greater biophysical potential. but larger infrastructure constraints (Ulimwengu and others 2(09). and Angola is in a similar situation. Cameroon. which is gradually running out of oil. is probably Africa's best example of strong agricultural growth in recent years, although it too has yet to fulfill its true potential in agriculture. The message here is that mineral abundance does not rule out an important role for agriculture. Indeed. one of the primary uses of increased oil revenues in Africa should be the promotion of agricultural and rural development. A third objection often raised in the African context is that even if agriculture­ based strategies were a good idea SO years ago, it may now be too late to pursue this strategy because so many Africans have already left depressed rural areas in search of jobs in big cities (Bryceson 20(2). Indeed our own results above might seem consistent with the conjecture. However, a closer look at the data suggests that this argument is probably overstated. In table 4 we again examine the argu­ ably more reliable 2003 DHS data on employment for Nigerian men and women, since Nigeria is the most urbanized country of any size in sub-Saharan Africa. Several facts stand out. First. the proportion of the urban population share is probably exaggerated in Nigeria and perhaps elsewhere in Africa (Headey, 78 The World Bank Research Observer. vol. 25, 110. 1 (February 2010) Bezemer, and Hazell 200S). The nationally representative DHS survey puts the rural population share at around 60 percent. which is consistent with a recent study that uses the spatially dis aggregated data of GIS-type techniques to improve upon the UN's widely criticized urbanization estimates (Uchida and Nelson 200S). Hence. more accurate data would imply that there is still a large rural population in Nigeria and other African countries that may not need any signifi­ cant relocation in order to participate in agricultural activities. Recent in-depth studies suggest that participation in the agricultural labor market in rural Africa are far greater than large-scale household surveys suggest (World Bank 2008). A second fact discernible from table 4 is that a large proportion of Nigeria's male labor force in rural areas is unemployed (24 percent) or employed in low skilled nonfarm sectors, while most rural women are typically in low skilled services (55.7 percent). In conjunction with significant tracts of unused fertile land, these figures would imply that the prospects for a growing agricultural sector to provide mean­ ingful employment for rural Nigerians stuck in low return nonfarm activities look favorable, especially since there are typically very few barriers to entry in agricul­ tural labor markets. And in most other African countries the rural population share is much higher than in Nigeria. In short the vast majority of Africans (especially the poor) still live in rural areas and have adequate access to land. On these grounds, at least, agriculture-based strategies are certainly not redundant. A fourth objection to agriculture-based development strategies in Africa is that they ignore endogenous growth processes, including the economies of scale associated with manufacturing, and the agglomeration externalities that come with urbanization (Krugman 1991; see Henderson 2003 and Quigley 2008 for excellent reviews). These arguments merit further research because the size of agglomeration economies are particularly hard to estimate. However. on the basis of existing evidence and theory. their existence would only seem to qualify mar­ ginally the importance of agriculture in Africa, especially in the medium term (see Headey, Bezemer, and Hazell 200S for a fuller discussion). This is because Table 4. DHS-based Estimates of Shares in 2003 Not employed Agriculture Professional Sales. services Skilled manual Unskilled manual Rural Men Rural 24.0 40.5 9.7 10.8 11.3 3.6 100.0 Total 27.1 27.8 12.5 13.5 15.2 3.8 62.8 Women Rural n.a. 27.9 5.7 55.7 7.7 1.9 100.0 Total n.a. 20.4 9.6 56.6 9.6 2.5 64.5 n.a. Not applicable. Notes: Data are the percent distribution employed in the 12 months before the survey. Key statistics are highlighted in italics. Source: DHS (2009). Headey. Bezemer. and Hazell 79 rapid urbanization is generally associated with rising urban unemployment for the reasons listed above, as well as significant problems with public service deliv­ ery (Hewett and Montgomery 2001). African experiences like Nigeria's show that urbanization alone is simply not enough to sustain durable and widely shared growth. Moreover, agricultural development should not always be thought of as competing with nonagricultural growth processes. Agriculture can actually support the growth of cities by increasing demand for nonfarm services, by keeping food prices low (which in turn keeps real urban wages competitive), and by freeing up scarce foreign exchange for industrial imports rather than food imports. Because of these spillovers it is no coincidence that Asia turned into the world's factory after experiencing a smallholder-based Green Revolution. and that the reverse is true in sub-Saharan Africa where urban food prices are often sur­ prisingly high and manufacturing is generally uncompetitive (Headey, Bezemer. and Hazell 2008). Finally, most of the existing arguments about the sectoral role of policies relate to comparisons of the expected benefits of agricultural and nonagricultural growth in terms of growth and poverty reduction. But in theory an optimal devel­ opment strategy requires allocating investments until marginal cost-benefit ratios are equalized across sectors. In reality, however, we have very little data on the returns to various public expenditures, especially in Africa. What data there are generally indicate high social returns to agricultural R&D and rural infrastructure (see Fan 2008 for various case studies in Asia and Africa). but against this evi­ dence runs a significant tide of mostly anecdotal assessments of the weakness of Africa's agricultural ministries (Headey, Benson, and Kolavalli 2009) and the poor state of its agricultural R&D institutions (Pardey and others 2006).9 Perhaps the only encouraging view vis-a.-vis these assessments is that Asia's Green Revolutions were primarily driven by government elites rather than their own poorly regarded agricultural ministries (World Bank 2008).10 In this view the binding constraint is not the capacity of agricultural ministries but the political will of the governing elite. Development Strategies for Managing the Transition out of Agriculture This article has identified a startling divergence in the agricultural exit paths of developing countries. While economic growth is traditionally accompanied by more rapid agricultural exits, fast-growing East Asian countries have generally only experienced modest ones. Slow-growing Africa, in contrast, has often wit­ nessed surprisingly rapid agricultural exits, and in some oil-rich countries~ Nigeria. Cameroon, Gabon~extremely rapid ones. To some extent these different 80 The World Bank Research Observer. vol. 25, no. 1 (February 2010) paths are the result of different endowments, but very different policy regimes also explain the divergence, especially the success of the Green Revolution in Asia and its failure in Africa. Despite this historical divergence, we have argued in this paper that Asia is still facing some daunting employment problems. In Asia there is a paradoxical food situation today. On the one hand, there are millions of increasingly affluent Asians who are rapidly diversifying and enriching their diets. Yet despite this growing food affluence for many, about 800 million Asians still live in abject poverty, and hunger and malnourishment are surprisingly persistent. These people desperately need better livelihood opportunities. In the nonfarm sector, rural-urban migration (especially from less favored areas) could often be the best solution, but recent evidence suggests that Asia's manufacturing sectors are struggling to create enough jobs (Khan 2007). This suggests that Asia's rural economies need to continue providing new job opportunities. In agriculture the most promising prospect for poor Asian farmers is to tap into urban-led economic growth, particularly the shift toward more affluent Asian diets. Yet the key lesson of the Green Revolution is that, if left to market forces alone, many poorer regions and poor people are likely to be left behind in modernization processes (Rosegrant and Hazell 2000). Asia's high population density also implies that the RNF sector may have a comparative advantage in employment creation because transaction and labor costs in rural Asia are generally low. Yet Asia's RNF sector is still been hindered by the neglect of government policies, especially those relating to rural credit, education, and infrastructure (Haggblade, Hazell, and Reardon 2007a). So for all their success to date, Asian governments have unfinished business in the war against rural poverty. In Africa, this war has scarcely begun. Despite rapid urbanization in a few countries, most of Africa's poor still reside in rural areas and still rely on agricul­ ture to eke out a living. Even in Nigeria, Africa's largest economy and the one in which the agricultural exodus is starkest, agricultural growth still has consider­ able potential to reduce rural and urban poverty. So the real question in Africa is arguably not whether agriculture has potential, but how that potential can best be exploited. The main lesson Africa can learn from East Asia is that labor­ intensive agricultural growth in egalitarian tenure systems is extremely pro-poor and can provide the foundation for successful nonfarm diversification (Bezemer and Headey 2008). Yet Africa's capacity to simply transplant the Asian Green Revolution model is limited. Africa is not only geographically more diverse, its political economy is also different. Asia's agricultural initiatives were sparked by a combination of famines, higher food prices, and nationalist politics. In most of Africa's more fertile countries the production of food staples has often been ignored because basic food security was less of an issue historically, because natural resources and Headey, Bezemer. and Hazell 81 traditional cash crops offered more lucrative earnings for government coffers, and because nationalist tendencies were undermined by ethnic conflicts. For these and other reasons food security has typically only been regarded as a high pri­ ority in drought-prone areas such as Ethiopia, Malawi. and the Sahelian countries. However. one of the few beneficial effects of the global food crisis of 2007-08 is that the political support for improving national food security has undoubtedly been reinvigorated in a number of African countries. The question is whether this political momentum will last and whether agriculture's champions in the continent can use that momentum to develop more coherent and more sustainable agricultural development strategies. 11 Notes Derek Headey is Research Fellow at the International Food Policy Research Institute (IFPRI). 2033 K Street NW, Washington. DC 20006-1002. USA: phone: -1 202-862-8103: fax: +1 202-467­ 4439: email address: d.Headey@cgiar.org. Dirk Bezemer is Assistant Professor at the Faculty of Economics & Business, University of Groningen. The Netherlands. Peter B. Hazell is Visiting Professor at Imperial College. Wye Campus. London, UK. The authors would like to thank Jose Funes for excellent research assistance, as well as participants at a seminar given at IFPRI. particu­ larly Xinshen Diao, Shenggen llano and Xiaobo Zhang. 1. See Temple and Woessman (2006) for a review. 2. Note. however. that GDP data is biased by the omission of the informal economy; which seems to be predominantly in the nonfarm sector. For example, despite no economic growth over 1960-2000, Nigeria appears to have experienced a large exodus of workers out of agriculture, but relatively little decline in agriculture's share in GDP. Much of this could be attributed to the informal sector. See Headey, Bezemer, and Hazell (2008) for further discussion. 3. Other developing regions are also of interest, although the Middle East and North Africa and most Latin American countries generally started form higher income levels. However, many of these countries appear to have urbanized very quickly. Whilst comparisons between Latin America 30 or 40 years ago and Africa today would make for interesting research, in this paper we focus on the starker contrast between Africa and developing Asia. 4. South Korea is not included: it did not follow the standard Asian path in that it urbanized very quickly. partly because of a weaker nonfarm sector relative to comparable countries such as Taiwan (Otsuka 2007). 5. We also note that Nigerian data is complicated by the high rates of unemployment reported in both urban and rural areas (above 20 percent). We have allocated unemployment according to the nonfarm shares of the remaining categories. 6. An obvious caveat to this concern is that supply begets its own demand. However, Asia's growth model has heavily relied on external demand, which is not inelastic. Moreover, the least devel­ oped Asian countries are facing a challenge that previous Asian Tigers did not encounter because the latter tended to grow in sequential phases with quite complementary shifts in economic struc­ tures-the so called 'flying geese' phenomenon-and among them only Japan could truly be called populous by Asian standards. In contrast developing Asia is now characterized by a large number of countries-China, India, Indonesia, Pakistan. Vietnam, Bangladesh-that are highly populous, at similar stages of development. and all in need of significant job creation outside of agriculture. 7. Specifically we assume a two-sector economy with agriculture (A) and nonagriculture (N). By definition, the annual growth in employment (l) equals the annual growth rates in output of agri­ culture (t) and nonagriculture (gN). multiplied by their initial employment shares (s'\ SN), and 82 The World Bank Research Observer. vol. 25. 110. 1 (February 2010) their (full) employment elasticities with respect to output (£A = 0.5; £1' 0.3). Rearranging this identity, we can derive the nonagricultural gmwth required to achieve full employment based on three key characteristics of the economy: (1) the relative labor intensities of agriculture and nonagri­ culture; (2) the share of agricultural employment 70 percent); and (3) the agricultural growth rate (which is set equal to population growth rate, assumed to be 2.9 percent). The full employment nonagricultural growth rate is then given by: 8. Of course this statement ignores the role of linkages. It is possible that exogenously deter­ mined oil revenue might have caused increased demand for agricultural produce or increased prices. hence stimulating agriculture. But it is difficult to imagine that when industry grows by 25 percent. it causes agriculture to grow by 50 percent. implying a multiplier of two. 9. Much of the quantitative evidence on expenditure effectiveness relates to budgetary measures, such as the ratio of actual expenditures to planned expenditures for various ministries. By this measure public eX'Penditure reviews in Africa typically reveal that ministries of agriculture fair poorly relative to many other key ministries such as health and education. See Headey. Benson. and Kolavalli (2009) for some discussion. 10. Background research on agricultural ministries for the 2008 WDR was conducted by Regina Birner and colleagues at the International Food Policy Research Institute. but these authors could not find a single example of a reformed agricultural ministry that was Widely regarded as a success story (personal communication). This state of affairs is reflected in the common joke that if you removed the agricultural ministry. the rural population wouldn't notice-and vice versa. 11. For a number of examples of recent work on agricultural strategies in Africa see International Food Policy Research Institute's (IFPRI) publications homepage (http://www.ifprLorg/ pubs/pubs.htm). the World Bank's Agricultural and Rural Development website, as well as the recent review of Ethiopia's Agricultural Demand-Led Industrialization(ADLI) strategy by Dercon and colleagues, summarized in Dercon and Zeitlin (2009). References Anderson, Dennis, and Mark Leiserson. 1980. "Rural Nonfarm Employment in Developing Countries." Economic Development and Cultural Change. 28(2): 227-48. Anriquez, Gustavo. and Genny Bonomi. 2007. "Long-Term Farming and Rural Demographic Trends." Background Paper for World Development Report 2008. The World Bank, Washington, DC. Bates, Robert H. 1981. Markets and States in Tropical Africa: The Political Basis of Agricultural Policies. Berkeley, CA: University of California Press. Benjamin, Dwayne, Loren Brandt. and John Giles. Forthcoming. "Income Inequality During China's Economic Transition." In Loren Brandt. and Thomas Rawski eds .. China's Economic Transition: Origins. Mechanisms. and Consequences. Cambridge: Cambridge University Press. Berry. R. Albert, and William Clinc. 1979. Agrarian Structure alld Productivity ill Developing Countries. Baltimore: Johns Hopkins University Press. Besley, Timothy, and Robin Burgess. 2004. "Can Labour Regulation Hinder Economic Performance? Evidence from India." The Quarterly Journal of Economics, 119(1): 91 134. Bezemer, Dirk, and Derek D. Headey. 2008. 'i\griculture, Development and Urban Bias." World Development. 34(8): 1342-64. Headey, Bezemer, and Hazell 83 Bhalla. G.S.. and Peter B. Hazell. 2003. "Rural Employment and Poverty Strategies to Eliminate Rural Poverty within a Generation." Economic and Political v\kekly. 38(33): 3473-84. Bhattacharya. B.B.. and S. Sakthivel. 2004. "Economic Reforms and the Jobless Growth in India in the 1990s." Working Paper E/245/2004. Delhi University Institute of Economic Growth. Delhi. Binswanger. H.P.. and K. Deininger. 1997. "Explaining Agricultural and Agrarian Policies in Developing Countries." Journal o.fEconomic Literature, 35(4): 1958-2005. Binswanger, Hans p', and John McIntire. 1987. "Behavioral and Material Determinants of Production Relations in Land-Abundant Tropical Agriculture." Economic Development and Cultural Change. 36(1): 73-99. Bloom, David E.. and Jeffrey G. Williamson. 1998. "Demographic Transitions and Economic Miracles in Emerging Asia." World Bank Economic Review. 12(3): 419-55. Bocquier. Philippe. 2004. "World Urbanization Prospects: An Alternative to the UN Model of Projection Compatible with Urban Transition Theory." DIAL Working Paper DT/2004/08. Developpement Institutions & Analyses de Long terme. de Brauw. Alan. 2002. "Are Women Taking over the Farm in China?" Department of Economics Working Paper 199. Williams College. de Brauw, Alan. and Gero Carletto. 2008. "Improving the Measurement and Policy Relevance of Migration Information in Multi-topic Household Surveys." Unpublished manuscript. Washington. DC: International Food Policy Research Institute. Brockerhoff. Martin. and Ellen Brennan. 1998. "The Poverty of Cities in Developing Regions." Population and Development Review. 24(l): 75-114. Bryceson. D.E 2002. "The Scramble in Africa: Reorienting Rural Livelihoods." World Development. 30(5): 725-39. Buvinic. Mayra, Catherine Gwin, and Lisa M. Bates. ] 996. Investing in Women: Progress and Prospects for the World Bank. Washington. DC: Overseas Development Council in cooperation with the International Center for Research on Women and Johns Hopkins University Press. Chen. Shaohua, and Martin Ravallion. 2007. "China's (Uneven) Progress against Poverty." Journal o.!Development Economics. 82(1): 1-42. Chenery. Hollis B. 1979. Structural Change and Development Policy. New York: Oxford University Press. Chotikapanich. Duangkamon. D.S. Prasada Rao. and Kam Ki Tang. 2007. "Estimating Income Inequality in China using Grouped Data and the Generalized Beta Distribution." Review of Income and Vikalth. 53(1): 127-47. Christiaensen. Luc. Lionel Demery, and Jesper KiihL 2006. "The Role of Agriculture in Poverty Reduction: An Empirical Perspective." World Bank Policy Research Working Paper Series 4013. The World Bank. Washington. DC. Cohen. B. 2004. "Urban Growth in Developing Countries: A Review of Current Trends and a Caution Regarding Existing Forecasts." World Development. 32(1}: 23-51. Collier. PauL 2007. The Bottom Billion: Why the Poorest Countries Are Failing and What Can Be Done about It New York: Oxford Cniversity Press. Collier. P.. and Venables. A. 2007. "Rethinking Trade Preferences: How Africa Can Diversify its Exports." In D. Greenway ed .. The World Economy: Global Trade Policy 2007. Hoboken. NJ: Wiley. Collier. P.. A. Hoeffler. and D. Rohner. 2007. Bellond Greed and Grievance: Feasibility and Civil War. Oxford: CSAE. Deaton. A.. and V. KozeL 2005. "Data and Dogma: The Great Indian Poverty Debate." The World Bank Research Observer, 20(2): 177-99. 84 The "World Bank Research Observer, vol. 25. no. 1 (Hebruary 2010) Dercon, S. 2009. "Rural Poverty: Old Challenges in New Contexts." World Bank Research Observer. 24(1): 1-28. Dercon. S.. and A. Zeitlin. 2009. "Rethinking Agriculture and Growth in Ethiopia: A Conceptual Discussion". Paper prepared as part of a DFID study on Agriculture and Growth in Ethiopia. Centre for the Study of African Economies (CSAE). Oxford University. (http://www.economics.ox. ac.uk/members/ stefan.dercon/policy3thiopiaagriculturework.htm). Djurfeldt. G.. H. Holmen. M. Jirstrom, and R. Larsson. (eds). 2005. The African Food Crisis: I,essons from the Asian Green Revolution. Wallingford, CABl. DHS (Demographic Health Surveys). 2009. Measure DRS Stat-Compiler. Demographic Health Surveys. sponsored by USAID. (http://www.statcompiler.com). Diao. Xinshen. Peter Hazell, Danielle Resnick, and James Thurlow. 2007. "The Role of Agriculture in Development: Implications for sub-Saharan Africa." International Food Policy Research Institute. (http://www.ifprLorg/publication/role-agriculture-deveiopment-O). Diao, X., D. Headey, and M. Johnson. 2008. "Toward a Green Revolution in Africa: What Would It Achieve, and What Would It Require?" Agricultural Economics, 39(supplement):539-50. Du, Yang, Albert Park. and Sangui Wang. 2005. "Migration and Rural Poverty in China." Journal of Comparative Economics, 35:688-709. Dudkj', Kirk, and Bernard Pillet. 1998. "Fertility Levels, Trends. and Differentials in Sub-Saharan Africa in the 1980s and 1990s." Studies in l;'amily Planning, 29(1): ] - 22. Eastwood, Robert, and Michael Lipton. 2004. "Rural-Urban Dimensions of Inequality Change." In Giovanni Andrea Cornia ed.. Inequality, Growth, and Poverty in an Era of Liberalization and Globalization. Oxford: Oxford University Press. Fan, Shenggen. 2008. Public Expenditures, Growth, and Poverty; Lessons from Developing Countries. Baltimore, MD: Johns Hopkins University Press. Fan, Shenggen, and Connie Chan-Kang. 2005. "Is Small beautiful? Farm Size, Productivity. and Poverty in Asian Agriculture." Agricultural Economics, 32(1): 135-46. Fan, Shenggen, and Peter Hazell. 2001. "Returns to Public Investments in the Less-favored Areas of India and China." American Journal of Agricultural Economics, 83(5): 1217-22. Fan, Shenggen, Connie Chan-Kang, and Anit Mukherjee. 2005. "Rural and Urban Dynamics and Poverty: Evidence from China and India." FCt-;'U Discussion Papers 196. International Food Policy Research Institute. Washington. DC. FAO (Food and Agriculture Organisation). 2009. AGROS'DIT. Rome: Food and Agriculture Organization. Fay, M.. and C. Opal. 1999. "Urbanization without Growth: A Not-so-uncommon Phenomenon." World Bank Policy Research Working Paper 2412. Washington. DC: The World Bank. Foster, A., and M. Rosenzweig. 2004. '!\gricultural Productivity Growth, Rural Economic Diversity, and Economic Reforms: India. 1970-2000. Economic Development and Cultural Change, 52(3): 509-42. GRUMPS (Global Rural-Urban Mapping Project). 2008. Global Rural-Urban Mapping Project (GRUMP): Urban/Rural Population Grids. Center for International Earth Science Information Network, Columbia University; International Food Policy Research Institute; the World Bank; and Centro Internacional de Agricultura Tropical. Gugler, Josef. 1982. "Overurbanization Reconsidered." Economic Development and Cultural Change, 31(1): 173-89. Haggblade, S., and E. Gabre-Mahdin. 2004. "Success in African Agriculture: Results of an Expert Survey." World Development, 32(4): 745-66. Headey. Bezemer. and Hazell 85 Haggblade. S.. P.B.R. Hazell. and 'r: Reardon. 2007a. "Strategies for Stimulating Equitable Growth in the Rural Nonfarm Economy." In Steven Haggblade. P.B.R. Hazell. and T. Reardon eds.. Transforming the Rural Nonfarm Economy. Baltimore, MD: Johns Hopkins University Press. 2007b. Transforming the Rural Nonfarm Economy. Baltimore. MD: Johns Hopkins University Press. Harris. J.R.. and M.P. Todaro. 1970. "Migration. Cnemployment and Development; A Two-Sector Analysis." American Economic Review. 60(1): 126~42. Hazell. Pet.er. and Xinshen Diao. 2005. "The Role of Agriculture and Small Farms in Economic Development:' Presented at The Future of Small Farms workshop. Wye College. United Kingdom. June 26-29. Headey. Derek D.. Todd Benson. and Shashidhara KolavallL 2009. "Why African Governments Under-invest in Agriculture: Results from an Expert Survey." Conference paper presented at the International Association of Agricultural Economists Conference. Beijing. China. (http :llageconsearch.umn.edu/handle/ 51818). Headey. Derek D.. Dirk Bezemer. D.. and Peter B. Hazell. 2008. 'i\gricultural Exit Problems: Causes and Consequences." Discussion Paper 802. International Food Policy Research Institute. Washington. DC. (wvllw.ifprLorg/pubs/dp/ifpridp00802.asp). Henderson. Vernon. 2003. "The Urbanization Process and Economic Growth: The So-What Question."lournal of EcorlOmic Growth. 8(1): 47~7L Henderson. Vernon. Adam Storeygard. and David N. Wei I. 2009. "Measuring Economic Growth From Outer Space." Working Paper 15199. National Bureau Of Economic Research. Cambridge. MA. (http; /1 www.nber.org/papers/w15199). Hewett. P.C.. and M.R. Montgomery. 2001. "Poverty and Public Services in Developing-country Cities." Policy Research Division Working Papers 154. Population Council. New York. Hirschman. Albert O. 1958. The Strategy of Economic Development. New Haven: Yale University Press. Holz. C.A. 2004. "China's Reform Period Economic Growth: Why Angus ~faddison Got It Wrong and What that Means." Social Science Division Working Paper 0504012, Hong Kong University of Science & Technology. Iill (International Labour Organization). 1996. FAO Estimates Derived from Population Data and Economically Active Population 1950~2010: 4th Revision Geneva: International Labour Organization. _ . _. 2008. lvletlwdological Description of IT.l) Estimates alld Projections of the Economically Active Population: 1.980~2020. 5th edn. Geneva; International Labour Organization. (http://laborsta.iIo. org/ applv81 data/EAPEP IEAPEP~methodology. pdf). Ivanic. M.. and W Martin. 2008. "Implications of Higer Global :Food Prices for Poverty in Low­ Income Countries," .1gricultural Economics, 39(supplement):405 ~ 16. IWMI (International Water Management Institute) and Earthscan. 2007. Water for Food. Water for Life: A Comprehensive Assessment of Water iVlanagement in Agriculture. London and Colombo: Earthscan and International Water Management Institute. Johnson. Michael. Peter Hazell. and Ashok GulatL 2003. "The Role of Intermediate Factor Markets in Asia's Green Revolution: Lessons for Africa?" Americanloumal of Agricultural Economics. 85(5): 1211 16. Johnston. Bruce R, and P. Kilby. 1975. Agriculture and Structural Transformation: Economic Strategies in Late-developing Countries. London: Oxford University Press. Joshi. P.K.. Ashok Gulati. and Ralph Cummings. Jr. 2007. Agricultural Diversification and Smallholders in South Asia. New Delhi: Academic Foundation. 86 The World Bank Research Observer. vol. 25, no. 1 (February 201 ()) Kanbur. Ravi. and Xiaobo Zhang. 1999. "Which Regional Inequality: Rural-Urban or Coast­ Inland? An Application to China." Jourtlill of Comparative Economics. 27(1999): 686-701. 2005. "Fifty Years of Regional Inequality in China: A Journey Through Central Planning, Reform and Openness." Review of Development Economics. 9(1): 87-106. Karp. Larry. 2007. "Managing Migration from the Traditional to Modern Sector in Developing Countries." Background Paper for World Development Report 2008. The World Bank. Washington. DC. Kelley. Allen C., and Robert Schmidt. 2003. "Economic and Demographic Change: A Synthesis of Models. Findings. and Perspectives." In Nancy Birdsall. Allen C. Kelley. and Steven Sinding eds .. Population Matters. New York: Oxford University Press. Khan. Azizur Rahman. 2007. ilsian Experience on Growth. Employment and Poverty. Geneva: UNDP and 110. Kherallah. Mylene. Christopher Delgado. Eleni Gabre-Madhin. Nicholas "Hnot. and Michael Johnson. 2002. Reforming A.gricultural Markets in iljriCil. Baltimore, MD: Johns Hopkins University Press. Kochhar. Kalpana. Utsav Kumar. Raghuram Rajan. Arvind Subramanian. and Ioannis Tokatlidis, 2006. "India's Patter of Development: What Happened. What Follows?" IMF Working Paper 06/ 22. Washington. International Monetary Fund. Krueger. Anne 0 .. Maurice Schiff. and Alberto Valdes. 1991. Political Economy of ilgricultural Pricing Policy. Baltimore. MD: Johns Hopkins University Press. Krugman. Paul R, 1991. "Increasing Returns and Economic Geography." The Journal of Political Economy. 99(3): 483-99. Kuznets, Simon, 1973. "Modern Economic Growth: Findings and Reflections." A.merican Economic Review. 63(3): 247 - 58. Laidlaw. Karen A.. and Edward G. Stockwell. ] 979. "Trends in the Relationship between Urbanisation and Development in Africa," The Journal of Modem iljrican Studies, 17(4): 687-94. Lewis. Arthur W 1954, "Economic Development with enlimited Supplies of Labour," The A1anchester School, 28(2): 139-91. Lipton. M, 1977. Why Poor People Stay Poor: A Study of Urban Bias in World Development. I,ondon: Temple Smith. 2009. "Can Small Farmers Survive. Prosper. or Be the Key Channel to Cut Mass Poverty?" Electronic Journal of Agricultural and Development Economics. 3(1): 58-85. Maddison, Angus. 1998. Chinese Economic Performance ill the Long Run. Paris: Development Centre of the Organisation for Economic Co-operation and Development. Malthus, Thomas R. 1798. iln Essay on the Principle of Population and il Summary View of the Principle of PopulatioH. Harmondsworth: Penguin. Mehra. Rekha. and Sarah Gammage. 1999. "Trends. Countertrends. and Gaps in Women's Employment." World Development. 27(3): 533-50. Mellor. J.W 1976. The New Economics of Growth: A Strategy for India and the Developing World. Ithaca. NY: Cornell Cniversity Press. 'Vfilanovic. Branko. 2005. "Half a World: Regional Inequality in Five Great Federations." World Bank Policy Researeh Working Paper 3699. The World Bank, Washington, DC. Mukherjee. Anit. and Xiaobo Zhang, 2007. "Contrasting Rural Nonfarm Policies and Performance in China and India: Lessons for the Future," In Steven Haggblade. Peter Hazell. and Thomas Reardon cds" Transforming the Ruml Nonfarm Economy, Baltimore. "10: Johns Hopkins University Press, lIeadey. Beumer. and Hazell 87 Munshi. Kaivan. and Mark Rosenzweig. 2005. Why is Mobility in India so Low? Social Insurance. Inequality. and Growth." cm Working Paper 121. Centre for International Development, Harvard University. Nagayets. Oksana. 2005. "Small Farms: Current Status and Key Trends." Information brief prepared for the IFPRIIODI/Imperial College Research Workshop on "The Future of small farms," Wye. UK, June 26-29. Otsuka. Keijiro. 2007. "The Rural Industrial Transition in East Asia: Influences and Implications." In Steven Haggblade, Peter Hazell. and Thomas Reardon eds., Trans/orming the Rural Non/arm Economy. Baltimore. MD: Johns Hopkins University Press. Pardey, P.G., N. Beintema, S. Dehmer, and S. Wood. 2006. '1\gricultural Research: A Growing Global Divide?" Agricultural Science and Technology Indicators Initiative. Washington. DC. International Food Policy Research Institute. Quigley, John M. 2008. "Urbanization. Agglomeration and Economic Development." In Michael Spence. and Robert M. Buckley. and Patricia Clarke Annez eds.. Urbanization and Growth. Washington, DC: World Bank Publications. Ranis, Gustav. and John C.H. FeL 1961. '1\ Theory of Economic Development." American Economic Review, 51:533-65. Ravallion, Martin, and Gaurav Datt. 1996. "How Important to India's Poor Is the Sectoral Composition of Economic Growth?" World Bank Ecollomic Review, 10(1): 1-25. Ravallion, Martin, Shaohua Chen. and Prem Sangraula. 2007. "New Evidence on the Urbanization of Global Poverty." World Bank Working Paper WPS4199. The World Bank. Washington, DC. Rawski. T.G. 2001. "What Is Happening to China's GDP Statistics?" Chilla Economic Review, 12(4): 347-54. Reardon, Thomas. 1997. "Using Evidence of Household Income Diversification to Inform the Study of the Rural Nonfarm Labor Market in Africa." World Development. 25(5): 735-47. Rodrik. D. (2007). One Economics, Many Recipes: Globalizatioll. Institutions and Economic Growth. Princeton. NJ: Princeton University Press. Rosegrant. Mark. and Peter Hazell. 2000. Trans/orming the Rural Asian Economy: The Unfinished Revolutioll. Oxford. New York: Oxford University Press on behalf of the Asian Development Bank. Sahn. D.E., and D.C. Stife!. 2003. "Urban-Rural Inequality in Living Standards in Africa." Journal 0/ African Economies, 12(4): 564-97. Schneider, Friedrich. 2005. "Shadow Economies around the World: What Do We Really Know?" European Journal 0/ Political Economy, 21:598-642. Schultz, Theodore W 1964. Trans/orming Traditional Agriculture. New Haven: Yale University Press. Sen, Amartya K. 1967. "Surplus Labour in India: A Critique of Schultz's Statistical Test." The Economic Journal. 77(305): 154-6l. Smith, Adam. 1776. ATl Inquiry into the Nature and Causes of the Wealth 0/ Nations, London: Methuen and Co, Sovani. N.V. 1964. "The Analysis of 'Over-Urbanization.''' Economic Development and Cultural Change, 12(2): 113-22. Summers, Robert, and Alan Heston. 2002. Penn World Table Version 6.1. Pennsylvania: Center for International Comparisons at the University of Pennsylvania. Temple, Jonathan, and Ludger Woessman. 2006. "Dualism and Cross-Country Growth Regressions." Journal a/Economic Growth. 11(3): 187-228. Timmer. c.P. 2002. '1\griculture and Economic Development." In B. Gardner, and G. Rausser eds., Handbook 0/ Agricultural Economics, vol. 2A. Amsterdam: Elsevier. 88 The World Bank Resean:h Observer, vol. 25. 110. 1 (February 2010) Timmer. M.P.. and G.}. de Vries. 2007. '1\ Cross-Country Database for Sectoral Employment and Productivity in Asia and Latin America. 1950-2005." GGDC Working Paper, Groningen Growth and Development Centre, University of Groningen. (www.ggdc.net/databases/lO_sector.htm). Todaro, Michael P. 1997. "Urbanization. Unemployment. and Migration in Africa: Theory and Policy." PDR Working Paper 104. Population Development Council, New York. Uchida. H.. and A. Nelson. 2008. '1\gglomeration Index: Towards a New Measure of Urban Concentration." Background Paper for the World Development Report 2009. Washington, DC: The World Bank. Ulimwengu, John, Jose Funes, Derek Headey. and Liang You. 2009. "Paving the Way for Development: The Impact of Road Infrastructure on Agricultural Production and Household Wealth in the Democratic Republic of Congo." Paper presented at the 2009 Annual Meeting of the Agricultural and Applied Economics Association, Milwaukee, Wisconsin, July 26-28. UN (United Nations). 2009. World Population Prospects: The 2008 Revision Database. New York: United Nations. (http://esa.un.org/unpp/index.asp?panel= 2). Wang, X., and 1. Meng. 2001. '1\ Reevaluation of China's Economic Growth." China Economic Review, 12:338-46. Winters, p', '1: Essam, B. Davis. A. Zezza, C. Carletto. and K. Stamoulis. 2008. "Patterns of Rural Development: A Cross-country Comparison Using Microeconomic Data." ESA Working Paper 08­ 06, Food and Agriculture Organization. Rome. Wood. Adrian. 1997. "Openness and Wage Inequality in Developing Countries: The Latin American Challenge to East Asian Conventional Wisdom." World Bank Economic Review. 11(1): 33-57. World Bank. 1979. World Development Report 1979. New York: Oxford University Press . ._ ..~.... 1984. World Development Report 1984. New York: Oxford University Press. _._. 2000. World Development Report 2000: Entering the 21st Century. Washington, DC: The World Bank. _~_. 2OCH. Rural Poverty Report. Washington. DC: The World Bank. _~_. 2008. World Development Report 2008: Agriculture for Development. Washington, DC: The World Bank. _.__ . ._. 2009a. World Development Report 2009: Resilaping Ecollomic Geography. Washington DC: The World Bank. ~. __. 2009b. World Development Indicators Online. Washington. DC: The World Bank. :t.hang. Xiaobo. and Xiaopeng Luo. 2007. "How to Make Pro~Poor Policy Work? Lessons from China's Rural Reform." Global Development Network Conference. Beijing. Headey. Bezemer, (lnd Hazell 89 Notes Are ALL the Sacred Cows Dead? InlpLications of the FinanciaL Crisis for Macro- and FinanciaL PoLicies Asl1 Demirgi.i{:-Kunt • Luis Serven --------------- -- -- -- -- -- -- ----­ ... ... ... ... ... The recent global financial crisis has shaken the corifidence of industrial and developing countries alike in the very blueprint of the financial and macropolicies that underlie the ~stern capitalist systems. In an effort to contain the crisis from spreading, the auth­ orities in the United States and many European governments have taken unprecedented steps of providing extensive liquidity, giving assurances to bank depositors and creditors that include blanket guarantees. structuring bail-out programs that include taking large ownership stakes in financial institutions, and establishing programs for direct provision of credit to nonfinancial institutions. Emphasizing the importance of incentives and ten­ sions between short term and longer term policy responses to crisis management, the authors draw on a large body of research evidence and country experiences to discuss the implications of the current crisis for financial and macroeconomic policies going forward. JEL codes: GOI, G21, G28, G32, E52, E58, F32 The finanCial turmoil that started as a meltdown in structured securitization instruments in the summer of 2007 in the United States and the United Kingdom has quickly spread to the rest of the industrial world. and has now become a full­ blown global financial crisis. In an effort to contain the crisis from spreading, the authorities in the United States and many European governments have taken unprecedented steps of providing extensive liquidity. giving assurances to bank depositors and creditors that include blanket guarantees, structuring bail-out pro­ grams that include taking large ownership stakes in financial institutions, and establishing programs for direct provision of credit to nonfinancial institutions. In some developing countries, there is talk about reintroducing capital controls as a policy of last resort in the event of extensive bank runs and capital outflows. The World Bank Research Observer The Author 2010, Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development I THE WORlD BANK. All rights reserved, For permissions, please e-mail: journals,permissions@oxfordjournals,org doi;lO,1093/wbro/lkp027 Advance Access publication February 4,2010 25:91-124 Also questioned is the wisdom of a monetary policy narrowly focused on goods prices without taking into account asset price inflation, and prudential regulation that does not recognize systemic vulnerabilities. Hence the crisis has shaken the confidence of industrial and developing countries alike in the very blueprint of the financial and macropolicies that underlie the Western capitalist systems. It is not surprising that many analysts are already declaring capitalism, and the mainstream policy view associated with it, to be dead. We argue that the "sacred cows" of financial and macropolicies are very much alive. We seek to make clear that (i) the ongoing crisis does not simply reflect a failure of free markets, but is a reaction of market participants to distorted incen­ tives; and (ii) managing a systemic crisis requires policy decisions to be made at different stages of the crisis-the immediate containment stage as well as the longer-term resolution and structural reforms that follow-which often entail dif­ ficult trade-offs between re-establishing confidence in the short term and contain­ ing moral hazard in the long term. Keeping in mind the importance of incentives and tensions between short-term and longer-term policy responses to crisis management, we address the following questions about the implications of this crisis for financial and macroeconomic policies going forward: • Are blanket guarantees inevitable to halt a systemic crisis? • Should governments bail out and take ownership of financial institutions? • Should governments regulate finance much more aggressively given the fail­ ures in market discipline? • Should monetary policy target asset prices? • Should countries resort to capital controls to contain the crisis? Crises recur in part because people forget the lessons from previous crises. While every crisis is different. past crises also provide important lessons that need to be learned to prevent policymakers from reinventing the wheel every time a new crisis erupts. We draw on past research and country experiences to address the issues that are at the forefront of policy debate today. Use of Blanket Guarantees in Containing a Systemic Crisis Crises go through different stages. The first is called the "containment" stage in which crises often emerge unexpectedly and evolve very quickly. 1 This is the stage that attracts the most attention, with bank runs, emergency liquidity loans. and weekend crisis meetings. Time is of the essence and the need for speed generally takes over good judgment. How should the authorities judge whether liquidity 92 The World Bank Research Observer; vol. 25. no. 1 (February 2010) support or official guarantees should be employed in the hope of preventing col­ lapse, but potentially at high long-term costs? Part of the answer lies in better crisis preparedness. Systemic crises are infrequent events. Hence, the incumbent policymakers often claim there is "no playbook" for handling crises. This lack of experience and knowledge leads to trial-and-error policymaking and copying of policy responses-and often mistakes-that are being employed elsewhere. In the latest crisis the U.S. and European governments provided extensive assur­ ances to bank depositors and creditors that included blanket guarantees in many cases. Some developing countries copied these arrangements, providing blanket guarantees in order to prevent capital outflows and assure the public about the safety of their banking systems. 2 But other countries have resisted the need to do SO.3 It is important to recognize that careful crisis containment strategies are very difficult to devise in the midst of an actual turmoil. Political pressures to rescue powerful interests are often too difficult for the authorities to resist. Because the crisis seriously threatens the political future of the incumbent governments, the usual short-termism in policy decisionmaking is even more exaggerated. Avoiding such mistakes requires that criSis-management decisions are made in an open debate outside of an actual crisis. Accountability would be improved by requiring that regulators establish and regularly test a well-publicized benchmark plan for crisis resolution (Caprio, DemirgiiG-Kunt. and Kane 2008). What Should Be Done in the Containment Stage? Walter Bagehot's (1894) classic policy advice for managing liquidity during a sys­ temic crisis is for the central bank to lend freely to solvent banks-but in order to minimize the subsidizing of risk-taking (moral hazard), the loans are to be made at penalty interest rates and only on good collateral. Put differently; the advice is for governments to avoid lending to insolvent banks at all, even on good collat­ eral. and certainly not at below-market interest rates. Unfortunately. as the recent events illustrate. modern governments pay only lip service to this principle. For governments to embrace Bagehot's advice, they need to be able to distinguish relatively quickly between deeply insolvent banks and those that are solvent enough to be salvageable. They also need to have the strength to resist the press­ ures that a crisis usually generates to rescue powerful interests. In a banking crisis, just like at a battlefield, regulatory authorities need to run to the aid of wounded deposit-losing institutions and temporarily stabilize their condition by providing liquidity. Effective crisis containment requires effective triage: treatment-worthy institutions need to be identified and provided with enough liquidity to restore public confidence in their ability to continue in Demirgii~-Kunt and Serven 93 operation. Unless emergency response teams are assembled and trained in advance. it is difficult to conduct good triage--a point which again stresses the importance of crisis preparedness. Also very important are information problems, as the recent sub prime crisis amply illustrated. Distinguishing viable institutions at short notice becomes more challenging in environments characterized by low levels of transparency. Therefore it is the duty of regulators to identify and remedy gaps in information well in advance, and recognize the gradual reduction in transparency that comes with financial engineering and regulatory arbitrage and to nip it in the bud by demanding improvements. Regulators also need to encourage the use of instru­ ments and the development of markets that would help to yield more accurate assessment of risks, both in and out of crisis situations. 4 When transparency is allowed to deteriorate. information problems can tie the hands of authorities and limit their ability to engage in efficient containment strategies. For example, advocates of bailing out insolvent institutions to halt a systemic crisis argue that only sweeping guarantees and extensive support can stop the panicky flight of depositors and other creditors. This is of course true if the crisis entails a series of self-fulfilling runs as envisioned in Diamond and Oybvig (1983). However, most modern financial crises. including the recent one. are driven instead by fundamental weaknesses in economic balance sheets, which reveal themselves initially as liquidity problems. It must be recognized that the short-term benefits of guarantees will vary with the fiscal strength of the guaranteeing government. To hasten the end of an insol­ vency-driven banking crisis and to constrain the spread of insolvencies in the short term, the government must manifest a substantial capacity for absorbing losses. This is not a luxury most countries can afford since most governments do not have the required fiscal capacity. Depending on the depth of the systemic insolvency, such support may not even halt the spread of the crisis, but merely delay healthy adjustments. This begs the question of whether social costs and adverse distribution effects could be reduced by following an alternative strategy. Even in the midst of a financial crisis, it is inefficient to set aside long-term goals completely. The manner in which a crisis is resolved affects the frequency and depth of future crises through the significant impact it has on market disci­ pline. Providing extensive liquidity support and guarantees to insolvent institutions subsidizes their gambling on their own resurrection and distorts risk­ taking incentives, undermining market discipline and spawning future crises (Kane and Klingebiel 2004; Calomiris, Klingebiel. and Laeven 2005). If insti­ tutions can count on crisis resolution to be handled in this way then they will be more willing to risk insolvency, and safety-net subsidies will mainly flow to insti­ tutions that take excessive risks at the expense of taxpayers. The short-term benefits of such bailouts have been oversold. Such policies seldom actually speed 94 The World Bank Research Observer. vol. 25, no. 1 (February 2010) the recovery of a nation's real economy from a financial crisis or lessen the decline in aggregate output. Instead, providing liquidity support for insolvent institutions often prolongs a crisis. It does this by distorting risk-taking incentives so extensively that sound investments and healthy exits are delayed and additional output loss is generated. Honohan and Klingebiel (2003) and Claessens, Klingebiel. and Laeven (2005) measure the impact of different crisis management strategies on the ultimate cost of resolving financial distress in a broad set of countries. They find that providing generous support-in terms of open-ended liquidity support and blanket deposit guarantees-not only increases the ultimate fiscal cost of resolving crises, but also that it does not speed the recovery, instead prolonging the duration of the crisis. Using a sample of 42 banking crises Laeven and Valencia (2008) also show that blanket guarantees have little impact on domestic deposits (they pretty much continue their trend), but more often than not exacerbate the decline in foreign liabilities. Indeed, announcement of a blanket guarantee is almost taken as a signal to start a run on a currency. Not surprisingly guarantees are often accompanied by increases in liquidity support, increasing the fiscal cost signifi­ cantly. Indeed it is important to note that empirical evidence suggests that govern­ ments incur most of the fiscal costs of resolving the crisis during the containment phase. Honohan and Klingebiel (2003) show that much of the variation in the fiscal costs of crises is explained by differences in the way a government handled its liquidity crisis, with the highest costs associated with governments that pro­ vided open-ended liquidity support and blanket deposit guarantees. Alternatives to Blanket Guarantees Providing blanket guarantees poses many challenges to government authorities. The first important challenge is to convince creditors and depositors that they have the political will and fiscal capacity to afford the cost of such guarantees. This may be quite difficult for many governments of developing countries. If the emergency response is seen as inadequate, it may quickly compound the pro­ blems, requiring emergency funding from external sources, such as resource-rich governments or the International Monetary Fund. Even if the guarantees provided are deemed credible and the crisis is contained, the governments still face a second set of important challenges. These include: the need to control the additional (guaranteed) debt that insolvent institutions will continue to attract; to make sure the guaranteed institutions invest these new resources prudently; to reduce or eliminate the guarantees once the containment stage of the crisis is over. Regulators are likely to find it very difficult to address the moral hazard created by these guarantees. Because fully guaranteed insti­ tutions can attract funding independent of the risks they take, managers of Demirgii9-Kunt and Servin 95 insolvent institutions can be easily tempted to abuse their government assistance by gambling on their resurrection. There is already a large literature which establishes that in normal times overly generous safety-net policies and deposit insurance will lead to moral hazard and financial instability. 5 The longer the guarantee remains in place. and insolvent banks are allowed to operate. the more difficult it will be to curb these excessive risk-taking incentives. Furthermore. once installed. such guarantees are difficult to claw back and. more importantly. they seriously undermine the credibility with which future safety net arrangements can be limited. While introduction of blanket guarantees may be tempting for policymakers in the short term, they are not inevitable. Provision of such extensive guarantees represents an extreme measure that is best resisted before other alternatives are exhausted. Ideally, policymakers must be ready to take the time to separate hope­ lessly insolvent institutions from potentially viable ones, and to provide liquidity support. guarantees. or "haircuts" in a way that would protect taxpayer interests. Hence an alternative strategy to an indiscriminate blanket guarantee is to take a "banking holiday" over several days in order to identify insolvent institutions and to recommend and impose preliminary haircuts on uninsured depositors and non­ deposit creditors before they can liquidate their claims (Kane and Klingebiel 2004). Using the holiday to prepare a program of limited guarantees and to write down the uninsured debt can restore public confidence both in the government's ability to deal with the crisis and in the banking system itself. Baer and Klingebiel (1995) examine the aftermath of pre-1992 systemic crises in a number of countries and find that in cases where the governments assigned losses to depositors of insolvent banks. the positive effects of reducing depositor uncertainty quickly overcame the negative effects that surviving banks experi­ enced from depositor write-downs. Fairness requires that small depositors-who are often more than covered by explicit deposit insurance schemes-have immedi­ ate access to their funds. By the same token, at the end of the holiday. larger uninsured depositors should also be allowed immediate fractional access to their transaction balances. Clearly the speed with which the authorities can deal and resolve these situations depends on the extent to which they have engaged in con­ tingency-planning and crisis-management simulations. A holiday that lasts for weeks or months is called a "deposit freeze." and this reduces depositors' liquidity and the nation's aggregate money supply; it may also have long-term adverse effects on depositor confidence. 6 Hence to minimize these adverse effects, insured depositors should be granted access to their funds as soon as this is feasible. It is also important to note that it is not necessary for banks to close for triage to begin. In a systemic crisis. bank holidays may be prolonged because it may be difficult to complete the investigation of all banks quickly. Hence. banks may continue to operate during inspections, though the deposit 96 The World Bank Research Observer, vol. 25, no. 1 (February 2010) insurer may be given a right to void large loans and withdrawals made from a bank within a certain period of its closure. This would make large transactions subject to clawback. but it would not interfere with the business of ordinary households while the inspections are going on. Broader time-out strategies for creditors that follow bankruptcy proceedings can also be employed. Productive assets can be conserved by instituting a grace period during which major creditors do not receive payments of principal or inter­ est due on existing bond or loan contracts. but use the time to work out a replace­ ment contract structure with the help of courts. mediators. or both. Forcing private creditors to renegotiate unenforceable contracts is called "bailing-in." and. like haircuts imposed on uninsured depositors. is intended to trap creditors that financed weak institutions into participating more fully in loss-sharing. Whichever strategy is used. ultimately the damage the crisis causes to the country's financial sector and its real economy is reduced by separating the insol­ vent from the viable institutions as quickly as possible and by providing support and allocating losses in ways that protect taxpayers and avoid subsidizing the go-for-broke strategies of insolvent institutions. Prior crisis planning and com­ mitment to these plans are important steps in being able to retain a long-term perspective in the containment stage of any crisis. The Role of the State in the Financial System After the panic abates. confidence is restored. and markets start functioning again. the crisis moves into its resolution stage. With the crisis contained. policies must now be chosen to deal with the undercapitalization and insolvencies that are often revealed. Past research and experience provide many valuable lessons on how best to deal with systemic insolvencies, an issue we discuss below. Given the intensity of the recent crisis. direct interventions in the financial system have been so massive that by the end of 2008 governments will be the largest shareholders in most developed economies' financial industries. reversing a trend of state retreat over the last 20 years. 7 With $500 billion or more invested. this is equivalent to roughly state ownership of a quarter of the indus­ try's market value, which strongly contrasts with the ideology of Western capital­ ist systems. Indeed. the extent of these interventions and the increasing government stake in financial institutions have led to questions as to whether financial systems should be privately owned in the first place. shaking the com­ mitment of developing country policymakers to ongoing bank privatization pro­ grams. If private banks are prone to excessive risk-taking leading to crises and costly bailouts, is it not better to have the governments own and operate the financial system in the first place? Demirgur-Kunt and Serven 97 In answering this question it is important to recognize that bank nationaliza­ tions are very common ways of dealing with systemic financial crises. Indeed. gov­ ernments have always taken ownership positions in banking either deliberately or indirectly as a result of banking crises. In almost every major banking crisis in recent history-in East Asia. Latin America. and other countries-governments have become temporary caretakers of financial institutions. 8 Hence. the recent crisis is not the exception in this regard. But the way in which this is done has important implications for maintaining and restoring a functioning financial system and minimizing the short term and long term costs of such interventions. 9 But the fact that governments find themselves involved in resolution of insolven­ cies does not imply this role should be permanent. On the contrary, all evidence suggests that they would be well-advised to do otherwise. Bureaucrats as Bankers? Government ownership of banking has been popular throughout history,lo Early proponents of state control argued that the government can better allocate capital to highly productive investments. Gerschenkron (1962) was among the first to argue that private banks would not be able to overcome deficiencies in infor­ mation and contracting in weak institutional environments. State ownership also makes appropriation of the surplus from finance and directing credit much easier. making it attractive for policymakers. Moreover. there is the concern that private ownership and concentration in banking may lead to limited access to credit by different parts of society, so hampering the development process. Indeed. govern­ ment banks are often expected to expand access, making financial services more broadly available. A final argument is rooted in the fragility of finance. and that private financial institutions are too prone to excessive risk-taking, which is diffi­ cult to keep under check: a popular sentiment that resurfaces with every signifi­ cant crisis. By now there is a significant amount of empirical research that suggests state ownership of banks is associated with less financial sector development. lower growth and lower productivity, and that these negative effects are more pro­ nounced at lower levels of income with less financial sector development and with weaker property rights' protection (Barth. Caprio. and Levine 2001; La Porta, Lopez-de-Silanes, and Shleifer 2002). Despite explicit mandates for govern­ ment banks to expand outreach, in banking systems dominated by state banks there are fewer bank branches and automated teller machines. Customers in such systems face lower fees but they also experience poorer service quality (Beck, Demirgii<;-Kunt, and Martinez Peria 2007. 2008). There are some exceptions in that certain state-owned banks. designated as development finance institutions. have become effective providers of know-how and have had a useful catalytic 98 The World Bank Research Observer. 1'01. 25. no. 1 (February 2010) function in "kick-starting" certain financial services, for example in Latin America (De la Torre, Gozzi, and Schmukler 2007). But even then the most suc­ cessful of these initiatives are expected to be privatized. Instead, the bulk of the evidence suggests that state-owned banks tend to lend to cronies, especially around the time of elections, as vividly illustrated by the recent empirical studies of Cole (2004). DinG (2005), and Khwaja and Mian (2005). For example, Cole has shown that state banks in Indian states ramped up agricultural lending in tightly contested districts in election years. DinG showed that increased lending by government-owned banks right before elections is not specific to India but can be observed in data from 22 developing countries. Drilling down to the individual loan level. Khwaja and Mian found evidence from Pakistan that politically active borrowers were able to secure larger and cheaper loans from state-owned banks and that they defaulted on these loans much more than other business borrowers. Given the extent to which lending policies are politicized. it is not surprising that state ownership appears to heighten the risk of crises instead of reducing it. If anything, research suggests that greater state ownership is associated with various measures of financial instability, including a greater probability of banking crises (Caprio and Martinez Peria 2000; Barth, Caprio, and Levine 2001; La Porta, Lopez-de-Silanes, and Shleifer 2002). These results and other related evidence explain why many countries embarked on privatization programs, selling their state banks. Indeed, evidence also suggests bank privatization, if well-designed, can significantly increase bank performance (Clarke, Cull. and Shirley 2005; Megginson 200S). To be sure privatization is diffi­ cult and can lead to problems in weaker institutional environments. Hence the sequencing is important: moving slowly but deliberately with bank privatization, while preparing state banks for sale and addressing the weaknesses in the overall incentive environment and regulations, seems to be the preferred strategy. Ultimately, gains from privatization-if designed properly-can be substantial since the alternative of maintaining large state ownership can significantly under­ mine real sector reforms and deter economic development. 11 So what explains the poor performance of bureaucrats as bankers? As often the case in finance, incentive problems are at the root of this issue since bureaucrats do not face incentives designed to reward efficient resource allocation. Not only do government officials often lack the expertise to be effective managers, they also face conflicts of interest due to their desire to secure their political base and reward supporters, which often goes against efficient resource allocation. These problems become worse with fewer checks and balances and in poorer insti­ tutional environments, explaining why state ownership is more damaging at lower per capita income levels ( Keefer, 2000; La Porta, Lopez-de-Silanes, and Shleifer 20(2). Demirgiir-Kunt and Servin 99 Hence while governments may inevitably find themselves as stakeholders in financial institutions as the outcome of systemic crises, they would do well to see this as a temporary arrangement and plan for their exit as part of the crisis resol­ ution exercise. Despite its weaknesses, a well-functioning private financial system is crucial for promoting development, and substituting government provision of financial services for that of the market is likely to lead to inferior outcomes. Economic growth does not resume on a sustainable basis until productive assets and banks are back in the hands of well-capitalized private parties. But how should financial crises be resolved and solvency of institutions restored? In other words, how should bail-out programs be designed in the event of crises, and what should be the government's role in this process? Bureaucrats as Caretakers If bureaucrats are not good bankers in good times, they are not likely to do better in bad times, given that the tendency for political forces to dominate economic judgments will be even stronger. As we have seen in the latest crisis, systemic crises often involve the injection of substantial sums and the determining of the financial fate of powerful interests. But systemic crises often require comprehen­ sive solutions by the government. So how should governments behave? While governments should be prepared to act in a systemic crisis, the approach and the actions they take still need to be designed to reduce conditions for moral hazard and the likelihood of a subsequent crisis. This should be done by imposing real costs on all responsible parties and getting the resources back in productive use as soon as possible (Demirgii<;-Kunt, 2008). To see the importance of paying attention to incentives. consider what happens to firms outside the financial sector that fall into a state of insolvency. Those in control of the firm find it difficult or impossible to raise new funds in any form. They can no longer act on profitable investment opportunities and may be forced to sell important assets. Creditors recognize this situation may distort the incen­ tives of managers, making them more susceptible to fraud and moral hazard. which at the very least reduces the incentives of the owner and managers to exert effort. In a market economy, the solution is bankruptcy. As long as the firm is economically viable. it makes sense to continue its operations, but only after restructuring. This restructuring generally wipes out existing equity holders, while debt holders often have a portion of their claims converted to equity. Alternatively debt holders can agree to cut down the face value of their debt in exchange for some warrants. Old management is often replaced. a substantial portion of the firm's assets are sold, and workers are laid off. This restructuring is not a mere reworking of the firm's balance sheet. but represents very real changes to the way it does business, perhaps even in the business it does. 100 The World Bank Research Observer. vol. 25. no. 1 (February 2010) The happy result for the economy is that resources continue in their best use, while all responsible parties incur some costs for the firm's poor performance. To a large extent dealing with financial insolvencies should follow the same general principles. Admittedly the financial sector is special due to its particular fragility, the possibility of contagion, and the major macro-implications when it is in systemic distress. However, although these differences may justify different approaches, they do not suggest that incentives matter any less. Any government involvement should be designed to protect the interests of the taxpayers, impose losses on the responsible parties, and use the private sector to pick the winners and losers. For example any plan that purchases bad assets from troubled finan­ cial institutions, or recapitalizes without extracting some claim from the insti­ tutions, amounts to a transfer from taxpayers to shareholders, which is the group that keeps the residual value of the entity. Recapitalization of the banking system should be designed so that those banks that need assistance with recapitalization are helped in an incentive-compatible way. Such a plan would limit taxpayers' loss exposure, and, in environments with good information and contract enforcement. leave resolution of bad assets to the banks themselves. But how should the banks be recapitalized? Just as bureaucrats are likely to make poor bankers, the selection of individual winners and losers is also what markets, not governments, do best. One way is for the authorities to inject funding only to those institutions that are able to meet the following criteria (World Bank 2001; Honohan 2005): • Those institutions that are in a strong position should be able to raise capital privately, say from a syndicate of private banks. If the institution has difficulty raising capital, it should at least be able to obtain some proportion of it-at least half of it or more--from the private sector before applying for govern­ ment recapitalization assistance. Only those institutions that can secure private sector funds would be eligible for the plan. This will ensure that the private sector plays an important role in picking the survivors.12 • Those institutions that are eligible receive government assistance in the form of preferred stock. Preferred stock status would force private parties rather than taxpayers to bear the first-tier losses. • Any bank participating in the plan will have to suspend all dividend pay­ ments (and restrict the amount and structure of its compensation plans for its senior managers) until the government is fully "bought out." The bank will also agree to comply with strict regulations on its leverage, risk-taking, transparency, and disclosure in order to participate in the plan. This will give incentives to banks to retire their preferred stock as soon as possible. These are tough criteria and only desperate banks will agree to these terms. Which is exactly the point: government assistance should only be injected into Demirgiir-Kum and Servin 101 banks in dire straits, yet simultaneously to those with a real chance of survival. As long as the amount of funding is such that some banks fail, this approach removes government from decisions as to which banks survive. The availability of private sector funding serves to identify the candidates, and restrictions on differ­ ent ways to take these funds out of the banks, combined with greater transpar­ ency, makes it more likely that the banks will not be looted or engage in gambling with taxpayers' money. By openly stating the terms on which it will assist banks, and ensuring that those terms provide good incentives for the restructured bank going forward, the government will have made the best use of market forces while minimizing its direct ownership involvement. Many of these features characterized the u.s. Reconstruction Finance Corporation's (RFC) program of taking temporary preferred equity positions in banks after the great depression. 13 Another alternative that has gained popularity after its successful application in Spain in the early 1980 s is a centralized approach where banks' non perform­ ing loans are carved off into an asset management company (AMC) which con­ tinues the restructuring process. The carving-out of an insolvent bank's bad loan portfolio and its organizational restructuring under new management and owner­ ship may be appropriate when large parts of the bank's information capital is dys­ functionaL The bad loan portfolio may be sold back into the market or disposed of by a government-owned AMC. A survey by Klingebiel (2000) shows, however, that the AMCs formed in devel­ oping countries have not been as successful in restructuring nonperforming loans. The effectiveness of AMCs has been quite mixed, better when the assets to be disposed are primarily real estate, less good when they are loans to large politi­ cally connected firms. In countries with developed private markets and insti­ tutions, individual banks-if well capitalized-would be in a superior position to engage in restructuring their own assets. But policymakers in weak institutions should not expect to achieve the same level of success in restructuring as those in more developed economies, and they would do well to design simple resolution mechanisms that offer little discretion to government officials (Honohan and Laeven 2005). In summary. governments are not good at providing financial services in normal times or in crises. and those governments that find themselves as bankers as a result of such crises should focus on their exit strategy as quickly as possible, using the market to identify winners and losers. And although drawing on public funds to recapitalize some banks may be unavoidable in systemic crises, they must be used sparingly for leveraging private funds and incentives (World Bank 2001). Hence, although finance is prone to excesses and crises, there is a substantial lit­ erature that suggests government ownership is not the answer. Where govern­ ments have a very important role to play in finance, it is in providing the 102 The World Bank Research Observer, vol. 25. no. 1 (February 2010) regulatory and supervisory arrangements that help reinforce incentives that limit this excessive risk-taking and fraudulent behavior. Design of Prudential Regulations One important implication of the recent crisis is the widespread calls for reforms of regulation and supervision. The initial reaction to the emerging crisis was one of disbelief: how could the crisis emerge in countries whose supervision of credit risk had been thought to be the best in the world? Indeed. the regulatory stan­ dards and protocols of these countries were in the process of being emulated worldwide through the international capital accords, known as the Basel standards. Basel II-which is currently in its implementation stage--grew out of concern that the Basel I accord was unable to address the range of risks in bank activities. as evidenced by the growth of securitization. 14 Basel II is built on three pillars: (1) minimum regulatory capital requirements for credit risk. operational risk, and market risk; (2) the supervisory review process: and (3) market discipline and dis­ closure. The minimum capital requirements are determined by either external ratings from ratings agencies for smaller banks or by outputs from the larger banks' own internal ratings models. 15 Many interpreted the crisis as a vivid example of market failure, evidence that there is no such thing as market discipline, reinforcing calls for stronger regu­ lations through improvements in the Basel II accord. But the crisis also spawned a growing argument about the role that the Basel I accord may have played in causing the crisis. Indeed, it is no secret that Basel I contributed to the growth in securitization by assigning lower capital charges and thus giving incentives to institutions to move their assets into off-balance-sheet securitization vehicles. While advocates claimed that Basel II. had it been implemented earlier, could have lessened or prevented the turmoil. critics of the Basel approach to capital regulation pointed out that the crisis has simply reconfirmed fundamental flaws that have been evident in this approach. 16 The financial turmoil challenged the Basel II framework in important ways. First. the events raised serious questions about setting capital requirements based on external ratings. These ratings proved excessively optimistic and confirmed long-standing concerns about the conflicts of interest that arise out of having the issuers pay the agencies for ratings required by regulators. More importantly. credit ratings are not appropriate for setting capital requirements. Ratings are based on expected default rates. yet capital is intended for unexpected losses. Ratings can be useful for establishing loss reserves for particular assets, but they do not consider correlations among assets. hence they are not helpful in assessing Demirgu(:-Kuflt and Serwin 103 how a bank's net worth or its portfolio of assets may vary. Therefore the amount of capital required for an institution's safety has to be linked explicitly to measures of volatility of its earnings. which is not information that ratings provide. Second, the crisis raised serious concerns about the accuracy of internal risk models employed by even the largest and most sophisticated market participants. These models proved inadequate and illustrated how financial models and data­ sets can be manipulated to provide desired outcomes. Third, the way the crisis emerged in subprime lending but spread to other secu­ rities revealed problems with inadequate documentation, the disconnect between the source of risks and the bearers of those risks. and lack of knowledge about how risks are ultimately distributed. I 7 All these point to weaknesses in the dis­ closure provisions included in the market discipline pillar of the Basel accord, and how fundamental issues of transparency were not addressed. Given the intensity and persistence of the crisis, many proposals for regulatory reforms are emerging. Some proposals originate from sources that interpret recent events as evidence that market discipline is not reliable and want to focus on designing tougher and more comprehensive regulations. These proposals range from re-establishing extensive activity restrictions, to speedier implementation of Basel II, to revising the accord in order to address the weaknesses discussed above. to enhanced opportunities for further official intervention. IS Others continue to believe in the reliability of market signals and market disci­ pline, and argue that the Basel approach is fundamentally flawed. On the hypoth­ esis that investors would be quicker to recognize changes in risk and risk premiums, such sources propose an alternative approach where official supervi­ sion would focus on generating and using better market signals. The main features of this approach can be summarized as follows. I 9 While the Basel Committee has been responsive to its critics by trying to make the capital requirements more reflective of the ways in which risks and vulnerabilities are assessed, this has led to greater and greater complexity. Increased complexity in bank regulation reduces transparency but increases the scope for regulatory arbit­ rage and forbearance, without necessarily increasing accuracy. Indeed, these critics argue that while the task of computing correct economic capital for banks is very difficult and complex, bank capital regulation need not be. An alternative is to rely in large part on the market itself to provide a measurement of risk, together with the enforcement of simple rules such as the leverage ratio and prompt corrective action, which are not subject to manipulation. Supervisors must not only draw on-but also help develop-informative market signals such as those imbedded in the prices of credit default swaps and subordinated debt. By requiring large institutions to engage in these markets, regulators would be able to incorporate market-generated information to their risk assessments. Hence this alternative combines supervisory and market oversight. helping to generate 104 The World Bank Research Observer. vol. 25, no. 1 (February 2010) and harness the information markets are capable of producing. The job of the supervisors would be significantly easier if this could be done effectively. However. as Caprio. Demirgu<;-Kunt, and Kane (2008) emphasize, prices in credit default swap or subordinated debt markets can be a strong source of disci­ pline only under two conditions: (1) market participants do not expect to be bailed out when trouble develops, and (2) investors have access to regular flows of high-quality information. This underlines the importance of establishing incen­ tives that would lead both supervisory authorities and market forces to operate more effectively. Caprio, Demirgu<;-Kunt, and Kane argue that this crisis exemplifies not just the limits of market discipline, but the power of government-induced incentive distor­ tions-and the limits of official supervision as commonly practiced. The failure of private parties to exercise sufficient due diligence was rooted in the failure of gov­ ernment supervisors to challenge decisions made by private accountants and credit-rating organizations. Authorities neglected their duty of examining and publicizing the implications that these decisions might have for safety-net loss exposure. By tolerating a decline in transparency. supervisors made it difficult to recognize and price the risk expansion, not only for themselves but also for the market participants. 20 Hence, Caprio, Demirgu<;-Kunt. and Kane propose reforms that improve the chain of incentives under which market discipline and official supervision operate. which include reforms for lenders, credit rating organizations, securitiza­ tion. accounting. and government officials. Perhaps most important among these are their proposals for enhancing accountability of government officials through better crisis preparedness. greater use of market information to track risks and subsidies, publicizing estimates of safety-net subsidies, and deferred compensation schemes. First, as already discussed above, crisis preparedness is important to avoid short-termism in crisis-management. Accountability would be greatly improved by requiring that regulators establish and regularly test a well-publiCized bench­ mark plan for crisis resolution. Second, regulators need to draw on market signals to overcome information problems and improve their ability to track risk in and out of crises. Requiring the largest banks to issue at regular intervals a series of credit default swaps or unin­ sured subordinated debt would provide such signals. since the holders of these instruments would apply the market discipline that pillar three of Basel II seeks to harness. 21 Third, the safety net needs to be strengthened by making authorities more accountable for its cost. This requires the development of a system of fair-value accounting for intangible safety-net subsidies to establish political accountability for controlling them. Important institutions and their supervisors must be Demirgu,-Kllnt and Serven 105 required to model, estimate, and expose to outside review the value of this intan­ gible source of income-both in individual institutions and in the aggregate. Fourth, the decisionmaking horizons of government officials can be lengthened if employment contracts included a fund of deferred compensation that the heads of supervisory authorities would have to forfeit if a crisis occurred within a couple of years after leaving their office. Calomiris and Kahn (1996) show that such a system worked well in the 19th-century Suffolk banking system, where claims to deferred bonuses were paid only after losses were deducted. While the discussion of how best to reform regulation and supervision of finan­ cial institutions is likely to continue, it is important to keep in mind that ulti­ mately the goal of financial regulation and supervision is not to reduce financial institution risk-taking, but to manage the safety net so that private risk-taking is neither taxed nor subsidized. This goal implies that supervisors have a duty to see that risks can be fully understood and fairly priced by investors. No one should expect that, in a risky world, risk-neutral regulation and supervision can elimin­ ate the risk of financial crises; what it can do is to reduce their frequency and cost. Monetary Policy, Asset Prices and Macroprudential Regulation Most observers agree that lax monetary policy in the United States in the early 2000s helped fuel the housing bubble, at least in its initial stages. In light of the devastating effects of its bursting, one major policy question looking forward is the proper role of monetary policy. Should central banks respond systematically only to inflation in goods prices, as they do at present. or should they also respond systematically to inflation in assets prices? To put it more bluntly. should monetary authorities attempt to "prick bubbles" through monetary tightening? Not all bubbles are alike. Some create risks to the financial system, while others do not. 22 Bubbles that threaten financial stability typically involve a feedback loop between asset prices and credit conditions: a credit expansion raises asset prices. which in turn encourages further lending and hence further asset price rises, and so on. When the bubble bursts. asset prices collapse and the loop goes in reverse, eroding the balance sheets of financial institutions, causing a credit crunch and a fall in economic activity. But not all bubbles have these features-for example the dot-com bubble of the late 1990s in the United States was not associated with a credit boom. and its crash did not weaken lenders' balance sheets to a significant extent. Hence the pursuit of financial stability poses a greater need for the monet­ ary authority to react to some bubbles than to others. Attempts by the authorities to deflate an asset price bubble-through monet­ ary policy or other means-in the absence of obvious symptoms of inflationary or 106 The World Bank Research Observer. vol. 25, no. 1 (Z,ebruary 2010) financial distress face serious political-economy obstacles. Lenders and borrowers riding the bubble euphoria are likely to condemn such seemingly unwarranted courses of action, and politicians may oppose it under the view that, rather than a bubble, rising asset prices actually reflect the beneficial effects of improved pol­ icies. 23 Deflating a bubble amounts to suppressing an event, and it is impossible to prove ex post that the event would have occurred in the absence of policy action (Wyplosz 2009). Political economy concerns aside, the monetary policy response to bubbles has been debated mainly in industrial countries in the context of inflation targeting regimes, under which monetary authorities react solely to inflation forecasts and the output gap (the "Taylor rule"). But the relevance of the question is much broader. It is of special significance to a growing number of emerging economies that have in recent years adopted inflation targets to guide monetary policy. More generally it matters to all countries whose monetary authorities are entrusted with inflation control and macroeconomic stability. The conventional view of inflation targeting assigns no role to asset prices in the conduct of monetary policy, except to the extent that changes in asset prices signal changes in expected inflation (Bernanke and Gertler 2(00). In this view, the authorities should deal with asset price bubbles only as their consequences, if any. arise in terms of the inflation objective-for example by supplying the needed liquidity in the event of a bubble burst. In other words, the authorities should not attempt to "prick" bubbles.24 An alternative view advocates a more proactive response of monetary policy to asset prices. Of course. asset prices may rise or fall for many reasons, including changes in fundamentals-that is, reassessments of the anticipated future pro­ ductivity of the assets-so in principle monetary policy should react to deviations of asset prices from their underlying fundamentals. that is, to asset bubbles, rather than to deviations from any particular target level (Cecchetti, Genberg. and Wadhwani 20(3). Formally. while the monetary authorities' objective func­ tion would continue to be defined only in terms of goods inflation, their reaction function should include not only inflation forecasts and the output gap, but also measures of asset price misalignment. In theory, adding more arguments to the monetary policy reaction function should allow the authorities to do no worse, and possibly to do better, than in the standard framework. But the approach remains untested, and several objections have been raised against its practical feasibility. First, asset price bubbles are not easy to spot in real time-at least at their early stages when policy action might be most useful to stop them. Identifying a misalignment of asset prices requires a reliable assessment of their fundamentals. But available models of asset price fundamentals are inherently imperfect, and in addition some fundamentals--for example investors' perceptions of risk and Demirgii,-Krmt and Serven 107 future asset productivity-are not readily observable. A simpler alternative would be to use arbitrary rules of thumb, such as a threshold above which asset price increases would be deemed to involve bubbles. However, simple rules of thumb would likely lead to frequent identification errors, especially in narrow and illiquid asset markets (such as those of most emerging countries). characterized by high asset price volatility.25 And reacting to a misidentified bubble may be very costly-for example tightening in response to an asset price rise that in reality is driven by improving fundamentals will both hamper growth and interfere with the role of asset prices in allocating resources. Of course, one possible solution to this problem is to react only to bubbles once they become self-evident, but by then it may be too late for monetary policy to mitigate their effects. Indeed. if the authorities tighten when the bubble is already nearing collapse on its own, the contractionary effects of the tightening could compound. rather than mitigate, those of the bubble collapse. Yet while this view that bubbles are hard to spot ex ante is certainly correct, one could argue that other standard tasks of monetary policy. such as forecasting output and inflation over a multiyear horizon. are not less difficult. Second. what asset prices should the monetary authority watch? Not all asset prices are synchronized, and at anyone time bubbles may be present in just a few of the many assets available in the economy. Some observers suggest that the price of housing may be the best candidate for close monitoring, given that housing values have major wealth effects on spending-and hence are more rel­ evant than other asset prices for inflation and the output gap-and that housing cycles of boom and bust are more frequent than equity market cycles. 26 A broader view holds that the monetary authority should focus on the prices of assets held by highly leveraged financial intermediaries. given their key role in the boom-bust credit cycle and the propagation of financial turmoil in the current crisis (Adrian and Shin 2009). Third. little is known about the timing and magnitude of the effects of monetary policy on asset prices. and hence its ability to prick bubbles. For example, some research suggests that very large interest rate hikes may be necessary to bring housing prices down to any noticeable extent-so large in fact as to result in huge output losses (Assenmacher-Wesche and Gerlach 2008). But other recent research indicates that monetary policy has a sizable effect on the acquisition of assets by leveraged intermediaries (Adrian and Shin 2009) and on bank lending standards (Maddaloni. Peydro, and Scopei 2008). The implication is that a timely monetary tightening in the boom might have been effective at containing the cyclical expan­ sion of leverage, credit, and asset prices. and prevented the deterioration of lending standards, two key ingredients in the gestation of the current crisis. Overall. the usefulness of monetary policy to prick bubbles remains controversial. One additional concern is that gearing monetary policy to deflate hard-to-identify 108 The World Bank Research Observer. vol. 25. /l0. 1 (February 2010) bubbles may detract from its transparency and predictability. especially in emerging markets that are still at the early stage of establishing the credibility of their inflation targeting regimes or. more generally. their commitment to price stability. To approach the problem from a different angle. it is important to recall that monetary authorities in most countries face two different objectives: price stability and financial stability. With monetary policy as their only instrument. the auth­ orities can find themselves in situations where the objective of price stability requires a policy change in one direction, while that of financial stability points in the opposite direction. Ideally a second instrument should be devoted to financial stability. The best option is a prudential regime capable of dampening financial cycles of boom and bust by preventing feedback loops between asset prices and credit supply. This is what has been termed "macroprudential" regulation. Its thrust is to complement regulators' traditional focus on the risk management of the individual financial institution with a focus on the risk management of the financial systenl as a whole. Macroprudential regulation aims at dealing with increases in systemic vulnerabilities due to 0) periodic business downturns that effect all financial institutions; (ii) increases in the number of financial institutions that have become too large. too interlinked. and therefore too difficult to fail and unwind. The first is an age-old problem. which is exacerbated by the amplitude of the business cycle. as discussed above. The second has become increasingly more important as safety-net subsidies provided to large. interconnected firms have increased over time. giving them more incentives to become even larger and more interlinked. While monetary authorities possess a variety of tools to pick up the pieces after a financial crash, they lack the regulatory instruments to contain the lending boom that usually precedes it. However. provisions. leverage ratios. loan-to-value ratios, and additional capital buffers can all be designed to be countercyclical, that is. to move inversely with the business cycle in order to make financial inter­ mediaries hold more liquid assets in good times so that they can be run down in bad times (Goodhart 2008a, 2008b; Goodhart and Persuad 20(8). The idea is to switch the basis of capital adequacy requirements from levels of risk-weighted assets to their rates of growth-hence requiring additional capital and liquidity when bank lending and asset prices are rising fast-and relaxing such require­ ments in the downturns. This can be seen as an alternative (or a complement) to monetary policies to prevent the growth of asset bubbles (and their busts). To date, however, this kind of instrument has seen little use. except for Spain's coun­ tercyclical "dynamic provisioning," as well as the introduction of time-varying, loan-to-value ratios in a few small economies. Aside from practical questions-for example. over what periods applicable credit growth rates should be calculated-these proposals also raise other poten­ tial difficulties that require deeper consideration. 27 As prudential requirements Demirgii(;-Kunt and Serven 109 are tightened in the upswing, they will encourage disintermediation to less-regu­ lated entities (or countries), and this may weaken the effectiveness of the regu­ lations. In addition, the cost of intermediation will rise in the boom, possibly constraining growth and financial innovation. Lastly. macro prudential regulation adds to the informational burden on the regulators and the complexity of their task. As with monetary policy. significant political will is required to enable the authorities to tighten regulation in the upswing. Although reliance on well­ defined rules (rather than reliance on discretionary regulatory changes) and inde­ pendence from government may be helpful in this regard. it is still questionable whether without appropriate incentive reforms this will be possible. Finally. to deal with too large or too interconnected firms. reform proposals aim to increase supervisory scrutiny (potentially through a college of supervi­ sors) or additional capital charges imposed on these institutions. However, proper identification and increased regulation of these institutions require cred­ ible estimates of the safety-net subsidies that are provided to them. Estimating and publicizing safety-net subsidies as advocated by Caprio, Demirgi.i<;-Kunt. and Kane (2008). as discussed in the previous section, would be one step in this direction. These and similar ideas underlie an active debate on macroprudential regu­ lation that has already generated a number of reform proposals (Kashyap. Rajan. and Stein 2008; Acharya and Richardson 2009; Brunnermeier and others 2009). However. their specifics differ. and a consensus on the best way to go has yet to be achieved. In the meantime. what can monetary policy do? An emerging view holds that. even if bubbles cannot be accurately identified. monetary policy can contribute to financial stability by becoming more "symmetric" over the financial cycle-that is. becoming more restrictive during a financial market boom. just like it almost invariably becomes accommodative at times of asset price crashes (Papademos 2(09). This view is supported by theoretical models in which monetary policy reacts to credit growth or other indicators of the financial cycle to dampen boom-bust episodes (Bordo and Jeanne 2002a. 2002b; Christiano and others 20(8). In light also of the recent empirical findings cited earlier, that monetary policy affects significantly the volume and the quality of the assets held by lever­ aged financial institutions, the implication is that there might be a role for balance sheet aggregates in the determination of monetary policy. But that role would be due to reasons of financial stability, rather than the classic reasons of price stability behind the once-popular targeting of monetary aggregates. (Adrian and Shin 2009) However, at this stage little is known about the complexities of implementing such policy or about its likely effects. for the obvious reason that it has never been applied. These questions are currently the subject of active research. 110 Tile World Bank Research Observer. vol. 25, no. 1 (February 2(10) Capital Controls The financial turmoil of the current crisis propagated rapidly to emerging markets around the world. which suffered. to varying degrees. sharp increases in external borrowing premiums and abrupt declines of capital inflows. as well as large falls in their currencies in the face of a flight to safety by international investors. Thus. even though the crisis originated in the North. the financial symptoms in emer­ ging markets have been similar-albeit in general less acute-to those of the homegrown crises of the 1990s. Propagation of this financial shock-which adds to the real shock accruing through the global slowdown-has been facilitated by the deepening of direct and indirect financial links across economies brought about by international financial integration. 28 As in previous episodes of global turmoil, the sudden stop in capital flows has put emerging markets under stress and has hit especially hard countries that were running large current account deficits, had developed large currency and maturity mismatches. or both-as was the case in several Eastern European econ­ omies. In this context. some observers have advocated controls on capital outflows to stem the creditor run and relieve the pressure on foreign reserves and exchange rates?9 Indeed. rigorous exchange controls were a key feature of Iceland's 2008 emergency package (supported by the IMP), amid the collapse of its currency and banking system. Among emerging markets. only a few have resorted to controls on outflows so far. Ckraine. Russia. and Belarus introduced some restrictions on outflows. while Ecuador imposed an exit tax. 3D Controls on capital outflows. which date back to Germany in the 1930s, have a long tradition as a crisis containment tool. They seek to prevent the disorderly retreat of investors at times of turmoil 31 and thereby protect the stock of reserves of the Central Bank, relieve pressure on the exchange rate, and provide some room for monetary policy-which is severely constrained when capital mobility is high and policymakers also try to pursue exchange rate targets. There have been numerous episodes of outflow controls in crisis times-for example Spain. in the context of the 1992 ERM crisis; Venezuela. at the time of the banking crisis of 1994; Malaysia and Thailand, on occasion of the 1997-98 East Asian and Russian crises; and Argentina, at the time of collapse of the Convertibility Plan in 2001. These controls took a wide variety of forms. They frequently targeted "speculat­ ive" transactions and exempted current account transactions, flows related to foreign direct investment, or both. They ranged from unremunerated deposit requirements on banks' foreign asset holdings (Spain) to comprehensive prohibi­ tion of outflows (Venezuela and Argentina). But how effective were they? The question has been the focus of a multitude of studies: see for example the over­ views by Ariyoshi and others (2000) and Magud, Reinhart. and Rogoff (2007). Demirgiic-Kunt llnd Server! 111 The case of Malaysia is perhaps the one that has attracted the most attention. Observers agree that the controls succeeded in limiting outflows and segmenting onshore and offshore markets, providing some breathing space for monetary and financial policy. There is much less agreement regarding the effective contribution of the controls for easing the cost of the crisis. as some observers contend that the worst of the global turmoil was already over at the time the controls were insti­ tuted and that Malaysia's subsequent recovery path was no better than that of countries that did not introduce controls, while others argue that the controls did allow a speedier recovery than would have otherwise occurred given the higher vulnerability of the Malaysian economy relative to that of other East Asian countries. 32 Analyses of other episodes of controls on outflows yield mixed conclusions. In Spain and Thailand. the controls helped contain outflows and relieve pressure on the exchange rate only temporarily. and eventually both countries had to abandon their pegs-Spain through an ERM realignment, and Thailand by float­ ing the exchange rate. In Venezuela, the controls did not fully succeed in contain­ ing outflows either, but there is evidence that they helped to ease pressure on the exchange rate and gain some degree of monetary policy autonomy which. through lower interest rates, reduced the immediate fiscal cost of the country's banking crisis. A concrete way to assess the action of controls is by examining the differential between the prices of identical assets traded in onshore and offshore markets. Among such assets are the American Depositary Receipts (ADRs) issued abroad by an increasing number of large emerging market firms. The onshore-offshore price differentials on ADRs have been recently examined by Levy-Yeyati, Schmukler, and van Horen (2009) for several episodes of inflow and outflow con­ trols. The finding is that controls on outflows generate a positive differential between onshore and offshore prices as investors buy equity at home and sell it abroad in order to transfer their wealth out of the country (inflow controls have the opposite effect).B This suggests that even if capital controls fail to bring out­ flows to a halt, they usually do succeed in temporarily segmenting asset markets, by creating "no-arbitrage" bands that effectively decouple rates of return at home from those prevailing abroad. However, there is also evidence that this market segmentation weakens with the passage of time (Kaminsky and Schmukler 2001). The reason is that capital controls develop leakages as investors find arbitrage strategies and loopholes to circumvent them. 34 The search for loopholes at times of crisis is driven by inves­ tors' anticipations of big return differentials between domestic and foreign assets, especially in the event of a large exchange rate depreciation. These anticipated differentials often dwarf the increase in the cost of taking capital out imposed by the controls. Evasion mechanisms for controls on outflows range from the 112 The World Bank Research Observer. vol. 25, no. 1 (February 2010) traditional overinvoicing of imports and underinvoicing of exports to misreporting of capital account transactions in order to make them fit under those categories of flows left unrestricted by the authorities. Deeper international financial inte­ gration. as well as a higher degree of development of domestic financial markets, make the enforcement of controls harder, as they broaden the menu of evasion strategies available to sophisticated (and typically large) investors. 35 As a result, effective enforcement of controls typically requires continuous monitoring and adaptation efforts by the authorities. In this regard, the international experience shows that comprehensive controls are easier to enforce because they are not vul­ nerable to evasion through those outflows left unrestricted. However. comprehen­ sive controls also deter ;;nonspeculative" transactions, such as those related to foreign direct investment. as was seen in the Malaysia episode. Controls on outflows also entail costs in terms of investors' confidence. Because they are usually imposed ex post, as an emergency measure, they typically lead to violation of explicit or implicit contracts. As a result they may raise the perceived riskiness of inward investment and reduce access to foreign capital, and this adverse effect may last well after the removal of the controls on outflows. as inves­ tors considering future capital inflows factor into their risk-return assessment an increased perception of the likelihood that they may be unable to take their capital out when needed, at least without cost. While this credibility effect is hard to quantify, there are strong indications that it was at work in the experience of Malaysia, in which the evidence suggests that the controls on outflows subsequently had a deterrent effect on capital inflows. including long-term ones (Goh 2005). Lastly, capital controls create rents whose allocation is at the discretion of public officials and which therefore opens the door to cronyism and corruption around exceptions and loopholes. In the Malaysia episode, for example. there is evidence that politically connected firms did much better under the controls than unconnected firms. 36 To summarize. the evidence indicates that controls on capital outflows are effective only temporarily and are more vulnerable to evasion when they are selec­ tive rather than comprehensive. They also involve costs in terms of investor confi­ dence and augmented potential for rent-seeking. All this. however. does not mean that controls on outflows should never be used under any circumstances. In fact, when a run on domestic assets gets under way, the authorities may have no option but to impose controls to prevent the collapse of the financial system, a free fall of the currency, or both. The lesson instead is that controls should only be used as a last resort, and only for brief periods to give the authorities a respite from corrective measures. But the more frequent the resort to controls, and the longer they are held in place, the more likely they are to become a substitute for the actual measures and lead eventually to an even bigger crash. DemirgIi,-Kunt and Serven 113 From a broader perspective. controls on capital outflows are just a rudimentary tool of last resort to cope with the vulnerabilities associated with international financial integration. While integration offers countries new opportunities in terms of access to foreign financing and enhanced risk diversification. it also increases countries' vulnerability to external shocks-as the current crisis has clearly shown. How to manage this vulnerability remains a hotly debated issue. 37 Conceptually, the fragilities associated with capital account openness arise from two main sources: currency mismatches and maturity mismatches. For develop­ ing countries, currency mismatches are due to their inability to borrow externally in domestic currency (the "original sin" of Eichengreen and Hausmann 1999). which forces them to bear the exchange risk associated with foreign-currency borrowing. In turn, maturity mismatches can leave borrowing countries vulnerable to creditor runs in a way that is analogous to bank runs. These runs were at the core of the East Asian crisis and have been a recurrent factor in other emerging market crises (Broner. Lorenzoni and Schmukler 2007). They played also a key role in the propagation and amplification of the subprime crisis in the United States (Brunnermeier 2009). Maturity mismatches may reflect the reluctance of lenders to enter into long-term contracts when moral hazard and asymmetric information are pervasive, or when the macrofinancial framework is deemed fragile. The appeal to public and private borrowers is that in such conditions short-term borrowing is cheaper than long-term borrowing. 38 However. private external borrowing decisions fail to take into account their social costs-that is their contribution to raising the aggregate risk of a creditor run or a currency collapse (Caballero and Krishnamurty 2003; Korinek 2009). In fact. major currency and maturity mismatches did arise in a number of countries in the run-up to the global crisis. For example. in some Eastern European countries households actively engaged in the "reverse carry trade" by borrowing in low-interest rate currencies. oblivious to the exchange risk involved; hence foreign-currency tlnancing of residential mortgages became ubiquitous. Worse yet. corporations often borrowed short term in foreign currency to tlnance long-term domestic-currency assets. Banks lent in foreign currency to borrowers lacking foreign currency assets or income. effectively replacing currency risk with credit risk in their balance sheets (Buiter 2009). These mismatches have contribu­ ted to the magnification of the adverse impact of the financial turbulence trig­ gered off by the crisis. The lesson, however. is not that external borrowing should be restricted. In the case of currency risk. the key is to achieve an eftlcient distribution of currency risk within the economy by ensuring. through regulatory means. that it is appro­ priately priced and therefore borne by those best able to do so (for example agents holding foreign currency assets. including exporters). For instance. regulations 114 The World Bank Research OIJsen'er, 1'01.25. no. 1 (february 2010) should discourage banks from lending foreign currency to borrowers without foreign currency assets or income. Likewise, adequate prudential regulation of large financial and nonfinancial borrowers can help limit maturity mismatches. Alternatively, the government could contain aggregate exposure to currency and maturity mismatches by accumulating short-term foreign assets to offset. at least in part. private sector actions-a strategy adopted by a number of emerging markets after the East Asia crisis. In the longer term, institutional changes supporting credible nominal stability should help open the door to domestic-currency borrowing, even if only margin­ ally at first-as seen in countries like Chile, Mexico, Poland, and South Africa­ thereby helping to mitigate currency mismatches. Credible macrofinancial stab­ ility should also allow enhanced access to borrowing at longer maturities. Controls on capital inflows are often advocated as a potential alternative (or a complement) to prudential regulatory measures for preventing the build-up of currency and time mismatches. Depending on the specifics, controls on inflows could help deter overborrowing-limiting exchange rate appreciation, asset price bubbles, and excessive risk-taking at times of large inflows-and!or alter the com­ position of flows against short-term inflows ("hot money") susceptible to rapid reversal. Since the early 1990s, recourse to controls on capital inflows has become fre­ quent as a number of emerging countries have tried to navigate global inflow booms. A partial list includes Indonesia (where controls were introduced in 1991), Malaysia (1994), Thailand (1995, and again in 2006), Brazil (1994), Chile (1991), Colombia (1993, and again since 2004), Mexico (1990), and the Czech Republic (1992). As in the case of outflows, controls on inflows included a variety of instruments-from explicit taxes on foreign borrowing by local firms and fixed-income investments by foreigners (Brazil) to unremunerated reserve and minimum-stay requirements for foreign borrowing by domestic firms (Chile and Colombia, as well as Thailand over 2006-08)-and limits on banks' external borrowing (Indonesia. Malaysia, Thailand, the Czech Republic, and Mexico). Several of these countries matched the introduction of controls on inflows with a removal of barriers to outflows and with various reforms in the prudential frame­ work of the domestic financial system. The case of Chile in the 1990s has been closely scrutinized by a host of empiri­ cal studies. The controls were based on a mandatory unremunerated one-year deposit (or encaje) at the Central Bank equal to a given fraction (initially set at 20 percent) of eligible inflows. The two distinguishing features of the scheme were (i) its implicit cost was higher for short-term inflows than for long-term ones, and (it) the cost was determined by the prevailing level of world interest rates. Initially the regime was applied only to foreign loans (except trade credit). As investors succeeded in finding a variety of loopholes. it was gradually extended to Demirgri,'-Klmt lind Sen'en 115 encompass nondebt creating flows as well--even including foreign direct invest­ ments deemed "speculative." The scheme was eventually eliminated as inflows evaporated at the time of the 1998 Russian crisis. 39 In spite of the often-heard view (especially outside Chile) that the inflow controls were highly successful. their effectiveness remains disputed. On the whole. the large literature concerned with the Chilean experience suggests that (i) the controls were successful at giving the Central Bank some degree of monetary autonomy. allowing it to keep domestic interest rates above international levels-although by most esti­ mates this effect was quantitatively small and temporary; (ii) the composition of inflows was somewhat altered in favor of longer-term flows-although here the evi­ dence is weak; (iii) no clear result emerges regarding the impact on total capital inflows; (iv) there was no discernible effect on exchange rates. Evidence from other episodes is mixed. Colombia's unremunerated reserve requirement, which was in effect over 1993-2000. and then again in 2007, was designed along the same lines as Chile's. Moreover, during 2004-06. as well as in 2007. Colombia also imposed restrictions on short-term portfolio investment (initially including its outright prohibition).4o But there is little evidence that these measures made any difference for the volume of inflows, the share of short­ term flows in the total. or the level of the exchange rate. In contrast, there is some indication that short-term inflows were reduced in the episodes of Malaysia and Thailand in the 1990s. In Brazil the controls were seemingly ineffective in all dimenSions. Thailand's recent imposition of unremunerated reserve requirements in 2006 was followed by a shift of portfolio inflows away from debt instruments and toward equity. and a substantial differential between offshore and onshore interest rates. 41 As with controls on outflows, these episodes show that controls on inflows quickly develop leakages. The most common evasion mechanism was the mis­ statement of the purpose of the inflow; see Carvalho and Garcia (2006) for the case of Brazil and Nadal-de-Simone and Sorsa (1999) for Chile. The result was a weakening of the controls over time-especially rapid in the case of Brazil-in spite of the authorities' constant efforts to close loopholes and deter unwanted flows. Controls on capital inflows also entail other costs. Persistent barriers to capital inflows may deter the development of local financial markets, which in turn may hamper efficiency and growth. 42 And they also raise domestic financing costs by allowing local interest rates to remain above international levels. Microeconomic studies have found large adverse effects of Chile's higher domestic interest rates on the availability and cost of firms' financing. 43 Moreover, because larger firms could better afford the cost of evasion strategies, or had direct access to foreign financing, the cost of the controls-in the form of a higher cost of borrowing­ was disproportionately borne by small and medium-sized firms. 116 The World Bank Research Observer, vol. 25. no. 1 (}ebruary 2010) Finally, what about the ability of controls on capital inflows to prevent volatility and crises, which was the main rationale for their use? Overall, there is not much evidence that they were of great help in this regard, although identifying the counterfactual is admittedly hard. Many observers have noted that the encaje did not prevent Chile from suffering a major sudden stop on occasion of the Russia crisis in 1998, although the scheme may have helped Chile mitigate the "normal" ups and downs of capital inflows in the 1990s (Edwards 1999). In ret­ rospect perhaps this should not be surprising. At times of acute financial turmoil, the run for the exit is not limited to the unwinding of short-term domestic asset positions-whose build-up the encaje and similar mechanisms try to deter-but includes also that of longer-term ones, and involves domestic and international investors alike. The only deterrent to investor exit then is the possibility that the entry cost would have to be incurred again in the future, should investors decide to rebuild their positions. But in past crisis episodes most countries have in fact dismantled their inflow controls. so not even this deterrent remains. This puts into perspective the capacity of inflow controls to deliver stability in times of extreme financial turmoil such as is the case at present. Conclusions The intensity and the breadth of the on-going financial crisis have surprised nearly everyone. Perhaps. more importantly. the policy responses to the crisis have led to considerable confusion and shaken the confidence of the development community in the wisdom of financial and macroeconomic poliCies that underpin Western capitalist systems. We have drawn on a large body of analytical research, econometric evidence, and country experience to argue that the "sacred cows" of financial and macropo­ licies are still very much alive. For the most part. the confusion arises from not being able to recognize incentive conflicts and trade-offs inherent in short-term and long-term responses to a systemic crisis. Policies employed to contain a crisis-often in haste to re-establish confidence and with disregard of long-term costs-should not be interpreted as permanent deviations from well-established policy positions. The fact that governments may end up providing blanket guaran­ tees or owning large stakes in the financial sector in an effort to contain and deal with the crisis does not negate the fact that generous guarantees over the long term are likely to backfire or that government officials make poor bankers. Financial crises often do expose weaknesses in the underlying incentive frame­ works and the regulation and supervision systems that are supposed to reinforce them. But finance is risky business and it is naive to think that regulation and supervision can, or should, completely eliminate the risk of crises, although they Demirgu9- Kunt and Serwln 117 can make crises less frequent and less costly. Neither monetary policy nor capital controls can substitute for well-designed prudential regulation. Despite their inherent fragility. financial systems underpin economic develop­ ment. The challenge of financial sector policies is to align private incentives with public interest without taxing or subsidizing private risk-taking. Public ownership or too aggressive regulation would simply hamper financial development and growth. But striking this balance is becoming increasingly complex in an ever more integrated and globalized financial system. Notes Ash Demirgil~-Kunt is Chief Economist. Financial and Private Sector Development Network and Senior Research Manager, Development Research Group, the World Bank. 1818 H Street NW, Washington DC 20433, USA; email address:Ademirguckunt@woridbank.org.Luis Serven is Senior Adviser for Macroeconomics and Growth. the World Bank; email address: Lserven@worldbank.org. Useful comments from Alan Gelb, Edward Kane. Michael Klein. Justin Lin. Martin Ravallion. and Sergio Schmulder are gratefully acknowledged. Annamaria Kokenyne kindly contributed infor­ mation on recent capital control measures. The authors' findings, interpretations. and conclusions do not represent the views of the World Bank, their Executive Directors. or the countries they represent. 1. Honohan and Laeven (2005) bring together research and first-hand crisis experience to catalog lessons on a variety of issues that regularly arise in crises. 2. Hungary and Jordan are such examples. 3. See for example Switzerland. the Czech Republic, Poland. and Slovakia. 4. See 'Design of Prudential Regulations' below for further discussion of these instruments and markets. 5. See Demirgil~-Kunt and Detragiache (1998, 20(2), DemirgilG-Kunt and Huizinga (2004), Demirgilc-Kunt and Kane (2002). and DemirgiiG-Kunt. Kane. and Laeven (2008a. 2008b). 6. Argentina's reduced deposit mobilization after its 2Ot)} deposit freeze is an example. 7. "Bank Bail-outs: Quids pro-quo." The Economist, November 22. 2008. pp.84~6. 8. Indonesia, the Philippines, South Korea. Mexico. and Chile are examples. See Caprio and Klingebiel (1996) for a complete list. 9. Whether or not the control of the bank completely passes into public hands during the resol­ ution stage is secondary to the fact that after the resolution the bank should be well-capitalized. Problems often emerge because governments are unable to do the financial restructuring properly; and banks that conlinue to operate with insufficient capital have every incentive to resume reckless risk-taking. 10. See World Bank (2001) for a discussion. 11. Similar trade-offs are considered in liberalizing financial systems. since premature liberaliza­ tion without adequate attention to developing institutions increases financial fragility. See Demirgil~-Kunt and Detragiache (1998). 12. Notice there was also significant private equity injected into financial institutions in the latest crisis. The figures reached US$ 500 billion by September. 2008, which is larger than in any other crisis in history. 13. It is difficult to evaluate the success of programs such as the RFC, but in this case it is cred­ ited with having contributed to a recovery of confidence and output (until monetary tightening reversed both in 1937). The government recovered its initial capital and did not bail out nonviable banks. See Mason (2000) for an analysis. 118 The World Bank Research Observer, 1'01.25. no. 1 (rebruary 20W) 14. The Basel committee itself notes that Basel II arose in response to the growth of securitiza­ tion. giving the impression that this growth was an exogenous development rather than at least in part a response to Basel I and its "Ioan-by-Ioan" approach to assessing a firm's overall risk exposure. 15. See Caprio, Demirgii<;-Kunt, and Kane (2008) for an in-depth discussion of Basel I and Basel II accords and recommendations for reforms. 16. See Caprio. Demirgii<;-Kunt, and Kane (2008) and statements of the Shadow Financial Regulatory Committee, which can be found on the American Enterprise Institute (AEI) website. 17. For a discussion of tensions between the goals of expanding access and maintaining stability, see World Bank (2007). IS. These include proposals by Krugman (2008), the Institute for International Finance. the Counterparty Risk Management Policy Group (200S). and the Financial Stability Forum. Goodhart (2008a) and Goodhart and Persuad (2008) recommend authorities set countercyclical capital requirements, introducing a leverage ratio that would move inversely to the business cycle. 19. See American Enterprise Institute statements and Caprio. Demirgii<;-Kunt, and Kane (2008). 20. Investigating the impact of compliance with Basel Core Principals on bank soundness. Demirgii<;-Kunt, Detragiache, and Tressel (2008) show that compliance with information provision and transparency is the most robustly associated with the financial strength of institutions. 21. Subordinated debt has advantages in terms of improving market discipline since subordi­ nated debt holders have incentives to negotiate covenants that would protect their interests. However, credit default swaps (CDS) have advantages in tracking risk since. unlike subordinated debt (which are illiquid instruments), CDS are continuously traded. providing useful signals on a timely basis. 22. Indeed. bubbles can be welfare-enhancing to the extent that they provide stores of value that would otherwise not be available to savers. The best example is that of fiat money in Samuelson's overlapping-generations model; see also Ventura (2004) for a recent open-economy example. 23. Goodhart (2008b). 24. This view is also stated by Mishkin (2008). 25. How fast should asset prices rise for us to conclude safely that a bubble is at work? Too low a threshold would lead to the (incorrect) detection of too many bubbles. Even modest changes in key fundamentals-such as investors discount rates or their anticipations of the rate of growth of future dividends--can lead to large changes in the prices of long-lived assets over short periods. In turn, too high a threshold (say 20 or 40 percent per year) could fail to detect many bubbles altogether. For example during the recent U.S. housing bubble real-estate price increases remained consistently below 15 percent per annum, according to the Case-Shiller U.S. national home price index (see http;//www.standardandpoors.com/home/en/us). 26. The role of housing prices is stressed by Bordo and Jeanne (2002a. 2002b) and Cecchetti. Genberg, and Wadhwani (2003). 27. Goodhart (2008b). 28. These links have been documented by extensive theoretical and empirical research. See for example Calvo (1999). Broner. Gelos. and Reinhart (2006). and Didier. Schmukler. and Mauro (2008). 29. Calls for a reconsideration of controls on capital outflows in the context of the current crisis span a broad range of observers, from leading academics like Calvo (2008) and Subramaniam and Rodrik (2008) to the World Council of Churches (2008). 30. Also, Indonesia and China tightened reporting requirements. Argentina has also recently sought to prevent sales of American Depositary Receipts abroad by domestic residents. 31. In this sense. controls on outflows are similar in spirit to a bank holiday or a suspension of stock market trading. 32. See Ariyoshi and others (2000), Dornbusch (2001). Kaplan and Rodrik (2001) and Magud. Reinhart, and Rogoff (2007). 3.3. Kotice that these transactions do flot entail a net capital outflow or a reserve loss. as the gross outflow implied by the nonrepatriation of the proceeds of the asset sale abroad is matched by Demirgiir,KlInt and Serven 119 the gross inflow implied by its purchase with foreign funds. without any net effect on the foreign reserve stock; see Levy-Yeyati. Schmukler. and van Horen (2004). However. this kind of inter­ national asset migration may have consequences for the tax collection capacity of the local auth­ orities. as domestic residents effectively shift assets out of the domestic jurisdiction. 34. This applies also to the case of intlow controls: see Carvalho and Garcia (2006) for a detailed account of evasion strategies employed by investors in Brazil. 35. As evasion strategies often involve setup costs (as well as the possibility of penalties). they are more easily available to large. wealthy and/or well-connected investors. In fact. evidence from several crisis episodes in Latin America shows that large investors typically liquidate domestic assets ahead of the full-blown crisis and the imposition of controls. so that the burden of controls falls mainly on small investors-while flight from domestic assets by large investors helps trigger the eventual crash. This is one of the mechanisms identified by Halac and Schmukler (2004) through which the cost of financial crises is disproportionately borne by small investors. 36. Johnson and Mitton (2003). 37. The perils of financial integration in a world of imperfect markets have been stressed by many distinguished observers. going as far back as Tobin (1978), and including Harberger (1980), Diaz-Alejandro (1985), Bhagwati (1998). Cooper (199S). and Stiglitz (2002). 3S. Broner. Lorenzoni and Schmukler (2007) show that this maturity premium is quite substan­ tial in the case of emerging markets. 39. Nadal-de-Simone and Sorsa (1999) provide a detailed description of Chile's controls on inflows. 40. Concha. Galindo, and Quevedo (2008) provide a detailed analysis of Colombia's experience with controls on capital inflows. 41. McCauley (200S) documents Thailand's recent experience. as well as those of other Asian countries. 42. See for example Bekaert. Harvey and Lundblad (2003). 43. Forbes (2007). References Acharya, A.. and M. Richardson. 2009. Restoring Financial Stability: How to Repair a Failed System. New York: Wiley. Adrian. Tobias, and Hyun Shin. 2009. "Money. Liquidity and Monetary Policy." American Economic Review 99(2): 600-5. Ariyoshi, A.. K. Habermeier. B. Laurens. and 1. Otker-Robe. 2000. "Capital Controls: Country Experiences with their Use and Liberalization." IMF Occasional Paper 190. Assenmacher-Wesche, K.. and S. Gerlach. 2008. "Ensuring Financial Stability: Financial Structure and the Impact of Monetary Policy on Asset Prices." University of Zurich Institute for Empirical Research in Economics Working Paper 361. Baer. Herb. and Daniela Klingebiel. 1995. "Systemic Risk when Depositors Bear Losses: Five Case Studies." In G. Kaufman, ed. Research in Financial Services: Private and Public Policy. vol. 7. Stanford. CT: JAI Press: 195-302. Bagehot, Walter. 1894. Lombard Street. London: Kegan, Paul, Trench. Trubner & Co. Barth. James R.. Gerard Caprio. Jr., and Ross Levine. 2001. "Banking Systems around the Globe: Do Regulations and Ownership Affect Performance and Stability?" In Frederic Mishkin, ed .. Prudential Supervision: What Works and What Doesn·t. Chicago: University of Chicago Press: 31-96. Beck. T., A. Demirgii<;-Kunt, and M. Martinez Peria. 2007. "Reaching Out: Access to and Use of Banking Services across Countries." Journal of Financial Economics 85:234-66. 120 The World Bank Research Observer, vol. 25. no. 1 (February 2010) _ _. 2008. "Banking Services for Everyone? Barriers to Bank Access and Use around the World." World Bank Economic Review. 22(3): 397-430. Bekaert. G.. C. Harvey. and C. Lundblad. 2003. "Equity Market Liberalization in Emerging Markets." Journal of Financial Research 26:275-99. Bernanke, B., and M. Gertler. 2000. "Monetary Policy and Asset Price Volatility." N'BER Working Paper 7559. Bhagwati. J. 1998. "The Capital Myth: The Difference between Trade in Widgets and Trade in Dollars." Foreign Affairs 77: 7 -12. Bordo. M.. and O. Jeanne. 2002a. "Monetary Policy and Asset Prices: Does 'Benign Neglect' Make Sense?" International Finance 5:139-64. 200lb. "Boom-busts in Asset Prices. Economic Instability and Monetary Policy." CEPR Discussion Paper 3398. Broner. E, G. Gelos. and C. Reinhart. 2006. "When in Peril, Retrench: Testing the Portfolio Channel of Contagion." Journal of International Economics 69:203- 30. Broner, E, G. Lorenzoni. and S. Schmukler. 2007, "Why Do Emerging Markets Borrow Short-term?" NBER Working Paper 13076. Brunnermeier, M. 2009. "Deciphering the Liquidity and Credit Crunch 2007-08." Journal of Economic Perspectives 23(1): 77 -100. Brunnermeier. M.. A. Crockett, C. Goodhart. A. Persaud. and H. Shin. 2009. "The Fundamental Principles of Financial Regulation." Geneva Reports on the World Economy 11. Buiter, W. 2009. "The Return of Capital Controls. (http://www.voxeu.org/index.php?q=node/3104), Caballero, R,. and A, Krishnamurty, 2003. "Excessive Dollar Debt: Financial Development and Underinsurance," Journal of Finance 58(2): 867-94, Calomiris, Charies, and Charles Kahn, 1996, "The Efficiency of a Self-Regulated Payments System: Learning from the Suffolk Banking System," Journal (If Money. Credit and Banking 28(4): 766-97, Calomiris, Charles. Daniela Klingebiel. and Luc Laeven, 2005, "Financial Crisis Policies and Resolution Mechanisms: A Taxonomy from Cross-l'ountry Experience." In Patrick Honohan, and Luc Laeven, eds" Systemic Financial Crises: Containment and Resolution, New York: Cambridge University Press: 25-75. Calvo, G.. 1999, "Contagion in Emerging Markets: When Wall Street Is a Carrier," (http://www,bsos, umd.edu/econ/ciecrp8,pdf). ___,2008, "A New Bretton Woods System Should Curb Boom and Bust." In R. Baldwin. and B, Eichengreen, eds.. What G20 Leaders Must Do to Stabilize our Economy and Fix the Financial System, (http://www.voxeu.orglindex.phptq=nodeI254 3). Caprio, G" and D, Klingebiel. 1996. "Dealing with Bank Insolvencies: Cross Country Experience," Policy Research Working Paper 1620, The World Bank. Washington. D.C, (updated database pub­ lished in Honohan and Laeven 2(05). Caprio, G.. and Maria Soledad Martinez Peria. 2002, 'Avoiding Disaster: Policies to Reduce the Risk of Banking Crises," In E, Cardoso. and A. Gala!, eds .. Monetary Polley and Exchange Regimes: Options for the Middle East. The Egyptian Center for Economic Studies. Cairo. Egypt. Caprio. G,. Ash DemirgiiG-Kunt. and Edward Kane. 2008. "The 2007 Meltdown in Structured Securitization: Searching for Lessons not Scapegoats." World Bank Working Paper 4756. Carvalho, B.. and M, Garcia, 2006, "Ineffective Controls on Capital Inflows under Sophisticated Financial Markets: Brazil in the Nineties," NBER Working Paper 12283, Demirguf,-Ktmt and Saven 121 Cecchetti, S.. H. Genberg, and S. Wadhwani. 2003. 'l\sset Prices in a Flexible Inflation Targeting Framework." In W Hunter, G. Kaufman, and M. Pomerleano, eds., Asset Price Bubbles. Cambridge, MA: MIT Press. Christiano, L., C. Ilut. R. Motto, and M. Rostagno. 200S. "Monetary Policy and Stock Market Bubbles." ECB Working Paper 955. Claessens, Stijn, Daniela Klingebiel. and Luc Laeven. 2005. "Crisis Resolution, Policies and Institutions: Empirical Evidence." In Patrick Honohan, and Luc Laeven, eds .. Systemic Financial Crises: Containment and Resolution. ~ew York: Cambridge University Press: 169-96. Clarke, George, Robert Cull, and Mary Shirley. 2005. "Bank Privatization in Developing Countries: A Summary of Lessons and Findings." Journal of Banking and Finance 29: 1905 - 30. Cole, S. 2004. "Fixing Market Failures or Fixing Elections? Elections. Banks, and Agricultural Credit in India." Harvard Business School. Cambridge, MA. Concha. A., A. Galindo. and C. Quevedo. 200S. 'l\n Assessment of Another Decade of Capital Controls in Colombia: 1995-200S." Unpublished manuscript. Cooper, R. 1998. "Should Capital Account Convertibility Be a World Objective?" Princeton Essays in International Finance 207. Counter party Risk Management Policy Group. 2008. "Containing Systemic Risk: The Road to Reform." The Report of the CRMPG III. August. (http://www.crmpolicygroup.orglindex.htmi). De la Torre, Augusto. Juan Carlos Gozzi. and Sergio Schmulder. 20D7. "Innovative Experiences in Access to Finance: Market Friendly Roles for the Visible Hand?" Policy Research Working Paper 4326. World Bank, Washington. DC. Demirgli<;-Kunt. Ash. 200S. "Resolution of Systemic Financial Problems." Viewpoint ~ote. World Bank. September. Demirgli<;-Kunt. Ash. and Enrica Detragiache. 1998. ''The Determinants of Banking Crises: Evidence from Developing and Developed Countries." IMF Staff Papers 45: 81-109. .___. 2002. "Does Deposit Insurance Increase Banking System Stability? An Empirical Investigation." Journal of Monetary ECOIlOmics 49: 1373 -406, DemirgiiG-Kunt. Ash, and Harry Huizinga, 2004. "Market DiScipline and Deposit Insurance," Journal of Monetary Economics 51(2): 375-~99, Demirgii<;-Kunt. Ash. and Edward Kane. 2002. "Deposit Insurance around the Globe? Where Does It Work?" Journal of ECOtlOmic Perspectives 16(2): 175-95. Demirgu<;-Kunt. Ash, Enrica Detragiache. and 1: Tressel. 2008. "Banking on the Principles: Compliance with Basel Core Principles and Bank Soundness." Journal of Financial Intermediation 17(4): 511~42, Demirgii<;-Kunt. Ash, Edward Kane, and Luc Laeven, 2008a. Deposit Insurance around tile World: Issues of Design and Implementation. Cambridge, MA: MIT Press, ___. 2008b, "Determinants of Deposit Insurance: Adoption and Design." Journal of Financial fntermet/iation 17(3): 407 - 38. Diamond. Douglas W, and Philip H. Dybvig. 19S3. "Bank Runs. Deposit Insurance, and Liquidity." Journal of Political Economy 91 :401-19, Diaz-Alejandro, C. 1985, "Good-bye Financial Repression, Hello Financial Crash."" Journal of Developmen t Economics 1 9. 1- 24, Didier, T., S. Schmukler, and P. Mauro. 2008, "Vanishing Contagion?" Journal of Policy Modeling 30(5): 775-91. Din<;. Serdar. 2005. "Politicians and Banks: Political Influences on Government-Owned Banks in Emerging Countries." Journal or Financial Economics 77(2): 453-79. 122 The World Bank Research Observer, vol. 25, no, 1 I I-ebruary 2010) Dornbusch, R. 2001. "Malaysia: Was it different?" NBER Working Paper 8325. Edwards, S. 1999. "How Effective Are Capital Controls?" Journal of Economic Perspectives 13:65-84. Eichengreen. B., and R. Hausmann. 1999. "Exchange Rates and Financial Fragility." Proceedings. Federal Reserve Bank of Kansas City: 329-68. Forbes. K. 2007. "One Cost of the Chilean Capital Controls: Increased Financial Constraints for Smaller Traded Firms." Journal of!memational Economics 71:294-323. Gerschenkron. Alexander. 1962. Economic Backwardness in Historical Perspective: A Book of Essays. Cambridge. MA: Belknap Press of Harvard University Press. Goh, S. 2005. "New Empirical Evidence on the Effects of Capital Controls on the Composition of Capital Flows in Malaysia." Applied Economics 37:1491-503. Goodhart. C. 2008a. "The Regulatory Response to the Financial Crisis." CESIfo Working Paper 2257. 2008b. "Central Banks' Function to Maintain Financial Stability: An Uncompleted Task." In R. Baldwin, and B. Eichengreen. eds .. What G20 Leaders Must Do to Stabilize our Economy and Fix the Financial System. (http://www.voxeu.org/index.php?q=node/2543). Goodhart. C., and Avinash Persaud. 2008. "A Party Pooper's Guide to Financial Stability." Letter to the Editor. The Financial Times. June 4. Greenspan. A. 2004. "Risk and Uncertainty in Monetary Policy." American Economic Review 94: 33-40. Halac. Marina, and Segio Schmukler. 2004. "Distributional Effects of Crises: The Financial Channel." Economia Jounlal of the Latin American and Caribbean Economic Association 5: 1- 6 7. Harberger. Arnold C.. 1980. "Vignettes on the World Capital Market." flmerican Economic Review 70:331-7. Honohan, Patrick. 2005. "Fiscal. Monetary and Incentive Implications of Bank Recapitalization." In Patrick Honohan, and Luc Laeven. eds .. Systemic Financial Crises: Containment and Resolution. New York: Cambridge University Press: 25--75. Honohan. Patrick. and Daniela Klingebiel. 2003. "The fiscal Cost Implications of an Accommodating Approach to Banking Crises." TournaI of Banking and Finance 27:1539-60. Honohan. Patrick. and Luc Laeven. 2005, Systemic Financial Crises: Containment and Resolution. New York: Cambridge University Press. Johnson. S., and T. Mitton. 2003. "Cronyism and Capital Controls: Evidence from Malaysia." Journal of Financial Economics 67:351-82. Kaminsky. G.. and S. Schmukler. 2001. "Short- and Long-run Integration: Do Capital Controls Matter?" In Susan Collins. and Dani Rodrik. eds .. Brookings Trade Forum 2000. Washington, DC: Brookings Institution Press: 125-78. Kane. Edward, and Daniela Klingebiel. 2004. ':Alternatives to Blanket Guarantees for Containing a Systemic Crisis." Journal of Financial Stability 1:31-63. Kaplan, E., and D. Rodrik. 2001. "Did the MalaYSian Capital Controls Work?" NBER Working Paper 8142. Kashyap. A.• R. Rajan. and J. Stein. 2008. "Rethinking Capital Regulation." Unpublished Manuscript. Keeler, Philip. 2000. "When Do Special Interests Run Rampant? Disentangling the Role of Elections. Incomplete Information and checks and Balances in Banking Crises." Policy Research Working Paper 2543. World Bank. Washington. DC. Khwaja, Asim Ijaz, and AUf Mian. 2005. "Do Lenders Favor Politically Connected Firms? Rent Provision in an Emerging Financial Market." Quarterly Journal of Economics 120(4): 13 71-411. Demirgii~-KuTlt and Servin 123 Klingebiel. Daniela. 2000. "The Use of Asset Management Companies in the Resolution of Banking Crises: Cross-Country Experience." World Bank Policy Research Working Paper 2284. Washington. DC. Korinek, A. 2009. "Excessive Dollar Borrowing in Emerging Markets." Unpublished Manuscript. Krugman, Paul. 2008. "Partying Like it's 1929." New York Times. March 21. Laeven, Lue. and Fabian Valencia. 2008. "The Use of Blanket Guarantees in Banking Crises." IMF Working Paper WP/08/250. La Porta, Rafael. Plorencio Lopez-de-SUanes. and Andrei Shleifer. 2002. "Government Ownership of Commercial Banks." The Journal of Finance 57(1): 265- 301. Levy-Yeyati, E.. S. Schmukler. and N. van Horen. 2004. "The Price of Inconvertible Deposits: The Stock Market Boom During the Argentine Crisis." Economics Letters 83:7-14. ___. 2009. "International Integration through the Law of One Price: The Role of Liquidity and Capital Controls." Journal of Financial Intermediation 18:423-63. Maddaloni. A.. J. Peydro, and S. Scopei. 2008. "Does Monetary Policy Affect Bank Credit Standards?" Unpublished Manuscript. Magud. N., C. Reinhart. and K. Rogoff. 2007. "Capital Controls: Myth and Reality." Unpublished Manuscript. Mason. Joseph R. 2000. "Reconstruction Finance Corporation Assistance to Financial Intermediaries and Commercial and Industrial Enterprise in the U.S.. 1932-1937." World Bank. Washington DC. McCauley, R.N. 2008. "Managing Recent Hot Money Inflows in Asia." ADB Institute Discussion Paper 99. Megginson. William. 2005. "The Economics of Bank Privatization." Journal of Banking and Finance 29:1931-80. Mishkin, R. 2008. "How Should we Respond to Asset Price Bubbles?" Speech delivered at the Wharton Financial Institutions Center, May 15. Nadal-de Simone. E, and P. Sorsa. 1999. 'l\. Review of Capital Account Restrictions in Chile in the 1990s." IMF Working paper 99/52. Papademos. L. 2009. "Monetary Policy and the 'Great Crisis': Lessons and Challenges." (http:// www.ecb.intipress/key/date/2009/htm!lsp090514.en.html.). Stiglitz, J. 2002. Globalization and its Discontents. New York, NY: \V.w. Norton. Subramaniam. A.. and D. Rodrile 2008. "We Must Curb International Flows of Capital." Financial Times, February 25. Tobin, J. 1978. 'l\. Proposal for International Monetary Reform." Eastern Economic Jouma14:154-9. Ventura, J. 2004. "Bubbles and Capital Flows," llnpublished Manuscript. World Bank. 2001. Finance for Growth: Policy Choices in a Volatile World. Policy Research Report. Washington DC. ___. 2007. Finance for All? Policies and Pitfalls in Expanding Access. Policy Research Report, Washington DC. World Council of Churches. 2008. "14-11-08 Statement on the Global Financial Crisis," (http:// www.oikoumene.org/index.php?id=6419). Wyplosz, C. 2009. "What Do we Know about Monetary Policy that Friedman Did Not Know?" Unpublished Manuscript. 124 The World Bank Research Observer. vol. 25. no. 1 (February 2010) The 2007 Meltdown in Structured Securitization: Searching for Lessons, not Scapegoats Gerard Caprio, Jr., Asl1 Demirgij~-Kunt, and Edward J. Kane The intensity of the crisis in financial markets has surprised nearly everyone. The authors search out the root causes of the crisis. distinguishing them from scapegoating explanations that have been used in policy circles to divert attention from the underlying breakdown of incentives. Incentive conflicts explain how securitization went wrong, why credit ratings proved so inaccurate. and why it is superficial to blame the crisis on mark­ to-market accounting, an unexpected loss of liquidity. trends in globalization. and dereg­ ulation in financial markets. The authors' analysis finds disturbing implications of the crisis for Basel II and its implementation. They conclude by drawing out lessons for developing countries and identifying reforms that would improve incentives by increasing transparency and accountability in government and industry alike. JEL codes: G21, G28.G32 Since August. 2007, after a long period of relative quiet in world markets/ a spreading financial crisis has nearly monopolized the flow of economic news. Beginning during a period of strong world macroeconomic growth and low inter­ est rates, the crisis appears to have surprised financiers and regulators alike. The turbulence was triggered by a sudden and widespread loss of confidence in securi­ tization and financial engineering and by the manifest failure of respected statisti­ cal models for assessing and pricing credit risk. Most astonishingly. these now-doubtful techniques had previously been hailed as the cornerstones of modern risk management. The turbulence proved greatest in countries whose supervision of credit risk had been thought to be the best in The World Bank Research Observer The Author 2010. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development I TIlE WORLD BAKK. All rights reserved. For pennisslons. please e-mail: journals.perrnissions@oxfordjournals.org doi:1O.l093/wbrollkp029 Advance Access publication February 17. 2010 25:125-155 the world. Indeed, the regulatory standards and protocols of these countries were in the process of being emulated worldwide. As the crisis unfolded, the world witnessed a series of unprecedented events, including a previously unthinkable rate of default on AAA instruments; the first run on a UK. bank in 150 years; and an explicit extension of the u.s. safety net to cover a major insurance company, the entire investment banking industry, and two giant government-sponsored housing-finance enterprises (Fannie Mae and Freddie Mac). These events were followed by the demise of a number of commer­ cial and investment banks and a sharp worldwide plunge in equity stock prices that was especially pronounced in the financial sector. Reverberations quickly spread beyond the two financial-center countries to other industrial countries including Australia. Ireland. Iceland. Germany. and many other countries. first by financial channels and then by the collapse of world trade. By April. 2009 the IMF was estimating total losses at over $4 trillion, about two-thirds of which would redound to banks around the world (IMF 2009. p. xi). Inquiring minds yearn to know how this crisis could have occurred in the 21st century, and especially how it could have originated in the United States, home to arguably the most sophisticated financial system in the world. Increases in leverage and risk taking. which were key factors in the crisis, were not limited to the aforementioned countries but seen also. in particular. in the Baltics and other Eastern European countries. Developing countries had been closely emulating the approaches to financial sector regulation taken in the high­ income countries. as seen in the decision by many of the former to adopt the Revised Capital Framework. or Basel II. Although most emerging markets had not yet adopted significant use of securitization in their domestic financial systems. they were well on the way to greater reliance on ratings organizations. Additionally, the failures in the incentive and information environment that were the hallmark of this crisis are relevant for developing countries. as seen in the Mexican and East Asian crises of the 1990s. Both rich and poor countries' finan­ cial systems share a vulnerability to asset bubbles. and a key message from the current crisis is that all countries need to focus on the vigilance of their regulat­ ory authorities. Beyond dramatic macroeffects, already seen in collapsing trade volumes and output growth, this crisis will have two important channels through which it will influence developing countries. First, influence will be felt through the upcoming reforms of the architecture of financial regulation. While we cannot predict future policies, how this reform should proceed is the focal issue of our analysis. Second, developing countries will be affected long into the future by the lessons that they draw from this crisis, regardless of the regulatory changes made in the high­ income countries. We regard it as critical that these lessons can and should be implemented in low- and middle-income countries. Complex methods for 126 The World Bank Research Observer, vol. 25. no. I (February 2010) regulating risk-taking have failed miserably in this crisis, so it is time for simpler yet effective approaches to regulation and supervision. Promptly uncovering the true roots of this crisis is important because false explanations are quick to gain a toehold. As a crisis matures and then begins to recede policymakers and pundits often latch onto simplistic theories of what hap­ pened, why it happened, and what should be done to see that similar events do not happen again. Sadly, the story that official theories are beginning to tell and the policy solutions that these flawed theories recommend tend to be dictated not by the economics of crisis generation, but by self-interested jockeying by groups and individuals that are anxious either to shift blame away from themselves or to see that national safety nets remain an important source of subsidies to large and complex institutions. Our study seeks to make it clear that the principal source of financial instability is not to be found in the aberrant behavior of a few greedy individ­ uals or in a sudden weakening of important institutions of a particular country at a particular time. Rather systemic financial fragility is marked by an undermining of the effectiveness of financial regulation and supervision in every country in the world. often involving contradictory political and bureau­ cratic incentives. Supervisory agencies overlook those occasions when financial institutions and their customers overleverage themselves in creative ways; they also close their eyes to the unbudgeted costs of the loss exposures that excess leverage passes onto financial safety nets until it is too late for anyone to control the damage that results. To understand the sources of instability we need to remember that regulation and supervision must be viewed as an endless game of action and response (Kane 1977). In this game the regulated side is able to move more often and more quickly than the regulatory side can. As outsiders. regulators are at a disadvan­ tage in monitoring and enforcing financial discipline, so they inevitably find themselves trying to catch up with their regulatees. Every move they make gener­ ates a series of new and creative moves by financial institutions who seek to mini­ mize the burdens regulations ultimately place on them. The profit-making orientation of financial institutions also ensures that their moves are not only swifter. but also more complex and harder to anticipate than those of regulators. This dynamic perspective helps us to ask and answer a series of central questions about the origins of the latest crisis; • Where did modern financial engineering, securitization. and risk manage­ ment go wrong? • Why did rating organizations not uncover the dangers. and who should bear responsibility for over-rating securitized debt? • Can mark-to-market accounting cause a crisis or is it merely a messenger? Caprio, Demirgii(:-Kunt. and Kane 127 • What did financial globalization contribute to the crisis. and should links among national markets be restrained in some way? • Has Basel II failed already, or would its wider implementation have stopped the spread of the current turmoil? • What are the lessons from this and past crises for developing countries? Finally mention should be made of two important issues that are beyond the scope of this paper, The first is the role of monetary policy in the run-up to, and the implosion of. bubbles. and the associated policy issues, For an analysis of this we refer the reader to Roubini (2005). Posen (2006), Taylor (2009). and DemirguG-Kunt and Serven (2010). The second issue is the authorities' crisis res­ olution policies. These are not covered here. since it would be premature to try to summarize their triumphs and mistakes while they are still shaping and reshaping their response to the crisis. How Did Securitization Go Wrong? The first reaction to the breakdown of structured securitization was disbelief: How could something as universally applauded as financial engineering go so wrong? Securitization was supposed to identify risks accurately and parcel them out to parties who could easily bear the risks they assumed. But everyone now realizes that promoters of modern techniques of risk management promised a great deal more than they could ultimately deliver. For centuries, loans were the most illiquid parts of a bank's balance sheet. Loan sales were limited by fears rooted in asymmetric information and adverse selection. Unless the sales contract specifically protected the buyer against the sellers' informational advantage, the original lender would be tempted to sell off its worst loans (its so-called "lemons") and hold back its solid loans for itself. These same issues made it hard for a buyer to resell loans when funds were needed. Securitization provided an indirect way to sell loans, one that could offer buyers a number of useful safeguards. It also made it easy for a buyer to reverse its position later. Concerns about the quality of the loans chosen to back claims were muted when the original lender retained a "first-loss" position and also when the securities were highly collateralized. In the United States. securitization took off in the 1970s. Its rapid development was greatly assisted by guarantees provided to investors by the Federal National Mortgage Association (Fannie Mae) and the Federal Home Loan Mortgage Corporation (Freddie Mac). These government-sponsored enterprises (GSEs) regu­ larized market practices by standing ready to buy and securitize mortgages that conformed to the particular standards that they set (Gorton and Pennachi 1995). 128 The World Bank Research Observer. vol. 25. no. 1 (February 2010) Subsequently, private securitizers developed protocols for trading claims to cash flows from other types of standardizable loans. The largest markets cover mort­ gages that were too big for the GSEs to purchase ( so-called "jumbos"), credit-card debt, student loans. and automobile loans. The 1988 Basel Accord (Basel I) also helped to generate a supply of loans that could be securitized and resecuritized. The accord tied two layers of bank capital requirements to an arbitrarily risk-weighted sum of their assets. The arbitrariness of the weights gave banks an incentive to move assets into off-balance-sheet secur­ itization vehicles because capital charges on credit lines with which sponsors sup­ ported these vehicles were lower than the charges levied against on-balance-sheet positions. Basel II, which was just being implemented as the crisis was unfolding, further encouraged the holding of highly rated instruments and a reliance on risk models that turned out to underestimate the fragility of banks' portfolios. By turning the cash flows from a pool of illiquid underlying assets (such as mortgages) into tradable bonds, securitization created liquidity-and that liquidity promised to make the financial system better diversified and more resilient. Instead of bankers having to hold onto and support every loan they originated until it matured or defaulted, securitization allowed risks to be stripped from the loans and disbursed beyond the traditional geographical areas in which a particu­ lar lender had been operating to investors in any country of the world. The gain from this innovation was the reduced cost of mortgages and more affordable home ownership for a range of marginally less-creditworthy individuals. Securitization might not greatly degrade credit quality, as long as quality is transparent and well priced to all parties along the securitization chain. Breakdowns in assessing and pricing mortgage risk played an important role in the seizing up of asset-backed markets in 2007. While securitization is simple enough, the slicing of cash flows from various mortgages and other instruments into tranches, and the pressure for high ratings over time, led securitizers to fashion very complicated structures of cash-flow disbursement. The relation of particular tranches to the underlying asset pool was often very opaque. This made tranched claims difficult to value and susceptible to sudden changes in risk perception. 2 Large banks could sell almost any pool of loans, securities, or reven­ ues into the securitization process. Buyers would slice the claims they used into a series of at least three subordinated tranches: senior (AAA). mezzanine (BBB). and an unrated residual or equity tranche. In the event of defaults on individual loans. the equity tranche would absorb losses until the equity was used up. After that. the BBB tranche would absorb further losses until it too was exhausted. Subordinated tranching of risk does not mean that the loans in the securitized pool were divided into bands based on the credit risk of particular loans. Rather, senior tranches had the first claim on whatever cash flow the whole asset pool might generate. Buying a senior tranche offered protection against losses by Caprio, Demirgiir;-Kunt, and Kane 129 assigning them-in a probabilistic sense-to the junior tranches. Investment­ grade credit ratings by credit rating organizations (CROs) awarded to the senior tranches suggested they were safe even when the underlying collateral was all subprime. 3 The root problem with securitization-as with loan sales-is that outsourcing the funding side of an originator's balance sheet undermines its incentives to monitor the quality of the loans it originates. Troubled loans become the property and problems of someone further down the transaction chain. As the demand for highly rated tranches intensified and securitization became more complicated and less transparent, underwriting incentives weakened because securitizers and CROs performed little actual due diligence. Low interest rates and increasing housing prices encouraged an overly friendly regulatory environment both for highly lever­ aged mortgages and for securitization structures based on them. Because securitization caused more subprime mortgages to be written, it expanded access to home ownership substantially.4 Unfortunately, federal regula­ tors and Congress were cheerleaders in this process, even though higher volumes went hand-in-hand with lower standards and severe mispricing of risk. 5 Dell' Ariccia, Igan, and Laeven (2 008) show that standards weakened the most for borrowers whose risks were highest. Increases in the volume of loan applications by subprime borrowers were associated with an increased rate of approval and lower loan quality.6 In contrast. for borrowers classified as prime, increased appli­ cations produced more rejections. How can one explain the growth of securitization? Demand was high both because interest rates on the safest securities were at or near record low levels and because of the capital relief afforded to banks for holding highly rated instru­ ments. as well as the regulatory requirements on other intermediaries to hold the same. The Securities and Exchange Commission (SEC),s sanctioning of certain ratings agencies (see below) no doubt gave investors some comfort. On the supply side. in addition to the quest to expand home ownership. risk-shifting created arbitrage profits for institutions able to service this demand. Securitization was simply the latest innovation through which financial institutions could simul­ taneously collect fees from investors and arbitrage loopholes in bank regulation and supervision. By placing important tranches of risky loans through and with foreign and nonbank firms. large commercial and investment banks layered the institutional character. and broadened the geographic span, of their funding arrangements. Moreover, they did this in ways that made these institutions ever more complicated. ever more interlinked, and therefore ever more difficult to fail and unwind in the event of a crisis. All of this increased their potential claims to intangible safety-net subsidies. Investors in complex claims on securitized pools of loans tended not to rely on either the lender's or their own due diligence and until well into 2()()7 many investors deemed it reasonable to allow credit rating 130 The World Bank Research Observer, vol. 25, no. 1 (FelJruary 2010) organizations to assess the risks for them. With supervisors closing their eyes to the fraying of contractual incentives for lenders and credit raters to fulfill their duties, few investors saw any reason to doubt that they were purchasing well­ rated and well-priced securities. Investors' mistaken belief that their growing demand for highly rated investments was being properly serviced allowed the demand side of the market to expand steadily. Outsized commissions and fees earned on securitizations assured a steadily growing deal supply. Compensation systems in commercial and investment banking paid large bonuses tied to immediate profits and required no payback if losses occurred in subsequent years. Even when compensation took the form of stock options. blackout periods before such options could be exercised were too short to align employee incentives with the long-term interest of the firm. The development of securitization in emerging markets has not flourished at the same rate, mostly due to underdeveloped capital markets. Securitization may be beneficial for them. however. if it is kept simple and well-regulated. A prudent framework of securitization should include two key features. First. leverage should be effectively limited by capital requirements by requiring loans to remain on a bank's balance sheet. Second, loan origination should be straightforward and fully transparent. 7 Why Were Credit Ratings so Inaccurate? While risk-management mistakes, low interest rates. and some kind of asset-price bubbles are features of most crises, this crisis may be remembered as one in which long-successful systems for using debt ratings to control institutional risk­ taking failed massively. Although blame must be apportioned across the entire chain of securitization activity, U.S. CROs come in for special criticism because investors have seen an embarrassingly large number of downgrades and defaults for highly rated securi­ ties. In helping potential counterparties to assess the creditworthiness of individ­ ual bond issues. CROs earn profits by producing classificatory information that regulators find helpful and that investors and guarantors use to compare credit spreads on issues of risky debt. However, CRO revenues come not from the inves­ tor or regulatory side, but from fees that issuers pay CROs for analyzing the credit quality of different issues; although accurate ratings benefit investors and issuers alike, issuers are asked to pay because. once it is announced, a security's credit rating becomes public knowledge. This asymmetric arrangement poses an obvious conflict of interest for CRO managers. Borrowers have an incentive to play different CROs against one another and to hold out for higher-than­ appropriate ratings. For issuers and securitizers, the counterincentive to seeking a Caprio. Demirgii~-Kunt. and Kane 131 corrupt rating is that they also need to employ a CRO that has a well-established reputation for honest and accurate work. Until the 1970s, ratings were generally in low demand (Sylla 2001). eROs' value added was uncertain, and as recently as the 19 60s these firms employed only one to two dozen analysts, with their then meager revenues coming mostly from sales of research (Partnoy 2001). Ratings were long criticized as lagging behind the business cycle--issuing upgrades late in cyclical upswings and downgrades late in slowdovvns in ways that did not help investors to anticipate or protect themselves against a rise in defaults. This is amply demonstrated by the avalanche of ratings downgrades in the 19 30s. 8 However, in 1975 the SEC created the designation of a "Nationally Recognized Statistical Ratings Organization (NRSRO),., and over time ratings-based government rules restricted the decisions of a variety of actors (pension funds, insurance companies, banks, municipalities, and so on). Government reliance on ratings encouraged private organizations to incorporate ratings into their own governance procedures or to advertise that only investments above a certain rating would be held. With financial intermediaries either strictly required to hold only highly rated instruments or allowed to hold less capital against highly rated securities, ratings understandably grew in demand, as did the pressure for "rating inflation." By the late 1990s, when the structured finance business expanded sharply, the eRa industry expanded apace. By 2006 Moody's, which won a large share of this business, was generating over $6 million per employee and employed over 20,000 persons worldwide. Because many regulatory agencies, investors, and bond insurers rely on eRa credit ratings to substitute for their own due diligence, the contract interest rate an issuer has to pay falls whenever its credit rating rises. In turn. for established eROs, the time and effort required to build a reputation for reliability. and the bureaucratic difficulties to be surmounted in being named an NRSRO, create a dual barrier for would-be new entrants into the eRa industry. Reinforced by the tendency of established firms to acquire lesser players, these barriers give them a leg-up in foreign venues as well. The resulting oligopolistic market structure helps to explain why major ratings organizations do not compete either in the models they use to assess credit risks or in the criteria they use to map the forecasts their models produce onto different rating classes. This similarity in methods means that errors are likely to be similar too. The core problem in the securitization crisis is to understand how and why securitizers, eROs, and bond insurers drasti­ cally over-rated and oversized the highest-quality tranches of structured-finance obligations. Part of the explanation lies in the conflict that managers and line employees of such firms faced between preserving the long-run value of their firm's reputation and chasing bonuses and wage raises that short-run revenue expansion can gen­ erate. Errors in classification are slow to reveal themselves. They can only be 132 The World Bank Research Observer. vol. 25, no. 1 (February 2010) established after a long and variable lag. This lag means that. to keep a firm's reputation strong over the long run, compensation structures must include fea­ tures that promise to reward employees for taking the long view and penalize them for succumbing to short-termism. Given the high proportion of revenues earned in recent years at the top three ratings firms (Moody's, Standard & Poors, and Fitch) from rating securitizations, individual managers and analysts must have been sorely tempted to risk the firm's reputation to secure or retain the repeat business of the biggest issuers, and it is doubtful that salary structures fully neutralized this temptation. According to Portes (2008), 44 percent of Moody's 2006 revenue came from advising issuers first on how to collateralize and to assign (that is to slice or "tranche") cash flows from pools of securitizable assets to get a desirable package of ratings, and then going on to rate the credit risk of the various packages that it and other eROs helped to construct. With compen­ sation in much of the financial sector increasing and being linked to short-term paper profits with little attention to risk it is hardly surprising that the same occurred in eROs. What's Different about Rating Structured Instruments? In principle, each rating should be interpreted as an interval estimate: that is as the sum of a point estimate and a two-sided margin for error. When a eRO does a good job of rating bonds or complex securities, the observed value of default and loss rates in different rating classes correlate closely with the riskiness of the grade that securities in each category had previously received. Because securitized instruments are claims on a fixed pool of individual assets, servicers who manage the cashflows can do little to mitigate the potential impact of adverse events on investor returns. 9 Even if point estimates of loss exposure were the same for a bond and a securitized claim, their margins for error would be very different. This means that it was misleading for eROs to employ the same set of letter grades to rank the through-the-cycle loss exposures of the tranches of structured deals and ordinary bonds. Even on ordinary bonds, ratings are lagging indicators whose changes tend to come too late to help investors avoid losses when an issuer's credit standing weakens or to achieve gains when an issuer's prospects improve. This leads scho­ lars to question whether on most deals eROs add enough informational value to justify their existence (Sylla 2001). However, because of the growing complexity of structured instruments, there can be no doubt that ratings were central to the successful placement of synthetic securities. The SEe and other regulators effec­ tively ceded to eROs their public-interest responsibility for monitoring and disclos­ ing investor loss exposures in structured financial instruments. Investors flocked to the highly rated tranches of structured securitizations precisely because they Caprio, Demirgii):-KuTlt, and Kane 133 promised miraculously to combine AAA and AA ratings with extraordinarily high yields-and regulators did not challenge this promise. As noted earlier, these high yields came mostly from blending in returns from the lower-rated com­ ponents of structured instruments. Although actual and proposed reforms seek to rework the details of eRa and issuer interactions. the process of rating complex structures of securitized debt differs critically from that of rating a simple bond issue. The process of rating a structured product is a sequence of bilateral negotiations that starts with the issuer specifying the mix of credit ratings it is looking for. eROs compete by speci­ fying the subordination structure and level of credit support needed to obtain the ratings desired. That a give and take between eROs and securitizers did occur is suggested by the high concentration of eRa forecasts for structured deals that lie at "notches" just above the thresholds that would move the different tranches into the next lower ratings class (Mason and Rosner 2007). This implies that the associated interval estimates on most issues regularly dipped at least into the next-lowest rating class. Assessing the risk of a portfolio of infrequently traded and innovative instru­ ments and monitoring factors that change this riskiness over time pose difficult problems for data verification and analysis. eROs could and should have identified and addressed these problems more carefully, and in particular. using a prudent­ man standard, they should have discounted the margins for error assigned to complex mortgage securitizations for the modeling, sampling, legal. and documen­ tation risks that investors were asked to assume. If the industry had been less oligo­ polistic. competitive pressure likely would have led independent parties to be tasked with auditing the models and criteria on which individual eRa ratings were based and to fact-check the data used to estimate model parameters. Most importantly. conscientious outside reviewers would have insisted that eROs update their models and rating methods as soon as evidence began to develop that loan pools in the 2005 and 2006 vintages were defaulting at unprecedentedly early dates. Without the protection of ratings-based legal "safe harbors," fund managers seeing the same events would have had their decisions exposed to adjudication in court. 10 To lessen ratings volatility. eROs prepare what they term "through the cycle" ratings. The models eROs use for this task were known to incorporate unverifiable and overly convenient assumptions about correlations, worst-case scenarios, and marketability that were bound to break down (that is to lose applicability) in periods of severe financial distress. Many deal structures were new and yet to be tested against the stresses that might develop during a business-cycle downturn. For example. ratings on instruments based on negative-equity or optional­ payment mortgage loans to low-income households used experience-based data on default frequency and loss given default drawn entirely from a period of macro­ economic expansion and soaring house prices. To simulate through-the-cycle 134 Tile World Bank Researcil Observer. vol. 25. no. 1 (February 2010) experience, observed defaults on innovative instruments ought to have been sup­ plemented by synthetic data designed to introduce effects that might unfold in times of price decline and market stress (Altman and Rijken 2006). eROs Need to Take Responsibility for their Mistakes . .. Formally a CRO's aggressive declaration that an adequately documented "true sale" of a particular loan pool had taken place was a key step in moving the assets off originator and securitizer accounting balance sheets. But CROs apparently felt no duty to describe how fully the ownership of the pool could be documented. Hence their judgments on this matter could have no legal standing in any case. The quality of CRO analysis was and is further undermined by CRO efforts to avoid legal responsibility for any mistakes. Despite their intense and critical invol­ vement in designing securitization structures, CROs claim only to be expressing an "opinion." They insist that the constitutional right of free speech protects them from lawsuits for damages suffered by investors who chose to rely on what might turn out to be incompetent or negligent opinions. To create a foundation for this defense, CROs routinely incorporate language into their reports stating that it is "unreasonable" for anyone to rely on their "mere opinions." which should not be construed by anyone as "investment advice," Ironically, for investors and regula­ tors, the reputational damage CROs have absorbed from massively over-rating structured securitizations has imparted to these disclaimers an element of unin­ tended truth that has undermined the value of their brands and is forcing them to rebuild confidence in the value of their work. Because CRO fees were so large. and because synthetic securities could not legally have been sold in large quantities to many investors without the blessing of high ratings, the courts might impose liability on CROs in any case. Whenever someone (say. a lawyer) collects a large fee for communicating his or her opinion to another party. the distinction between opinion and advice seems to break down. The sheer size of the fees collected for forming and issuing opinions about the riskiness of complex securitizations renders hollow the claim that users should not-and therefore would not-rely on them. In fact, CROs had to foresee and value that reliance, as it explains why they were being remunerated so welL They should share responsibility with any securitizer and insurer of these deals who distorted or failed to verify the value of the analysis on which CRO "opinions" ultimately were based . . . . but the Authodties also PLayed a RoLe On the grounds that they were helping innovative U.S. firms to compete effectively in global markets. federal supervisors refused to take on the political and practical Caprio. Demirgu}'-Kunt. and Kane 135 challenge of establishing and maintaining their ability to see and discipline com­ plicated risk exposures. By tolerating the decline in transparency that came with structured finance; by not recognizing CRO incentives and that they were using poorly tested models and issuing aggressive legal judgments about whether non­ recourse "true sales" of the underlying loans had actually taken place; and by not requiring CROs to discount their ratings on these instruments for the modeling and documentation risks inherent in structured finance-supervisors made it dif­ ficult for themselves and other market participants to recognize and discipline these risks. As a result investors were fed overly optimistic estimates of the credit quality of the instruments they purchased. The flaw in relying more on CROs is critical for developing countries, since, as noted below, prior to the crisis many of them announced plans to rely more on ratings in determining bank capital requirements, in a rush to adopt Basel n. Going forward, the problem is to find reliable ways to express and value differ­ ences in risk on structured instruments. One way is for CROs to bond the quality of their work by subjecting it to effective independent review (Goodhart 2008a) and setting aside some of their fees in a fund from which third-party special masters or expedited civil judgments could indemnify investors for provable harm in instances where the independent reviewers find that negligence or misfeasance occurred. Alternatively, requirements that various intermediaries and fund man­ agers hold rated instruments. or received relief from capital requirements by doing so, could be dropped, and the category of NRSRO. which implies a govern­ ment sanctioning of the ratings process, could be abandoned. Did Mark-to-market Accounting Cause the Crisis? In the United States, accounting rules consist largely of generally accepted accounting principles (GAAP) and generally accepted auditing standards (GAAS). GAAP sets rules that constrain (but do not fully determine) the information systems that financial-institution managers may use to document their firm's economic condition and performance. GAAS sets procedures to be used in exam­ ining and verifying a firm's reports and records for compliance with GAAP. In a world of diverse and evolving circumstances, reporting systems inevitably convey options about where to book assets and liabilities and how to measure risk exposures and changes in value. The existence of these options and the ways that particular institutions use them to conceal losses is only imperfectly understood by outsiders. From a statistical point of view, accounting income and net worth are merely estimates of a firm's economic income and ownership capital. Because contemporary reporting systems only provide imprecise pOint estimates of the current and future earning power of a firm, careful users of accounting data must 136 The World Bank Research Observer. vol. 25, no. 1 (February 2010) acknowledge the existence of a margin for error. In different firms and in the same firm at different times. measures of income or net worth may be biased up or down and may vary substantially in their exposure to estimation error. GAAP has never required a firm to report interval estimates or to disclose helpful sup­ plementary information about the degree of imprecision or bias inherent in the methods its managers adopt. Observers who fail to acknowledge the many reporting options that fair-value accounting still allows blame it for causing the crisis. They argue that thin markets can cause a downward spiral in asset prices by encouraging institutions to sell troubled assets quickly and "forcing" them to take writedowns that under­ state the "true" value of the underlying assets. However. in creating and deepen­ ing the securitization crisis. the most serious impact of accounting rules did not emanate from the values chosen to represent various on-balance-sheet positions. It came instead from using off-balance-sheet extensions of commercial and investment banks to warehouse risks that, for reputational reasons, would have to be brought back onto the balance sheet if and when cumulative losses developed. Especially at large and complex financial institutions, individual managers have strong incentives to discover and to exercise reporting options that overstate their capital and understate their exposure to loss. This expands their ability to extract implicit subsidies that risk-taking can generate from implicit safety-net support. Concealment processes may be characterized as simple and complex forms of arbitraging the supervisory system. Whatever their other economic benefits, inno­ vative instruments are designed in part to create or expand concealment options. Moreover, trade associations and managers of systemically important institutions routinely use their political and economic clout to lobby standard-setting bodies for accounting rules that make it harder for government supervisors to monitor and to diSCipline their important exposures to loss. Under the historical-cost valuation principles in use during the 1980s, U.S. authorities allowed and even encouraged economically insolvent "zombie" insti­ tutions to hide their insolvency and to roll over their debts solely on the strength of government guarantees (Kane 1989). The rules did not make them record deterioration in market values, even of assets and liabilities for which perfect sub­ stitutes were trading regularly in an organized market. These rules (which still govern positions in a U.S. depository institution's "banking book") encouraged dis­ tressed or ruined financial institutions to endeavor to grow out of their insolvency by pursuing long-shot strategies that created stockholder value by shifting respon­ sibility for large loss exposures onto financial safety nets. In the United States a major purpose of moving to fair-value accounting was to require some of the developing losses at troubled financial firms to be recognized and resolved more promptly than in the past. It is mischievous for loss-making or Caprio. Demirgilc-Kunt. and Kane 137 undercapitalized firms to blame fair-value accounting for causing the trouble they encounter in rolling over their debt. At best, fair-value accounting is a messenger that makes private counterparties and officials charged with managing a coun­ try's financial safety net aware more quickly of their need to guard against the possibility that a particular firm (such as Bear Stearns) might be seeking to fund endgame "gambles for resurrection" at their expense. Realistically, careful exploi­ tation of the many reporting options that fair-value principles convey still allows a clever manager to greatly understate developing losses. By the time accounting evidence of insolvency can emerge, well-informed interval estimates of a firm's economic net worth would be deeply in the red. It is misleading for critics to claim that, because temporarily disorderly markets may overshoot equilibrium prices, fair-value accounting "forces" financial insti­ tutions to book paper losses that have no practical importance. First. under fair­ value principles, many portfolio positions are "marked to model" rather than to an actual transactions price. This creates incentives for managers of distressed firms to ask their quantitative staff to adjust model outcomes until they produce prespecified results. Personnel responsible for modeling decisions can do this by expanding or contracting either their samples of data points or the parameter space of their models to eliminate uncomfortable valuation outcomes. Attempts to defraud investors and creditors with models or assumptions that violate "prudent­ man" standards of negligence should be settled in the court system. Second, risk managers are free to move assets that are sensitive to changing credit spreads, either into off-balance-sheet entities or from their firm's "banking book" to its "available-for-sale" portfolio. As in the past, GAAP asks that impair­ ments in borrower credit that affect banking-book assets be translated into explicit additions to loss reserves. Increases in loss reserves reduce reported earnings and (if large enough) erode accounting net worth as well. While assets that are classi­ fied as available-for-sale have to be fair-valued, the writedowns do not pass through current earnings and charges taken directly against GAAP net worth are not incorporated into Basel measures of regulatory capital that supervisors were using to determine capital adequacy. Finally. in estimating GAAP capital in an environment of illiquid or panicked markets, banks are allowed to ignore impair­ ments implied by observable credit spreads simply by declaring that in manage­ ment's considered view the impairments are only temporary. Notwithstanding that mark-to-market accounting did not contribute to the crisis, U.S. authorities in early 2009 decided to ease these rules in an effort to relieve the banks of pressure. Notably this contrasts with the advice given to developing countries in past crises, where the IMF and a variety of international experts have advised developing country authorities to take account of market values in determining the viability of banks. 138 The World Bank Research Observer. 1'01. 25, no. 1 (February 2010) Did Financial Globalization Exacerbate the Crisis? The globalization of financial markets and institutions tends to heighten compe­ tition between alternative regulatory systems (Kane 1999, 2008). Although econ­ omists often treat regulation merely as a tax on institutional income, financial institutions understand that regulation is a service that generates benefits as well as costs. Regulatory benefits include improving customer confidence and conven­ ience. Supporting bank efforts to accumulate and exercise market power benefits banks, while resisting these efforts benefits society. Because regulation requires resources to produce, both the efficiency of its production and its pricing can vary. In a world in which financial markets are globalized, services that provide regu­ latory benefits are available both from foreign suppliers and from domestic regula­ tors of differently chartered firms. Rules and enforcement systems are continually tested and reshaped by changes in the net regulatory burdens that other jurisdic­ tions offer. This means that a worldwide market for regulatory services exists. Regulation is supplied competitively and accepted voluntarily to the extent that entry and exit opportunities exist for banks willing to incur the transaction costs of switching all or part of their regulatory businesses to another supplier. Competition has the benefit of lowering net regulatory burdens for the regulated financial institutions. While heightened international competition has tended on balance to displace poor systems of regulation by better ones, the maximum improvement in any country is limited by switching costs and by the level of best­ practice regulation that can be found elsewhere. In the current crisis, securitization helped to bring firms that were supervised in different regulatory systems into sharper competition with one another. In this environment, competition not only encouraged deregulation, it also and more importantly reduced the effectiveness of supervision. Securitization put pressure on particular regulatory enterprises to relax their scrutiny of innovative financial instruments as a way of defending or extending their bureaucratic turf. In the United States the worst offender was the Office of Thrift Supervision (OTS). The Treasury's Office of the Inspector General confirmed that the OTS allowed several thrifts to backdate capital infusions specifically to avoid showing a capital deficiency. Other banking supervisors helped their clientele by legitimizing cutting-edge ways to hide and transfer risk without fully exploring the threat that formally uninsured "shadow" affiliates (such as structured investment vehicles) and complex new contracting structures (such as doubly collateralized debt obli­ gations) imposed on individual-country safety nets. Whenever a regulator authorized an innovative entry by a foreign or nontradi­ tional firm. it also had to relax restraints that might make it hard for its tra­ ditional clients to compete with the new entrants. Institutions pressed politicians to make this happen promptly. In most countries, defects in accountability led Caprio, Demirgur-Kunt, and Kane 139 supervisors of commercial and investment banks to assess the risks of innovative instruments of risk transfer with less watchfulness than these instruments deserved. With structured securitizations, the SEC, banking supervisors, mortgage insurance firms, and investors jointly outsourced their duty of vigilance to apprai­ sers. accountants, and CROs. They did this despite knowledge of these firms' obvious conflicts of interest and outsized delays in recognizing problems or down­ grading distressed securities in past downturns (Portes 2008). While supervisors relaxed entry restrictions, they resisted the exit of domestically important com­ mercial and investment banks by standing ready to let unprofitable clients be sup­ ported by safety-net bans and guarantees. The goal of financial reform should be to induce nondiscriminatory and effi­ cient patterns of regulation and supervision. Regulators should be made accoun­ table not just for producing a stable financial economy, but for providing this stability fairly and at minimum long-run cost to society. In practice. this would require embracing market-based standards of supervisory performance designed to identify undercapitalized institutions promptly and to require them to shrink, raise more equity capital. or pay higher interest rates for their debt. The globaliza­ tion and information revolution that is underway in finance today makes it short­ sighted to require taxpayers to subsidize weak institutions and inefficient patterns of real investment. Regulatory efforts to respond to the globalization of financial institutions and markets have been led by the Basel Committee on Banking Supervision. This Committee meets regularly to discuss ways of harmonizing national standards for banking supervision. The Committee's stated objective is to eliminate perceived cross-country competitive inequalities and to improve financial stability by pro­ moting comprehensive risk management and consistency in regulatory standards across countries for multinational firms. Its major accomplishment is to negotiate the Basel I and II capital accords. Has Basel II Failed Already? The crisis has spawned a growing argument about the role Basel I may have played in causing the crisis and about whether Basel II, had it been implemented earlier. could have lessened the turmoil. This debate is critically important for developing countries, many of whom are on record as planning to adopt the revised capital framework. Basel I distinguished two types of capital: Tier One Capital (core capital, roughly the same as stockholder equity) and Tier Two Capital (which includes some hybrid forms of debt). It also defines a formula for risk-weighting categorized "buckets" of similar asset holdings and summing up the weighted values to form an aggregate measure of risk exposure called "risk- 140 The World Bank Rcsean'h Observer. vol. 25. no. 1 (February 2010) weighted assets" (R\N\). Banks are required to hold at least 8 percent of RWA in capital, at least half of which must be in Tier One (equity plus retained earnings). The weights employed in Basel I gave banks no credit for the extent to which they might have diversified or hedged the risks in their loan portfolio; risk was evaluated on a loan-by-Ioan basis, rather than at the portfolio level, taking into account covariances. The formulas also did not make any effort to account for operational, interest-rate, or exchange-rate risks, though market risks were incor­ porated later. Finally, Basel I failed to link the risk weights it applied to particular assets to the risk premiums that can be observed in loan markets. These weaknesses provided opportunities for arbitrage that contributed to the current turmoil: if regulatory demands for capital generated a compliance burden at a particular bank. its managers could eliminate this burden by selling or secur­ itizing a sufficient amount of assets. For example, under Basel L mortgages held on a bank's balance sheet were subject to a 50 percent weight, while securities backed by mortgages received only a 20 percent weight. No risk weights were assigned to loans that were sold to special-purpose bank-sponsored securitization conduits, which were regarded as "off balance sheet," or to short-term lines of credit with which sponsors supported these conduits. Proponents of Basel II argue that the new accord will ameliorate weaknesses in Basel L since Basel II is more granular and mitigates securitization incentives by reducing the capital charge for mortgages held on the balance sheet to 35 percent and by imposing a capital charge on short-term lines of credit. Nevertheless. long­ standing concerns about the Basel approach have been reinforced by the recent turmoil. Basel II-which was negotiated in 2004, but is still in the process of being implemented-was born out of widespread dissatisfaction with the obvious short­ comings in Basel 1. 11 The new agreement is much more complex than Basel I and rests on three mutually reinforcing pillars: (1) minimum capital requirements; (2) supervisory review of banks' capital adequacy; (3) strengthened market diSCipline of capital adequacy. Besides increasing the number of risk categories in pillar one. Basel II proposes to use a mix of statistical methods and expert opinion to track a bank's changing exposure to insolvency risk over time. It also envisions improved disclosure as a way to generate complementary market discipline on bank capital positions. However. Basel II does not improve on Basel L either in how it measures capital or in the arbitrary target ratios it sets: the definitions of Tier One and Tier Two Capital and the rules limiting their components 12 are unchanged; and the Committee continues to insist on the 4 and 8 percent minimum ratios of capital to risk-weighted assets without any rationale as to why either level is appropriate. The new accord sets out to make capital more sensitive to credit risk in one of two ways: through reliance on external credit ratings issued by rating organiz­ ations (the Standardized Approach) or through reliance on internal ratings based Caprio, Demirgiir-Kunt, and Kane 141 on banks' own risk models (the Internal-Ratings-Based Approach, IRB). The Standardized Approach requires banks to allocate their exposures to risk buckets and resembles Basel I, except that it incorporates a wider range of weights and asks countries to choose a set of external rating organizations and use their assessments of risk to determine country-level capital requirements. As of 2008, 105 countries' authorities have stated their intent to adopt Basel II (PSI 2008)­ and in its present evolution developing countries. lacking the data and skills for reliance on banks' models. would have to place greater weight on ratings in setting capital requirements. The IRB Approach makes use of banks' own internal risk systems for specifying minimum capital requirements. subject to the requirement that their internal models satisfy regulatory eligibility conditions. 13 Under pillar 2. Basel II grants national regulators substantial discretion over the implementation of these options: countries could end up with widely divergent levels of required capital. This would generate increased opportunities for regulatory arbitrage and under­ mine effective capital control. Indeed. standards under Basel II could cease to be global standards at all. which would defeat the original purpose of the accord. as well as the raison d'l'ire of the Basel Committee. Most importantly. recent events challenge the appropriateness of both the Standardized and IRB Approaches of measuring capital charges against credit risk. The use of ratings to set risk weights in the former encourages ratings inflation because ratings organizations face revenue-based incentives to relax ratings requirements. In addition. credit ratings do not confront the issue that capital requirements are supposed to address. Loan-loss reserves are tasked with accounting for anticipated losses. Capital requirements are intended to provide a buffer against unexpected risks. It does not make sense to use credit ratings to set capital requirements, since they convey no information about the volatility of an asset's return around its mean exposure to loss. Ratings may be useful for estab­ lishing loss reserves for particular assets, but they say nothing about how a bank's net worth or its portfolio of assets might vary in value with unexpected events. The amount of capital that must be set aside to achieve a particular target level of safety for a particular institution has to be linked explicitly to measures of the volatility of its earnings. The internal models employed by even the largest and most sophisticated market participants also failed to track risk accurately. The models proved inadequate in that they systematically underestimated the types of risks in complex securitizations that produced large losses and substantial downward revi­ sions in earnings. The Basel Committee undermined accurate modeling by decid­ ing to treat five years of data as an adequate sample span. This time period is too short to capture a full business cycle. Any serious attempt at risk modeling should have confronted the aberrant behavior of housing prices in the United 142 The WiJrld Balik Research Observer. 1101. 25, no. 1 (February 2010) States and other countries. as well as evolving features of the market (for example new types of borrowers, lower downpayments, new mortgages in which the buyer had the option of determining payments in the initial years. and so on). Thus this experience demonstrates that it is a mistake to assess risk mechanically by using models estimated with data from periods when important features were changing. In addition. complex financial models and datasets can be manipulated to provide desired outcomes. Although Basel II employs a more detailed categorization of credit risks. 14 it fails to address "liquidity" risk and reputation risk. both of which have proved important in the current turmoil. Once doubts emerged about the accuracy and reliability of ratings and accounting net worth. holders of maturing short-term liabilities refused to renew their funding. Feedback from this so-called liquidity risk intensified credit and market risk. Reputation risk encouraged lending insti­ tutions to rescue off-balance sheet shadow entities they had sponsored. even though sponsors were not contractually obliged to assume these losses. An impor­ tant reason the crisis spread rapidly from sub prime loans to other securitization structures. and that the turmoil has persisted, is that the subprime meltdown revealed that serious contracting weaknesses existed at every stage of the process of securitized risk transfer. Inadequate documentation of underlying mortgages, the daunting complexity of securitized structures for allocating and reallocating cashflows from questionable loans. and the opacity of off-balance-sheet vehicles that purchase securitized instruments led investors to seek comfort not from an understanding of underlying cashflows. but from credit enhancements and the behavior of dealer spreads. Neither regulators nor market participants knew to whom risks were formally allocated, let alone on whom potential losses might finally fall. The disclosures envisaged in pillar three of Basel II have a long way to go before they can claim to unravel these issues. Finally. while Basel II assigns to national regulators the responsibility for moni­ toring and controlling insolvency risk. it does not develop any protocols for pre­ venting financial institutions from becoming insolvent. nor does it impose requirements for prompt corrective action (PCA) on supervisory authorities. The accord will eventually have to benchmark a pattern of actions that home and host authorities should take as the capital position of a client institution slips deeper and deeper below acceptable standards and is not promptly replenished. Detection and resolution of impending insolvencies is crucial. The central problem in financial regulation is to make sure that even in politically and econ­ omically stressful circumstances. regulators have robust incentives to protect tax­ payers by identifying troubled banks and forcing them to recapitalize before their capital can become exhausted. Basel's arbitrary "risk" -weighting process has already failed. This failure under­ scores the importance of enforcing PCA obligations and monitoring leverage per Caprio, Demirgiir-Kl1nt, and Kane 143 se. Even in the midst of the crisis, large financial institutions managed to keep from violating Basel I's risk-weighted targets by gaming authorities with sophisti­ cated (but faulty) risk-transfer techniques. For this reason during the turmoil, markets ignored RV\k\ ratios and focused instead on the more transparent ratio of assets to tangible net worth as the primary indicator of financial strength. We think that this is excellent advice for developing countries to follow, as argued below. Because the volatility of and correlation between returns on different assets tend to surge in crises. the risks that modern institutions take cannot be captured in a static formula. no matter how complex it might be. When static rules are also complex, they reduce transparency and generate loopholes that foster regulatory arbitrage and support acts of supervisory forbearance. To be effective, prudential regulation must be adaptive and it must combine supervisory stress tests with market oversight. To track the changing importance of particular risks in a timely fashion, supervisors therefore should use market signals. Current market turmoil underscores the inadequacy not only of Basel's static formulas, but also the dangers of taking accounting statements at face value, both of which are of great relevance for countries regardless of their income level. Safety-net subsidies increase effective leverage, weaken market discipline. and reduce the exposure of formally at-risk loss bearers in ways that render the usual accounting disclosures ineffective. Subsidy-induced innovations can only be coun­ tered by conscientious supervision. Supervisors must not only draw on-but help to develop-informative market signals, such as those imbedded in the prices of credit default swaps and subordinated debt. Deal making in these markets incor­ porates timely estimates of changing default probabilities and loss exposures. However. prices in these markets can only be a strong source of discipline under two conditions: (i) market participants do not expect to be bailed out when trouble develops; (ii) investors have access to regular flows of high-quality infor­ mation (Barth. Caprio, and Levine 2006). A crucial element in limiting expected bailouts is to assign political accountability for measuring and defending safety­ net costs to regulators and elected officials. Lessons for Developing Countries Although this crisis began in industrial countries and was associated in the United States with sophisticated financial innovations, both this crisis and its many antecedents offer a number of lessons for developing countries. When a crisis is in process, interest in drawing lessons from the past surges. Unfortunately, officials and market participants alike have vested interests in promoting views that are most beneficial for their reputation. balance sheet, or both. This is why 144 The World Bank Research Observer, vol. 25, no. 1 (February 2010) popular theories of past crises zero in on the bad behavior of a few convenient scapegoats. However, crises rarely result from the corrupt acts of a few greedy individuals or from a handful of isolated regulatory mistakes. Crises have their roots in longstanding structural flaws in the way that financial institutions and government officials interact. With rare exceptions, crises in developed and devel­ oping countries alike are caused or exacerbated by perverse incentives that make it worthwhile for politicians, regulators, and the private sector to ignore mount­ ing danger signals until it is too late to avoid a widespread meltdown in asset values. Modern Crises Are Often Revealed by Identifiable Shocks that End Booms or Bubbles in Important Macroeconomic Sectors, but the Underlying Distortions Were Building up for a Long Time The meltdown that began in 2007 is not the first time that financial institutions, in taking advantage of regulatory loopholes, engaged in reckless risk-taking that fueled a long-lasting bubble in asset prices that in one way or another had to burst eventually. Extraordinary risk-taking is easier to disguise and rationalize during bubble or boom periods. Increasing leverage based on unsustainable surges in the price of residential, commercial property, or both, on the one hand. and corporate stock, on the other, featured prominently in an end-of-the-century spate of crises: in Japan (1990s), Malaysia (mid-1980s). Mexico (1994). Sweden (1991-94), and East Asia (1997-98). The current crisis in Ireland and several Eastern European countries occurred without sophisticated financial products, and the underlying distortions-an unsustainable housing boom in the former, and widespread currency mismatching even at the household level in the latter­ were not difficult to detect. Historically, wherever a banking industry has existed, economic booms and asset bubbles have often preceded financial crises. Demirgii<;-Kunt and Detragiache (2005) survey a large literature that shows that the likelihood of crises increases with the strength and duration of economic booms and that banking crises are occasioned by shocks in asset prices. output. terms of trade, and interest rates. In addition. the same scholars (2002) assemble convincing evi­ dence that the character of a country's financial safety net plays a critical role in encouraging institutions to make themselves vulnerable to the particular shock that brought each crisis on. This is not to say either that crises would not occur in the absence of a safety net or that financial safety nets should be dismantled. 15 It is accurate to say that financial crises have become more frequent and more expensive (in terms of losses per dollar of deposits) as safety nets have expanded. 16 By permitting losses to spread to taxpayers rather than being borne solely by contracting parties, safety Caprio. Demirgu(:-KlInt. and Kane 145 nets displace market discipline (Calomiris 1995). By making it easier to attract deposits, safety nets encourage private parties to lever themselves more exten­ sively; and increasing leverage shifts more and more of the downside onto the national safety net. Financial Deregulation Is Often Blamed for Causing Crises, but the Fact and Character of Deregulation Is Itself Shaped by the Ways that Governments and Regulated Institutions Interact As financial regulations were relaxed in the 1970s and 1980s, and increasing reliance was placed on prudential supervision (with little accountability). the fre­ quency of crises did increase (Demirgiic;-Kunt and Detragiache 1999). However. deregulation does not necessarily provide greater opportunities for shifting private risk exposures onto the safety net. This happens only when authorities fail to adapt their systems of insolvency detection and resolution appropriately. In prin­ ciple. relaxing controls on interest rates, charter powers, and portfolio structure promised to improve the ability of banks to foster economic growth and economic justice. But coupling deregulation with inadequate supervision of leverage and asset quality is a recipe for disaster, because "desupervision" allows safety-net subsidies to be extracted by doubling and redoubling risks. 17 Blaming the current turmoil on financial deregulation without mentioning the role of deficient over­ sight suggests that rules-not incentives-are at fault (Stiglitz 2008; Krugman 2008). During the period of deregulation. most industrialized countries intro­ duced many new rules. The imbalance between the attention paid to rules and incentives is well illustrated by the 2004 Basel II agreement which devotes 16 pages to issues of market discipline and 225 pages to spelling out formulas and strategies imbedded in pillar one and options for national discretion authorized in pillar two. Over Time, Regulation-induced Innovation Leads to Progressively More Complex and Less Transparent Forms of Risk-shifting Financial crises are often driven by breakdowns in innovative financial instru­ ments or arrangements designed to exploit loopholes in a country's risk controls. Recent examples of risk-shifting include: overexposure to foreign exchange risk (Chile. 1981; Mexico. 1995; Nordic countries, early 1990s; Turkey, 1994; East Asia. 1997); aggressive lending to politically important foreign markets (in the developing country debt crisis of the 1980s); complex deal-making (in the securi­ tization crisis). In seeding the current crisis, institutions abused derivative instru­ ments-whose existence had been rationalized as vehicles that would diversify and hedge risk-in order to magnify safety-net loss exposures. Abusive trading of 146 The World Bank Research Obsen'er. vol. 25. no. 1 (February 2010) derivatives instruments fueled the Orange Country fiasco (Jorion 1995) and the growth of credit default swaps (CDS). Because Institutions Can Count on Crisis Resolution to Be Mismanaged, Safety-net Subsidies Flow to Institutions Willing to Risk Insolvency Walter Bagehofs classic policy advice for managing liquidity during a systemic crisis is for the central bank to lend freely to solvent banks-but to minimize safety-net subsidies, the loans are to be made at a penalty interest rate and only on good collateral. Put differently, his advice is for governments to avoid lending to insolvent banks at alL even on good collateral. and certainly not at below-market interest rates. Gnfortunately, modern governments pay only lip service to this principle. Supervisory authorities find it hard to mobilize the political and budget­ ary support needed to follow the Bagehot strategy. In their study of 12 recent crises, Kane and Klingebiel (2004) found that all but one country adopted a crisis-management strategy that combined blanket guarantees with extensive and immediate liquidity support for insolvent institutions. 18 Advocates of using liquidity injections to halt a systemic crisis argue only that sweeping guarantees and extensive liquidity support can stop the panicky flight of depositors and other institutional creditors to less risky venues. But this begs the question of whether social costs and adverse distribution effects could be reduced by following an alternative strategy (Kane 2001). Incentive Conflict not only Explains How a Particular Crisis Develops, but How the Manner in which a Crisis Is Resolved Affects the Frequency and Depth of Future Crises Even in the midst of a financial crisis, it is inefficient to set aside long-term goals completely. Providing extensive liquidity support and guarantees to insolvent insti­ tutions subsidizes "gambles for resurrection" and strengthens the risk-shifting incentive schemes that spawn crises and guarantee their recurrence. Without incentive reform, short-sighted methods of crisis resolution create the expectation that they will be used again when the next crisis inevitably arrives. This expec­ tation undermines market discipline and financial stability in future periods. The short-term benefits of liquidity injections have been oversold. Such policies seldom actually speed the recovery of a nation's real economy from a banking crisis or lessen the decline in aggregate output. Honohan and Klingebiel (2003) measure the impact of different crisis management strategies on the ultimate cost of resolving distress in 40 different financial crises. They find that blanket guaran­ tees, open-ended liquidity support. and regulatory forbearance significantly increase the ultimate fiscal cost of resolving crises. They also find that spending Capriu. Demirflil~-Kunt. (md Kane 147 more to support distressed institutions does not speed the recovery. Instead. pro­ viding liquidity support for insolvent institutions tends to prolong a crisis. It does this by distorting risk-taking incentives so extensively that sound investments and healthy exits are delayed and additional output loss is generated. as discussed in Demirgih;-Kunt and Serven (2010). Past Crises Provide Important Lessons, but They Fall on Deaf Ears History provides clear lessons about how to minimize the frequency and cost of financial crisis. The refusal to embrace these lessons underscores the existence of persistent defects in the incentives that govern the ways in which politicians. reg­ ulators. and financial institutions interact. Authorities routinely underestimate the frequency and depth of crises. When crises do occur. they prefer to resolve the conflicting pressures under which they must function by treating troubled insti­ tutions generously, claiming that it is their duty to minimize potential short-term contagion at all cost. Underinvesting in crisis preparedness implants a preference for improvisation. But improvisation leads to inefficient and myopic solutions. Instead we argue that there are clear principles for regulatory reform derived from recent crises and aimed not at reallocating regulatory and supervisory auth­ ority. but at establishing incentives that would lead supervisory authority and market forces to operate more effectively. We stress that the current crisis exempli­ fies not just the limits of market discipline. but the power of government-induced incentive distortions-and the limits of official supervision as commonly prac­ ticed. The failure of private parties to exercise sufficient due diligence was rooted in the failure of government supervisors to challenge decisions made by private accountants and eROs. They neglected their duty of examining and publicizing the implications that these decisions might have for safety-net loss exposure. By tolerating a decline in transparency. supervisors made it difficult to recognize and price the risk expansion not only for themselves, but also for the market partici­ pants. Many developing countries share this trait of a low degree of financial sector transparency, and the current crisis demonstrates not only the importance of increasing it, but the continuous nature of the battle to do so as well. Authentic Reform Must Address the Contradictory Political and Bureaucratic Incentives In the recent crisis these incentives led regulators and supervisors first to out­ source their due diligence and then to resolve the crisis in inefficient ways. Incentive reform is politically difficult because existing defects in supervisory incentives did not come about by accident. They reflect the political preferences of regulated institutions and other politically powerful market participants. No 148 The World Ballk Research Observer. voL 25. no. 1 (/+bruary 2010) matter how drastically a proposed reform may redistribute supervisory authority, unless it also establishes accountability and transparency for the costs of safety­ net management. effects will prove more ostensible then real. For this reason, our reform proposal focuses on improving the chain of incentives under which market discipline and official supervision operate, While we have many other recommen­ dations suitable for industrial countries, here we concentrate on those that are most relevant for developing country authorities. Lender Reform. Compensation for loan officers must be linked to long-term performance rather than to short-term profits. Whenever authorities see compen­ sation in the financial sector growing rapidly they must suspect that great risk-taking and risk-shifting is occurring. Governments can reinforce (rather than undermine) market discipline by assisting in the dissemination of information about contract evolution and by encouraging development of better information systems. In developing countries, accurate information is often quite scarce, and officials should consider ways to encourage or compel greater information disclos­ ure by intermediaries. eRG Reform. CRO reform, at the top of the industrial country agenda, will have a significant impact on developing countries, and should incorporate two main elements: (i) withdrawing government blessings from their work and (ii) improv­ ing CRO accountability for ratings decisions. Because the NRSRO designation pro­ vides an explicit government blessing and introduces barriers to entry into the ratings business, it should be eliminated. Most importantly, references to ratings should be removed from all SEC and bank regulations, including Basel II. Government rules that rely on CRO ratings reduce investor incentives to conduct sufficient due diligence before making investments-indeed it is likely that far less securitization would have occurred if those managing other people's money were not protected by the requirement that they hold highly rated instruments. At the same time, such rules reduce the accountability of government regulators and supervisors for neglecting their duty of oversight. By outsourcing due diligence to CROs, regulators shift the blame for the safety-net consequences of ratings mis­ takes away from themselves. Clearly, CROs must be made more accountable for the quality of ratings they provide. Individual CROs can only recover the damage their brand has suffered by taking responsibility for their mistakes. For securitized claims, this could be done by requiring CROs to publish an ex ante margin for error with each credit rating and to publish the data used in establishing complex claims so that outsiders can fact-check their inputs and challenge and improve the modeling. This provides a nonbureaucratic way to subject ratings decisions to effective independent review (compare Goodhart 2008a). Caprio. Demirgiit;-Kunt. and Kane 149 Reform of Basel II. Because credit ratings and sophisticated risk-management models have been discredited, it is clear that Basel II must be reworked signifi­ cantly. Some have suggested that the Basel Committee acknowledge that risk­ management standards have changed so much that it is necessary to move directly to a new agreement: Basel III. Ideally, pillar two of the new system would include a simple leverage requirement and PCA rules for structured early inter­ vention into the operations of loss-making financial institutions. This would further enhance supervisory and regulatory accountability. Framers of the new accord should find ways to use market signals from CDS and subordinated-debt markets to estimate and disclose how regulatory decisions in different countries affect safety-net costs in other countries. We think that the Basel Committee's drive to set ever more sophisticated ways to set risk-weights is a mistake, as revealed in the current crisis. Instead, they should abandon this task and adopt simple maximum leverage ratios, which include all on- and off-balance-sheet items. Developing country authorities would have a simple system that they could implement easily. and they may want to proceed to this model on their own if Basel tarries. Accordingly authorities should pay much less attention to the parts of Basel II that rely on credit ratings and models. Reform of Government Accountability. For individual countries, systemic crises are infrequent events. This means that incumbent policymakers seldom have direct experience in working through the crisis-driven stresses generated by lobbyists for insolvent institutions. While improvisation works for well-practiced jazz musi­ cians, the current crisis amply illustrates that, in crisis management. it deterio­ rates into methods of wholesale financial-institution rescue. Crisis-management decisions are full of errors because they are made in stressful circumstances by unpracticed policymakers who feel that their career in government service is on the line (Kane and Klingebiel 2004). While even high-income-country policy­ makers make mistakes, developing country authorities often have less skilled staff on which they can rely, due not just to a skill issue in the country but also by compensation differentials when compared with the private financial sector. Where skills are scarce, advanced planning is even more important. Decisions about which institutions to rescue and how to save them tend to follow the path of least resistance, This invariably entails bailing out deeply insolvent institutions and extending the dimensions of the safety net in a way that, by further subsidiz­ ing risk-taking. sow the seeds of future crises. To avoid short-term ism , crisis-management decisions should be made in an open debate outside of an actual crisis. Accountability would be improved by requiring that regulators establish and regularly test a well-publicized benchmark plan for crisis resolution. The events of the current crisis confirm that not 150 Thc Wodd Bank Resemrh Obsfrl'cr, vol. 25, no. 1 (trbnwry 2010) planning for crises prolongs and deepens the disruption by tempting regulators to subsidize loss-making institutions at taxpayer expense. The damage a crisis works on a country's financial sector and its real economy can be reduced by taking actions that promptly estimate and allocate losses during the early stages of a crisis. The critical first step is to allow time for forensic accountants to separate the hopelessly insolvent institutions from potentially viable ones. Authorities need such information to deal with insolvencies in ways that protect taxpayers and avoid subsidizing further risk-taking. In and out of crisis, regulators need to draw on market signals to help them to track risk. A promising way to do this is to require at least the largest banks to issue at regular intervals a series of CDS. to issue large amounts of credibly unin­ sured subordinated debt. or both. 19 Transactions prices for these instruments can be incorporated into information systems that the supervisors can use to improve their assessments of safety-net loss exposures. This is because holders of these instruments would apply the market discipline that pillar three of Basel II seeks to harness. Most importantly it is necessary to strengthen the safety net by making auth­ orities more accountable for its costs. Interactions between large banks and super­ visory agencies could be made more transparent if both were required to estimate and to disclose the amount of these subsidies on their books. This requires the development of a system of fair-value accounting for intangible safety-net subsi­ dies. If the probability of bailouts can be reduced. regulators can use CDS prices to estimate individual-institution and aggregate values of safety-net subsidies (Kane 20(8). For example the spread at which CDS trade can be used to strip out the value of the safety net subsidies imbedded in that firm's market capitalization. In government enterprises, decision-making horizons could be lengthened if employment contracts included a fund of deferred compensation that heads of supervisory agencies would have to forfeit if a crisis occurred within three or four years of leaving their office (Kane 2(02). Calomiris and Kahn (1996) show that such a system worked well in the 19th century Suffolk banking system. where claims to deferred bonuses were paid only after losses had been deducted. The public embarrassment of having to forfeit compensation. in addition to the finan­ cial penalty. would incentivize top supervisors to use market signals more effi­ ciently and help them to resist political pressure to bail out insolvent firms. The less transparent a country, and the greater the degree of corruption, the more it needs to worry about the financial incentives facing financial sector regulators. None of these accountability enhancements-better crisis preparedness, greater use of market information to track risks and subsidies, publicizing estimates of safety-net subsidies. or offering deferred compensation-would be costly to implement. Yet current and past crises suggest that the return to implementing them would be enormous. Caprio. Demirgiif-Kutlt. am/ Kane 151 Lastly. developing countries need to continue to build the infrastructure that a modern financial system requires, most importantly a good information environ­ ment led by sound accounting and auditing, efficient contract enforcement. and clear incentives. As noted above, improving the information and incentive environment needs to be a continuous effort, as financial sector players in rich and poor countries alike regularly contrive to conceal risk-taking. With weaker checks and balances in government. often critical shortages of supervisory skills. and a generally less independent media, financial sector oversight is more prone to breakdowns in poorer countries. Although the specific methods of avoiding regulation that were part of structured securitization were more confined to high income countries. the general lessons on regulatory arbitrage and the need for continued supervisory vigilance are common to all. Developing country auth­ orities can let industrialized countries experiment with new instruments but they need to ignore pressures to adopt the latest regulatory and product innovations until they have been amply tested. Notes Gerard Caprio Jr. is Professor of Economics, Williams College, 305 Schapiro Hall, 24 Hopkins Hall Drive. Williamstown. MA 01267, USA; email address: Gerard.caprio@williams,edu. Ash Demirgii9-Kunt is Chief Economist. Financial and Private Sector Development Network and Senior Research Manager, Development Research Group, the World Bank; email address: Ademirguckunt@worldbank.org. Edward J. Kane is Professor of Finance at Boston College; email address: edward.kane@bc.edu. The authors are grateful to Stijn Claessens, Enrica Detragiache, Ramon DeGennaro, Robert Eisenbeis. Alain Ize, Emmanuel Jimenez, Philip Keefer, Joseph Mason, James Moser, Roberto Rocha. Luis Serven, Tressel Thierry. James Thomson, Dimitri Vittas, and three anonymous referees for valuable comments, Their findings, interpretations, and conclusions do not represent the views of the World Bank, their Executive Directors, or the countries they represent. l. From 1999 to 2007 the only Significant financial crisis occurred in Argentina. a country that has experienced numerous crises throughout its history. This crisis had major consequences for only a single neighboring country (Uruguay). 2. See Mason and Rosner (2007) and Brunnermeier (2009) for a description. 3. Recently one investment bank sold some of its super-senior holdings for only 22 cents on the dollar, and at least one critic claimed that their true value was closer to zero (Roubini 2008). 4. The value of subprime mortgages originated in the United States shot up from $190 billion in 2001 to $600 billion in 2006. Much of this growth was fueled by securitization: as a percentage of subprime mortgage originations, the volume of sub prime issuance rose from 50 to 80 percent over the same interval (Economist May. 2008). 5. The report by the Federal Reserve Bank of Boston (2003). "Closing the Gap: A Guide to Equal Opportunity Lending", contains a variety of statements urging banks to make every effort to facili­ tate such lending. such as by being aggressive on lending standards. appraisals. and loan to income ratios (pp. 22-6). 6. Also see Keys and others (20] 0) for consistent results. 7. Joint Statement of the Shadow Financial Regulatory Committees of Asia, Australia-New Zealand, Europe, Japan, Latin America, and the United States on "Making Securitization Work for Financial Stability and Economic Growth" (2009). 152 The World Bank Research Observer. vol. 25. no. 1 (February 2(10) 8. Prior to World War I. corporate default rates were significant and ratings agencies did seem to supply some modest function. though even the most authoritative study (Hickman. cited in Sylla. 2(01) fails to lay to rest the critique that ratings might at most have been modestly anticipating declines in performance and did not do demonstrably better than market indicators. In the post­ World War rr period through the early 1970s. the default rate on corporate bonds declined to only about 1% percent. so ratings were not critical to investors' decisions. 9. If a corporation that has issued a bond gets into trouble. it can make a variety of adjustments that will enable its bond to perform in line with ratings. A fixed pool of securities cannot adjust: it remains a claim on cashflows from a variety of sources. 10. It is interesting to speculate on how many fund managers would have successfully defended their position with the claim that they did not know what they were investing in (but neither did their competitors) or that they knowingly invested in pools of high-risk mortgages confident that housing prices would rise endlessly. 11. The Basel Committee. created by the G-IO central banks in 1974. now includes Argentina. Australia. Belgium. Brazil. Canada. China. France. Germany, Hong Kong SAR. India. Indonesia. Italy, Japan. Korea. Luxembourg. Mexico. the Netherlands. Russia. Saudi Arabia. Singapore. South Africa. Spain. Sweden. Switzerland, Turkey, the United Kingdom, and the United States. The Committee itself notes that Basel II arose in response to the growth of securitization. giving the impression that this growth was an exogenous development rather than at least in part a response to Basel I and its "loan-by-Ioan" approach to assessing a firm's overall risk exposure. 12. The Committee continues to regard subordinated debt as inferior to shareholder equity, but only because such debt is limited in its maturity. This neglects the better incentives that debt estab­ lishes in monitoring risk. Whereas equity value varies directly with the returns promised from increasing risk, debtholders' returns can be harmed by increasing risk. This leads fir~1-loss debt­ holders to exert a more conservative influence on corporate governance than equity holders. 13. The IRB Approach has two variants: I;oundation and Advanced. Under the Foundation Approach, instead of relying on external assessments of creditworthiness, banks are able to use their own estimates of probabilities of default for each borrower. These borrower-specific factors. supplied by each bank, are then combined with supervisory-determined estimates of loss given default. exposure at default, and effective maturities to arrive at regulatory risk weights. If a bank satisfies the stricter eligibility conditions to qualify for using the Advanced Approach. then it can place even greater reliance on internal credit systems by using not only their estimates of the probability of default but also their own estimates of loss given default. exposure at default. and effective maturities. 14. Under pillar one. Basel II combines the evaluation of capital adequacy for credit risk. oper­ ational risk (defined as the risk of loss resulting from inadequate or failed internal processes, people. and systems, or from external events). and market risk (de1ined as the risk of losses in on- and off-balance-sheet positions arising from market movements in interest rates. foreign exchange, and securities or derivatives prices) with incentives for banks to invest in better risk management pro­ cesses to qualify for the discretion allowed under each more-advanced approach. Interest-rate risk in the banking book is relegated to the supervisory review process of pillar two. 15. Early bubbles occurred before safety nets commonly were extended to the financial sector. During the South Sea bubble, John Martin famously said: "When the rest of the world are mad, we must imitate them in some measure." The size of the equilibrium "measure" is what the safety net influences. Unlike modern securities firms and commercial banks, Martin could not have hoped that the government would rescue his bank. since the Bank of England had not at that time begun to operate as a central bank. 16. Costs of resolving crises has risen significantly over time: in the late 19th and early 20th century, they ran at about 2 percent of GOP; in modern times they have averaged five to six times this figure, with some cases reaching the range of 20 to 50 percent of GOP. Caprio. Demirgiir-Kullt. and Kane 153 17. Examples of deregulation with little supervision include Malaysia in the 1970s-which fea­ tured a buildup of real estate exposures to 50 percent of bank lending-and the U.S. Savings and Loans. which were allowed to gamble in high risk investments even after they were insolvent. 18. The crises studied occurred in: Argentina. 2001: Ecuador, 1998; Finland, 1991; Indonesia. 1997: Japan, 1991; Korea, 1997; Malaysia, 1997: Mexico, 1994; Russia. 1998; Sweden. 1991: Thailand. 1997; and Turkey. 2000, Only Sweden refused to supply extensive liquidity support to insolvent institutions, 19. As Calomiris (1997) and Evanoff and Wall (2001) explain. expectations of implicit support would contaminate the signal. References The word processed describes informally reproduced works that may not be commonly available through libraries, Altman. Edward. and Herbert Rijken. 2006. '1\ Point-in-Time Perspective on Through-the-Cycle Ratings." Finallcial Analysts Journal 62(1): 54-70. Barth. James R.o Gerard CapriO, Jr. and Ross Levine. 2006. Rethinking Bank Regulation: Till Angels Govern. New York: Cambridge University Press. Brunnermeier. Markus K. 2009. "Deciphering the Liquidity and Credit Crunch 2007-2008." Journal of Economic Perspectives 23(1): 77 -100. Calomiris, Charles. 1995. "Financial Fragility: Issues and Policy Implications." Journal of Fillancial Services Research 9:241-57. ___. 1997, The Postmodem Bank Safety Net. Washington, DC: AEI Press. Calomiris, Charles, and Charles Kahn. 1996. "The Efficiency of a Self-Regulated Payments System: Learning from the Suffolk Banking System. Journal of Money. Credit and Banking 28(4): 766-97. Dell' Ariccia, Giovanni. Deniz Igan, and Luc Laeven. 2008. "Credit Booms and Lending Standards: Evidence from the U.S. Subprime .Mortgage Market." IMF Working Paper WP 08/106. Demirgu<;-Kunt, Ash, and Enrica Detragiache. 1999. "Financial Liberalization and Financial Fragility." In B. Pleskovic. and J. Stiglitz, eds., Proceedillgs of the World Bank {1nnual Conference on Developmellt Ecollomics. Washington, DC: World Bank. ..._ _. 2002. "Does Deposit Insurance Increase Banking System Stability? An Empirical Investigation." Journal of Monetary Economics 49(7): 1373-406. ___. 2005. "Cross-Country Empirical Studies of Systemic Bank Distress: A Survey." National Institute Economic Review 192:6883. Demirgu<;-Kunt, Ash, and Luis Serven. 2010. '1\re All the Sacred Cows Dead? Implications of the Financial Crisis for Macro ami Financial Policies." World BaTIk Research Observer. Evanoff. Douglas. and Larry Wall. 2000. "Subordinated Debt and Bank Capital Reform." Federal Reserve Bank of Atlanta Working Paper 2000-24, November. Federal Reserve Bank of Boston. 2003. "Closing the Gap: A Guide to Equal Opportunity Lending." (http://www. bos.frb.org! commdev I commaffl closingt. pdf J. FSI (Financial Stability Institute). 2008. "2008 FSI Survey on the Implementation of the .'Jew Capital Adequacy Framework in non-Basel Committee Member Countries: Summary of Responses to the Basel II Implementation Survey." (http://www.bis.org/fsi/fsiop2008.htm). Goodhart. Charles. 2008a. "How, If at All, Should Credit Rating Agencies (CRAs) Be Regulated?" Unpublished manuscript. London School of Economics, July 7. Gorton. Gary, and George Pennachi. 1995. "Banks and Loan Sales: Marketing Non-Marketable Assets." Journal of Monetary Eco1lomics 3 S( 3): 389-411. 154 The World Blink Resclll'ch Obsem'l; vol. 25. 110, 1 (February 2010) Honohan. Patrick. and Daniela Klingebiel. 2003. "The Fiscal Cost Implications of an Accommodating Approach to Banking Crises." Journal of Banking and Finance 27:1539-60. IMF (International Monetary Fund). 2009. Global Financial Stability Report: Responding to the Financial Crisis and Measuring Systemic Risks. Washington. DC: International Monetary Fund. Jorion. Philippe. 1995. Big Bets Gone Bad: Derivatives and Bankruptcy in Orange County. The Largest Municipal Failure in U.S, History. San Diego: Academic Press. Kane, Edward J. 1977. "Good Intentions and Unintended Evil: The Case Against Selective Credit Allocation." Journal of Money Credit and Banking 9:55-69, ___, 1989. The S&L Insurance Mess: How Did it Happen? Washington. DC: Urban Institute Press, ___, 1999. "How Offshore Financial competition Disciplines Exit Resistance by Incentive­ Conflicted Bank Regulators," Journal of Financial Services Research 16(2/3): 265-91. ___, 2001. "Dynamic Inconsistency of Capital Forbearance: Long-Run vs, Short-Run Effects of Too-Big-to-Fail Policymaking," PacifiC-BaSin Finance Journal 9:281-99, 2002. "Using Deferred Compensation to Strengthen the Ethics of Financial Regulation," Journal of Banking and Finance 26: 191 9 - 33. ___, 2008, "Ethical Failures in Regulating and Supervising the Pursuit of Safety-Net Subsidies," Processed, Boston College, Kane, Edward. and Daniela Klingebiel. 2004, ''Alternatives to Blanket Guarantees for Containing a Systemic Crisis." Journal of Financial Stability 1: 31-63, Keys, Benjamin, Tanmoy Mukherjee, Amit Seru, and Vikrant Vig, 2010. "Did Securitization Lead to Lax Screening? Evidence From Subprime Loans," Quarterly Journal of Economics 125(1), Krugman. Paul. 2008, "Partying Like It's 1929," New York Times, March 21. Mason. Joseph R.. and Josh Rosner, 2007, "Where Did the Risk Go? How Misapplied Bond Ratings Cause Mortgage-Backed Securities and Collateralized Debt Obligation Market Disruptions," (http:// ssrn,com/abstract= 102 7475). Partnoy, Frank. 200 L "The Paradox of Credit Ratings." University of San Diego Law and Economics Research Paper 20. Portes. Richard. 2008, "Ratings Agency Reform," Vox Website (http://www.voxeu,org). posted January 22. Posen, Adam S, 2006, "Why Central Banks Should Not Burst Bubbles." International Finance 9(1): 109-24, Roubini, Nourie!. 2005, "Why Central Banks Should Burst Bubbles." Processed. (http://www,cfr.org/ publication! 109 8 8/internationaUinance,html), 2008, "Super-Senior Tranches of CDOs are Worth Much Less than 22 Cents on the Dollar: Another Ponzi Scheme of 'Selling' Toxic Garbage with More Leverage," Roubini Global Monitor. July 29, Stiglitz, Joseph, 2008. "Unfettered Markets Do Not Lead to Societal Well-Being," (http:// economistsview,typepad.com/economistsview/2008/02/joseph-stiglitz,html), Sylla, Richard, 2001. '~i\ Historical Primer on the Business of Credit Ratings." Paper presented at the conference, "The Role of Credit Reporting Systems in the International Economy." Washington, DC: The World Bank, Taylor. John B. 2009, Getting OJ{ Track: How Government Actions and Interventions Caused, Prolonged, and Worsened the Financial Crisis, Stanford: Hoover Institution Press. Caprio, Demirgu(:-Krmt. and Kane 155 r j. .-. other key staple foods also rose sharply in 2008. The price increases for cassava (161 percent), matooke (54 percent), and sweet potato (44 percent) were particu­ larly steep in Northern Uganda, and as a result households in this area were worst affected. Moreover, since Northern Gganda had the largest concentration of the poor to start with, this increase in food prices is likely to have worsened regional inequalities. The fact that the distributional impact of rising food prices affects poor house­ holds partly based on where they live is also clearly illustrated by recent work on Mozambique. Arndt and others (2008) using simulations based on nationally representative data show that urban households and those who live in the south of the country are adversely affected by rising food prices, while those in the north and center of the country, especially better-off farmers. typically benefited. Prior to concluding our review of poverty impacts it is worth emphasizing two points. First, we focused our discussion on the impact on consumption poverty mirroring the focus of recent analytical work on the food price spike. Yet the impact on worsening malnutrition. school drop-outs. and adverse gender conse­ quences of households struggling to access suffiCient food is just as. if not more. important, as the early evidence from Liberia and Sierra Leone show. Second, as has been shown, impacts differ substantially between geographical areas within a country, which poses a dilemma for policymakers. On the one hand. the desire to help households cope with the increase in food prices may lead policymakers to implement or expand safety net interventions in the hardest hit areas. On the other hand. these hard-hit areas may not necessarily be among the poorest in the country. Hence, the effectiveness of the country's overall poverty reduction strategy may be jeopardized if public resources are diverted from the poorest areas to less poor areas to address the food crisis. and trade-off's between immediate responses and medium-term measures to reduce chronic poverty may emerge. PoLicy Responses to Rising Food Prices in Sub-Saharan Africa In this section we review the results of a survey on the policy responses to the sharp rise in food prices in 2008 before examining two specific instruments--Iow­ ering foodgrain taxes and expanding food-based safety nets--in more depth. Policies in Surveyed Countries Countries vary widely in the type of policies or programs they are able to intro­ duce. or scale-up, to respond to rising commodity prices. As discussed earlier there are essentially three broad types of policies that are used to respond to a 166 The World Bank Research Observer. vol. 25, no. I (February 2010) sharp rise in food prices. The first set includes policies which attempt to stabilize rising prices by affecting the aggregate supply and demand balance. These include reducing taxes on foodgrains (import tariffs and sales taxes), using food grain stocks to increase supply, using some form of subsidy on essential items, and imposing export restrictions on staple food items. A survey of 120 country teams carried out by the World Bank in March, 2008 shows that in Sub-Saharan Africa the most common 'economy-wide' policy response was to reduce foodgrain taxes--either tariffs, VAT. other sales tax, or a combination of these measures. On the other hand, the most common response outside Sub-Saharan Africa was some form of subsidy on essential items, which is what over half of these countries used to stabilize domestic prices (see figure 1). These subsidies have a long history in several countries and vary significantly in the extent to which they create distortions and generate fiscal pressures. For instance in Eastern Europe formal or tacit agreements between producers and the government on either actual prices or profit margins are common for basic staples. These restrictions can create producer disincentives. In the Middle East and North Africa, universal subsidies on key items are an important part of the social compact between the State and the citizens in several countries. These are known to be fiscally expensive and crowd out spending on other social programs. In South Asia, several countries have targeted subsidies on "inferior goods" (for example coarse rice in Bangladesh) which are used during crises and typically create the fewest distortions. Only 30 percent of Sub-Saharan African countries used some form of subsidy. This lower share is most likely due to the greater fiscal and administrative constraints in Africa relative to non-African countries. In some ways these constraints are a blessing in disguise for Sub-Saharan Africa­ widespread use of these subsidies makes them difficult to remove, reduces the Figure 1. Economy-wide Policy Responses to Rising Food Prices I- Africa ~ Rest of the WOrld] til 60% .!:! 1:~: '0 30% ~ .3 20% = 10% ~ ~ ~ 0% Reduce Release Export Introduce I None foodgrain foodgrain restrictions extend taxes stocks subsidies Wodon and 7,aman 167 incentives for the development of alternative better targeted safety net programs, and can risk curbing an agricultural supply response. Around a third of countries in Sub-Saharan Africa used foodgrain stocks to increase domestic supply and curb prices-this ratio is similar to the global average. The transportation, storage. and rotation of physical stocks require con­ siderable expertise and these strategic grain reserves have had their share of cor­ ruption problems. Yet the pressures arising from the virtual suspension of rational pricing mechanisms on global foodgrain markets in early 2008. with consequent knock-on impacts in local markets. has led to a renewed interest in building up physical reserves, which can then be released during crises. Export restrictions were as common outside Sub-Saharan Africa (23 percent of countries) as they were for countries within Africa. Governments typically intro­ duce such measures due to pressures from the domestic political economy, though it is unclear whether these measures have any impact on domestic prices. What is clear is that when these are imposed by large grain exporting countries they have serious food security implications for neighboring countries and for regional com­ modity markets. The second broad type of policy response revolves around using existing safety net instruments to either increase benefit levels or increase beneficiary coverage. However, while many countries in Sub-Saharan Africa have food-based transfer programs (figure 2). the coverage of these programs tends to be small (in part due to lack of financing). relative to needs. In Burundi, Central African Republic. Ghana. Liberia, and Togo school feeding programs were expanded. while in Guinea and Sierra Leone public works programs were set up. As Figure 2 shows from survey responses by World Bank country teams in March, 2008. some countries in Sub-Saharan Africa have cash based programs. For example, Ethiopia was able to respond by expanding the benefit levels of its major cash transfer Figure 2. The Distribution of Safety Net Programs : • Africa lSI Rest of the World .~ 60% +--AA~----- 1: 50% + - - - I : ' $ & ! - - - - - ­ ::I 8 40% 'S 30% '" J 20% 5 10% ~ ~ 0% Cash Food for Food School transfer work ration! feeding stamp 168 The World Bank Research Observer. vol. 25, no. 1 (February 2010) program. 3 Informal networks and family based support mechanisms are an important part of the safety net, though these were strained during the food price spike. The ability to scale up safety net programs during these crises is limited by both fiscal constraints-worsened by economy-wide measures such as reducing taxes or increasing subsidies-and capacity constraints. These include weaknesses in targeting vulnerable groups as well as food and cash management issues. Yet given the range and frequency of shocks in Sub-Saharan Africa strategic choices will need to be made so that the fiscal space and requisite capacity for scaled up safety nets is created. The third type of policy measure involves supporting domestic food production, though impacts here typically accrue over the medium term. Scaling up or intro­ ducing free or subsidized input distribution (for example Malawi), or the govern­ ment import of fertilizers (for example Ethiopia and Tanzania), are common short-term responses, though they often have significant fiscal consequences. Over the medium run African countries, through the New Partnership for Africa's Development (NEPAD) mechanism. have committed to increasing their investments on agricultural research and extension as well as on irrigation and new technology. In practice countries combined economy-wide policies and safety net programs as their immediate response to the food price hike. Country examples are illustra­ tive of this mix. Liberia's response revolved around reducing import taxes on food­ grains and scaling up targeted feeding programs. Kenya imported 3 million metric tons of maize and subsidized fertilizer. Guinea introduced a targeted consu­ mer subsidy for rice and expanded an existing school feeding program to urban areas. Cameroon sharply reduced \¥\'T and customs duties on basic food staples as well as on imports of agricultural inputs while raiSing civil service wages. There are no studies which examine the cumulative effectiveness of these policies. We therefore limit ourselves to reviewing recent work on the extent to which the poor are likely to be targeted by two types of policy-indirect tax cuts and food-based safety nets-which are commonly used in Sub-Saharan Africa. Economy-Wide Polides: The Case of Indirect Tax Cuts As figure 1 illustrates almost half of the governments in Sub-Saharan Africa have reduced taxes levied on food items, such as import taxes and VATs, in order to deal with the increase in food prices. An IMF survey shows that most of these cuts were in import tariffs as they were easier to administer than VAT (IMF 2008). Wodon and others (2008b) estimate the extent to which the poor are likely to benefit from a reduction in indirect taxes. The authors provide data on the consumption of various imported foods for the same set of West and Central African countries discussed earlier. The share of rice consumption in the bottom Wodon and Zaman 169 40 percent of the population varies from 11 percent in Mali to 32 percent in Sierra Leone. averaging around 20 percent. This means that if one considers the bottom 40 percent as the poor. out of every dollar spent by a government for reducing indirect taxes on rice. and assuming that the indirect tax cuts results in a proportionate reduction in consumer prices. only about 20 cents will benefit the poor on average. From the consumer perspective the extent of the benefit will clearly also depend on the extent of initial tariffs and the magnitude of the tariff cut. In West Africa most import tariffs on food staples were below 15 percent. For most of the other imported foods for which indirect tax cuts were implemented, the proportions of those foods consumed by the poor tend to be even lower than that for rice. If we therefore assume that the share of imported foods represented at most 10 percent of average household consumption then the real income gain even from a complete elimination of tariffs would be no more than 1.5 percent. Thus, while reducing taxes is a popular "stroke of the pen" measure. it suffers from several weaknesses. First, it can be costly in budgetary terms if these are per­ manent cuts. For instance the reduction in taxes following the rise in food prices was as much as 1.1 percent of GDP in Liberia and 0.8 percent in SenegaL and equivalent to 7 percent of tax revenues in Guinea Bissau, worsening an already tight fiscal balance. Second, for many food items, much of the benefit of the tax cuts will accrue to the non poor and the real income gains for the poor will be marginal. Third, compared to reducing VAT or a sales tax, lowering import tariffs may well hurt domestic producers in the short run. and in some circumstances reducing import tariffs may increase poverty. Hence the decision to reduce tariffs would need to balance the benefits for the poor in terms of the reduction in retail prices and political economy gains. with the costs outlined here. Food-Based Safety Net Programs As figure 2 shows food-based safety net programs are more common in Sub­ Saharan Africa compared with other regions. In this section, we focus on the tar­ geting performance of three types of food-based safety nets-food aid as typically distributed by humanitarian agencies. school feeding programs. and public works which often make payments in kind (but may also provide cash benefits). Using nationally representative household survey data a recent assessment of a World Food Programme (WFP) feeding programme in Burundi shows that there was little difference in the likelihood of receiving food aid between various groups of households (Diang'a. Wodon. and Zoyem 2009). The share of total food distri­ bution obtained by the poor was in fact slightly lower than the share of the poor in the total popUlation. The main shortcoming appeared to be that this program did not specifically target areas in the north where food insecurity was most 170 The World Bank Research Observer, vol. 25. no. 1 (February 2010) severe-which WFP now does. Clearly. using poverty maps is important for making the most out of scarce resources. Analysis from a 2007 nationally representative household survey in Liberia shows that 22 percent of the population received some form of food aid. School feeding was the most common (74 percent were recipients of food aid). followed by food-based public works projects and nutritional supplementation. Similar to Burundi. estimates of the targeting performance of these programs suggest that nonpoor households are essentially as likely to benefit from food aid as poor households. There are differences in targeting estimates acroSs programs (school feeding programs are slightly pro-poor. while other programs are slightly in favor of the nonpoor). but these differences are not large (Tsimpo and Wodon 2008b). Gilligan and Hoddinott (2007). using longitudinal data from Ethiopia. compare the targeting and food security outcomes of a feeding program and a Food for Work program. They find that the feeding program was better targeted to the extreme poor, while the benefits from the public works program accrued mainly to the middle and upper end of the distribution. Del Ninno, Dorosh, and Subbarao (2007) review the experience with food aid in two East African countries (Ethiopia and Zambia) and two South Asian countries (India and Bangladesh). Their results suggest that to be effective food aid needs to be timed to avoid local producer disincentive effects and be channeled in a manner which helps create local level infrastructure-all characteristics of a well-functioning public works program. The drawback with such programs is that participants have to give up other employment in order to participate in these programs. and hence they are more suitable for shocks. or lean seasons. which create unemployment. Adato and Haddad (2002) find that among a sample of public works projects in South Africa. about 90 percent outperformed an untar­ geted transfer scheme. Teklu and Asefa (1997.1999) find that in rural Botswana and Kenya, the poor are more likely to participate in public works programs than the nonpoor and have a substantial positive impact on their income. At the same time. a substantial number of nonpoor individuals also participate in the schemes. so that targeting performance could be improved. Using data from Chad, Ghana, Liberia, and Rwanda Wodon and others (2008c) define the overall leakage rate in public works programs as the share of program outlays that are likely not to raise the incomes of the poor. either because program participants are not poor or because of the income foregone from alternative employment. The authors show that if public works programs are randomly placed within the country, the leakage rates are potentially high, varying from 50 percent to close to 75 percent in all four countries. On the other hand. if the programs are geographically targeted. targeting performance can improve substantially. In Ghana. for example. the simulated leakage rate for a hypothetical randomly placed public works program is very high. at 73 percent. Wodon {lnd Zamall 171 while it could be as low as 18 percent if resources were only concentrated in the Upper West region of the country. This review shows that the food-based safety net programs more commonly used in Sub-Saharan Africa can be effectively targeted to the poor with clear benefits to household welfare, but that there is considerable variation across pro­ grams and contexts. The choice of program depends on a range of factors, such as the type and duration of the shock being faced by the community, the adminis­ trative capacity, and budgetary constraints. School feeding programs may be par­ ticularly effective for retaining poor children in school following a real income shock such as a sharp increase in food prices. Public works programs may be more effective during lean seasons or following lay-offs during an economic downturn. Timing is also crucial. For instance administrative structures need to be in place to enable interventions during the lean seasons. thereby avoiding households having to consume their seeds and selling their assets. In many countries, given large differences in consumption and nutrition levels between regions. geographic targeting could be used to improve program effectiveness. Furthermore in order to address chronic food insecurity for vulnerable groups, regular fortified food provision. and other essential nutritional interventions aimed at reaching children under five and pregnant mothers. are required year­ round. Conclusion The first objective of this article was to review the evidence regarding the poten­ tial impact of the recent increase in food prices on poverty focusing on Sub­ Saharan African countries. The second objective was to document the policy responses adopted by governments to cushion the poor from the immediate impacts of this crisis. and to assess the extent to which the poor may have been reached by these policies. Two main findings stand out from our review. First, due to the higher share of net consumers relative to net producers in most countries, higher food prices leads to increased poverty even when second­ round wage effects are taken into account. The bulk of the welfare loss from higher prices accrues to those who are already poor, and the evidence from Sub­ Saharan Africa shows that increases in the poverty gap can be significant. Recent analysis in Sierra Leone and Liberia. albeit using nonrepresentative samples, shows that households cut back on rice consumption and significant numbers cut back their consumption of micronutrient rich foods. The poorest families reported sending fewer children to school and avoiding hospitals. Hence even if these price increases were temporary-and it is unclear to what extent this is the case-­ these price shocks can have long-lasting impacts. 172 The World Bank Research ObserFer. vol. 25. no. 1 (February 2(10) Second, governments potentially have various tools at their disposal to deal with the immediate impact of the increase in food prices, with important differ­ ences in the effectiveness of these tools, The types of policies which are used to deal with crises vary considerably across continents. In Eastern Europe and the Middle East for instance there is a long history of universal subsidies and controls on producer prices. These policies clearly have significant fiscal costs and require a fairly extensive state apparatus to implement, which may partly explain why these instruments are less common in Sub-Saharan Africa. On the other hand targeted subsidies (using products typically consumed by the poorest) deployed during sudden shocks, or lean seasons, can be an effective way to protect the poor while not creating excessive fiscal costs or economic distortions. In Sub­ Saharan Africa, almost half the countries surveyed following the increase in food prices reduced taxes on foodgrains. We argue that the bulk of the benefits from a cut in tariffs accrued to the nonpoor, and the paltry real income gains for the poor, are most likely not worth the tax revenue foregone. In addition to economy-wide policies, governments have scaled up safety net programs in an effort to cushion the poor from the impact of higher food prices. Safety net programs tend to be better targeted than economy-wide policies, though our review suggests that they can also suffer from significant leakages to the nonpoor and face fiscal and administrative constraints. The limited evidence available on feeding and public works programs suggests that their targeting per­ formance can be improved considerably, for instance through the use of geo­ graphic targeting. Moreover in most Sub-Saharan African countries the coverage and generosity of safety net programs falls significantly short of the size and needs of their vulnerable populations. Hence it is clear that in practice most governments in Sub-Saharan Africa have limited instruments to address events such as the 2008 spike in food prices and yet there are strong political pressures for them to do so. Given the absence of well-targeted safety nets with adequate coverage in many countries, it could be that some of the "economy-wide" measures that were adopted by governments were necessary both for reasons of political economy as well as for practical reasons. Moreover some of these economy-wide options are less costly than others, and time-bound and targeted subsidies, for instance, may be as cost effec­ tive as any other alternative policy option to protect the poor. Yet experience from other countries shows that investments in the building blocks of safety net pro­ grams could yield tangible improvements in the lives of a large proportion of poor households at the time of the next crisis. Specifically. investments in targeting methods and data, payment systems, accountability mechanisms, and monitoring are required (see Grosh and others 2008 for more details). These can create the basis for a safety net menu which includes regular year-round programs for vul­ nerable groups (for example infants, pregnant women, the disabled) as well as Wodon and Zaman 173 temporary programs which scale up during lean seasons or during shocks. Finally. although medium to long term initiatives to boost food production have not been discussed here, these initiatives are clearly necessary as part of an overall package of policies to stimulate pro-poor growth and reduce household vulnerability to shocks. Notes Quentin Wodon is an adviser in the Development Dialogue on Values and Ethics in the Human Development Network. Hassan Zaman (corresponding author) is a lead economist in the Poverty Reduction Group in the Poverty Reduction and Economic Management (PREM) Network: email address: hzaman@worldbank.org. The authors benefited from discussions and comments with. among others, Douglas Addison, Harold Alderman, Antonella Bassani. Louise Cord, Shanta Devarajan, Hinh Dinh. Wilfried Engelke, Louise Fox, Delfin Go. William Martin, Ana Revenga, Sudhir Shetty. Kenneth Simler, Linda Van Gelder. and Jan Walliser. The opinions in this paper are those of the authors only. and need not represent those of the World Bank, its executive directors, or the countries they represent. 1. Ivanic and Martin also show that the effect of a relatively small 10 percent change in prices can be a first-order approximation for the impact of a larger change. though some results vary sig­ nificantly depending on the extent of the clustering of households around the poverty line. In rural Peru. for instance, the impact of a 20 percent price rise on the poverty headcount is five times greater than that of a 10 percent rise, 2. For more detailed country studies. see for example Joseph and Wodon (2008) on Mali: Wodon, Tsimpo. and Coulombe (2008) on Ghana: and Tsimpo and Wodon (2008a) on Liberia. 3. Field interviews in 2008 revealed that due to the sharp erosion in purchasing power due to high inflation. the vast majority of participants in the Productive Safety Net Program would have preferred in-kind payments instead of the cash transfer (Hobson 200Y). References The word processed describes informally reproduced works that may not be commonly available through libraries. ACF (Action Against Hunger). 2009. "Feeding Hunger and Insecurity: The Global Food Crisis." Briefing Paper. Adato. M.. and 1. Haddad. 2002. "Targeting Poverty through Community-Based Public Works Programmes: Experience from South Africa." Journal of Developlnent Studies 38(3}: 1-31}. Alderman, H.. 1. Hoddinott. and B. Kinsey. 2001}. "Long Term Omsequences of Early Childhood Malnutrition," (hjord Economic Papers 58(3): 450-74. Arndt. C.. R. BenficCl, N. Maximiano. A. Nucifora, and J. Thurlow. 2008. "Higher Fuel and Food Prices: Economic Impacts and Responses for Mozambique." International Food Policy Research Institute. Barrett, CD., and P.A. Dorosh, 1 YYI}, "Farmers' Welfare and Changing Food Prices: Nonparametric Evidence from Rice in Madagascar." American Journal of Awicuitllml Economics 78(3): 65!l-1}9. 174 The World Bank Researc/, Obsen'cr, 1'01. 25. 110, 1 (February 20l 0) Benson. T. 2008. An Assessment of the Likely Impact of Rising Global Food Prices on Ugandan Households. Kampala: IFPRI. Budd. J.W 1993. "Changing Food Prices and Rural Welfare: A Non-Parametric Examination of the Cote d·[voire." Economic Development and Cultural Change 41(3): .587-603. Christiaensen. L.• and 1. Demery. 2007. Down to Earth: ilgriculture and Poverty Reduction in Africa. Washington. DC: lVorid Bank. Coulombe. H.• and Q. Wodon. 2008. "Geographic Impacts of the Food Price Crisis in Sub-Saharan Africa." Processed. World Bank. Washington. DC. Deaton. A. 1989. "Rice Prices and Income Distribution in Thailand: A Non-Parametric Analysis." The Economic Joumal99 (39.5): 1-37. Del Ninno. c., P.A. Dorosh. and K. Subbarao. 2007. "Food Aid. Domestic Policy and Food Security: Contrasting Experiences from South Asia and Sub-Saharan Africa." Filod Policy. 32(4): 413-3.5. Dessus. S" S. Herrera. and R. De Hoyos. 2008. "The Impact of Food Inflation on Urban Poverty and its Monetary Cost: Some Back of the Envelope Calculations'" Agricultural Economics 39:417- 29. Dewbre. J.• C. Giner. W Thompson. and M. Von Lampe. 2008. "High Food Commodity Prices: Will They Who Will Pay?" Agricultural Economics 39:393-403. Diang'a, E.. Q. Wodon. and J.-P. Zoyem. 2009. « L'insecurite alimentaire et l'aide alimentaire au Burundi.» Processed. World Bank. Washington DC. Easterly. W. and S. Fischer. 2001. "Inflation and the Poor." Journal of Money. Credit and Banking 33(2): 160-78. Fafchamps. M.. C. IJdry. and K. Czukas. 1998. "Drought and Saving in West Africa: Are Livestock a Buffer Stock?" Jonrnal of Development Economics .55(2): 273-305. Gilligan. D.o., and J. Hoddinott. 2007. "Is There Persistence in the Impact of Emergency Food Aid? Evidence on Consumption, Food Security, and Assets in Rural Ethiopia." American Journal of AgriCHltural Economics 89(2): 225-42. Grosh. M.. C. Del Ninno. E. Tesliuc, and A. Ouerghi. 2008. For Protection and Promotion: The Design and Implementation of EiJective Safety Nels. Washington. DC: World Bank. Heady. D.. and S. Fan. 2008. ':Anatomy of a Crisis: The Causes and Consequences of Surging Food Prices." I1gricLlltllmi Ecollomics 39:375-91. Hobson. M. 2009. "The Food Price Crisis and its Impact on the Ethiopian Productive Safety Net Programme in 2008." Humallitarian Exchange 42. Overseas Development Institute, London. IMF (International Monetary Fund). 2008. Pood and Fuel Prices~Recent Developments, Macroeconomic Impact alld Policy Responses. Washington. DC: International Monetary Fund. Ivanic. M.. and W Martin. 2008. "Implications of Higher Global Food Prices for Poverty in Low­ Income Countries." Agricultural Economics 39:40.5-16. falan. J.. and M. Ravallion. 2()02. "Household Income Dynamics in Rural China," Discussion Paper 2002110. United Nations University, World Institute for Development Economics Research. Joseph. G.. and Q. Wodon. 2008. ':Assessing the Potential Impact on Poverty of Rising Cereals Prices: The Case of Mali." Policy Research Working Paper 4744. World Bank. Washington. DC. Klytchnikova. 1.. and N. Diop. 20()6. "Trade Reforms. Farm Productivity. and Poverty in Bangladesh." Policy Research Working Paper 3980. World Bank. Washington. DC. Poulton. C.. J. Kyddd. S. Wiggins. and A, Dorward, 2006, "State Interventions for Food Price Stabilization in Africa: Can it Work?" Food Policy 31:342-.56. Rashid, S. 2002. "Dynamics of Agricultural Wage and Rice Price in Bangladesh: A Reexamination." Discussion Paper 44, International Food Policy Research Institute. Washington DC. i4!Jdon and Zaman 175 RavalUon. M. 1990. "Welfare Changes of Food Price Changes under Induced Wage Responses: Theory and Evidence for Bangladesh." O.\jord Economic Papers 42:574-85. ___ . 2000. "Prices. Wages and Poverty in Rural India: What Lessons Do the Time Series Data Hold for Policy?" Food Policy 25(3): 351-64. Simler. K. 2009. "The Impact of Higher Food Prices on Poverty in Uganda." Processed. World Bank. Washington DC. Singh. 1.. L. Squire. and J. Strauss. 1986. Agricultural Household Models: Extensions and Applications. Baltimore: Johns Hopkins University Press. Teklu. T.. and S. Asefa. 1999. "Who Participates in Labor-Intensive Public Works in Sub-Saharan Africa? Evidence from Rural Botswana and Kenya." World Development 27(2): 431-8. ___. 1997. "Factors Affecting Employment Choice in a Labor-Intensive Public Works Scheme in Rural Botswana." Economic Development and Cultural Change 46(1): 175-86. Tsimpo. C.. and Q. Wodon. 2008a. "Rice Prices and Poverty in Liberia." Policy Research Working Paper 4742. World Bank. Washington. DC. 2008b. "Targeting Performance and Impact of Food Aid in Post-Conflict Countries: Evidence from West and Central Africa." Processed. World Bank. Washington. DC. WFP (World Food Programme). 2009. "Trends in Staple Food Prices in Vulnerable Countries." Issue 3 (March). Rome. Wodon. Q.. C. Tsimpo. and H. Coulombe. 2008. "Assessing the Potential Impact on Poverty of Rising Cereals Prices: The Case of Ghana." Policy Research Working Paper 4740. World Bank. Washington. DC. Wodon. Q.. C. Tsimpo. P. Backiny-Yetna. G. Joseph, E Adoho. and H. Coulombe. 2008a. "Potential Impact of Higher Food Prices on Poverty: Summary Estimates for a Dozen West and Central African Countries." Policy Research Working Paper 4745. World Bank, Washington. DC. ___. 2008b. 'i\.re Indirect Tax Cuts to Protect the Poor from Higher Food Prices Well-targeted?" Processed. World Bank. Washington. DC. Wodon. Q.. P. Backiny-Yetna. H. Coulombe. and C. Tsimpo. 2008c. "Targeting Performance of Labor Intensive Public Works in Africa: Simulations for Four Countries." Processed. World Bank. Washington. DC. World Bank. 2008. Rising Food and Fuel Prices; Addressing the Risks to Future Generations. Washington. DC: World Bank. Zaman. H.. C. Delgado. D. Mitchell. and A. Revenga. 2008. "Rising Food Prices: Are There Right Policy Choices?" Development Outreach. Washington. DC: World Bank. 176 The World Bank Research Observer. vol. 25. no. 1 (Pebruary 2010) ORPORATE SERVICES We offer tailored services to industry. Our extensive services include: Supplements Reprints E-prints Sponsored Bulk Subscriptions Advertising - both print and online Corporate Services Email: special.sales@oxfordjournals.org Tel: +44 (0) 1865 353827 Fax: +44 (0) 1865 353774 For more information please visit our website www.oxfordjournals.org/corporate_services ERTISING Readership of our journals is assured by the high quality of our editorial content. This provides an excellent base for advertising your products and services to targeted key audiences. For more information about advertising opportunities including: Display advertising Website/banner advertising Recruitment/classified advertising Inserts Sponsored eTOCS (Email Table of Contents) Journals Advertising Email: jnlsadvertising@oxfordjournals.org Tel: +44 (0) 1865 354637 Fax: +44 (0) 1865 353774 For more information please visit our website www.oxfordjournals.org/corporate_services currenlt( offers a UI1icp! text. It~ laS . . help ~ . . .. , . . , and can be used levels d technical ~ It Is suItatJIe for dIuraam tead*w d ur.deilliiduate and post waduate courses h econometrils. For experlenailr "I!S d econometric prosrams, it offers a variety d IIMrIate ~ ••~AJ.llU methods. multivariate techniques fa cointesratlon. principal components, canonical correlations and multivariate volatility look modellins. and provides a Iqe number d cfiasnostIc and £520.001 non nested tests not readltt available in other packaSes. The Windows ComniIII:;[aI interaction d excellent IJIPhIcs and estimation capabilities iI Instalatlon 1-5 UIIII Microfit 5.0 allows important econometric: research to be £2000.00 I 978-0-19-92311 U carried OIl h a matter d days rather than weeks. Windows Commercial Mulllple Installalon 6-10 users • Major update d bestselling econometrics £3750.00 I 978-0-19-923113-3 software package designed specifically for Windows Commercial Multiple econometric modelling of time series data Installation 11-20 users £6000.00 I 978-0-19-923114-0 • Easily used at different levels d technical sophistication by students, academics. and PI_ nota: .11 prices Ilbove exclude VAT practitioners Printed user's manual .User interface has been completely (al$o available on software as Help revamped to improve accessibility and provide Files) clear instructions to users £35.00 I 978-0-19-956353-1 • Much enhanced graphic module allows numerous graph types and an u«estricted Upgrades ~e available fa registered users number of plots per screen of Mic:rofit 4.0 / 4.1 Please contact. with details d ywr previous version: Electronic Publishing Customer ServIce Tel: +44 (0) 1865 353534 0:\ lORD Fax: +44 (0) 1865 353308 Emait mlaoflt.lnfoOoup.com Hofmann WASHINGTON DC