75984 MARCH 2013 • Number 110 Promoting Shared Prosperity in South Asia Ejaz Ghani, Lakshmi Iyer, Saurabh Mishra The geography of poverty has changed. More than 70 percent of the world’s poor live not in low-income countries, but in middle-income countries. In 2008, nearly 570 million people lived on less than US$1.25 a day in South Asia, compared to 385 million in sub-Saharan Africa. In addition, nearly 70 percent of the poor people in South Asia live in the lagging regions. Improving the living standards of these regions is crucial to achieving the goal of shared prosperity. Economic growth is not sufficient to enable the lagging regions of South Asia to catch up with the leading regions, in terms of proportional reductions in poverty rates. Policies must be specifically targeted toward achieving greater growth and poverty reduction in these regions. One particular policy channel to achieve shared prosperity is pro-poor fiscal transfers. For the most part, interstate fiscal transfers in South Asian countries do promote equity through transfer of resources to poorer regions, but this outcome usually occurs when pro-poor redistribution has explicit rules and transparency. Further, simply directing financial resources to lagging regions may not be sufficient, and may need to be complemented with increases in capacity, transparency, and participation to facilitate accountability at the local level. Poverty is the worst form of violence. —Mahatma Gandhi The global geography of poverty has changed over the last two Where Do the Poor Live? decades. More than 70 percent of the world’s poor at US$1.25 a day now live not in low-income countries, but in middle-in- Most of the poor people in India and other South Asian coun- come countries (Kanbur and Sumner 2011). This raises a tries live in the lagging regions (states with per capita income number of big questions. Is economic growth not sufficient to below the national average). Figure 1 shows that poverty mass pull everybody out of poverty? Can middle-income countries (or the number of poor people) is largely concentrated in the or parts thereof suffer from poverty traps? If so, then what lagging states. Leading states (with per capita income above can one do to promote shared prosperity? This note examines the national average) have poor people too, but rapid econom- these questions in detail, using both a national and a subna- ic growth has helped them to manage poverty better. Nearly tional lens, and in a global setting, drawing on the recent 70 percent of the poor people in India and South Asia live in World Bank report, The Poor Half Billion in South Asia (Ghani the lagging regions, so improving the living standards of these 2010). regions is crucial to achieving the goal of shared prosperity. 1 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise Figure 1. Poverty Mass Concentration in India Leading Lagging Areas Areas Poverty Mass Low Poverty Mass is Concentrated AFGHANISTAN High Poverty Mass* Khyber in Lagging Regions Pakhtunkhwa No Data Punjab Punjab Uttarakhand Far NEPAL Arunachal Pradesh Western Mid Western PAKISTAN Haryana Delhi Western Sikkim Balochistan Central Eastern BHUTAN Rajasthan Uttar Pradesh Assam Sindh Nagaland Bihar Meghalaya Rajshahi Sylhet Manipur Jharkhand Dhaka West Tripura Gujarat Madhya Pradesh Bengal Barisal Mizoram Khulna INDIA Chhattisgarh Orissa MYANMAR Maharashtra BANGLADESH Andhra Pradesh Goa Karnataka Andaman and Nicobar Tamil Kerala Nadu Northern North Central * High poverty mass is de�ned as subnational areas of a country having a large poor North Western Eastern SRI LANKA population. In Bangladesh, high Central Western ¯ poverty mass is > 15 million Uva persons; in India and Pakistan, Sabaragamuwa > 20 million persons; in Nepal, Southern > 1.5 million persons; and in Sri Lanka, > 0.6 million persons. MALDIVES Note: Source Laggingfrom : Reprinted leading andThe BillionIn states: Poor Half in most developing South Asia—What and industrial Is Holding Back Laggingcountries, Regions? lagging regions are de�ned as those areas where growth and Note income : Lagging seriously and lag behind leading states: averageand in most developing national peformance. industrial In regions countries, lagging this map, a state are defined as is de�ned those as lagging areas where if its growth and per capita income income seriously is average below lag behind national peformance. In this map, a state is defined as lagging if its per capita income is below the national average. the national average. One way to achieve shared prosperity is through econom- gions. This can be achieved by improving the business envi- ic growth. The first part of this note shows that current pat- ronment for the private sector, supporting market integration, terns of economic growth do not appear to be enough. Lag- improving connectivity, and ensuring macroeconomic stabil- ging regions have slower growth than leading ones. While ity. The government can also intervene directly in sectors or there is poverty reduction in both leading and lagging regions, locations in which the private sector is reluctant to invest. lagging regions do not display greater poverty reduction in Another strategy is to promote pro-poor fiscal transfers proportional terms. Policies should therefore aim to acceler- and devolve power to local governments who may then imple- ate the growth rate and poverty reduction of the lagging re- ment policies more suited to local conditions. The second 2 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise part of this note examines whether the current fiscal decen- 1, column 1). If the absolute change in the headcount ratio tralization programs are working to the benefit of the poor is regressed on the initial poverty level, it shows a negative regions in South Asia. and statistically significant coefficient, even after control- ling for growth rates of per capita gross domestic product Poverty Convergence across Countries (GDP). and Regions Poverty itself can become a constraint to growth through Figure 2. Convergence in Poverty Headcount Ratios (absolute channels such as lower savings or investment rates; dismal level) education or health outcomes; poor access to credit or prop- 4 erty rights; incomplete insurance markets, which increase the annualized change in poverty rate Uzbekistan risks of crop failures, floods, and droughts; and high conflict Djibouti rates. Such “poverty traps� limit the choices of economic 2 Tanzania agents, whether individuals, households or firms, to fully ex- Rwanda Bolivia Guinea-Bissau Côte d’Ivoire press their economic potential. These traps can start a vicious Guyana Colombia Mongolia Nigera Argentina South Africa cycle, with no income growth feeding into greater poverty, 0 Latvia Paraguay Pakistan Lesotho Zambia Malaysia India Burundi Morocco Sri Lanka Niger which in turn reduces growth even further. Tunisia Brazil Mexico Costa Rica Philippines Cambodia Sierra Leone Madagascar Thailand Uganda Mozambique So, does evidence show that countries suffer from pover- Kenya Nicaragua Ghana Lao PDR Bangladesh Honduras Armenia Nepal Guinea ty traps? Not really. Figure 2 shows that countries with ini- -2 Guatemala Central African Rep. Indonesia tially higher levels of poverty experienced greater absolute re- Swaziland Vietnam China ductions in the poverty headcount ratio—the proportion of Cameroon the population living below the international poverty line of -4 US$1.25 a day. 0 20 40 60 80 100 These trends are further verified in regressions report- initial poverty rate ed in table 1 and found to be statistically significant (table Source: World Bank 2009b. Notes: Poverty rate is for US$1.25. Time period for change in poverty is for 1977–2007. Number of observations is 91. Table 1. Poverty Convergence across Countries Annualized change in Annualized change in poverty Annualized changes in headcount ratio poverty gap gap squared Absolute Proportional Absolute Proportional Absolute Proportional Initial poverty > 10% Indicator 1 2 3 4 5 6 7 Initial poverty -0.02*** -0.02*** (level) (0.005) (0.010) Initial poverty (log) -0.006 (0.01) Initial poverty gap -0.046*** (level) (0.006) Initial poverty gap -0.024*** (log) (0.013) Initial poverty gap -0.05*** squared (level) (0.006 Initial poverty gap -0.045*** squared (log) (0.015) Growth rate of GDP -0.25** -0.31** -0.01*** -0.104* -0.01*** -0.05 -0.022** per capita (0.12) (0.15) (0.004) (0.061) (0.007) (0.037) (0.009) Observations 82 56 82 82 82 79 79 R-squared 0.97 0.32 0.13 0.48 0.21 0.60 0.37 Source: World Bank 2009b, 2009c. Notes: Robust standard errors are reported in parentheses. *** represents significance at 1 percent, ** represents significance at 5 percent, * represents significance at 10 percent. Poverty rate is defined as percentage of population living in households with consumption or income per person below the international US$1.25 poverty line. Annualized poverty change is for 1977–2007. Poverty gap is defined as the mean distance below the poverty line as a proportion of the poverty line. Squared poverty gap is defined as mean of the squared distances below the poverty line as a proportion of the poverty line. 3 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise Using the reduction in headcount ratio is obviously an erty (table 1, columns 4–7). Although the poorest show the important way to measure economic progress. But there are largest increases in consumption, they are so far from the pov- two issues with using this measure. First, countries with low erty line that this progress does not show up in equivalent levels of poverty cannot display high values of poverty reduc- changes to the headcount ratio. tion. The existence of a “floor value� of zero therefore may not Poverty Convergence in South Asia capture all the progress in a country starting from low levels of poverty. At the subnational level, examining poverty convergence/di- One can partially control for the existence of such a floor vergence shows that trends within India and across South value by restricting the sample to countries that had a head- Asian regions are similar to the earlier results for the global count poverty ratio greater than 10 percent in the initial pe- sample. At the subnational level, states with higher levels of riod. The results regarding convergence are robust to this re- poverty experienced greater absolute reductions in the head- striction (table 1, column 2). These results provide a basis for count ratio (table 2, columns 1 and 2). As in the global sample, cautious optimism regarding progress in poor countries; poor however, they did not experience greater proportional reduc- countries may not be trapped in poverty forever. tions in the headcount ratio (table 2, columns 3 and 4), sug- The second caveat to using absolute poverty change is the gesting that more needs to be done in the lagging regions to fact that reducing poverty from a high level might be easier align them with the leading ones. When looking at measures than reducing poverty from an already low level. The latter of poverty depth and severity, once again the initially worse-off might reflect more entrenched factors leading to poverty, in- regions showed greater improvements in these measures than cluding low levels of education or health problems or even the initially better-off regions (table 2, columns 5 and 7). persistent beliefs about the value of effort (for example, 12– Unlike the global sample, however, the lagging regions in 13 percent of the U.S. population has been under the national India do not show proportionally better performance in re- poverty line for more than a decade). Using a percentage ducing the PG or the SPG measures (table 2, columns 6 and change in poverty (or equivalently, a change in log poverty) 8). This is a worrisome sign that poverty reduction in the lag- gives a greater magnitude to poverty reductions starting from ging regions of India and South Asia needs to be accelerated, a low base. It also avoids the floor value problem. especially with regard to extreme poverty. When one considers percentage changes in poverty, there Overall, these results provide for cautious optimism on is no longer any significant degree of convergence in poverty shared prosperity and progress in the lagging regions. There is across countries (table 1, column 3). In other words, although little evidence of persistent poverty traps at a regional level in poorer countries do reduce absolute numbers of people living South Asia, or even among developing countries as a whole. in poverty, they do not reduce them proportionally faster However, there is also no room for complacency. Lagging re- than richer countries. This fact has also been documented in gions are not catching up with the leading regions in terms of Ravallion (2012), who ascribes the lack of convergence to two per capita income or in health indicators, or even in propor- reasons. The first is that initially poor countries do not grow as tionate terms for poverty reduction. fast as not-so-poor countries, and the second is that, for a giv- As discussed earlier, one solution is to implement poli- en rate of growth, poverty reduction in proportional terms cies to accelerate growth in the lagging regions. Another is to appears to be slower in poorer countries. redistribute resources across regions in an effort to improve In addition to the headcount ratio, other measures of living standards. The next section reviews fiscal decentraliza- poverty that take into account how far households are from tion arrangements in South Asia and examines whether these the poverty line can also be examined. This is particularly im- result in greater transfers to the lagging regions. portant in measuring extreme poverty, which may not be re- Fiscal Decentralization Arrangements in flected fully in a simple comparison of headcount ratios. The South Asia two measures to consider are the Poverty Gap (PG) Index and the Squared Poverty Gap (SPG) Index—where the poverty gap A typical cross-country measure of fiscal decentralization is is the average distance of an individual from the poverty line the share of total revenues or expenditures that are collected (with those above having a gap of zero). by subnational governments. India is quite decentralized Regressions using the PG or SPG measures show strong compared with the world average: subnational (state and lo- evidence of convergence across countries, with poorer coun- cal) governments collect 34 percent of all government reve- tries showing the largest reductions in these indices—both for nues and are in charge of 52 percent of total government ex- absolute reductions as well as proportional ones. This result is penditures (table 3). encouraging. Even if poorer countries are not able to achieve Like many large federal nations, states in India receive huge reductions in the headcount ratio, they do appear to transfers from the central government through a variety of show the largest reductions in the depth and severity of pov- mechanisms. The first consists of tax shares and grants decided 4 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise Table 2. Poverty Convergence in India and South Asia Annualized change in Annualized change in headcount ratio headcount ratio Annualized change in Annualized change in (absolute) (proportional) poverty gap poverty gap squared India, India, India, India, South Asia India South Asia India absolute proportional absolute proportional Indicator 1 2 3 4 5 6 7 8 Initial poverty -0.03*** -0.02*** (level) (0.01) (0.01) Initial poverty -0.01 -0.002 (log) (0.01) (0.01) Initial poverty gap -0.03*** (level) (0.01) Initial poverty gap 0.001 (log) (0.01) Initial poverty gap -0.04*** squared (level) (0.01) Initial poverty gap -0.003 squared (log) (0.01) Growth rate of -0.04 -0.06 -0.26 0.31 0.45 -0.10 0.03 0.02 GDP per capita (0.07) (0.07) (0.30) (0.31) (1.19) (0.33) (0.72) (0.39) Observations 29 18 29 18 18 18 18 18 R-squared 0.21 0.14 0.05 0.02 0.31 0.003 0.46 0.002 Source: World Bank staff calculations using the National Sample Survey Organization’s (NSSO) 55th and 61st round, http://mospi.nic.in/Mospi_New/site/Home.aspx. Note: Robust standard errors are reported in parentheses. *** represents significance at 1 percent, ** represents significance at 5 percent, * represents significance at 10 percent. South Asia includes states in India (1994–2005), Pakistan (1999–2005), and Sri Lanka (1996–2002). Odisha is deleted from the sample for the India region. South Asia: Odisha, Tripura, and Northern have been removed. by a nonpolitical Finance Commission, which places weights menting state development plans. The Planning Commission on factors such as the state’s area (10 percent), population (25 takes into account the state population and the gap between percent), per capita income (50 percent), and other factors in- state per capita income and the national average, among other cluding the state’s own revenues as a fraction of state domestic factors, when deciding state-level allocations. Third, various product (7.5 percent).1 The second source of funding is from central government ministries give grants to their counterparts the Planning Commission, which is in charge of formulating in the states for specified projects either wholly funded by the national five-year plans and makes grants and loans for imple- center (central sector projects) or requiring the states to share a proportion of the cost (centrally sponsored Table 3. Extent of Fiscal Decentralization in South Asia and the World schemes). These grants are wholly discretion- ary and often are not coordinated with Plan- % of government % of government Transfers to subna- revenue raised by expenditure by tional governmental ning Commission transfers, although they are subnational subnational units as a share of meant to serve similar objectives. Country governments governments subnational revenues In addition to these explicit transfers India 33.6 52 39 from the central government to the states, a Bangladesh <2 3–4 64–70 number of “hidden� or “implicit� transfers arise from the large subsidies provided by the Pakistan 7.3 30.3 81.1 central government for food and fertilizer. In Sri Lanka 7 12 82.5 2007–8, the central government had bud- World 21.7 29.1 32.5 geted about 1.34 percent of overall GDP to be China 59.7 81.5 35 paid out in subsidies, the bulk of which paid Canada 52.2 59.7 21.3 for subsidized food sales through the Food United States 41.1 49.3 28.9 Corporation of India, and for fertilizers. Ad- Mexico 23.5 23.1 47 ditionally, subsidized borrowing resources for the states are provided by either the central Sources: India, Canada, Mexico, and world figures are from Government Finance Statistics (GFS 1999); U.S. and Mexico figures are from GFS 2006; world average is based on the 41 countries in the GFS database. government or government-owned financial 5 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise Figure 3. Social Service Expenditures across Indian States, 2006–7 sion is that these food subsidies are not significantly higher 2,000 in poorer regions. The second-largest source of subsidies, fer- social services per capita per capita social services expenditure �tted values Maharashtra tilizer, benefits richer regions much more than poorer re- Kerala gions, because the richer regions tend to consume more fer- 1,500 Punjab tilizer (figure 4f). If the subsidies are meant to improve Gujarat Haryana Karnataka Rajasthan Tamil Nadu welfare programs and investment levels in lagging regions, Andhra Pradesh 1,000 Jharkhand West Bengal they need to be targeted to those regions, rather than to a Odisha Bihar Chhattisgarh specific good or service that may be consumed more heavily Madhya Pradesh Uttar Pradesh in richer states. 500 In Pakistan and Sri Lanka, poorer regions are obtaining a 10,000 20,000 30,000 40,000 higher level of per capita fiscal transfers. In Sri Lanka, the real gross state domestic product per capita highest levels of per capita funding have been allocated to the Source: Ministry of Finance 2008, 2009. North-East Province, the center of a long-running Tamil sepa- ratist movement. In this sense, fiscal transfers appear to ad- institutions. The largest component of fiscal transfers in India dress a political problem as well. In Bangladesh, however, no comes from the tax-sharing schemes, but the discretionary explicit mandates direct resource transfers toward poorer re- transfers and the subsidies put together are almost as large as gions. The World Bank’s recent Public Expenditure Report the tax shares. on Bangladesh raised the concern that poorer regions are be- Achieving Equity through Fiscal Transfers ing allocated lower levels of per capita development funding (World Bank 2009a). In most countries, fiscal transfers ensure equity across sub- national regions. This equity is important for economic and Helping Lagging Regions political reasons. Poorer regions have a lower base of eco- The focus on shared prosperity does not necessarily imply re- nomic activity to tax, and typically these regions spend con- ducing inequality, but rather emphasizes the need for a social, siderably less on social services, including education and economic, and institutional arrangement that maximizes the health care. Figure 3 shows social spending per capita across incomes of the less well off. Moreover, policies and programs states of India. that are consistent with these objectives have to be fiscally sus- Achieving horizontal equity through fiscal transfers can tainable. The right policies for a country are those that are ensure a level playing field. This equity can be particularly im- part of a detailed social contract that aims to end poverty per- portant if the government services are important inputs into manently, along an intertemporally balanced growth path and future growth potential, such as in developing a healthy and in an environmentally sustainable fashion. educated workforce. Growing regional disparities can cause The well-being of lagging and poorer states is constrained political tensions, and fiscal transfers can offset some of these by the total availability of resources. One way of overcoming disparities. the resources problem would be to divert resources from ex- Does the system of fiscal decentralization in India trans- isting subsidy-oriented programs toward safety nets, educa- fer more revenue to poorer regions? When looking at the dif- tion, health, and infrastructure. Another is to accelerate the ferent components of fiscal transfers in India, it is clear that privatization of central public sector enterprises (PSEs) and horizontal equity is being achieved only through the tax-shar- earmark the proceeds from these sales specifically to the de- ing schemes of the Finance Commission (figure 4a). The state velopment of much needed economic and social infrastruc- plan grants administered by the Planning Commission do not ture in the lagging regions. This would be a much more effec- seem to be directed toward the poorer states (figure 4b), tive way of helping the poorer states than the traditional whereas the discretionary schemes show higher per capita ex- approach of pushing existing PSEs to make commercial in- penditures in the richer states (figure 4c). vestments in the less developed states. Such initiatives have Food subsidies per capita are roughly uniform across done little in the past for the economic development of the poor and rich states, that is, if it is assumed that all subsidies area and have often increased the probability of driving the are used for the sale of food by the Food Corporation of India PSEs into sickness. On the other hand, privatizing existing (figure 4d). Conversely, if one allocates food subsidies on the central PSEs, and using the proceeds to build social and eco- assumption that all the subsidies are spent in food procure- nomic infrastructure in the backward states will increase the ment at above-market prices, then the highest levels of subsi- efficiency with which existing PSE assets are used, while si- dies are given to the leading states of Punjab, Haryana, and multaneously helping to improve the efficiency of resource Maharashtra (figure 4e). The true picture is probably a mix use in the poorer states and hopefully leveraging a greater flow of production and consumption subsidies, but the conclu- of private investment. 6 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise Figure 4. Different Types of Fiscal Transfers to Lagging Regions in India, 2005–6 a. Tax shares b. State plan schemes Orissa 300 1,200 Bihar Odisha Chattisgarh 250 Gujrat 1,000 Uttar Pradesh Madhya Pradesh Andhra Pradesh Haryana Rajasthan 200 Madhya Pradesh 800 West Bengal Tamil Nadu Chattisgarh Andhra Pradesh Jharkhand Karnataka Kerela Bihar Jharkhand West Bengal 150 Karnataka Tamil Nadu 600 Gujrat Rajasthan Kerela Haryana Punjab Punjab Maharashtra Maharashtra 100 Uttar Pradesh 400 10,000 20,000 30,000 40,000 10,000 20,000 30,000 40,000 per capita GDP (1999–2000 prices) per capita GDP (1999–2000 prices) tax share tranfers per capita �tted values state plan tranfers per capita �tted values c. Discretionary schemes d.Food subsidy Tamil Nadu 700 Punjab 4 Andhra Pradesh Odisha Karnataka 600 3 West Bengal 500 West Bengal Karnataka Madhya Pradesh Jharkhand Kerela Rajasthan Kerela Maharashtra Jharkhand 2 400 Odisha Gujrat Rajasthan Andhra Pradesh Bihar 300 Chattisgarh Tamil Nadu Haryana 1 Haryana Madhya Pradesh Uttar Pradesh Punjab Gujrat Maharashtra Chattisgarh 200 Bihar Uttar Pradesh 10,000 20,000 30,000 40,000 10,000 20,000 30,000 40,000 per capita GDP (1999–2000 prices) per capita GDP (1999–2000 prices) discretionary transfers per capita �tted values per capita food subsidy (sales) �tted values e.Food subsidy (purchases) f. Fertilizer subsidy Punjab 6 Punjab 40 Haryana 30 4 Andhra Pradesh 20 Haryana Karnataka Gujrat 2 Uttar Pradesh Chattisgarh Chattisgarh Tamil Nadu Maharashtra 10 Madhya Pradesh Rajasthan West Bengal Odisha Odisha Andhra Pradesh Bihar Jharkhand Kerela Maharashtra Bihar West Bengal Madhya Pradesh 0 0 KarnatakaTamil Nadu Uttar Pradesh 10,000 20,000 30,000 40,000 10,000 20,000 30,000 40,000 per capita GDP (1999–2000 prices) per capita GDP (1999–2000 prices) per capita food subsidy (purchases) �tted values per capita fertilizer subsidy �tted values Sources: Ministry of Finance 2008; Food Corporation of India for food subsidy allocation. Note: States in India can obtain resources from the central government in three main ways. The first consists of tax shares and grants decided by a nonpolitical Finance Commis- sion. Centrally sponsored schemes give various ministries grants to their counterparts in the states for specified projects either wholly funded by the center (central sector projects) or requiring the states to share a proportion of the cost (centrally sponsored schemes). These are wholly discretionary and often are not coordinated with Planning Commission transfers. In addition to these explicit transfers from the center to the states, a number of hidden or implicit transfers arise from the large subsidies provided by the central government for food and fertilizer, the existence of subsidized borrowing resources for the states from the central government or government-owned financial institutions, and tax exportation. The largest component of fiscal transfers in India comes from tax-sharing schemes, for which the Finance Commission has an explicit mandate to help poorer states. Conclusion The current patterns of economic growth are not sufficient to Given that South Asia has the world’s largest number of peo- enable the lagging regions of South Asia to catch up with the ple living below the poverty line, and that the majority of the leading regions in terms of proportional poverty reduction. poor are in lagging regions, policies must be specifically tar- 7 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise geted toward achieving greater growth and poverty reduction estimated per capita revenue and the per capita revenue of in these regions. Haryana, the second highest state in terms of revenue per One particular policy measure to achieve shared prosper- capita). ity is pro-poor fiscal transfers. The performance of South References Asian countries in achieving horizontal equity through fiscal transfers is mixed. For the most part, the systems of interstate Ahluwalia, M. 2000. “Economic Performance of States in the fiscal transfers in South Asian countries do transfer a greater Post Reforms Period.� Economic and Political Weekly May (6): 1637–48. amount of resources to poorer regions, suggesting that they Bardhan, Pranab. 2002. “Decentralization of Governance and De- are working to achieve greater equity. However, this outcome velopment.� Journal of Economic Perspectives 16 (4): 185–205. usually occurs when interstate fiscal transfers are transparent Chen, Shaohua, and Martin Ravallion. 2008. “The Developing and have explicit rules, and this is not always the case. World Is Poorer Than We Thought, But No Less Successful in Further, simply directing financial resources to lagging the Fight against Poverty.� World Bank Policy Research Working regions may not be sufficient and may need to be comple- Paper 4703, Washington, DC. mented with increases in capacity, accountability, and partici- Galasso, E., and M. Ravallion. 2005. “Decentralized Targeting of an Anti-Poverty Program.� Journal of Public Economics 89: 705–27. pation at the local level, so that poor regions can make full use Galiani, Sebastian, Paul Gertler, and Ernesto Schargrodsky. 2008. of these resources. “School Decentralization: Helping the Good Get Better, But Fiscal decentralization and other resource transfer poli- Leaving the Poor Behind.� Journal of Public Economics 92: cies can be complementary to policies that directly aim to ac- 2106–20. celerate the growth rate in lagging regions. This can be done Ghani, Ejaz, ed. 2010. The Poor Half Billion in South Asia—What Is by improving the business environment for the private sector, Holding Back Lagging Regions? Oxford University Press. Ghani, Ejaz, L. Iyer, and S. Mishra. 2010. “Are Lagging Regions supporting market integration, improving connectivity, and Catching up with Leading Regions?� In The Poor Half Billion in ensuring macroeconomic stability. The government can also South Asia—What Is Holding Back Lagging Regions?, ed. E. Ghani. intervene directly in sectors or locations in which the private Oxford University Press. sector is reluctant to invest. International Monetary Fund. 2006. Government Finance Statistics. Policy makers need to boost shared prosperity and take March. another look at the Millennium Development Goal para- Kanbur, Ravi, and Andy Sumner. 2011. “Poor Countries or Poor People? Development Assistance and the New Geography of digm. A new lens is needed—one that shifts the focus of policy Global Poverty.� http://kanbur.dyson.cornell.edu/papers/Kan- from national to subnational level, and from leading to lag- burSumnerPoorCountriesOrPoorPeople.pdf. ging regions, where poverty, gender disparity, and human Kraay, Aart. 2006. “When Is Growth Pro-Poor? Evidence from a misery are concentrated. Panel of Countries.� Journal of Development Economics 80: 198–227. About the Authors Lopez, Humberto, and Luis Serven. 2009. “Too Poor to Grow.� World Bank Policy Research Working Paper 5012, Washington, Ejaz Ghani is a Lead Economist in the Poverty Reduction and DC. Economic Management (PREM) Economic Policy, Debt, and Ministry of Finance, India. 2008. Indian Public Finance Statistics, Trade Department. Lakshmi Iyer is an Associate Professor at the 2007–8. New Delhi. Harvard Business School. Saurabh Mishra is at the International ———. 2009. Report of the Thirteenth Finance Commission (2010– Monetary Fund’s Research Department. 2015), Chapter 8. http://fincomindia.nic.in/ShowContentOne. aspx?id=28&Section=1. Note Ravallion, Martin. 2012. “Why Don’t We See Poverty Conver- gence?� American Economic Review 102 (1): 504–23. 1. These figures are for the 12th Finance Commission. The Seabright, Paul. 1996. “Accountability and Decentralization in 13th Finance Commission (Ministry of Finance 2009), Government: An Incomplete Contracts Model.� European whose recommendations apply from 2010–15, assigned Economic Review 40 (1): 61–89. World Bank. 2009a. Bangladesh Public Expenditure Review. Wash- weights as follows: state area, 10 percent; population, 25 per- ington, DC. cent; fiscal discipline, 17.5 percent (measured as the improve- ———. 2009b. PovcalNet. PovcalNet Online Poverty Analysis Tool. ment in the ratio of the state’s own revenue to total expendi- http://iresearch.worldbank.org/PovcalNet/index.htm. ture, relative to all other states); and fiscal capacity distance, ———. 2009c. World Development Indicators, World Bank. 47.5 percent (measured as the difference between the state’s ———. 2008. “Qualitative Decentralization Indicators.� World Bank. The Economic Premise note series is intended to summarize good practices and key policy findings on topics related to economic policy. They are produced by the Poverty Reduction and Economic Management (PREM) Network Vice-Presidency of the World Bank. The views expressed here are those of the authors and do not necessarily reflect those of the World Bank. The notes are available at: www.worldbank.org/economicpremise. 8 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise