JAMAICA 67809 Poverty and Social Impacts of Fiscal Reforms Report No. 67809-JM JAMAICA Poverty and Social Impacts of Fiscal Reforms Poverty Reduction and Economic Management Caribbean Country Management Unit Latin America and the Caribbean Region © 2012 International Bank for Reconstruction and Development / International Development Association or The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions ex- pressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. For permission to reproduce any part of this work for commercial purposes, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. GOVERNMENT FISCAL YEAR April 1 – March 31 CURRENCY EQUIVALENTS US$1.00 = J$87.40 (as of June 13, 2012) WEIGHTS AND MEASURES Metric System SELECTED ABBREVIATIONS AND ACRONYMS BOJ Bank of Jamaica IMF International Monetary Fund CARICOM Caribbean Community JDX Jamaica Debt Exchange CARTAC Caribbean Technical Assistance Center JLP Jamaica Labor Party CAS Country Assistance Strategy LAC Latin America and the Caribbean CCRIF Caribbean Catastrophe Risk Insurance MOF Ministry of Finance and the Public Services Facility CCT Conditional Cash Transfer MOU Memorandum of Understanding CDB Caribbean Development Bank MTF Medium-Term Framework CEM Country Economic Memorandum NEPA National Environmental Planning Agency CPI Consumer Price Index NROCC National Road Operation and Constructing Company CPS Country Partnership Strategy PATH The Programme for Advancement through Health and Education DBJ Development Bank of Jamaica PAYE Pay-As-You-Earn DMU Debt Management Unit PER Public Expenditure Review DPL Development Policy Loan PIOJ Planning Institute of Jamaica FRF Fiscal Responsibility Framework PFM Public Financial Management GCT General Consumption Tax PSED Public Service Establishment Division GDP Gross Domestic Product PSTU Public Sector Transformation Unit GNI Gross National Income SCT Special Consumption Tax GNP Gross National Product UNDP United Nations Development Programme GOJ Government of Jamaica USAID U.S. Agency for International Development IADB/IDB Inter-American Development Bank Regional Vice President Hasan A. Tuluy Country Director Françoise Clottes Sector Director Rodrigo A. Chaves Sector Manager Auguste Tano Kouame / Louise J. Cord Sector Leader Christine M. Richaud Task Team Leaders Denis Medvedev / Maurice Kugler ACKNOWLEDGMENTS This report was prepared by a team led by Denis Medvedev (Economist, LCSPE) and Maurice Kugler (Senior Economist, LCSPP) under the guidance of Auguste Tano Kouame (Sector Manager, LCSPE) and Louise J. Cord (Sector Manager, LCSPP). Françoise Clottes (Country Director, LCC3C) and Rodrigo A. Chaves (Sector Director, LCSPR) linked the team to the Bank’s overall strategy and steered them in that direction. The team included Christine Clarke (University of West Indies), Rafael de Hoyos (LCSHE), David Lawson (University of Manchester), Jann Lay (University of Goettingen), Michiel Paris (LCSPE) and Olga Romero (LCSPP). Florencia Liporaci (LCSPE), Patricia Holt (LCSPE) and Juliet Williams (LCSPE) provided valuable production and logistical support. The peer reviewers were Ambar Narayan (PRMPR), Collette Robinson (PIOJ), and Emily Sinnott (ECSH4). The team also benefitted from comments by Amparo Ballivian (LCSPP), Barbara Cunha (LCSPE), and Zafer Mustafaoglu (LCSPR). The team received valuable guidance through meetings with Jamaican authorities and stakeholders. Particular thanks are due to the residents of Trelawny and Bernard Lodge, Inswood communities who granted us their time through participation in interviews. The team is grateful to the Jamaican authori- ties for their continued cooperation, especially the Planning Institute of Jamaica. Table of Contents Executive Summary ................................................................................................................................. ii 1. Introduction ........................................................................................................................................ 1 2. Tax reform ............................................................................................................................................ 2 2.1 Reform background .......................................................................................................................... 2 2.2 Short-term impacts: tax incidence analysis ...................................................................................... 5 2.3 Long-term impacts: macro-micro simulation ................................................................................ 10 3. Public bodies rationalization ............................................................................................................ 18 3.1 Reform background ........................................................................................................................ 18 3.2 Quantitative results ........................................................................................................................ 20 3.3 Qualitative results ........................................................................................................................... 33 4. Conclusions ........................................................................................................................................ 44 Annex A: Tax incidence methodology and simulated tax rates........................................................ 47 Annex B: Description of the CGE Model ............................................................................................. 54 Annex C: Description of the Micro-Accounting Model...................................................................... 58 Annex D: Impact of discrete changes in independent variable(s) on the estimated probability .. 62 Annex E: Incorporating State Dependence.......................................................................................... 64 References ............................................................................................................................................... 65 i Executive Summary This report explores the poverty and distributional effects of a set of recent fiscal reforms by the Government of Jamaica. The specific policy actions—tax reforms and the privatization of sugar es- tates—have been identified through consultations between the Government of Jamaica and the World Bank team on the basis of the fact that these actions were, on the one hand, supported by the Bank through its Programmatic Fiscal Sustainability DPL series and, on the other hand, could be expected a priori to generate negative distributional effects. The report offers both quantitative and qualitative as- sessments of the potential poverty and distributional effects of these policy changes. The report uses tax incidence analysis to assess the poverty and distributional effects of tax reforms in the short term, while the longer-term impacts are modeled by linking a Jamaican household survey with a recursive dynamic computable general equilibrium (CGE) model. The impacts of public bodies rationalization are analyzed through both qualitative and quantitative lenses by complementing the empirical work with household and labor force surveys with community focus group interviews with affected workers. The report finds that relatively few households are likely to experience negative shocks to their in- come or consumption as a result of the reform measures analyzed here. The impact of the tax reform on the poor is limited due to the highly progressive nature of the fuel taxes and numerous tax exemp- tions for food items. Over the long term—assuming fiscal consolidation efforts are sustained—the tax reform and the Jamaica Debt Exchange are likely to substantially improve fiscal and debt sustainability, accelerate growth, and reduce poverty. Finally, privatization is likely to directly affect less than one per- cent of the Jamaican population. Although the affected households tend to be poorer than households not directly exposed to privatization, they are also more likely to be the beneficiaries of the conditional cash transfer program PATH. In addition workers in some sectors have been entitled to receive compen- sation for job losses (e.g. payments to laid-off sugar workers). Although few, workers in the sectors where privatization is taking place are at a higher risk from job losses than an average Jamaican worker. Public sector workers in the sugar and aluminum in- dustries are almost 25 percent more likely to lose their job than comparable workers in other sectors. Moreover, once unemployed, their chance of getting a job are significantly lower, suggesting that work- ers in the public sector have a particular set of skills that are not fungible (or marketable) enough to allow them to transit in and out of unemployment easily. In addition, the qualitative results revealed a series of worrisome issues affecting the communities in the neighborhoods of privatized sugar factories and therefore suggest that many of the issues identified in the 2006 social impact assessment of sugar reform, such as lack of information and limited access to training opportunities, continue to be relevant. Formal job training can make a significant difference in improving the likelihood of finding em- ployment. More skilled workers (as measured by the number of years of schooling) have a lower prob- ability of losing their job. And among the unemployed, education can help, but only in the medium term. Participation in training programs like HEART and NTA significantly reduces the probability of losing a job—among those working—and increases the probability of finding a job among the unem- ployed. Expanding the reach of these programs could therefore make a significant difference in helping workers transition out of unemployment. ii 1. Introduction This Poverty and Social Impact Analysis (PSIA) explores the distributional effects of a package of fiscal reforms initiated by the Government of Jamaica and supported by the World Bank under the Programmatic Fiscal Sustainability DPL series. The DPL series supports improved budget and debt management in order to reduce the debt overhang and create additional fiscal space for productive pub- lic spending, including social expenditures. The PSIA discusses the poverty and distributional impacts of the prior actions supported under the DPL, with a particular focus on two reform actions likely to have the most significant impacts: (1) tax reform and (2) public sector reform, focusing on rationalization of public bodies. The report offers both quantitative and qualitative assessments of the potential poverty and distri- butional effects of these policy changes. Tax incidence analysis is used to assess the short-term effects of tax policy changes, while the long-run impacts are analyzed in a forward-looking fashion by linking a Jamaican household survey with a recursive dynamic computable general equilibrium (CGE) model. The impacts of public bodies rationalization are analyzed through both qualitative and quantitative lenses by complementing the empirical work with household and labor force surveys with community focus group interviews with affected workers. The focus areas of this PSIA have been identified through consultations between the Government of Jamaica and the World Bank team. This PSIA identifies the particular channels through which the above reforms are likely to affect the various segments of the Jamaican society and explores the extent of the likely impacts. The results of this PSIA are expected to contribute to the policy dialogue in Jamaica in three key areas: (1) for policy actions that have already been implemented, such as major parts of the tax reform, the findings of the PSIA evaluate the expected impacts and aid in the design of future policies; (2) for policy actions that are still under implementation, such as the ongoing program of public sector rationalization, the findings of the PSIA contribute to ensuring that the implementation is carried out with minimal adverse distributional and poverty impacts; and (3) more broadly, the PSIA is expected to contribute to the broader policy debate about the costs and benefits of shifting towards a greater reliance on indirect taxation in Jamaica. These contributions—and the ample public understanding and buy-in they are expected to engender—are particularly important in the political economy context of Jamaica. The report is structured as follows: Section 2 analyzes the expected impact of changes in tax policy, Section 3 investigates the potential impacts of public bodies rationalization, and Section 4 offers some caveats and concluding remarks. Each section begins with a discussion of the reform background as well as the major supporters and opponents of the reform. The analysis in each section is presented with the least possible amount of technical details in order to maximize the appeal to a broader audience. For the interested reader, the methodological details of the empirical approaches em- ployed in this report are contained in the Annexes. 1 2. Tax reform 2.1 Reform background The tax regime in Jamaica is complicated, with nine major tax instruments applied at varying rates, numerous exemptions, waivers, and incentives. The main tax instruments in Jamaica include the income tax, stamp duty, general and special consumption tax (GCT and SCT), customs duty, transfer tax, property tax, education tax, travel tax, and betting and gambling tax. The three major taxes, the GCT, SCT, and income tax, account for approximately 70 percent of total tax revenues. The uniformity of the tax code continues to be adversely affected by numerous discretionary tax waivers, exemptions, and incentives, despite the recent removal of a number of GCT exemptions. Complexity has been a historical feature of the Jamaican tax system both in terms of the direct and indirect taxes. The complex structure of the tax system dates back to the days of independence in the 1960s, when the prevalent doctrine supported the extension of tax incentives to encourage private sector investment and activity at a time when the provision of public infrastructure was limited. During the 1970s period of state populism, income taxes were raised a total of five times and by 1985 there were six tax brackets with a zero threshold and the highest rate set at 80 percent. Midway during the decade of the 1980s, the tax rate for the highest bracket was reduced to 57½ percent. On the corporate income tax side, the regime offered preferential terms to the agricultural sector reflecting a general policy stance that elevated agriculture in attempt to promote the growing of Jamaican produce. Generally speaking, attempts at major tax reform began in 1986 and included the simplification of the personal income tax (PIT) with a single rate of 33⅓ per cent above the zero tax threshold and the implementation of the GCT. In 1992, the PIT rate was further reduced to 25 percent and the agricultural bias was also removed from the corporate income tax structure. In order to stabilize the revenue collection at the time of the global crisis, broaden the tax base, and improve the efficiency of tax collection, the Government of Jamaica (GoJ) introduced three tax packages during the FY 2009/10 (Table 1). The first package was introduced as part of the FY 2009/10 budget. The Government raised the excise tax on gasoline by J$8.75 per liter, and broadened the General Consumption Tax (GCT) base by eliminating exemptions on several items. The second package was introduced on September 29, 2009 and included increasing the GCT rate on telephone ser- vices from 20 percent to 25 percent and increasing the departure tax to J$1,800. The third package was introduced on December 23, 2010, and included raising the GCT rate from 16½ percent to 17½ percent, increasing the rate of GCT applicable to the tourism sector from 8.25 percent to 10 percent, applying an ad-valorem fuel tax of 15 percent, and raising the personal income tax rates on incomes above J$5 million. The fuel tax covered motor spirits (including E10/87, E10/90), automotive diesel oil, kerosene, and marine diesel oil. Unlike previous instances of tax reform—which were met with widespread public opposition and ultimately were rolled back—the most recent round generated little public controversy amid the broad understanding of the urgency and necessity of the policy actions. The history of tax reforms in Jamaica provides several examples of reforms that have been stalled and reversed due to popular 2 backlash and, at times, violent resistance. Riots due to fuel price increases in 1979, 1985, and 1999 led to looting, deaths, and eventually a reversal of policy. During the most recent round of tax reform, however, there were no public protests associated with the implementation of the tax packages. While the opposition raised concerns in several forums, including the national 2010 Budget Debate, about the potential negative impacts of the policy change, the comments related mostly to the manner in which the changes would be implemented rather than opposing the tax reform per se.1 The general population appeared to have an appreciation of the tough economic situation; in addition, the then expected IMF Stand-By Agreement (SBA) provided a level of authenticity to the need to implement measures to raise additional revenue in order to fund government operations. Newspaper articles attested to the fact that despite the concerns of the opposition, they approached discussions related to the issue with restraint. In addition, the trade unions advocated broadening the tax base in order to lessen the perceived dis- proportionate tax burden on employees subject to income tax deductions (e.g., the 350,000 workers subject to the pay-as-you-go (PAYE) labor income tax are estimated to make up only 28 percent of the labor force). 1 One area of concern was that the administration presented new measures such as the inclusion of additional items in the GCT tax base that were not properly advertised to the public and the list was unavailable for several days after the initial announcement. 3 Table 1: Revenue measures introduced during the FY2009/10 Date Expected revenue Description of revenue measures introduced impact (J$ million) Increase in income tax threshold -5,330 (Ministry Paper 26/09) Removal of income tax preferences 1,200 April 23, 2009 Reduction in stamp duty and transfer tax rates -644 Imposition of GCT on telephone instruments 736 Removal of GCT exemptions 7,500 Increase of SCT on petrol and CUF on petroleum products 13,328 Imposition of withholding tax on dividends for non-residents 1,341 Total expected revenue impact 18,131 Increase of SCT on cigarettes 1,840 (Ministry May 6, 42/09) Paper 2009 Increase of SCT on alcoholic beverages 530 Total expected revenue impact 2,370 September Increase of departure tax 609 (Ministry 29, 2009 107/09) Paper Increase of GCT on telephone calls and telephone instruments 1,100 Total expected revenue impact 1,709 Increase in the standard rate of GCT 3,600 (Ministry Paper 128/09, revised) Re-introduction of the ad valorem component of the SCT 9,400 Increase in SCT on cigarettes 1,400 December 23, 2009 Increase in the rate of GCT applicable to the Tourism Sector 1,200 Electricity for commercial and industrial customers 1,453 Pre-payment of GCT on value added merchandise at customs 2,900 Increase in income tax for high income earners 1,317 Increase in license fees for luxury vehicles 32 Removal of certain customs exemptions 25 Increase in Common External Tariff rate on luxury items 485 Total expected revenue impact* 21,812 Note: GCT stands for General Consumption Tax, SCT for Special Consumption Tax, and CUF for Customs User Fees. * For the first three sets of measures, estimated revenue impacts are provided for FY2009/10; for the last set (Ministry Paper 128/09), estimated revenue impacts are for a full calendar year. Source: Jamaica Ministry of Finance. 4 2.2 Short-term impacts: tax incidence analysis 2.2.1 Methodology and data This report captures the short-term distributional impacts of the GCT and fuel tax reform via tax incidence analysis. Tax incidence analysis is a simple partial equilibrium approach to assess the impact of changes in consumption taxes across the range of the income distribution. The incidence of consumption taxes is obtained by simply multiplying the GCT, fuel, or other consumption tax rate with the (net of taxes) value of each expenditure item consumed by each household, as reported in the 2007 Jamaica Survey of Living Conditions (JSLC).2 The implied tax burden under the initial, pre-reform conditions can then be contrasted with the tax burden under the new tax structure to determine which groups of households were more affected by the reform. Furthermore, the welfare impact can be ap- proximated by the change in the cost of purchasing the same consumption basket before and after the tax change (i.e., price effect due to the tax change, see Annex A for additional details). Tax incidence analysis is a partial equilibrium, short-term approach which does not take into ac- count second round behavioral effects. The dynamic effects, such as changing the composition of income between consumption and saving or the possibility of changing expenditure patterns due to relative price changes are ruled out. These shortcomings are balanced out by the highly detailed nature of the analysis, which permits the identification of changes in the tax burden at the level of each house- hold and each expenditure item. The approach is also a reasonable short-term approximation to the impact of the reforms because the GCT rate increase is too small to trigger much of a behavioral response and substitution options for fuel are very limited. Moreover, the longer-term effects of the reform are ad- dressed in the following section, which develops a forward-looking macro-micro simulation and allows for behavioral response of the households at the macro level. There are a number of additional important caveats to the incidence analysis. The approach assumes that value added taxes are entirely shifted from the seller to the consumer who then bears the full burden of the tax increase. Due to data limitations, the approach also does not take into account the fact that prices of exempted items include some GCT, as no credit is given for GCT paid on inputs for the exempt items. However, the resultant bias is likely to be small as zero-rated food items often mainly use zero-rated inputs. Similarly, because the data do not allow distinguishing between imported and domestic goods, the analysis does not account for changes in tariffs on imported goods. Finally, the approach does not consider indirect price effects, i.e., the fact that the increase in the price of fuel may have spillover effects on the costs of other items. The last assumption, common in tax incidence analysis, is justified because the focus of the analysis in this section is on the first-round effects and the general equilibrium model developed in the following section explicitly takes the spillover second-round effects into account. However, recognizing that fuel tax changes are likely to substantially affect the price of transportation and that the impact of the reform on the poor may be underestimated if only direct fuel costs are taken into account, the analysis includes an ad- ditional robustness check which allows for a full pass-through of rising fuel prices to transportation costs. 2 See Box 1 for a brief description of the survey data and Annex A for a list of expenditure items and the respective tax rates. 5 Box 1: Survey data description The micro-level analysis in this report is based on the 2005-2008 rounds of the Jamaica Labor Force Survey (JLFS) and the Jamaica Survey of Living Conditions (JSLC). The JSLC is a comprehensive household survey, which uses household consumption as the primary welfare metric. The survey has been conducted in yearly rounds since 1988 as a subset of the Jamaica Labor Force Survey (JLFS). The JLFS is conducted four times a year—in January, April, July, and October—and covers approximately 1.0-1.3 percent of the Jamaican population. The sample size for the JSLC is one-third of the households in the JLFS, or 0.33 percent of all households in Jamaica, and interviews are carried out face-to-face between May and August of each year. Because the JSLC is a subset of the JLFS—the same dwellings chosen for JLFS interviews in April are revisited in May-August for the JSLC interviews—the two surveys can be combined to extend the coverage of consumption surveys. The JSLC and LFS datasets were obtained from the implementing institutions already in processed form. The Planning Institute of Jamaica (PIOJ) and the Statistical Institute of Jamaica (STATIN)—the two institutions jointly responsible for survey implementation—have shared the data with the consumption aggregates, adult equivalence scales, the poverty line, and other post-interview variables already calculat- ed. Adult equivalence scales allow for calculation of poverty and distributional statistics while recogniz- ing that caloric intake requirements (and consequently, the income required to purchase the minimum adequate bundle of calories, i.e., to meet the poverty line) vary by age and gender. Therefore, a family with a large number of women and children may require less income to buy the minimum food basket than a household made up primarily of adult males. The JLFS, as opposed to the JLCS, is a revolving panel surveying households every quarter since 1992. Therefore, with the JLFS it is possible to see workers in and out of employment at different points in time, making this dataset an ideal one to estimate the probabilities of transition described above. The JLFS surveys more than 20,000 workers during the months of January, April, July and October. The Labor Force Survey is composed of 4 panels (groups of dwellings or PSUs A, B, C, and D as described in Table) that are surveyed for two consecutive quarters and then after a year. Every two years, all dwellings (PSUs) are replaced, so a panel can be followed for a maximum of 7 quarters (see panel A in Table). Structure of the JLFS Year Quarter Panels Surveyed 2005 January A, B April B, C July C, D October D, A 2006 January A, B April B, C July C, D October D, A Source: STATIN Note: The JLCS reports that “[t]he sample dwellings for the labour force are revised once in 3-4 years when a new sample of two PSUs from each revised sampling region is selected and new listings of dwellings prepared�. Additionally, the International Labour Organization (ILO) (http://laborsta.ilo.org/applv8/data/SSM3/E/JM.html), claims that the updating of the sample is “every 3 years on the basis of the new list- ings� and that the length of time for complete renewal of the sample is 1 year. Based on this information, statistical tests were performed to identify the same dwellings overtime concluding that after two years most of the dwellings are replaced. 6 2.2.2 Scenario results This section considers three alternative scenarios which illustrate the short-term distributional impacts of the FY2009/10 tax reform. The baseline scenario is the status quo pre-tax reform regime, approximated by the 2007 household survey. The first alternative scenario is the full package of tax amendments implemented by the Jamaican authorities in FY2009-2010, such as the increase of the SCT rate on petroleum and petroleum products (from J$7.36 to J$16.11 per liter of unleaded gasoline), the subsequent re-introduction of the ad-valorem component of the SCT on petroleum, the increase in the general rate of the GCT, the increase in the SCT on cigarettes, and the rise in GCT applicable to the tour- ism sector.3 The full impact of the reform is then decomposed into its two major components by separat- ing the change in fuel taxes and the change in the GCT base rate into different scenarios. Table 2: Tax incidence analysis scenarios Scenario Description Base (tax regime 2007) CGT rate 16.5 percent, fuel tax 13 percent, cigarette tax 85 percent, GCT on tourism 8.25 percent Tax reform 2010 CGT rate 17.5 percent, fuel tax 28 percent, cigarette tax 105 percent, GCT on tourism 10 percent Only fuel tax increase CGT rate 16.5 percent, fuel tax 28 percent, cigarette tax 85 percent, GCT on tourism 8.25 percent Only GCT CGT rate 17.5 percent, fuel tax 13 percent, cigarette tax 85 percent, GCT on tourism 8.25 percent The scenario results show that the tax reform increased the tax burden on the rich substantially more than on the poor. The results of the tax burden simulations are summarized in Table 3, which shows the average consumption tax rate paid by households in each consumption decile. This exercise illustrates that pre-reform consumption taxes were slightly progressive with a tax burden of 6.6 percent of total consumption for the poorest 10 percent of Jamaican households. This share rises to about 8 percent between the second and the third deciles and is higher still for households in the richer deciles. While the full FY2009/10 tax reform (the third column of Table 3) increased the consumption tax bur- den for all households, it did so considerably more for richer households. The tax burden for the poorest 10 percent rises to 7.1 percent – an increase of 0.5 percentage points – while it rises to 9.5 percent for the richest, an increase of 1.0 percentage points.4 The progressive impact of the tax reform is due primarily to the increase in the fuel tax. The decompo- sition of the full reform scenario into its GCT and fuel tax components shows that the progressivity is due to both the increase in fuel taxes and the overall increase in the GCT, but the impact of the fuel tax is more important. While the increase in the GCT (the last column of Table 3) is slightly progressive (with an increase 3 Because the fuel and cigarette taxes are specific taxes but the consumption quantities are not observed in the house- hold survey, the ad-valorem equivalents of the pre-reform levels were obtained from Edmiston and Bird (2004). 4 Note that these are decile averages, so individual households may experience even larger changes in the tax burden. 7 in the tax burden of about 0.3 percentage points in the lowest deciles and around 0.4 in upper deciles), the fuel tax increase affects those in the highest decile substantially more than the rest of the Jamaican population. Just the increase in the fuel tax accounts for 60 percent of the overall rise in the tax burden of the richest decile, but only 33 percent of the total increase in the tax burden of the poorest decile. Table 3: Tax incidence of base tax regime and alternative scenarios Taxes paid as percentage of total household consumption Scenario Baseline Full tax reform Only fuel tax Only GCT 1 6.59 7.11 6.76 6.91 Per capita expenditure decile 2 7.73 8.28 7.87 8.09 3 7.97 8.54 8.11 8.35 4 7.59 8.10 7.69 7.95 5 8.02 8.62 8.19 8.39 6 8.04 8.66 8.24 8.42 7 8.18 8.82 8.40 8.56 8 8.29 8.98 8.53 8.66 9 8.28 9.04 8.62 8.65 10 8.50 9.52 9.10 8.86 all 7.92 8.57 8.15 8.28 Source: Authors’ calculations with JSLC2007. In the near term, the increase in taxes results in a minor increase in poverty but also a decline in inequality. The increase in the tax burden shown in Table 3 leads to welfare losses for all households, with declines ranging from 0.5 to 1.25 percent of baseline consumption. However, because the negative impact is strongest for the households at the top of the distribution, poverty increases by just 0.2 per- cent of the population (Table 4).5 Moreover, because the negative impact of the reform on the better-off households is stronger than on the poor, inequality decreases by 0.1 Gini points. Table 4: Short-term impact of tax reform on poverty and inequality Scenario Baseline* Full tax reform Only fuel tax Only GCT Poverty headcount (%) 9.4 9.6 9.4 9.5 Poverty gap (x100) 2.3 2.5 2.5 2.5 Gini 38.1 38.0 38.0 38.1 Theil (GE1) 25.8 25.7 25.7 25.8 Source: Authors’ calculations with JSLC2007. * Note that the baseline poverty rate of 9.4 percent is different from the official poverty rate of 9.9 percent of the popu- 5 lation. The difference between the official poverty rate and the rate reported in Table 4 is due to rounding errors and slight imprecision in rebuilding the consumption aggregate from the bottom up. In other words, the consumption ag- gregate calculated by the authors as the sum of consumption of each expenditure item differs slightly from the official consumption aggregate reported in the household survey due to specific aggregation procedures employed by STATIN and PIOJ which could not be replicated by the authors. 8 The progressive distributional impact of the tax reform package is strongest at the two extremes of the distributions. The “reform incidence curves� in Figure 1 plot the change in real consumption per capita at (pre-reform) consumption per capita percentiles by comparing the cost of purchasing the same basket of goods before and after the change in taxes. The downward slope of the overall tax reform curve in the left panel of the graph shows that the distributional impact of the 2010 tax reform was quite progressive. Disaggregating the overall impact into the GCT and fuel tax components shows that the GCT increase only had a slight progressive impact, all of it explained by the relatively high share of exemptions in the consumption basket of the poorest decile. On the other hand, the fuel tax increase is highly progressive, affecting the top two deciles much more than the rest of the Jamaican population. Figure 1: Incidence of tax reform Change in real consumption per capita, percent 1 1 Change in real consumption per capita, percent 2010 fuel tax, direct effects 2010 reform, all measures 2010 fuel tax, incl. purch. transport Only 2010 fuel tax change .5 Only 2010 GCT change .5 0 0 −.5 −.5 −1 −1 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Consumption per capita percentiles s Consumption per capita percentile Source: Authors’ calculations with JSLC2007. Taking into account the impact of fuel tax increase on transportation costs does not change the finding on the progressivity of the reform. The poorest 20 percent of Jamaicans spend on average 7.3 percent of their consumption on transportation—a substantially higher share than the 5.3 percent spent by the richest 20 percent. Therefore, the extent to which higher fuel taxes translate into higher transportation costs could be an important determinant of the final distributional impact of the reform. In order to test the robustness of the earlier result to the inclusion of a potential rise in transportation costs, an alternative scenario increases the GCT rate on purchased transport from 16.5 to 23 percent.6 As shown in the right panel of Figure 1, in such a scenario the fuel tax increase loses some but not all of its progressive impact. 6 Such a scenario hence assumes a fairly high fuel-cost share of purchased transport of about 50 percent (and perfect transmission of the tax burden to consumers). 9 2.3 Long-term impacts: macro-micro simulation 2.3.1 Methodology and data In order to analyze the distributional impacts of tax reform over a longer term, this section devel- ops a recursive dynamic computable general equilibrium (CGE) model as well as the micro- accounting module used to translate the CGE results into poverty and inequality outcomes. The CGE analysis is carried out by contrasting a baseline simulation with a set of alternative scenarios for the years 2007-2020. The results of these simulations are subsequently mapped to the 2007 household survey (Jamaica Survey of Living Conditions) to explore the potential impacts of changes in the macro- economic and sectoral variables on household welfare, poverty, and the distribution of income. The CGE model of this chapter is a recursive dynamic extension of the model used by Bussolo and Medvedev (2008) to analyze labor supply dynamics and international competitiveness in Jamaica. At its core, the model is a standard World Bank single-country CGE model with a mostly neo-classical structure which is augmented by the addition of a labor-leisure tradeoff in the household utility func- tion, therefore allowing labor supply to be determined endogenously (see Annex B for model details). As discussed in Bussolo and Medvedev (2008), the introduction of an endogenous labor supply is particularly important in the case of Jamaica where labor force participation rates have been declining, to a large extent due to increased international remittances (see World Bank, 2007, and Kim, 2007). Beyond the endogenously determined labor supply, growth in the model is determined by savings/investment deci- sions (and therefore capital accumulation) and assumptions regarding productivity growth. Unlike more simple growth models, the multi-sectoral nature of the CGE model allows for more complex productivity dynamics including differentiating productivity growth between agriculture, manufacturing, and services and picking up the changing structure of demand (and therefore output) as growth in incomes leads to a relative shift into goods with higher income elasticities (manufactures and services). Moreover, the model has a diverse set of productive factors including factors and different skill categories of labor. CGE models, such as the one used in this report, are particularly appropriate for tax policy analysis given the high level of sectoral detail and the explicit link between fiscal performance and public debt trajectory. Effective sales (GCT) tax rates can vary in each of the 25 sectors included in the model, which also contains other tax instruments such as production taxes, labor taxes (PAYE), and income taxes (modeled separately for individuals and corporations). Since both current and capital government spending as a share of GDP are exogenous, calibrated to approximate actual outturns during 2008-2010 and maintained at recent historical averages thereafter, changes in tax rates lead to changes in the fiscal balance which must be offset by additional borrowing in domestic and international markets. Therefore, to be beneficial over the long run, a reduction in taxes must stimulate growth in order to offset the ini- tial negative impact on fiscal balances and debt accumulation. On the other hand, an increase in taxes can also be growth-inducing if such an increase leads to a reduction in public borrowing and therefore limits the crowding-out effect in the domestic bond market. The model has a base year of 2007, drawing on a collection of data sources. The base year data include a Social Accounting Matrix (SAM) constructed specifically for this exercise and employment by sector 10 and skill calculated from the 2007 Jamaica Labor Force Survey (LFS). The base year of 2007 was chosen in consultation with Jamaican counterparts as the most recent year that can be considered a stable period for Jamaica (since 2008 and 2009 were years of large external shocks and consequent distortions to rela- tive prices). The macro data for 2007, including national accounts, government financial statistics, and the balance of payments, were obtained from the Central Bank of Jamaica, Planning Institute of Jamaica (PIOJ), and the Statistical Institute of Jamaica (STATIN), while data on debt stocks was collected from the IMF. Sectoral data on value added and investment were also obtained from the above mentioned Jamaican sources, while the consumption vector has been estimated using data from the 2007 Jamaica Survey of Living Conditions (SLC). The input-output coefficients as well as the structure of indirect taxes have been taken from the 2005 Jamaica SAM constructed by Robinson and Willenbockel (2009).7 The model includes three representative household groups, defined according to the sector of em- ployment and skill level of the household primary earner from JSLC. The introduction of three rep- resentative households—unskilled rural, unskilled urban, and skilled—is important because the sources of income, both labor and other types, vary substantially by household type.8 For example, the two former groups receive very little income from bonds, while the latter group receives a substantial share of its income from interest paid on government bonds. Distinguishing across these three groups adds an important distributional dimension to the analysis, which otherwise would have been lost if all three groups were aggregated together. Following the base year, the model is solved year-by-year through 2020. In order to generate a dy- namic solution, certain assumptions have been made regarding the evolution of the model’s exogenous variables. BaU (business-as-usual, or “no-reform�) growth has been calibrated using the near- and medi- um-term assumptions of the IMF; the evolution of remittances, FDI, and the overall capital account also are drawn from IMF sources. For years beyond the IMF projection period (2015-2020), the assumptions extend the trends of several preceding years (2012-2014). The maximum labor supply available in each period evolves according to the World Bank population projections for the 15-64 age cohort, but the actual quantity of labor supplied in each period is determined endogenously by the model. The CGE simulations are complemented with a micro-accounting exercise similar to the tech- niques described in Bussolo et al (2008) and Ravallion and Lokshin (2008). The income distribu- tion of 2007, obtained from the 2007 JSLC (see Box 1), is shocked with CGE-generated changes in total household income for each representative household group, prices of consumption goods faced by each household group, and overall consumption per capita (as a consistency requirement) to produce a counter-factual income distribution in 2020 (the details of the micro-accounting approach are provided in Annex C). This allows the micro-accounting module to capture changes in both the average con- sumption and its distribution. Distributional changes are captured by the variation in income sources of different households (i.e., some households gain more than others because the returns to the different types of labor supplied by each household—skilled or unskilled, rural or urban—vary in the CGE model) 7 As is normally the case with combining multiple data sources, a final balancing procedure using a cross-entropy ap- proach had to be applied to the input-output coefficients. 8 Skilled primary earners are those that have some type of secondary school certificate. 11 and the differences in the types of goods consumed by each household (i.e., because each household’s consumption basket is slightly different from another, the overall price change experienced by each household will vary even if prices of individual goods change by the exact same amount). The overall change in poverty will then be an outcome of a combination of changes in the mean (how far the dis- tribution moves in one direction or another) and the shape of the distribution (whether inequality in- creases or decreases). The process is repeated for each scenario and, although solutions for intermediate years could easily be obtained, only the 2020 results are presented and discussed in this chapter. The micro-accounting approach rules out any behavioral response on the part of the households in the survey, but offers simplicity and tractability in return. A major caveat of the micro-accounting approach is that it does not produce any changes in household behavior; that is, although representa- tive household groups in the CGE model do change their consumption patterns in response to relative price changes, households in the survey maintain their initial consumption shares despite changes in prices. The advantage of this approach is simplicity, tractability of the results, and freedom from the relatively strict distributional and functional form assumptions imposed by more sophisticated mod- eling methods. On the other hand, this methodology is obviously less realistic than approach which allows households to change their consumption and labor supply decisions—although Bourguignon et al (2008) note that many times the additional structure tends to bring out further subtleties without necessarily changing the basic message of more simple approaches. Furthermore, modeling options for this report were also restricted by data limitations: while modeling labor supply decisions at the micro level could have been an interesting extension of the endogenous labor supply in the CGE model, this approach was not possible due to a large number of workers in the LFS with missing wages. Therefore, a simpler approach modeling welfare as a function of household consumption was chosen. 2.3.2 Scenario results The BaU (no-reform) scenario incorporates the adverse impacts of the global food, fuel, and finan- cial crises of 2008-09, and projects a smooth but gradual recovery in 2011-2020. The behavior of the main macro variables in the BaU scenario is summarized in the first few columns of Table 5. The com- bined impacts of the food/fuel and financial crises in 2008-10 lead to a cumulative loss of 4.6 percent of real GDP (relative to 2007). From 2011 onwards, the economy recovers gradually, achieving annual growth of 2 percent by 2012 and 2.1 percent in 2014 and each year thereafter. The baseline growth is plausible as the recent crisis created a considerable slack in the economy and per capita income grows at an average annual rate of 0.8 percent over the entire period 2007-2020. Following a sharp contraction in 2008-09, growth in exports and imports rebounds in 2010; after that, trade at constant prices grows at a rate similar to real GDP. The real exchange rate, which depreciated significantly in 2009, recovers gradually towards its 2007 level in the following years as the availability of foreign financing improves in the course of the global recovery. Under the no-reform conditions, poverty and inequality are likely to continue their downward trends but at a slower rate than in the past. World Bank (2011) showed that between 1997 and 2007, poverty in Jamaica fell from 19.9 to 9.9 percent of population, while the Gini coefficient declined from 38.3 in 2003 to 36.8 in 2007 (the 2007 outturns are reported in the second column of Table 6). However, 12 poverty spiked to an estimated 17.6 percent of the population in 2010 due to the severe adverse effects of the global crisis. The CGE model and the accompanying micro-accounting exercise do not take into account all of the adverse developments in the Jamaican economy during the 2007-2010 period and consequently the simulated 2010 poverty headcount of 12.4 percent (third column of Table 6) falls well short of the actual outturn. As the economy recovers from the crisis, poverty reduction is expected to resume, although not at the same rate as in the 1997-2007 period. By 2020, the moderate poverty head- count could fall by another 3.5 percentage points in the no-reform scenario, while the Gini coefficient could decline by half a point (fourth column of Table 6). The main reason the no-reform scenario does not deliver more poverty reduction is the relatively slow rate of growth in per capita consumption. Additionally, this scenario does not incorporate the pro-poor shift in relative prices observed between 2003 and 2007.9 Finally, poverty reduction is more difficult when the initial headcount is lower: it takes a much higher rate of growth to halve poverty from an initial headcount of 10 percent than from an initial headcount of 20 percent. The alternative scenario incorporates the major tax changes in fiscal years 2009 and 2010 as well as the Jamaica debt exchange (JDX) in February 2010. This scenario (JDX + tax reform) includes the two largest measures (in terms of expected revenue impacts) from the FY2009/10 tax reform: the increase in the fuel tax (both the SCT and the ad-valorem components) as well as the increase in the base GCT rate (Table 1). In addition, the scenario also takes into account the decrease in the rate of interest paid on domestic and foreign debt after the JDX. 9 Between 2003 and 2007, overall inflation was 57 percent and the prices of food and non-alcoholic beverages rose by 65 percent, while the cost of the consumption basket of the poor (i.e., the poverty line) increased by just 49 percent. 13 Table 5: Macro Summary Indicators, 2007-2020 No-reform JDX + Tax Reform 2007 2008 2009 2010 2015 2020 2008 2009 2010 2015 2020 National accounts (percent change y-o-y) GDP at constant prices -1.70 -2.50 -0.50 2.10 2.10 -1.70 -2.62 -0.41 2.34 2.48 Private consumption -0.7 -4.9 1.3 2.5 2.6 -0.7 -5.6 0.7 2.6 2.7 Investment 7.5 -38.8 10.6 1.3 0.0 7.5 -37.4 16.3 2.5 2.0 Exports -7.8 7.2 5.2 1.8 2.0 -7.8 6.5 3.6 1.9 2.2 Imports 0.4 -18.9 10.7 2.2 2.0 0.4 -19.3 11.0 2.5 2.5 Balance of payments (US$ million) Current account balance -1,968 -2,315 -864 -1,020 -1,328 -1,356 -2,315 -864 -1,020 -1,328 -1,356 (as percent of GDP) -14.5 -16.9 -7.1 -7.9 -9.3 -8.6 -16.9 -7.1 -7.8 -9.2 -8.3 Balance on goods and services -3,344 -3,776 -2,083 -2,301 -2,980 -3,292 -3,776 -2,078 -2,396 -3,194 -3,696 (as percent of GDP) -24.6 -27.6 -17.0 -17.7 -20.9 -21.0 -27.6 -17.0 -18.4 -22.0 -22.7 Exchange rate 68.95 68.32 72.11 69.26 69.20 69.33 68.32 71.87 68.87 68.74 68.79 Public finance (percent of GDP) Overall balance -4.1 -6.9 -8.7 -10.6 -12.5 -17.0 -6.9 -7.1 -5.3 -3.8 -3.4 Primary balance 6.6 4.4 5.2 3.9 6.5 7.2 4.4 6.8 5.9 8.4 8.8 Total revenue 23.3 23.4 22.7 22.9 23.2 23.3 23.4 24.2 24.7 24.9 24.8 Total expenditure 27.4 30.3 31.4 33.5 35.7 40.2 30.3 31.3 30.0 28.7 28.2 Recurrent expenditure 24.0 26.0 27.9 28.5 32.3 36.8 26.0 27.8 25.0 25.3 24.8 Capital expenditure 3.4 4.3 3.5 5.1 3.4 3.4 4.3 3.5 5.1 3.4 3.4 Debt 115 127 141 145 168 205 127 140 139 128 120 Foreign 51 67 78 81 93 111 67 77 79 77 75 Domestic 64 60 63 64 75 95 60 63 60 51 45 Memo (percent change y-o-y) Employment 1.638 2.430 1.125 1.020 0.325 1.638 2.362 1.138 1.034 0.346 Population 0.426 0.463 0.499 0.368 0.343 0.426 0.463 0.499 0.368 0.343 Source: Authors’ calculations. The improved fiscal performance in the alternative scenario is conditional on continued efforts at fiscal consolidation. In the no-reform scenario, the Government is assumed to maintain primary balances of around 6 percent of GDP in the medium term—consistent with the historical performance in the recent years. Nonetheless, this relatively strong performance is not sufficient to prevent further deterioration in the debt ratios, primarily due to the unsustainably high debt servicing costs (in the con- text of low GDP growth and limited capacity to increase fiscal revenues). In the alternative scenario, tax reforms and a containment of primary expenditure raise the primary balance to an average of 8 percent in the medium term. Only this further fiscal consolidation, together with substantially lower interest payments, is sufficient to reverse the increasing trend in the debt-to-GDP ratio and put the debt back on 14 a sustainable path. This underscores the importance of continued fiscal discipline for future continued debt and fiscal sustainability, particularly in light of recent (FY2011/12) slippages in fiscal performance. The strengthening in the fiscal position has important positive spillovers for real GDP growth. Although initially growth is impacted negatively by the tax increase (the contraction in 2009 is 2.6 percent rather than 2.5 percent without the tax increase), growth recovers quickly. By 2015, growth is 0.2 percentage points higher than in the no-reform and by 2020, this premium widens to 0.3 percent- age points. As a result, GDP per capita in 2020 is more than two percentage points above the BaU GDP of the same year. The positive growth spillover is mainly due to the decreased borrowing needs of the government, which has a positive impact on private investment through reduced crowding-out effects. Real consumption per capita, on the other hand, remains unchanged in the alternative scenario. This is because the JDX + tax reform scenario in reality represents a transfer of resources within the economy and does not generate new income. The gains in real GDP are observed because the economy’s resources are shifted towards accumulation of new capital stock, rather than financing of debt which does not create factors of production. However, the reduction in public borrowing is only made pos- sible by taxing households and enterprises at higher rates and lowering the rate of return on their bond holdings. Therefore, initially the policy shock represents an income loss to the households, which is then compensated by increased demand for investment goods—which stimulates domestic output and increases demand for domestic factors of production—and the efficiency gains stemming from the fall- ing capital-output ratio in the no-reform scenario. However, the efficiency gains are also limited by the choice of financing instrument, since increases in indirect taxes distort economic incentives and erode competitiveness. 10 Figure 2: Allocation of consumption spending by percentiles of the income distribution Consumption shares by welfare percentiles 100% 90% Services Agricultural goods Manufactured goods Petroleum-based fuel 80% (incl. food) 70% 60% 50% 40% 30% 20% 10% 0% 1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 Source: Authors’ calculations with Jamaica Survey of Living Conditions, 2007 10 This scenario is an example of a second-best outcome, when adding an extra distortion (higher taxes) may be ben- eficial in the presence of other distortions (high debt). 15 Although shifts into indirect taxation are usually regressive—because poor people consume a larger portion of their income—the incidence of the fuel tax is actually progressive, as shown in the previous section. As shown earlier, the consumption of gasoline (as a share of total consump- tion spending) is an increasing function of household welfare (Figure 2). For any household, gasoline consumption is a small share of total household budget; however, a poor household normally spends approximately 3 percent of its total expenditure on gasoline while the same share for a non-poor house- hold is 4.7 percent. Therefore, the increase in the petrol tax imposes a higher burden (in relative terms) on the rich households than on the poor.11 On the other hand, the incidence of a reduction of interest payments on domestic debt is quite progressive. There is no information in the JSLC or other readily available data sources on the distribu- tion of interest income from government bonds in Jamaica. However, the JSLC does contain a ques- tion on dividend, interest, and rental income, and this information has been used to allocate the bond interest receipts across the three representative household groups in the CGE model. The data indicate that households with skilled primary earners receive approximately 84 percent of all dividend, interest, and rental income in Jamaica, while households with unskilled primary earners in urban occupations receive 15 percent. Therefore, the distribution of interest income is heavily biased towards richer skilled households, and the poorest households—those with unskilled primary earners in rural occupations— receive less than 1 percent of total interest income. As a result, a reduction in interest earnings due to the JDX would affect rich households much more than poor households. Table 6: Micro Summary Indicators, 2007 and 2020 2007 2010 2020, 2020, JDX + no-reform Tax Reform Extreme poverty headcount (%) 2.86 3.46 1.54 1.54 Poverty headcount (%) 9.91 12.39 6.41 6.11 Poverty gap (x100) 2.48 3.01 1.44 1.40 Poverty gap squared (x100) 0.95 1.15 0.53 0.52 Gini 36.78 36.16 36.32 36.03 Theil (GE1) 23.75 22.92 23.13 22.74 Note: 2007 outcomes are actual, the rest (2010 and 2020) are simulated. Source: Authors’ calculations. The micro model results show that the reform scenario leads to additional poverty reduction and lower inequality, although the differences with respect to the no-reform scenario are quite small. The third column of Table 6 shows that in the JDX + tax reform scenario, the moderate poverty head- count could fall by an additional 0.3 percentage points, while the Gini coefficient could decline by more than a quarter of a point. Although these differences are very small, there are several reasons why one 11 In the absence of other concurrent reforms, the positive distributional effect of increased petrol taxes is negated by its negative income effect on average consumption, resulting in a 0.14 percentage points increase in moderate pov- erty in 2020. 16 may not expect to observe a big boost in poverty reduction in this scenario. First, as mentioned earlier, consumption per capita does not increase in this scenario relative to no-reform, which means that pov- erty reduction can come only from distributional change. Second, although some distributional changes are pro-poor—as discussed earlier, the increase in the petrol tax and the reduction in bond coupon pay- ments affect rich households more than the poor ones—other changes are biased against the poorest households. These include the overall increase in the GCT rate, which hurts poor households who spend all of their income on consumption, and the large increase in demand for investment goods, which do not require rural factors in the production process. As a result, the wages of unskilled rural households decline relative to the labor earnings of unskilled urban and skilled households. On balance, the pro- poor distributional changes outweigh the anti-poor ones but the overall distributional change—and hence poverty reduction—is not particularly large. 17 3. Public bodies rationalization 3.1 Reform background 3.1.1 Historical Context The divestment program, including the current public bodies rationalization plan, followed the nationalization of Jamaican industries shortly after independence and into the 1970s. These na- tionalizations reflected both the political thought of the day as well as a response to the global crisis and the associated capital flight. The administration platform in the 1970s articulated a position based on the notion of democratic socialism which was argued to require government ownership/control of the means of production, distribution and exchange. Government then proceeded to take over the banking system, public utilities sugar and bauxite. In addition there was a perception that the large amount of foreign investment would make the society dependent on foreign sources and this was viewed as being undesirable. Handa and King (1996) argue that this ownership scale and structure reflected the “com- manding heights� philosophy of the government at the time. In the 1980s, the Government began a process of reducing the role of state in the economy. During the 1970s, the population became increasingly disillusioned with the nationalization agenda as well as the increasing inability of the government maintain its interventionist stance in the economy. Capitalizing on the declining public sentiment towards this stance, the opposition at the time commit- ted to reducing the involvement of government in the economy in its manifesto preceding the 1980 election. This was to be accomplished through the privatization of state owned enterprises as well as targeted public sector retrenchment. In 1981, the Jamaican Ministry of Finance listed 400 entities on its records as being wholly or partially owned by the Government of Jamaica. These entities were operating in the areas of finance, tourism (where the government controlled approximately one-half of the rooms), minerals, petroleum, construction and agriculture. Several attempts at rationalization have since been pursued beginning in the 1980s under different institutional approaches (Table 7). A major portion of the privatization process in Jamaica occurred during the most intensive period of structural adjustment between 1989 and 1993, characterized by a relatively deep public sector retrenchment to the tune of 20,000 jobs. Altogether by 1996, only 15 percent of the entities that were publicly owned in 1980 were divested, however these represented the largest of the government holdings. The policy framework for the more recent round of divestments was set out in the Ministry Paper 34/1991. The MP 34 established the National Investment Bank of Jamaica (NIBJ) Limited as the co- ordinating and implementing agency for privatization activities in Jamaica. The Cabinet retained full responsibility for approving and reviewing the privatization program. The NIBJ was accountable to the Office of the Prime Minister which provided policy guidance for the implementation of the divest- ment process subject to MP 34. During its operation, the NIBJ completed 101 sales and approximately 30 to 40 leases. Although MP 34 provided for the sale or lease of government lands, the NIBJ deferred to the National Land Agency in this aspect of divestment. The NIBJ was formally merged with the 18 Development Bank of Jamaica (DBJ) in September 2006 with the DBJ assuming responsibility for the GOJ privatization processes. Table 7: Rationalization attempts in Jamaica, 1981-1994 Time Divestment 1981 – 1985 2 net divestments 1986 5 divestments 1986 – 1988 3 divestments per year 1989 – 1994 8 divestments per year Source: Compiled from Handa and King (1996) The latest round of reforms in this area has largely proceeded from a philosophy which redefines the role of the state in the economy but also takes into account the fiscal impacts of years of losses by public bodies. The recently formed Public Sector Transformation Unit (PSTU) in the Office of the Prime Minister has identified things that government should do and should pay for; what it should pay for but should not do; and what government should not do and should not pay for based on Oshorne and Plastrik (1997). The entities and processes that are inconsistent with this new role are therefore expected to be referred for closure or divestment. This included, for example, the Sugar Company of Jamaica, whose accumulated losses reached J$15.8 billion as of May 27, 2010. The proposed reforms have largely been met with broad support: in the case of sugar companies, for example, there has been a remarkable amount of political consilience around the current privatization efforts due to the mount- ing debt burden, the obsolescence of the factories and the perceived inability of the government to efficiently manage the industry. The impetus for divestments has been given further support by the dif- ficult fiscal situation, although there have been instances, such as the sugar industry and Air Jamaica, where entities that have been divested have reappeared on the governments’ books to the detriment of the fiscal accounts.12 3.1.2 Current Privatization Efforts Recently, the Government’s program has focused on privatization in sugar, mining, hotels, and air transport. The recently privatized entities include Air Jamaica, six publicly-owned sugar estates (St. Thomas Sugar Factory at Duckenfield, Long Pond and Hampden in Trelawny, Monymusk in Clarendon, Frome in Westmoreland, and Bernard Lodge in St. Catherine), and Pegasus Hotels. Air Jamaica became part of Caribbean Airlines in May 2010. The first three estates were sold in June 2009, while the sale agreement for the last three was signed in July 2010. The Sugar Company of Jamaica Limited is now a legacy company that was slated for closure in December 2010. Finally, a sale agreement for Pegasus was reached in September 2010. The most comprehensive assessment of the impact of industry change on socioeconomic issues was conducted for the sugar industry in 2006. This assessment took place prior to the imminent 12 The SCJ had previously been privatized in 1994 but was reacquired by the government when the private owners de- cided to dissolve the conglomerate partnership with the government. 19 change in the European Union’s sugar price regime with the objective of identifying the potential im- pact on the livelihoods of workers post-regime change (PIOJ, 2006). At the time of the study, the sugar industry was estimated to be the direct source of income and employment for 38, 000 Jamaicans and when considered in conjunction with the significant price cut of 39 per cent over a five-year period, would mean that the social impact could be quite substantial. The 2006 study focused on workers at the Long Pond and Bernard Lodge factories due to the above normal poverty levels and the low-levels of post-primary education among the workers. It was anticipated that the most affected groups was likely to be the non-administrative factory staff and the field workers. The respondents also expressed fear that the resultant idleness could increase the number of persons who engage in crime as a coping strategy. This would be accompanied by increased disease rates, poverty, hunger, community decay, and a decline in interpersonal relationships. They expressed hope that the process would involve the distribution of parcels of sugar land to workers and the provi- sion of housing. The 2006 study provided a broad set of recommendations to help mitigate the adverse impacts on the workers. The interventions identified included: an education campaign about the redundancy process and composition; a comprehensive survey to determine the skill profile and facilitate the devel- opment of training programs; the provision of information about alternative sources of employment incentives and relevant credit access and advice to encourage the start up of businesses; and, the con- tinued provision of social services such as health, education, counseling and nutritional support. The report estimated that the interventions would require expenditure of J$629million. 3.2 Quantitative results Privatization has both direct and indirect effects on household welfare. Figure 3 shows the short- run welfare effects of privatization as predicted by economic theory, drawing from La Porta and López- de-Silanes (1999). The direct effects operate through three main channels: firms, labor markets, and the equilibrium in the market for goods produced by privatized firms and consumed by households. The firm channel works primarily through increased competition, which has a positive effect in terms of the efficiency in the use of inputs, the use of new technologies and innovation in general (Sheshinski and López-Calva, 2003). The labor channel operates through a change in the input mix, usually reduc- ing the amount of workers to give space to more machinery and technology. Although these changes often bring about increases in unemployment (McKenzie and Mookherjee, 2003), they also usually result in higher labor productivity for those workers that remain employed. Finally, the increased com- petition combined with the efficiency gains from input reallocation would be enough to push prices of final products down while increasing the quantities produced as well as their quality (Birdsall and Nellis, 2002; McKenzie and Mookherjee, 2003). And all of these changes could have an impact on eco- nomic growth which could imply further changes in the way incomes are distributed across households (Bourguignon and Pereira de Silva, 2003). 20 Figure 3: Short run welfare effects of privatization Privatization Fiscal Revenues (+) Firms: Labour Market Effects: Market Equilibrium: 1. Competition (+) 1. Employment (–) 1. Prices (–) 2. Efficiency in the 2. Labour 2. Quantity (+) use of inputs (+) Productivity (+) 3. Quality (+) 3. Use of new 3. Wages (–) technology (+) Income Side Household Consumtion Side Welfare reduces Welfare Effects Welfare increases due to higher due to lower prices unemployment and higher quality and lower wages Source: Authors, based on the discussion in La Porta and López-de-Silanes (1999). The analysis in this report focuses primarily on the labor market channel. The industries recently privatized in Jamaica sell most of their output abroad (sugar, bauxite, and tourism). Therefore, few Jamaican consumers would benefit from increased competition and lower prices per se. And while in- creased efficiency could certainly benefit Jamaica—a country long characterized by extremely low pro- ductivity growth (World Bank, 2011)—this is both a longer-term and more macro-oriented outcome. On the other hand, the employment effects of privatization are likely to be visible immediately. Moreover, while the potential benefit of increased competition and lower prices could be shared by all consum- ers, the incidence of an employment shock is much discrete. Certain households will bear the majority of the labor costs of privatization through the job loss of their members and distributional outcomes may suffer to the extent that these households may have already been vulnerable prior to the reform. Therefore, identifying the group of workers affected by the reform—together with understanding the determinants of these workers losing or finding jobs—could help the authorities design policies to miti- gate the potential adverse impacts on household welfare. 3.2.1 Households directly affected by privatization One of the most important costs of privatization is the loss of jobs in the privatized industries. If the authorities can identify vulnerable groups (e.g., poor households) among those who work in an in- dustry to be privatized, they could provide these individuals with some sort of safety net to cushion the negative shock. This section uses the 2007 Jamaican Labor Force Survey (JLFS) and the Jamaican Living 21 Conditions Survey (JLCS) to identify the households with jobs, and hence incomes, linked to the public bodies subject to privatization (hereafter households at risk). The objective of this exercise is twofold. First, it attempts to get an order of magnitude of the privatization reform in terms of the number of households and individuals it could affect. Second, it compares characteristics between households who depend on a sector to be privatized (vulnerable households) and the rest of the population. Given the difficulty to identify in JLCS, at the 4-digit level of disaggregation, the public bodies to be privatized, when identifying the households that are part of the sector to be privatized, the exercise considers only three sectors: sugar, aluminum and ethanol.13 Table 8: Differences between Households with Information on Industry and the Rest Households Households All Significant Reporting Not Reporting Households Difference Industry of Occ. Industry of Occ. Consumption per capita (mean) 165,761 167,716 151,141 *** (1864) (2019) (4598) Poverty Rate (%) 9.91 9.38 13.92 *** (0.36) (0.38) (1.24) Gini coefficient 0.37 0.37 0.38 Path Beneficiary (%) 9.37 8.79 13.67 *** (0.37) (0.37) (1.30) Household size 3.30 3.46 2.44 *** (0.05) (0.05) (0.10) Years of schooling 8.70 8.84 8.05 *** (0.05) (0.06) (0.14) Gender of household head (Male) (%) 53.82 57.11 36.46 *** (1.16) (1.27) (2.82) Source: Authors’ calculations with JSLC data. The households considered by the analysis in this section could be a non-random sub-sample of all Jamaican households potentially affected by privatization. An important data limitation of the JLCS is that only 84 percent of all households report the industry where the principal earner works. Therefore a first question to answer is whether this subsample is representative of the entire population. Table 8 shows that households with reported sector of occupation have higher consumption per capita, lower 13 To identify the number of households and individuals at risk of losing their jobs (i.e. workers in public bodies subject to be privatized), we rely on the JLFS (2007); the JLCS is then used to compare characteristics of households at risk versus other households. In other words, the labor force survey is used to count the number of individuals (and households) at risk of losing their job, but the living condition survey is used to compare household characteristics. The use of two surveys is justified based on the type of information that we can extract from each of them and the limitations of the living conditions survey. 22 incidence of poverty, fewer female household heads and are more educated. Therefore, households that report a sector of occupation are, indeed, better-off than those who did not report and, statistically speaking, these two groups cannot be considered as being part of the same population. Reporting or not reporting the sector of occupation might be related to households belonging to the formal and informal sectors in Jamaica. In particular, those who do not report the sector of occupation are more likely to be part of the informal sector where income and education levels are lower. Therefore, the welfare effects of privatization that are discussed in the present study are based on changes taking place in a subset of the population, possibly capturing the effects taking place in the more formal sectors of the economy.14 The reform is likely to directly affect few Jamaican households. Figure 4 shows that, according to the JLFS, of the total population of 2,675,800 in Jamaica in 2007, 95,406 (3.5 percent) of them live in a household whose head was working in an industry where privatization is taking place. But the great majority of them (89,780) were already part of the private sector and only 5,626 lived in a household with a head employed in a public body to be privatized (households at risk). In other words, the priva- tization reform could directly affect the livelihood of as many as 5,626 Jamaican citizens (0.2 percent of the total population) living in 2,274 households whose heads are employed in public bodies subject to privatization. In the case of sugar, this number may somewhat underestimate the total employment effect because many small sugar farmers would sell their output to government-owned estates, without being employed by the estates directly. However, there a total of just 13,700 of small sugar farmers in all parishes with privatized estates, which means that the aggregate employment effects would still be limited even if a large proportion of these small farmers could not find new buyers for their sugar. Figure 4: Population potentially affected by privatization 89,780 5,626 Not directly related with a body to be privatized Household head employed in industry where privatization is taking place (private) Household head employed in 2,657,070 industry where privatization is taking place (public) Households with a direct link with a public body to be privatized are significantly poorer than the rest of the population but are also more likely to benefit from social assistance programs. Table 9 shows the average characteristics of households at risk versus other households. The figures shown in Table 9 are based on an identification of households at the 2-digit level of disaggregation since at the 14 Workers in public firms subject to privatization are most likely to be in the formal sector. 23 4-digit level only 21 households in the sample can be classified as household at risk. Although the aver- age characteristics of these 21 households are no different from the characteristics of other households in the same sector at the 2-digit level, their standard errors are significantly larger. Therefore, the com- parison of households at risk versus other households shown in Table 9 is based on a sectoral identifica- tion at the 2-digit disaggregation level. As expected, household heads of the poorer households at risk are less educated than other households, however, surprisingly, there are more male-headed households among the population at risk.15 Overall, households at risk have a level of consumption per capita 30 percent lower than other households. This massive difference in consumption explains the disparities in poverty rates with 18 out of 100 households at risk being below the poverty rate compared with less than 6 among other households. However, a higher fraction of the households exposed to privatization are PATH beneficiaries than the rest of the population. Thus, these vulnerable households are more likely to be covered by safety nets associated with conditional cash transfers than the rest of the population. Table 9: Characteristics of Households with a Direct Link with a Public Body to be Privatized Affected Other All reporting Households Households Significant Households Difference (2-digit) (2-digit) Consumption per capita (mean) 167,715 117,505 189,678 *** (2019) (2258) (2643) Poverty Rate (%) 9.38 17.84 5.67 *** (0.38) (0.9) (0.36) Gini coefficient 0.37 0.33 0.37 Path Beneficiary (%) 8.79 13.17 6.87 ** (0.37) (0.81) (0.41) Household size 3.46 3.38 3.5 (0.05) (0.1) (0.06) Years of schooling 8.84 8.34 9.08 *** (0.06) (0.09) (0.07) Gender of household head (Male) (%) 57.11 73.89 49.51 *** (1.27) (1.94) (1.55) Notes: (1) standard errors in parenthesis. (2)*** significant difference at the 1% level. (3) The second column “All reporting households� includes only households with information on the sector of occupation of the HH head. Source: Authors’ calculations with JSLC. The findings of this section suggest that, although the impact of privatization on average welfare and aggregate poverty indicators is likely to be marginal, many of the affected households are a part of the more vulnerable segments of the Jamaican society. The analysis presented so far shows that the privatization 15 This is surprising since, international evidence shows that, female-headed households tend to be poorer than male- headed households (see Bussolo and De Hoyos, 2009). 24 reform might affect the livelihood of a relatively small proportion of the population (0.2 percent). In the short term, the reform will not have a significant effect on aggregate measures of wellbeing such as wages, employ- ment and poverty. Nonetheless, data from JLCS shows that households whose head works in a sector subject to privatization are worse-off than other households. These results do not change if households at risk are identified using a 4- or 2-digit level of disaggregation. Hence, even if the privatization reform may be small in terms of its macroeconomic effects, the 2,000 or so households at risk are significantly poorer, less educated and hence more vulnerable than the rest and may need government assistance to cushion the shock. 3.2.2 Employment effects One of the most important linkages between the privatization reform and household welfare is via employment. However, due to data limitations it was not possible to observe the effects of privati- zation on employment. Instead, the available data were utilized to obtain insights into the functioning of the labor market in Jamaica and in particular what determines the probability of losing a job and, once unemployed, what determines the probability of finding a new one. The objective of the empirical strategy is to identify the characteristics that make an individual more likely to lose his or her job and, once unemployed, identify the characteristics that make him or her more likely to find a job and draw lessons for public policy based on these findings. In order to accomplish this objective, the present study analyzes all quarters included in the JLFS for years 2005, 2006, 2007 and 2008.16 Figure 5: Probability of transition by length between periods .045 .7 Unemployment to Employment Employment to Uemployment .6 .04 .5 .035 .4 .03 .3 0 2 4 6 8 Length between periods of observation (quarters) to Unemp to Emp Source: JLFS 2005, 2006, 2007 and 2008 More than half of unemployed Jamaicans manage to find a new job within a year after losing one. Given the structure of the JLFS, it is possible to analyze transitions of different lengths (in terms of quarters between observations of the same dwellings). In particular, labor transitions from employment to unemployment and vice versa, can be analyzed for the following lengths of quarters between observations 16 The complete JLFS panel for dwellings surveyed during 2009-2010 was not available at the time of writing this report. 25 of the same individual: 1, 3, 4, 5 and 7. Figure 5 presents the proportion of the employed (left axis) and unem- ployed population (right axis) that transit to unemployment and employment, respectively, by length between periods of observation. The figure shows that more than half of the unemployed population manages to find a job after 4 quarters (one year). During the same length of time, less than 4 percent of the employed population loses his or her job. After one year in employment, the probability of a job separation starts declining. Employment probabilities depend on personal characteristics like education and age. Figure 6 presents the probability of transition by years of schooling and age, respectively, distinguishing between short- (L1) and medium-run (L4) transitions.17 Interestingly, the relationship between the probability of transition and years of schooling is not monotonic. Ex-ante, one could have expected to see that more educated workers have a higher probability of transiting out of unemployment (dashed lines), however this is only true in the medium run (L4) and it is quite the opposite in the short run (L1). In other words, workers with more education have a higher probability of exiting unemployment after one year (80 percent) but a lower probability after three months (40 percent) than their lower-skilled counterparts—perhaps reflecting the fact that more skilled workers have higher reservation wages and/or higher savings which result in longer times spent looking for a new job. Figure 6: Probability of transition by… … years of schooling … age .05 .7 1 Unemployment to Employment .15 Employment to Uemployment Employment to Unemployment Unemployment to Employment .04 .6 .8 .03 .1 .5 .6 .4 .02 .05 .4 .3 .01 .2 .2 0 0 0 5 10 15 0 20 40 60 Years of Schooling Age Emp to Unemp L1 Emp to Unemp L4 Emp to Unemp L1 Emp to Unemp L4 Unemp to Emp L1 Unemp to Emp L4 Unemp to Emp L1 Unemp to Emp L4 Vertical line represent the average years of schooling in Jamaica Vertical line represent the average age in Jamaica Source: Authors’ calculations with JLFS. The relationship between the probability of losing a job and years of schooling is non-monotonic. For workers with years of schooling below the national average (close to 10), more years of schooling in- creases their chances of losing their job, however, for workers with schooling above average, additional schooling reduces their chance of losing their job. Finally, restricting the sample to the working-age population (between 15 and 65), Figure 6 shows that younger workers have a much higher probability of losing their jobs, both in the short- and medium-run (continuous lines). Additionally, given that an individual is unemployed, his or her probability of finding a job in the short-run increases substantially with age; after one year, age does not seem to have an effect on the probability of abandoning unemployment (dashed lines). This suggests that more experienced workers can afford to conduct longer job searches. 17 The curves in Figures 4 and 5 were smoothed using a local average regression kernel with optimal bandwidth. 26 The probability of transiting from employment to unemployment also varies across industries. According to Figure 7, workers in construction or mining industries have higher chances of losing their jobs (more than 6 percent) compared to agricultural or government workers (less than 3 percent). Dividing workers into those employed by the government versus the private sector shows that the for- mer face a probability of losing their job that is roughly a third (1.8 percent) of the probability faced by the latter (6.3 percent). This fact, by itself, suggests that, as the reform transfers workers from the public to the private sector, the overall risk of unemployment in the economy increases. Figure 7: Probability of transition by industry 10 Pr (Employment to Unemployment), % 8 6 4 2 0 F g g er n es . e t es m en nc & in rin io at ic ic om ,F in nm ct ra rv rv W tu ru re M su se Se C ac & er tu st In t& nd uf er ov on ity ul th an & sa G ric or C ic O e M tr sp el Ag nc ec ot an na El H Tr Fi Source: Authors’ calculations with JLFS. Workers with formal training have a lower likelihood of becoming unemployed. Workers that received training in their current employment have a probability of losing their job of 2.8 percent, compared to a prob- ability of 4.3 percent for those who did not get any formal training. This simple comparison of the probability of passing from employment to unemployment between workers that received training versus others seems to corroborate the hypothesis of Card and Sullivan (1988) that training is associated with both a higher prob- ability of sustaining employment and a higher probability of a successful job search when unemployed. Regression estimates confirm the previously reported links between unemployment and age, edu- cation, and training. In order to identify the variables which significantly affect the probability of losing and finding a job, respectively, we estimate a formal “mover-stayer� model. In order to test if the effects affecting the probability of transition are different after one quarter versus one year, separate es- timations are undertaken for the short- and medium-run, respectively. The transition functions are estimated using a standard probit model, accounting for survey weights and using robust standard errors.18 All results 18 The latter is important for accounting for unobserved individual-level heterogeneity. 27 are presented in terms of marginal effects, therefore the coefficients should be interpreted as the change in the probability of observing an outcome (losing or finding a job) given a change in the independent variable, with values of all independent variables being set at their mean. The results of estimating this model are shown in Table 10. Although the point estimates differ in the short versus the medium run, there is no change in signs. Therefore, the typology of the “vulnerable� worker will be the same regardless of the time length used in the analysis. According to the results in Table 10, young (or inexperienced) women, with unfinished primary school, in private sectors out of agriculture, with no special training and working in entities to be privatized are the most vulnerable workers. In other words, this type of workers are the most likely to lose their jobs. Table 10: Determinants of the probability of transition Pr(losing a job) Pr(finding a job) Short run Medium Short run Medium effects run effects effects run effects Sex (Male=1) -0.877 -2.080 18.626 15.842 Experience -0.173 -0.257 -0.653 -0.555 Experience squared 0.002 0.003 0.016 0.008 Years of schooling 0.327 0.335 -1.583 4.442 Years of schooling squared -0.034 -0.045 0.091 -0.219 Public sector worker -0.960 -1.534 -6.987 -2.915 Worker received training -0.736 -1.137 -1.487 6.176 Working in entity to be privatized 2.429 2.611 -6.706 -8.457 Industry Controls Mining and Quarrying 2.792 6.930 -17.996 -25.162 Manufacturing 3.787 5.832 -19.098 -27.346 Electricity & Water 13.715 10.543 -5.244 -32.382 Construction 11.739 14.797 -11.555 -29.068 Hotels, tourism 2.471 4.257 -23.875 -23.946 Transport/Communication 2.680 6.226 -13.346 -12.046 Finance & Insurance 1.565 4.259 -20.252 -19.383 Government Services 2.149 4.223 -16.266 -28.972 Other Services 5.524 7.985 -16.120 -21.218 Year Controls 2006 -0.116 -0.140 2007 -0.728 -1.706 2008 0.146 1.980 N 26,682 29,288 2,123 2,442 R2 7.43 7.14 5.4 3.53 Source: Authors’ estimates with JLFS. Even though the qualitative relationship between observables and the probability of losing a job are the same in the short and medium run, the effects strengthen in the medium run. For almost 28 all variables, the estimated coefficient is larger when the length of the transition is one year rather than one quarter. This result suggests that the vulnerability of disadvantaged workers is exacerbated in the medium to long run. Finally, the year controls included in the short-run estimation show the effects of 2008 economic crisis. While, in general, the probability of a job separation fell between 2005 and 2007, in 2008, this probability increased, regardless of the workers’ characteristics. Controlling for other factors, the probability of losing a job for workers with above-average years of schooling is substantially lower than for less-educated workers. Figure 8 shows the probability of losing a job—given that an individual is unemployed—at different values of years of schooling.19 At very low levels of education, the probability of losing a job increases slightly with additional years of schooling (both in the short and the medium run). However, for workers with 5 or more (4 for the medium-run) years of formal education, an additional year of schooling reduces their probability of losing a job. Figure 8: Estimated relationship between years of schooling and probability of losing a job 6 5 Pr (losing a job) 4 3 2 Short-run (after one quarter) 1 Medium-run (after one year) 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Years of schooling Source: Authors’ calculations with JLFS. Contrary to what we would have expected based on the results from the probability of losing a job, for workers with less than 20 years of experience, an additional year of experience reduces the probability of getting a job. This makes the youth less vulnerable or more likely to find a job (given that they are unemployed) compared to mid-age workers. This is true in the short- and medium-run. As shown by Figure 9, the effects of years of schooling on the probability of finding a job depend on wheth- er the individuals are followed and their labor status is captured after one quarter or one year. More educated workers are less likely to accept a job offer after one quarter of being unemployed, however, their probability of getting a job after one year of unemployment is much higher compared with poorly educated workers (see Figure 9). The reason behind these results is that workers with more human capi- tal (i.e. with more experience and years of schooling) also have a higher reservation wage, which, in the 19 See Appendix C for an explanation of how the probit results were used to construct Figures 7 and 8. 29 short-run, reduces their probability of accepting a job just a quarter after being unemployed. After one year of unemployment, workers with more human capital are more likely to find and accept a job. Although public sector workers face a lower probability of losing a job, they also find it much more difficult to gain new employment once unemployed. All individuals used in the estimation of the probability of finding a job were unemployed in the base period. Therefore, the variable “public sec- tor worker� and the industry controls were based on previous employment information. The result on “public sector worker� is interesting since, as already described, public sector workers more job security. However, once they lose their job, the probability of finding a new one is 6.98 percentage points lower than individuals who previously worked in the private sector. This result suggests that workers in the public sector have a particular set of skills that are not fungible enough to allow them to transit in and out of unemployment in an easy way. The result might also be the outcome of an inefficient allocation of labor in the public sector where the number and characteristics of the workers employed are not the result of a profit-maximizing criterion. Figure 9: Estimated relationship between years of schooling and probability of finding a job 65 Short-run 60 (after one quarter) Pr (finding a job) Medium-run 55 (after one year) 50 45 40 35 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Years of schooling Source: Authors’ calculations with JLFS. Workers in the industries with entities to be privatized not only face a higher probability of los- ing their job but also a more difficult time finding new unemployment. Once unemployed, their chance of getting a job after one quarter (year) is, 6.7 (8.5) percentage points lower than workers in other industries. These figures suggest that workers in the entities to be privatized are particularly vul- nerable to employment shocks. The positive impact of job training on employment described earlier remains robust when con- trolling for other determinants of finding or losing a job. In the short-run, workers who received specialized training, either by the employer or via the government training programs such as HEART or NTA, have a lower probability of getting a job, perhaps motivated by a higher reservation wage. However, after one year, those who received special training have a probability of getting a job more 30 than 6 percentage points higher than those who did not get specialized training. On the other hand, it should be acknowledged that those who have received HEART/NTA training may represent a non- random sample of workers; while the LFS data does not offer readily-available instruments to control for selection bias, its potential existence offers an important caveat to this finding. Table 11: Determinants of the probability of being employed Pr(being employed) Short-run effects Medium-run effects Previous employment status 48.902 29.206 Sex (Male=1) 1.922 2.958 Experience 0.150 0.197 Experience squared -0.001 -0.002 Years of schooling -0.377 -0.054 Years of schooling squared 0.039 0.032 Public sector worker 0.946 1.729 Worker received training 0.881 1.572 Working in entity to be privatized -2.901 -2.911 Industry Controls Mining and Quarrying -4.100 -8.818 Manufacturing -5.250 -7.820 Electricity & Water -12.581 -12.668 Construction -11.274 -16.003 Hotels, tourism -4.067 -5.626 Transport/Communication -3.629 -6.814 Finance & Insurance -3.013 -5.366 Government Services -3.009 -6.169 Other Services -6.427 -8.858 Year Controls 2006 0.126 2007 0.734 2008 0.003 N 28,805 31,730 R 2 35.83 21.85 Source: Authors’ calculations with JLFS. The above results are robust to controlling for persistence effects in employment or unemploy- ment. As demonstrated by Card and Sullivan (1988), one of the most important variables determining the probability of employment in t is the worker’s employment history, i.e. so-called state dependence.20 Table 11 shows the results of an alternative estimation approach which controls for state dependence 20 See Annex D for methodological details. 31 by including the individual’s lagged employment status in the set of independent variables. Being em- ployed the previous quarter increases the chances of being employed this quarter by almost 50 percent- age points. After one year, the information on previous (one year ago) employment is still large and significant with a point estimator or almost 30 percentage points. The results under this alternative specification are not different than the previous ones: being male, experienced worker in the public sector and having received specialized training increases the change of being employed, controlling for previous employment history. For workers with 5 or more years of formal schooling (remember that the national-wide average is close to 10), an additional year of schooling translates into a higher chance of being employed. Agricultural sector workers have a significantly lower probability of being unemployed compared to workers or individuals in other industries. Finally, individuals that are or were part of the aluminum or sugar industries are less likely to be employed by almost 3 percentage points. Taking all factors into consideration, workers in industries under privatization are about 25 per- cent more likely to lose their job than workers in other industries. The conditional probabilities reported in Table 10 and Table 11—showing that controlling for personal characteristics, workers in public entities to be privatized face a higher probability of losing their jobs—do not tell the full story as, for example, workers in these industries could be over-represented by young women with unfinished primary school and no special training. In order to address this question, a probability of losing a job was calculated for all active workers in the JLFS of October 2008 using the estimated coefficients from Table 10. Figure 10 shows the distribution of the predicted (or fitted) probability of losing a job, distin- guishing between workers in public entities to be privatized and workers in other industries.21 Overall, workers in the sugar and aluminum industries have an average predicted probability of losing their job of 4.6 percent compared with 3.7 percent for workers in other industries (vertical lines in Figure 10). 21 The areas below the densities in Figure 9 have been scaled to represent the proportion of the population in and out of the industries to be privatized, 5 percent and 95 percent, respectively. 32 Figure 10: Predicted probability of transiting from employment to unemployment 20 Workers in other industries 15 Workers in public entities under privatization Density 105 0 0 .05 .1 .15 .2 .25 Probability of transition Vertical lines represent the average probability for both types of workers Source: Authors’ calculations with JLFS. 3.3 Qualitative results “Sugar is sweet but working in this industry is not so sweet, as workers right continues to be abused� – Bernard Lodge Community Member This section summarizes the findings from a series of interviews in the sugar privatized areas of Trelawny and Bernard Lodge, Inswood. Participatory-based research interviews at the community, household and individual level took place in December 2010, one year after the privatization process, and were followed up with subsequent visits in June 2011 and March 2012. By covering a 16 month post-privatization period with three field visits, the qualitative assessment of this section provides a unique time dimension and fuller understanding of the social and poverty impact of the reforms on sugar communities. Although only two community groups were interviewed extensively, individual qualitative data was collected across several communities in Jamaica. Of the two communities inter- viewed, they were in close proximity to the privatized sugar factories, with one (Bernard Lodge) hav- ing been previously visited in 2006 for an assessment under an earlier study of sugar privatization in Jamaica. Overall we found the effects of the sugar privatization to be quite similar across communities. 33 The population samples used for qualitative fieldwork—both for the community and life histo- ries—are not nationally representative. As stated above, the purpose of the small scale ‘pilot’ type nature of the research was to triangulate some of the quantitative findings, whilst providing further understanding the key underpinning issues. Community interviews took place in two communities that were selected based on their proximity to privatized sugar factories, and thus findings may exemplify negative effects of the privatization process. Additional bias may also result from the fact that the com- munity interviews took place at the local community office. For example the recording effects at the community level may greatly reflect those that had an incentive to attend or could afford to transport themselves to the meeting. The individual participants of the life histories interviews were selected on a more random basis than the community participatory interviews. For the life histories, villages based on geographical access were selected and then a ‘random walk’ through the village, i.e., stopping at random households, was adopted for interviews. Although biases may still exist in such an approach, given the resources available they more than adequately meet the research objectives. 3.3.1 Community interviews Focus groups conducted prior to the privatization revealed concerns about substantial communi- ty-level impacts of the reforms. The outcomes of these focus group interviews are graphically sum- marized in Figure 11 (abridged from PIOJ, 2006). The figure highlights the anticipated impact factors from privatization, including increased stress, children being forced out of school, and an increase in the street population. The community also felt that such factors would then lead to an increased likeli- hood of killings, drug use and higher levels of illiteracy and migration from the area, all of which would culminate in the existence of a ‘no good community’. It was largely felt that communities that are pre- dominantly dependent upon single industries such as sugar, would potentially ‘collapse’ with few if any benefits resulting from privatization and the suggestion that some areas would become ‘ghost towns’. Yet, shortly after privatization, the interviewed communities were characterized by overwhelm- ing optimism for the future. In early 2011, interviewed communities expressed genuine ‘hope’ that the privatization of the local sugar factory would result in a more prosperous future; however, the un- derlying premise for such hope lay in ‘re-recruitment’ or ‘retention’, to the jobs formerly held at sugar factories. Expectations of re-employment in the sugar post-privatization were high and divestment was thought to be a good thing that would result in investment into the factories and improved manage- ment practices.22 However, given that many communities had received little or no information regard- ing the Sugar Transformation Unit (STU) payments, the processes associated with receiving payment (Box 2) and to what extent re-employment at local sugar factories would occur, some individuals started to ‘hustle’ by selling juice and small crops (peppers/lime/other) on the streets, while others became taxi drivers or resorted to localised prostitution. 22 This could potentially reflect a bias in the cohort attending the community meetings, as those who were able to attend were likely the ones with greater opportunities. Interestingly, such findings were largely confirmed when random one to one discussion occurred in villages. 34 Figure 11: Anticipated consequences of factory closure in St. Catherine Sickness No Good Community People leave Pressure on community Can’t bury people the community More Killing Can’t read More dead people Drugs or write Children More Street Poople not in Sick people School Loss of House Crime Lack of Stress Employment LOSS OF JOB Source: Innswood Focus Group, Adapted from PIOJ 2006 However, optimism and hopes for re-employment gradually diminished over time. By June 2011 the level of anger amongst former sugar workers was far higher than in the previous round of interviews: most were still waiting for STU payments, the previously ad hoc ‘hustling’ had become semi-permanent jobs for many, and those less physically able struggled the worst as localised informal community sup- port gradually reduced. By March 2012, there was an increased reality that re-employment would only be for the “favored few�. There was a general acceptance that many of the former sugar workers would not be re-employed and that re-employment was allocated on the basis of “who you know� and whether one had the skills to use tractors – indicative of the movement to use technology in the fields and to ensure the sugar industry reaches a minimum level of efficiency. 35 Box 2: Sugar Transformation Unit Payments The STU was formed as part of the implementation of the Government of Jamaica’s Sugar Adaptation Strategy that was slated to end in 2015, but this has been extended to 2020. The ongoing program is in three parts: • Component One: Completion of the privatization with the goal of developing a sustainable industry that is commercially viable • Component Two: Mitigate the fallout from the industry through the Sugar Dependent Area Development Program • Component Three: Debt reduction of approximately J$20bn. In addition, a sub-program relates to Sugar Area Development Project (2008 – 2013) for approximately J$2bn of which J$1bn is spent on Social Services and the balance on grant programs. The divestment program resulted in re- dundancy of all workers in December 2008 when notice pay was effected. Redundancy payments were made on June 16, 2009 was made to 7,200 workers to the tune of J$2.3bn. STU payments were determined by assessments at the sugar factories with the goal being to determine how to provide further assistance to the most vulnerable. The intent was to stabilize the area around the estates dur- ing the transition period of the divestment and as such they were initially expected to begin after privatization and therefore was not overly concerned with focusing on building long, lasting livelihoods. Given that of the 7,200 work- ers directly affected by the redundancy, approximately 2,500 were re-employed by the Sugar Company of Jamaica. Additionally, another 500 were indirectly employed by contractors. This left approximately 4,200 displaced workers. The grant system provided an in-kind grant J$170, 000 to females who were vulnerable and displaced (i.e. not re-em- ployed in any capacity) whereas the males were provided with J$150, 000. Those earning more than twice the national minimum wage (approx. J$397, 000 per annum) were ineligible for grants and could only benefit from social services such as the preparatory schools located at factories and clinics. All children of the vulnerable remained in the schools on property and their tuition was paid by the STU. Clinic attendance was also paid for by the STU via reimbursement. This grant program was implemented by the STU, which established offices and employed approximately 28 members of staff. A total of 3,200 vulnerable and displaced were registered over the period of 4 – 6 months by the STU based on the lists generated by the SCJ – the STU has no ability to adjust the list. In addition, approximately 1,000 small cane farmers were eligible. After registration, beneficiaries are allocated to STU staff, who work with approximately 140 – 150 members – the ideal number would have been 35 clients per employee but this was not possible given the staff complement. STU staff prepare a program for each beneficiary based on their interests in spending their grants. Approximately 25 to 30 percent indicated that they wanted to use their grant to pay down mortgages and another 5 - 10 wanted a housing grant. The balance wanted to either open small businesses including in agriculture or to pay for children to complete school. Pensioners were facilitated through the Credit Union through which they received an annuity of J$6, 000 per month that would last for approximately 24 months. As at the end of September 2010, one thousand two-hundred grants had been processed. In March 2011 the workers’ aspect of the grant program was completed. With respect to the 1,000 small farmers that are also being targeted by the program, the pro- cess has started however the priority is on addressing the workers. Due to the poor state of the factory at Long Pond, the Minister of Agriculture has granted permission for a partial grant of J$100, 000 regardless of gender to the 135 members of staff there. 36 Post-privatization, there was a broad recognition that the lack of available employment had major (negative) multiplier effects for spending/employment within the local community. Community members suggested that incomes have declined by more than 50 percent post-divestment. For example, group participants noted that many people ‘were unable to purchase basic foods’ – particularly single- occupied, former sugar industry households headed by individuals over 60 years of age. It was perceived that the current sugar privatization (or at least information relating to the process) is in ‘chaos’ – with suggestions that at least 50 percent of those previously employed were not able to find new employ- ment. Furthermore, community interview participants proposed that the overall number of jobs lost due to sugar privatization was more than double the direct job losses. Bars, cook shops, taxis, and other businesses closed because of the loss of income from the sugar workers. One wholesaler in Spanish Town (near the Bernard Lodge sugar estate) who used to cash sugar workers checks is now on the verge of closing.23 ‘…the economic situation has got gradually worse over the last 18 months…. At first when ‘the Chinese’ came the people were full of hope of re-employment but now it seems that this will not happen… although some still hope for a job from the Ethanol plant…’ - Wholesaler, Spanish Town. Job security and work patterns became increasingly uncertain. This was particularly noted for the period between the December 2010 and June 2011 community group interviews, when work had been suspended at the Bernard Lodge. Workers reported being sent home without any indication as to when work would be resuming and were informed from middle of June not to report to work “if they wake up and it was raining.� Unfortunately, in June 2011 the rain lasted for almost 2 weeks. Thus, community members felt as if growing cane was no longer a priority in Bernard Lodge. In a number of cases the lack of jobs prompted households to engage in risky coping strategies. Many households reported having no access to either public (e.g., conditional cash transfer program PATH) or private safety nets (e.g., remittances are reported to be non-existent in Bernard Lodge/Monymusk and in Clarks Town). Thus, households resorted to selling of assets, reducing food consumption to less than two ‘full’ meals per day, higher levels of distress borrowing from families and neighbors which may have reached a critical level in some instances due to uncertainty of future income streams (‘many people have borrowed against future STU payments’), and migration of key family members, particularly men who leave their families. In several cases elderly headed households who were formerly employed in the sugar industry were resorting to stealing cane from fields as a source of gaining nutritional intake: some members of the Innswood community reported eating one meal per day and “..jump[ing] in to the cane field to eat ‘cane’.. “ if they got hungry. However, interview participants did indicate that outward migration was limited and only occurred once households had exhausted their internal coping mecha- nisms (e.g., selling of assets/borrowing from other households/friends). 23 The wholesaler would request that customers purchase merchandise worth at least 10 percent of the value of the check in order to exchange the check for cash. 37 Figure 12: Causal flow actual (short run) consequences of sugar factory privatization Sickness Pressure on Community People leave the community Family Distress/Break Up/Extra Diversify Family/Community Jobs/Skills (–/+) Dependance/Possible Killing/Civil House Loss Disruption Drugs Children not in School Coping Mechanisms: Borrow/Sell Assets/ Reduce Consumption Crime/Prostitution Stress Lack of Employment Loss of Job/Insecure Re-employment PRIVATIZATION OF SUGAR FACTORY Source: Trelawny & Innswood Focus Groups, 2010 In the worst affected areas parents stated that they could only afford to send their children to school 2-3 days per week. Interviewees stated that “sending children to school is no longer a priority�; consequently, there has been an increased level of ‘loitering’ as a result and some students have dropped out of high school. One example given by the community forum was of a typical farm worker who on average earns J$9,000 per fortnight. From this ammount, approximately J$2,000 (10*J$200) would be paid for transport and J$2,500 (10*J$250) for lunch “… can you tell me how such individuals find the money to feed and send their children to school?� Gender-specific effects were also mentioned in the interviews. Female-headed households appear to be particularly vulnerable due to having above average numbers of children and higher dependency ratios, in the households, once families separate. Both community groups identified that divorce/separa- tion of families had increased since the period of sugar privatization started. This was particularly noted for families directly affected by the loss of jobs—which has had a series of gender effects with women in the household commonly ‘being the ones left to cope and look after the children.’ Although no direct information was sought relating to alternative income sources of such groups, it was commonly per- ceived that few income earning options existed for those severely affected. Increased levels of crime appear to be an issue. People in affected communities ‘live from day to day’ and, consequently, theft has increased. In Trelawny the reported deaths increased immediately after privatization because of ‘gangs’ but have subsequently reduced. In addition to the perceived impact of 38 increased crime, community members in general appear to have a relatively high tolerance for crime levels: “…we have a few rapes now and then and a few murders but we don’t have any crime and violence…�. Although few, if any, individuals have lost their homes following the privatization, the affected areas are already some of the more disadvantaged in the country with generalized lack of access to housing, proper sanitation, or land rights. In the Bernard Lodge area/estate most houses are con- sidered to be ‘shacks’, with houses made/’pieced together’ from wood/boards. Prior to the privatization process, Bernard Lodge Estate had previously been approached to renovate such houses but refused to re-invest, in the hope that such buildings would fall down and the tenants would move out. This was not an uncommon policy for many sugar estates across Jamaica. Hence, land tenure for residents is, and has been, very insecure. The resulting effects of this are that security associated with livelihood coping mechanisms is reduced, with fewer opportunities to diversity incomes, allow for crop cultivation, etc. As a result of being unable to access land formally, and delays associated with accessing STU funds, several groups of smallholder farmers have clustered together and started growing food crops for consumption and selling them locally to sustain themselves. However, this tends to be undertaken on land that is not owned by such groups and carries the obvious risks, such as eviction and destruc- tion of crops. On the other hand, the authorities suggest that issuing land titles to smallholders is now a priority and could be one method to combat the growing informalities associated with some of the ‘hustling’ that has increased. In addition, the communities sense that Government schemes to develop housing have been “talked about� without sharing substantive information; although such schemes have started there is a perception that they are not affordable for many residents and that the schemes need to be related to employment and not just an STU payments. Recently, residents were required to begin paying for housing, water, and electricity. Until 2011, residents on sugar estate owned land did not pay for electricity or water. However, in November 2011 families/individuals living in the houses owned by sugar estates were notified that they will have to pay for the electricity, although up to March 2012 no bills had been presented. In addition, all persons now living on the estates \ received a letter from the SCJ Holding indicating that everyone was required to pay $2,000 per month and upwards, depending on the type of property occupied. Several persons were relocated to accommodate the construction of the new offices for SCJ Holdings and other desired prop- erty was acquired by persons connected with the organization. Although government-sponsored housing assistance programs exist, accessing them has proven difficult for many sugar workers. The National Housing Trust (NHT) is a government agency which provides housing solutions for communities and offers the provision of home loans/mortgages. Over the years, sugar workers have had their NHT contributions deducted from their basic pay but it appears that this has not been paid over to NHT. The effects of this are having an increasing impact, as former sugar workers try and continue to ‘do business’ with the NHT, post sugar employment. However, as there is an information void—i.e. a person’s pay slip shows the amount deducted but formal NHT re- cords show a nil receipt—any benefits for such individuals are hard to receive. 39 “Claimants who have acquired housing under the Sugar housing program have been told that their benefit will be paid to the NHT, but in many instances NHT has not received any payments from STU. Local estimates suggest that less than 10% of claimants have received their benefits.� - Community Member, Spanish Town While some STU payments have been distributed, there is no clear idea who has received these and many are still waiting and borrowing against them. Across all three of the research visits, the issue of outstanding STU payments has been a key concern to all sugar workers. The first payment was made in December 2010/January 2011, but many of the people who worked as contractors did not receive any- thing as they had no direct employment contract with Bernard Lodge. Others were ineligible because they were re-employed on short term contracts (even if this were for a few days only), were slightly above the minimum wage level, or were pensioners who had passed the pension age. Following strikes by cane cutters, STU revised the qualification conditions in early 2011, allowing persons re-employed to get J$100,000. Despite the “relaxation� of the eligibility criteria, receiving STU payments has proven difficult and burdensome. Accessing the funds has remained a point of concern throughout the entire post- divestment period and very few individuals have received their STU payment in cash. The STU acquired a large quantity of items, such as motorcycles, grass cutters, chain saws, etc. that some former workers accepted as payment instead of cash. However, the bikes “were not provided with insurance and tax,� resulting in several bikes been confiscated by the police and forcing the recipients to look for funds to re- claim the bikes. It should be acknowledged, however, that in the Bernard Lodge /Monymusk area, some of the problems associated with accessing payments were due to issues of no registration for National Insurance Scheme (NIS), National Housing Trust (NHT) and Taxpayer Registration Number (TRN). The problems with receiving the STU payments have been accentuated by difficulties in accessing banking services. Prior to the redundancy exercise all workers were mandated to open bank accounts with NCB, but were then asked to open another account with The St Catherine Credit Union with a mandatory minimum deposit of $J2,000. However, a series of identification documents were required to open accounts, many of which (i.e., birth certificates, etc.) were not possessed by low skilled sugar workers. Communities such as those close to Bernard Lodge now consider that such processes require spending additional funds as each institution has their own registration procedures, causing delays and in many cases barriers to receiving payment . Given the difficulties experienced, communities felt that there was a role for other government agencies to play in the lack of funding and receipts of funds. Benefitting from the STU educational component also proved problematic as the funds ‘have rare- ly been paid at the agreed time’. The former sugar workers who have chosen to receive STU payments in the form of education grants often found that the STU direct payments to third parties, i.e., institu- tions at which they are studying, were not made on time. This has had adverse consequences, with some students having to defer their courses because the institution doesn’t accept a promissory letter from the STU. The process of the STU checking the validity of the course with the institution before making 40 payments commonly means that it can take up to 9 months before educational fees are paid. The result of this has been that students are commonly unable to sit exams, and incur re-sit fees. Despite the poor administrative arrangements around the STU educational payments, the educa- tional component was recognized as one of the most successful elements of the program. Several local community representative have noted that, once the education provisions/payments have been paid, this component of the scheme reflects a positive component to the STU process. “All of those who have chosen to back to school, and have completed or continuing with their studies, are in my view quite frankly the only success that I can assign to this STU program�, - Local community representative, March 2012. Lack of information is a common theme that has exacerbated many of the adverse impacts above. Communities and all households interviewed noted that there is very little information regarding the privatization process/re-employment, and most thought that re-employment would follow within six months of the initiation of the divestment program. It was felt that the government and new sugar fac- tory owners have clear responsibility, and interest, in relaying the future plans and working with the community. In the case of STU payments and other support programs, the communities could benefit from additional consultations to explain eligibility criteria as great many (former) workers continue to expect that ‘something’ will happen. All communities identified future job and skill training to be critical. Training through extension services/job creation/skill enhancement was emphasized by all groups as a key to the future job diver- sification. However, the potential to adopt new coping mechanisms, such as re-training in new skills, appears to be limited in many cases given the above average age of agricultural workers and their low qualifications/educational attainments. For example, although HEART/NTA training may assist with re-employment, as shown by the quantitative results earlier, a minimum of grade 9 level education is re- quired. In some cases the local SDC is now assisting workers to attain this minimum level of education. It is also increasingly apparent that re-training alone is not sufficient: in the latest round of interviews some of those who have re-trained form the sugar industry into hotel management/catering came to realize that once you have the academic qualification, getting the job requires experience. Thus, com- munity members identified opportunities for job training and on-the-job experience as essential com- ponents of transition to a new sector of employment. 3.3.2 Life histories The community interviews were complemented by individual life history interviews to pro- vide a further understanding of the coping mechanisms adopted by individuals and families. Characteristics from the quantitative and community based discussion were used to inform the design of a life history interview. The adoption of life histories allowed for the opportunity to provide compara- tive information about households as well as recording responses to open-ended questions that arise 41 during the course of interviews. The latter focused on critical incidents, events and factors identified by households and information that households identified as important but was not part of the question- naire design.24 The perceived welfare (as referenced by the self-judgment of the respondent) of a redun- dant sugar worker and, for comparison, a coffee farmer, are shown on the vertical axis of Figure 13 and Figure 14. Figure 13: Welfare timeline of a redundant sugar worker Welfare Sugar factory privatized – reemployed for 3 months then made redundant, but still hopeful about future employment at the factory Left school and obtained Diagnosed with empoyment had Drove lorries for the local HIV/AIDS – two kids in the sugar factory, ‘business was ostracised from late 1970’s early good, after an initial the community to Attended 1980’s. In the late downturn... I opened my star with, but school but not period lost job own bar...’ more recently well off and and started to beceme re- parents did not work in the sugar integrated. have a lot of industry money Born 1967 1970’s 1980 – 2006 2010 Source: Life history interview, 2010. The sugar worker was employed as a driver at the local sugar factory and re-employed for 3 months, post privatization, but had no work since. He is still hopeful of future employment and how the future will turn out and is ‘highly expectant’ of re-employment. He perceives his welfare at the present time to be at the lowest in his entire life; this has been caused by a series of shocks. Firstly he was diagnosed with HIV/AIDS in 2006. Secondly, as he was ostracized from the community, his bar produced less in- come than in the late 1990’s. Thirdly, the compound effects of even fewer bar customers due to rising unemployment and lower community incomes—which led, inter alia¸ to reduced social spending—fur- ther affected his income and welfare. The loss of his own employment in the sugar industry made the downward trajectory in his welfare even steeper. As is common with many households suffering ‘crises’, the sequence or serial shocks tends to result in a permanent loss of welfare more than inflicted by any single shock. 24 The life histories adopted a “best practice� approach drawing from the work of others, an extensive review of life history literature, advice from life history experts, and experience of the research team. Specifically the life histories traced an individual’s life from childhood to the present day, focusing on key events. In many instances the interviewee also drew a timeline at the end of the interview to triangulate the details of the interview, clarify any inconsistencies and identify incidents or processes not captured in the previous discussion. At the end of the interview, which normally lasted be- tween one to two and a half hours, the respondents were given an opportunity to ask the interviewers questions. 42 The redundant sugar worker views his current employment options as highly limited. Although there is no current ‘demographic burden’ (i.e. high numbers of school children to care for etc.), he con- siders re-employment options ‘for those at my age’ to be limited. His situation is not unique in that his primary goal for the future lay with re-employment at the factory where he has worked for the last 20 years and he has no other income sources available to him. Comparing the life history of the sugar worker with that of a coffee farmer reveals a different set of vulnerabilities but not very different outcomes. The second life history was undertaken with a coffee worker in the Blue Mountain region who has been employed at a local coffee grower for 15 years. This worker has managed to maintain a permanent salary for more than 20 years: ‘.. I was one of the lucky ones – I’ve had a permanent job working for others ….’. By his own admittance he has done ‘very well’, but the sequential hurricanes and lowering of coffee prices have made the ‘lives of many so vulnerable’. The adverse shocks experienced by the estimated 1,500 small scale (cropping 0.5 to 10 acres) farmers in Jamaica have included price shocks (prices declined from J$3,500 to J$,1500 per box), intermittent, and unexpected, cessation of coffee purchasing from the coffee buyers, and hurricanes. He has lost most of his ‘annual bonus’ (income from his own land), as fertilizer costs now exceed coffee prices and with two young children faces ‘a difficult few years’ ahead. This life history therefore suggests that even for Jamaicans employed permanently and consistently are affected by severe negative welfare shocks with limited ability to cope due to few alternative income sources, at least in the interviewed areas. Figure 14: Welfare timeline of a Blue Mountain coffee farmer Welfare ‘My 2 kids recent kids have kept me going’ (born 2005 & 2007). But price inflation has made things End of 2003 had $67,000 worse – “we savings bought 2nd car and 2006 USAID provide really struggle had 67 pigs but had to sell as 2 greenhouses for these days and Life was very pig food price increased. community- do not have good “I had a Several hurricanes started to supplement income many job and 2 kids, come – “cut off the road� with tomatoes etc, alternative not married, and no produce to sell Prior to 2007 could income bought 1 car in 2000, mother of “hassle�/“do a day’s options�. Early yrs: 1 of 3 labour for someone� children died (but siblings went to but then coffee price not together)� school parents fallen very kind and worked hard, “no problems� Born 1967 1998 2003 2005 2010 Source: Life history interview, 2010. 43 4. Conclusions The set of reforms analyzed in this report has been defined through consultations between the Government of Jamaica and the World Bank team. The specific policy actions—tax reforms and the privatization of sugar estates—have been chosen on the basis of the fact that these actions were, on the one hand, supported by the Bank through its Programmatic Fiscal Sustainability DPL series and, on the other hand, could be expected a priori to generate potential negative distributional effects. Furthermore, given the importance of tax reform for the overall reform program of the Government of Jamaica, the lessons and the methodology of the current PSIA are likely to prove useful to the Government in analyz- ing the impacts of future reforms. In this regard, the data and modeling tools used in Section 2 of this report (CGE-microsimulation analysis) have already been transferred to the Government during a PSIA modeling workshop held at the Planning Institute of Jamaica in December 2011. Relatively few households are likely to experience negative shocks to their income or consump- tion as a result of the reform measures analyzed in this report. The impact of the tax reform on poor households is limited due to excise taxes affecting items which are by and large demanded in lesser proportion by less well-off households. Furthermore, the short-term negative poverty impacts of the tax reform are likely to be small, and in the long term the improved fiscal outturns may well lead to faster growth and accelerated poverty reduction (assuming a continued commitment to fiscal consolidation). There are also few households that are likely to be directly affected by privatization. Moreover, these vulnerable households are more likely to be covered by safety nets (e.g., the PATH program) than the rest of the population. In addition workers in some sectors have been entitled to receive compensation for job losses (e.g. STU Payments). Hence, the privatizations will induce hard- ship on a small number of households, as revealed by the qualitative analysis. In this sense, to the extent that additional safety nets are called for to provide protection to workers vulnerable to job loss due to privatization, a relatively modest extension of PATH can be envisioned rather a completely new program. Workers in the sectors where privatization is taking place are at a higher risk from job losses than an average Jamaican worker. Controlling for other characteristics, public sector workers in the sugar and aluminum industries face a higher probability of losing their jobs. Overall, these workers are almost 25 percent more likely to lose their job than comparable workers in other sectors. Moreover, once un- employed, their chance of getting a job are significantly lower than others who previously worked in other industries and/or in the private sector. This result suggests that workers in the public sector have a particular set of skills that are not fungible (or marketable) enough to allow them to transit in and out of unemployment in an easy way and, therefore, that workers in the entities to be privatized are particu- larly vulnerable to employment shocks. More schooling does not necessarily help in finding a job, but receiving job training makes a sig- nificant difference. More skilled workers (as measured by the number of years of schooling) have a low- er likelihood of losing their job. And among the unemployed, education can help getting a job, but only in the medium term. In the very short term (within three months), more years of schooling is actually associated with a lower probability of finding a job, most likely due to a higher reservation wage among 44 workers with more human capital (i.e. with more experience and years of schooling). Participation in training programs like HEART and NTA significantly reduces the probability of losing a job—among those working—and increases the probability of finding a job—among the unemployed. Given their vulnerability in the labor market, the training program should give priority to young women with un- finished primary school out of the agricultural sector. The results of the ex ante modeling of distributional impacts of tax reform come with a set of im- portant caveats and qualifications. First, the results are not intended to serve as forecasts; the forecast- ing performance of the models used in this chapter has not been validated with historical data and, in fact, CGE models tend to underperform relative to other models in forecasting macroeconomic trends. Second, standard CGE critiques of fairly restrictive functional forms and little empirical validation of elasticity values also apply here; although care has been taken to use appropriate parameter values, many estimates are simply not available for Jamaica. Third, data limitations both on the macro and micro sides have led to a number of simplifications and approximations. For example, the distribution of bond interest income across households has been approximated with a distribution of all dividend and interest income, and the welfare function in the micro data could only be defined at the level of the household, rather than the individual. With these caveats in mind, the ex ante scenarios provide useful information to policymakers. First, the types of models used in this paper tend to do well in a simulation setting, highlighting the macro and micro marginal changes (both directions and relative magnitudes) due to shifts in policies. In particular, the comparative static version of the CGE model used in this chapter was shown to yield the same policy conclusions as a forward-looking, rational-expectations dynamic-stochastic general equi- librium (DSGE) model (see Bussolo and Medvedev, 2008). Moreover, the policy conclusions of the same model have also been shown to be robust within a reasonable range of key elasticity values. Finally, the micro-simulation literature shows that more sophisticated modeling techniques tend to add realism and detail to the more simple micro-accounting methods, but rarely reverse their basic conclusions. The positive poverty and distributional outcomes of the reform scenario presented in this report are conditional on continued efforts at fiscal consolidation. As mentioned in the previous paragraph, the numerical results reported in this study are the outcomes of ceteris paribus simulation scenarios rather than forecasting exercises. Most importantly, the results of the reform scenario depend on the achievement of high primary surpluses (through primary expenditure restraint) and capitalizing on the window of opportunity created by the JDX. In this regard, recent fiscal slippages represent a danger to the positive momentum created by earlier reforms and—if they remain unaddressed—may put the Jamaican economy closer to the unsustainable path of the no-reform scenario than the positive develop- ments envisioned in the reform simulation. The qualitative results suggest that many of the issues identified in the 2006 social impact assess- ment of sugar reform continue to be relevant. The 2006 report (Boxill et al, 2006) identified commu- nication, education and training, alternative livelihoods, and social services as four necessary compo- nents to diminish the adverse impacts of privatization on the sugar communities in Jamaica. However, as mentioned by participants of the community interviews commissioned for this study, information 45 remains scarce in the sugar-producing communities and access to education and training opportuni- ties— which this report shows to be of significant help in mitigating employment shocks—is limited for individuals who have not already completed a basic level of education. Furthermore, options for alterna- tive livelihoods are also scarce, although this factor is not unique to the sugar communities as Jamaica has struggled with diversification for a number of years (see, for example, World Bank, 2011). Finally, access to social services is likely to have improved since 2006, as Jamaica has made significant strides in increasing the access to, and benefits provided by, its conditional cash transfer program PATH. The qualitative results, while invaluable for putting a human face to a problem, are not nationally representative. The community interviews revealed a series of worrisome issues affecting the commu- nities in the neighborhoods of privatized sugar factories. However, even before the privatization, these communities were among some of the most poor and disadvantaged ones in Jamaica, and many of the problems plaguing the community could be traced back to structural issues predating the privatization. Similarly, the interviews took place during the period of a severe economic contraction in Jamaica, mak- ing the adverse effects of the crisis difficult to separate from the adverse effects of the privatization. This is corroborated by similarities in the recent difficulties experienced by a sugar worker affected by the privatization and a coffee worker who was not directly affected by this reform. 46 Annex A: Tax incidence methodology and simulated tax rates In order to investigate the short-term distributional implications of Jamaican value added tax reforms, this report relies on a partial equilibrium tax incidence analysis. This method uses budget shares as the underlying metric which can be calculated on the basis of household consumption and income surveys (Sterner, 2007). “By multiplying budget shares by the proportional increase in the corresponding prices attributable to the reforms, one obtains an estimate of the proportional change in household real in- come� (Coady, 2006, p.258). These income effects are commonly interpreted as short-run impacts of reforms, i.e. before consumer and producer have adjusted their behavior in response to the tax changes. The first step in the approach is to evaluate the tax burden by expenditure deciles. The average value ˆ added tax rate t d (as a share of total household consumption) of all household Jd in each per capita consumption decile d is calculated as shown in equation (A.1), where denotes household’s Jd (gross, i.e. including taxes) expenditure on good i, as reported in the survey and ti is the tax rate (or its proximate ad-valorem equivalent, as reported in Appendix 1). A.1A.1 A.2 The second step computes the real income change implied by each tax reform scenario. For each coun- terfactual tax reform scenario, the change in real per capita consumption of household is computed as follows: A.1B.1 A.2 A.2 A.1 A.1 B.2 The change of real household expenditure results from the relative price shift because of the dif- A.2 A.2 B.1 ference in the good-i-specific tax rate change in the counterfactual. B.3 B.2 B.1 B.1 B.4 B.2 B.2 B.5 B.3 B.6 47 B.4 B.3 B.3 Table A.1: Tax rates by scenario Item code Item Exempt Baseline Full tax reform Only fuel tax Only GCT 102 Coal 0.165 0.175 0.165 0.175 103 Kerosene 0.130 0.280 0.280 0.130 104 Wood 0.165 0.175 0.165 0.175 Other fuel for cooking or light- ing different than cooking gas 105 and electricity 0.130 0.280 0.280 0.130 Tobacco products ( cigars, cigarettes, chewing tobacco, 106 pipes,…..) 0.850 1.050 0.850 0.850 Meat, poultry or fish meals bought away from home (in- 107 cluding gifts) 0.165 0.175 0.165 0.175 108 Sandwiches, Burgers, Patties 0.165 0.175 0.165 0.175 Dairy Products e.g. milk, 109 Supligen, Nutrament E 0.000 0.000 0.000 0.000 Breakfast beverages e.g. tea, cof- 110 fee, milo etc. 0.165 0.175 0.165 0.175 111 Fruits, juices & vegetables E 0.000 0.000 0.000 0.000 112 Drinks – box, bottle, etc 0.165 0.175 0.165 0.175 Others eg. soups, vegetarian 113 meals etc. 0.165 0.175 0.165 0.175 201 Fresh or frozen beef E 0.000 0.000 0.000 0.000 202 Fresh or frozen pork E 0.000 0.000 0.000 0.000 203 Fresh or frozen mutton E 0.000 0.000 0.000 0.000 Offal – heart, kidney, liver, tripe 204 etc. E 0.000 0.000 0.000 0.000 Other fresh or frozen ( oxtail, 205 trotters, cow’s foot, hocks) E 0.000 0.000 0.000 0.000 Salted, cured or canned meat 206 (eg. pigtail) E 0.000 0.000 0.000 0.000 207 Fresh or frozen fish and shellfish E 0.000 0.000 0.000 0.000 208 Salted codfish E 0.000 0.000 0.000 0.000 Canned mackerel, sardines, 209 herring E 0.000 0.000 0.000 0.000 Other salted or canned fish and shellfish (eg. Mackerel, red her- 210 ring …) E 0.000 0.000 0.000 0.000 Fresh or frozen whole chicken 211 or parts E 0.000 0.000 0.000 0.000 212 Chicken neck or back E 0.000 0.000 0.000 0.000 Other poultry, fresh frozen 213 salted, cured or canned E 0.000 0.000 0.000 0.000 Liquid milk (including flavoured 214 milk) E 0.000 0.000 0.000 0.000 48 Item code Item Exempt Baseline Full tax reform Only fuel tax Only GCT 215 Condensed/Evaporated Milk 0.165 0.175 0.165 0.175 216 Powdered milk ( D.S.M ) 0.165 0.175 0.165 0.175 Food Drink (including Lasco, 217 Supligen, Enerplus, Nutrament) 0.165 0.175 0.165 0.175 218 Butter 0.165 0.175 0.165 0.175 219 Cheese 0.165 0.175 0.165 0.175 Other dairy products ( yogurt, 220 ice cream , …) 0.165 0.175 0.165 0.175 221 Eggs E 0.000 0.000 0.000 0.000 Oils and fats (vegetable oil, coconut oil, lard, margarine ( 222 chiffon) E 0.000 0.000 0.000 0.000 223 Bread E 0.000 0.000 0.000 0.000 Crackers and unsweetened 224 biscuits E 0.000 0.000 0.000 0.000 Other baked products (sweet- ened biscuits, cakes, buns, bullas 225 etc) E 0.000 0.000 0.000 0.000 226 Cassava bread / Bammy E 0.000 0.000 0.000 0.000 227 Flour E 0.000 0.000 0.000 0.000 228 Rice E 0.000 0.000 0.000 0.000 229 Cornmeal E 0.000 0.000 0.000 0.000 230 Dried peas and beans, Tofu 0.165 0.175 0.165 0.175 Breakfast cereals (cornflakes, 231 oats, hominy corn …) 0.165 0.175 0.165 0.175 Yams ( white, yellow, Negro, St. 232 Vincent, Lucea,….) E 0.000 0.000 0.000 0.000 233 Irish Potatoes E 0.000 0.000 0.000 0.000 Other roots and tubers (cas- sava, coco, sweet potatoes , 234 dasheen…) E 0.000 0.000 0.000 0.000 Other starchy fruits (Plantains, 235 green banana, bread fruit, …) E 0.000 0.000 0.000 0.000 Fresh vegetables, (tomatoes, carrots, lettuce, turnip, avocado, onion, corn on the cobs, string 236 beans, peas and beans..) 0.165 0.175 0.165 0.175 Frozen canned and dried 237 vegetables 0.165 0.175 0.165 0.175 238 Ackee E 0.000 0.000 0.000 0.000 Fruit and vegetable juices (fresh 239 or frozen) E 0.000 0.000 0.000 0.000 Fresh fruit (oranges, lime, apples, bananas, melons, pine- 240 apples, pears) E 0.000 0.000 0.000 0.000 241 Canned and dried fruits 0.165 0.175 0.165 0.175 49 Item code Item Exempt Baseline Full tax reform Only fuel tax Only GCT 242 Sugar E 0.000 0.000 0.000 0.000 Sweets (sugar, honey, sweeten- 243 ers, jams, jellies) 0.165 0.175 0.165 0.175 Soups (packaged, canned, 244 frozen) 0.165 0.175 0.165 0.175 Prepared meats and fish (curried 245 mutton, fish fingers, …) 0.165 0.175 0.165 0.175 Dry packaged foods (macaroni, 246 spaghetti, vegie chunks …) 0.165 0.175 0.165 0.175 Powders, flavoring and extracts (baking powder & soda, yeast, 247 coconut milk / powder, vinegar) 0.165 0.175 0.165 0.175 Sauces and relishes (ketchup, mayonnaise, pepper sauce, 248 pickles, …) 0.165 0.175 0.165 0.175 Condiments (salt, pepper, gin- ger, curry, pimento, cinnamon, 249 spices) 0.165 0.175 0.165 0.175 Nuts (peanuts, cashew, coconut, 250 …) 0.165 0.175 0.165 0.175 Baby food (milk food, cereals, 251 strained food, …) 0.165 0.175 0.165 0.175 Other food (chips, snacks, 252 cheese trix, ..) 0.165 0.175 0.165 0.175 Breakfast drinks (coffee, tea, 253 Ovaltine, Milo, …) 0.165 0.175 0.165 0.175 Non alcoholic beverages (coke, nectars, canned fruit drinks, powdered & frozen ,purified 254 water/flavoured bottled water) 0.165 0.175 0.165 0.175 Alcoholic beverages (rum, 255 whisky, wine, beer, sherry…) 0.230 0.230 0.230 0.230 Personal care supplies soap, toothpaste/brushes shaving 301 cream, razors & blades 0.165 0.175 0.165 0.175 Cosmetics (lotions, deodorants, 302 …) 0.165 0.175 0.165 0.175 Hair and body care (lotions, 303 dyes, etc) 0.165 0.175 0.165 0.175 Laundry supplies (soap bars/ powders, bleach, starch, clothes 304 pins, …) 0.165 0.175 0.165 0.175 Polishes, waxes, air fresheners, 305 insect sprays 0.165 0.175 0.165 0.175 Kitchen supplies (napkins, matches, garbage bags, dish 306 washing liquid, …) 0.165 0.175 0.165 0.175 50 Item code Item Exempt Baseline Full tax reform Only fuel tax Only GCT Toilet supplies (toilet paper, 307 cleanser, …) 0.165 0.175 0.165 0.175 Other household supplies (scouring pads, liquid cleanser, brooms, light bulbs, batteries, 308 …) 0.165 0.175 0.165 0.175 Home help services (cook, nurse maid, household help, gardener, 309 …) 0.165 0.175 0.165 0.175 Laundry and dry cleaning 310 services 0.165 0.175 0.165 0.175 Rental of equipment (radio, tele- 311 vision, …) 0.165 0.175 0.165 0.175 312 Cooking Gas 0.165 0.175 0.165 0.175 Furniture, indoor (chair, table, bed, mattress, baby crib, cabi- 313 net, …) 0.165 0.175 0.165 0.175 Furniture outdoors (lawnchair, 314 barbecue grill,…) 0.165 0.175 0.165 0.175 Furnishing (carpets, drapes, 315 sheets, towels, …) 0.165 0.175 0.165 0.175 Dinner ware (plates, cups saucers, glasses, knives, forks, 316 spoons, …) 0.165 0.175 0.165 0.175 Cook ware (pots, pans, skillets 317 …) 0.165 0.175 0.165 0.175 Other small kitchen equipment (ice box, toaster, mixer, hot 318 plate, .) 0.165 0.175 0.165 0.175 Large kitchen appliances (Fridge, stove, microwave, 319 freezer, water heater …) 0.165 0.175 0.165 0.175 Radio, TV, VCR, DVD, DSS,CD player, component set, com- 320 puter, printer, fax 0.165 0.175 0.165 0.175 Other small household equip- ment ( tools, camera, hair dryer, 321 suitcase, electric iron, fan…) 0.165 0.175 0.165 0.175 Repairs on furniture or house- 322 hold equipment 0.165 0.175 0.165 0.175 Medicines (pills, tonics, drugs, family planning supplies, herbal 323 medicine) E 0.000 0.000 0.000 0.000 Medical services (doctor’s fee, hospital care, prescriptions, 324 spectacles…) E 0.000 0.000 0.000 0.000 325 Health Insurance E 0.000 0.000 0.000 0.000 51 Item code Item Exempt Baseline Full tax reform Only fuel tax Only GCT 326 Shoes and sandals for adults 0.165 0.175 0.165 0.175 327 Shoes and sandals for children 0.165 0.175 0.165 0.175 Clothing material for adult 328 (Dacron, linen, cotton, silk …) 0.165 0.175 0.165 0.175 Clothing material for children 329 (Dacron, linen, cotton, silk …) 0.165 0.175 0.165 0.175 Adult clothing (suits, dresses, jeans, swim wear, underwear, 330 …) 0.165 0.175 0.165 0.175 Children clothing (shirts, trou- 331 sers, coats, jeans, pampers …) 0.165 0.175 0.165 0.175 Making and repair of clothes 332 (adult and children) 0.165 0.175 0.165 0.175 Accessories (watches, jewelry, 333 sunglasses, …) 0.165 0.175 0.165 0.175 Reading materials (Books, maga- 334 zines, newspapers, …) E 0.000 0.000 0.000 0.000 Stationary and writing equip- ment (pens, pencils, envelops, 335 stamps, …) E 0.000 0.000 0.000 0.000 Education expenses (tuition, 336 books, boarding, fees, …) E 0.000 0.000 0.000 0.000 Sporting activities (exercise equipment, bicycle, tricycle en- 337 trance fees, club membership) E 0.000 0.000 0.000 0.000 Other recreational activities (cinema, dance clubs, records, 338 tapes, DVD, CD, Cable rental .) 0.165 0.175 0.165 0.175 Purchased transportation (taxi, 339 bus, car, rental, air fare) 0.165 0.175 0.165 0.175 340 Gasoline, motor oil, diesel 0.130 0.280 0.280 0.130 Car / motor cycle repair, tires, 341 motor parts 0.165 0.175 0.165 0.175 342 Car / motor cycle insurance 0.165 0.175 0.165 0.175 343 Vehicles taxes, duties 0.000 0.000 0.000 0.000 Purchase of car, motor cycles for 344 personal use 0.165 0.175 0.165 0.175 Other transport expenses (mo- tor vehicle and driver licenses, 345 traffic tickets) 0.165 0.175 0.165 0.175 Vacation expenses (excluding 346 fares) (hotels, travel tax …) 0.083 0.100 0.083 0.083 Gardening and horticulture (plants, fertilizer, garden equip- 347 ment, home animals …) 0.165 0.175 0.165 0.175 348 Telephone 0.165 0.175 0.165 0.175 52 Item code Item Exempt Baseline Full tax reform Only fuel tax Only GCT Other consumption expendi- 349 tures (flowers, etc.) 0.165 0.175 0.165 0.175 Purchases for special occasions (parties, entertainment relating to weddings, funerals, bounce 350 about etc.) 0.165 0.175 0.165 0.175 53 Annex B: Description of the CGE Model The CGE model used in this report has its origins in a standard neoclassical general equilibrium model and is based on the World Bank’s prototype single-country model.25 It is also a recursive dynamic exten- sion of the CGE model in Bussolo and Medvedev (2008), who augmented the above prototype model by introducing an endogenous labor-leisure tradeoff. The main features of this model will be familiar to readers accustomed with the CGE literature and thus are summarized only briefly. The labor-leisure choice is pre- sented in more detail, with the discussion borrowing heavily from Bussolo and Medvedev (2008). Production. Output results from nested CES (Constant Elasticity of Substitution) functions that, at the top level, combine intermediate and value added aggregates. At the second level, the intermediate ag- gregates are obtained combining all products in fixed proportions (Leontief structure), and total value added is obtained by aggregating the primary factors. The nested structure of production allows for dif- ferent degrees of substitutability across inputs and is a standard feature in most CGE models. The full structure of production nests is shown in Figure B.1. Figure B.1: Production structure of the Jamaica CGE model Output p Aggregate intermediate demand Value Added =0 v Intermediate Capital + Labor Aggregate demand Land m kl Intermediate Labor Capital demand by region of origin l Skilled Labor Unskilled Labor Note: Although the model allows substitution between Land and the other primary factors, given that the data for separating land and other factors contributions to value added was not available, the nesting structure actually active in the current model does not include Land as a separate factor. 25 See van der Mensbrugghe (2005b) for detailed model documentation and van der Mensbrugghe (2005a) for the user’s guide. 54 Income distribution and absorption. Labor income and capital revenues are allocated to households according to a fixed coefficient distribution matrix derived from the original SAM. Private consump- tion demand and labor supply decisions are obtained through maximization of household specific util- ity function following the Linear Expenditure System (LES). The quantity of aggregate saving behavior is determined by a fixed marginal propensity to save, calibrated using the base year consumption, in- come, and saving. Household utility is a function of consumption of different goods and leisure. Once total value of private consumption is determined, government and investment demand are disaggre- gated into sectoral demands according to fixed coefficient functions. International trade. The model assumes imperfect substitution among goods originating in different geographical areas.26 Import demand results from a CES aggregation function of domestic and imported goods. Export supply is symmetrically modeled as a Constant Elasticity of Transformation (CET) func- tion. Producers allocate their output to domestic or foreign markets according to relative prices. Under the small country assumption, Jamaica is unable to influence world prices and its imports and exports prices are treated as exogenous. Assumptions of imperfect substitution and imperfect transformability grant a certain degree of autonomy of domestic prices with respect to foreign prices and prevent the model from generating corner solutions. Furthermore, they permit cross-hauling—a feature normally observed in real economies. The balance of payments equilibrium is determined by the equality of foreign savings (which are exogenous) to the value of the current account. With fixed world prices and capital inflows, all adjustments are accommodated by changes in the real exchange rates: increased import demand, due, for instance, to trade liberalization, must be financed by increased exports, and these can expand due to improved resource allocation. Import price decreases drive resources towards export sectors and contribute to falling domestic resource costs (or real exchange rate depreciation). Factor markets. Labor is divided into two categories: skilled and unskilled. These categories are consid- ered imperfectly substitutable inputs in the production process. The labor market skill segmentation27 has become a standard assumption in CGE modeling and it is easily justifiable for the case of Jamaica, where inequalities in educational endowments and access to education support this assumption. Skilled and unskilled labor types are then aggregated into a composite labor bundle which is then combined with composite capital (see production nest in Figure A.1). In the standard version, composite capi- tal and labor types are fully mobile across sectors, with the rental rate on capital and relevant wages clearing the factor markets. Capital supply in each year is fixed. Labor supply, for both the skilled and unskilled categories, is derived, as shown below, from utility maximization where individuals chose the optimal consumption level for both commodities and leisure time under their budget constraint. Labor-leisure tradeoff. The introduction of a consumption-leisure tradeoff in the household utility function follows the approaches of Barzel and McDonald (1973), de Melo and Tarr (1992), and Annabi (2003). Consider a Stone-Geary utility function and a budget constraint of the following form: 26 See Armington (1969) for details. 27 See Taubman and Wachter (1986) for a general discussion of labor market segmentation. 55 A.1 A.2 B.1 A.1 B.1 A.1 In this utility function, Ci denotes the consumption of good i with leisure (C0) being a normal good, are A.2 B.2 usually interpreted as consumption minima,28 and the share parameters μi (including μ0) must sum to A.2 unity. T denotes the total time a household has available for work and leisure activities, and the amount A.1 of resources available for non-leisure consumption is limited by non-labor income (y) and total wage A.1 income (ignoring saving and taxes for simplicity).29 Constrained maximization gives rise to the familiar B.1 linear expenditure system (LES) demand functions: B.1 B.3 A.2 A.2 B.2 B.2 B.4 B.2 B.1 The household labor supply is the difference between total time available and the time allocated to con- B.1 sumption of leisure, and substituting the budget constraint into the demand function yields: B.5 B.3 B.2 B.3 B.2 B.3 B.6 B.4 Partially differentiating the labor supply equation with respect to non-labor income and the wage rate B.4 yields the following elasticities: B.3 C.1B.3 B.5 B.4 B.5 where B.4 B.6 B.4 , defined at the level of household primary earner B.6 where is an identity operator equal to 1 when household h belongs to group i and 0 otherwise B.5 B.5 , B.5 defined at is the total value of goods and services consumed by household h C.1 is the total value of good c consumed by household h C.1B.6 is an identity operator equal to 1 when B.6household h be Whilewhere the labor supply is decreasing in non-labor income, the sign of the wage elasticity depends on the is the total value of goods and services co defined at the expenditures.30 ratio of non-labor income to the total “committed�, consumptionlevel of household primary earner where is the total value of good c consum , defined at the level of household primary earner C.1 28 Note that there is no theoretical requirement forwhen household hpositive. to group i and 0 otherwise is an identity operator equal to 1 any of the θi to be belongs C.1 29 Note that the price an identity operator equal to 1 wage rate W (i.e. consumed by household h0 otherwise is is the economy-wide when household h belongs to group i and is of leisure the total value of goods and services P0=W). where 30 This sign ambiguity allows for a backward-bending labor supply curve. is the total value of good c consumed by household h where , defined at the level household h is the total value of goods and services consumed by of household primary earner is the total value of good ,cdefined at the level of household primary earner consumed by household h 56 is an identity operator equal to 1 when household h belongs to group i and 0 otherwise is an identity operator equal to 1 when household h belongs to group i and 0 otherwise is the total value of goods and services consumed by household h Model closure. The equilibrium condition on the balance of payments is combined with other closure conditions in order to obtain a unique solution. First, both current and capital government expendi- tures are fixed in real terms as a share of GDP. This assumption can be modified in alternative scenarios, but is appropriate in the baseline because public goods do not enter the household utility function (and A.1 therefore changes in public expenditure affect household welfare only indirectly). Tax rates are also fixed at base year levels, and the primary balance is endogenous. The composition of public borrowing (i.e., the ratio of foreign borrowing to domestic borrowing) is fixed at the base year value, and this set A.2 combination of borrowing instruments clears the fiscal accounts. Second, aggregate investment—which together with an exogenous rate of depreciation determines the next period’s capital stock—is set equal to aggregate savings. The volume of available savings is determined by an exogenous level of foreign sav- ing including foreign direct investment (which evolves according to medium-term assumptions of the B.1 IMF), endogenous government saving, and households who save a fixed share of their post-tax income. The price of absorption is the numéraire. Dynamics. The model is solved in a recursive dynamic mode, in which a sequence of end-of period B.2 equilibria is linked with a swet of equations that update the main macro variables. There are three de- terminants of real GDP growth in the model: labor supply growth, capital accumulation, and increases in productivity. The maximum stock of labor available in each period (which, form equation A.3, will be greater than or equal to labor supply LS) grows exogenously at the rate of growth of the working age population (ages 15-64), obtained from World Bank population forecasts. The capital stock in each pe- B.3 riod is the sum of depreciated capital from the period before and new investment, which, as mentioned before, is determined by the available savings in the previous period. The behavior of the third com- ponent—productivity—is factor- and sector-specific. Labor and capital productivity in agriculture grow B.4 exogenously at one percent per year, broadly consistent with the econometric literature on productiv- ity growth in developing countries. For all other sectors, capital productivity remains fixed throughout the model horizon, while growth in labor productivity (which is assumed to be Harrod-neutral, purely labor-augmenting technical change) can be exogenous or endogenous depending on the type of simula- B.5 tion. The evolution of skill- and sector-specific labor productivity is given by the following equation: B.6 B.6 In the baseline scenario, also referred to as Business-as-Usual (BaU ) scenario, is endogenous while real GDP growth is fixed. This allows the user to calibrate the model to any given GDP growth rate, which C.1 in this chapter follows the near- and medium-term projections of the IMF. In scenarios other than BaU, is fixed in each period at the BaU solution level, and GDP growth becomes endogenous. Thus, in the absence of any shocks, the BaU GDP growth rate is reproduced exactly. In policy simulations, real GDP where growth may differ from BaU due to faster/slower accumulation of labor or capital, or shocks to the , defined at the level of household primary earner sector-specific productivity shift parameters for labor or capital ( ). In other words, variations in GDP growth across scenarios can be directly attributed to the simulated policy reforms, allowing for clear comparisons an identitythe simulations. 1 when household h belongs to group i and 0 otherwise is between operator equal to is the total value of goods and services consumed by household h is the total value of good c consumed by household h 57 Annex C: Description of the Micro-Accounting Model The micro-accounting model used in this chapter is a simple approach based on techniques de- scribed in Bussolo et al (2008) and Ravallion and Lokshin (2008). As described in much greater detail in Bourguignon et al (2008), the empirical tool is referred to as “micro-accounting� because no behavior is modeled at the micro (survey) level. Instead, the objective of the model is simply to translate the shocks observed in the macro model to the household survey in a consistent fashion, such that the resulting counter-factual distribution of income or consumption is consistent with the results at the macro level. Although this approach lacks theoretical and empirical sophistication of more complicated micro-simulation methodologies (e.g., Ferreira et al, 2008, or Robilliard et al, 2008), its advantages in- clude simplicity, transparency, and less stringent consistency requirements (with respect to the data un- derpinning the macro model). Moreover, as described in the following paragraphs, more sophisticated modeling approaches could not be readily used due to data constraints. The micro-accounting model creates a new, counter-factual distribution of welfare through shocks to link aggregate variables (LAVs). The LAVs are observed in both macro (CGE) and micro (survey) data, and therefore provide a bridge by means of which consistency between the two sides is assured. The LAVs used in this chapter are a fairly standard set and include variables meant to capture the change in the overall level of welfare in the population—such as overall population growth and changes in real consumption per capita—as well as variables meant to capture any potential redistribution of welfare within the population—such as changes in population structure, changes in prices of different types of consumption goods and changes in incomes based on the sector of employment and skill level of the household primary earner. The specific LAVs used in this chapter are listed in the Table below. 58 Table C.1: Link Aggregate Variables in the Micro-Accounting Model that affect aggregate Variables Real consumption growth welfare A.1 Overall population growth Population growth, unskilled rural household Population growth, unskilled urban household A.1 A.2 Variables that affect the distribution of welfare Population growth, skilled household Total income, unskilled rural household A.2 Total income, unskilled urban household B.1 Total income, skilled household Price of agricultural products consumed by unskilled rural household B.1 Price of petroleum products consumed by unskilled rural household B.2 Price of manufactured products consumed by unskilled rural household Price of services consumed by unskilled rural household B.2 Price of agricultural products consumed by unskilled urban household Price of petroleum products consumed by unskilled urban household Price of manufactured products consumed by unskilled urban household B.3 Price of services consumed by unskilled urban household B.3 Price of agricultural products consumed by skilled household Price of petroleum products consumed by skilled household B.4 B.4 Price of manufactured products consumed by skilled household Price of services consumed by skilled household B.5 The LAVs above define the household welfare function in terms of initial level of household con- B.5 sumption and changes in household size, household income, and prices of goods and services B.6 consumed by the household. Thus, we can express the economy-wide average household per capita (or B.6 adult equivalent) welfare at time t as follows: C.1 C.1 C.1 wherewhere , defined at the level of household primary earner , defined at the level of household primary earner is an identity operator equal to 1 when household h belongs to group i and 0 otherwise is an identity operator equal to 1 when household h belongs to group i and 0 otherwise is the total value of goods and services consumed by household h is the total value of value ofand services consumed by household h is the total goods good c consumed by household h is size of household good c in terms of by household is thethe total value of h (either consumed persons or adulthequivalents) is the size of population group composed of i-type households is the total post-tax income (labor, capital, and transfers) of an i-type household is the price of good c consumed by an i-type household D.1 59 D.2 Note that equation C.1 defines the income-generation process at the household, rather than individual level. This distinction underpins the choice of micro-simulation methodology in this chapter: rather than being defined as the sum of incomes of each working member (conditional on the characteristics of each worker), household income is defined as an inseparable total, conditional only on the charac- teristics of the primary earner. As mentioned in the first paragraph of this annex, this modeling choice was determined largely by data constraints: due to a large number of missing or zero labor income reported by employed individuals in the Labor Force Survey (LFS), more than half of all households (57 percent) in the Survey of Living Conditions (JSLC) did not have any labor income despite having multiple working members. As a result, it was not possible to specify a model of individual earnings and labor supply—such as the one developed by Bourguignon and Ferreira (2005)—for these households, and using only households for which this information was available would have substantially distorted the original survey design and, consequently, the poverty and inequality statistics. Therefore, the clas- sification of households into unskilled rural, unskilled urban, and skilled groups was done according to the characteristics of the household primary earner, as reported in the JSLC. The model operates on total household income, rather than labor income only. This distinction is important because many models of the type used in this chapter pass on only labor earnings from the macro model down to the micro. In this case, however, we use total income because of the criti- cal importance of income sources other than labor to the simulation results. For example, the debt exchange leads to a large decrease in bond earnings for households which hold bonds. Consequently, ignoring this important channel would have introduced a substantial distributional bias towards richer (skilled) households who capture nearly all of the bond income (because less-wealthy unskilled house- holds hold few bonds). Due to these considerations, the household income generation process—taking into account all income sources—was modeled at the macro level for three representative household groups—unskilled rural, unskilled urban, and skilled—and shocks to total income were transmitted to the household survey. The impacts of changes in prices of consumer goods on household welfare are modeled as first-or- der effects without allowing households to re-optimize their consumption bundle. In other words, households in the survey continue to devote the same share of their consumption budget to each con- sumption category c in time t as they did in time 0, meaning that only price (and not quantity) effects are taken into account. However, because the macro model does include three representative house- holds (depending on the source of income of the primary earner) and the macro model allows for a re- optimization of the consumption bundle, price changes faced by the households in the survey do vary across the representative groups (but not within them). This approximation is consistent with other approaches taken in the majority of studies in the macro-micro simulation literature (e.g., Bourguignon et al, 2008), and avoids the difficulties of specifying and calibrating a household-specific demand system while still capturing a substantial share of the impact of price changes on household welfare. The simulation approach developed above is implemented sequentially, with no feedbacks to the macro model. Consistent with other micro-accounting applications, the results of the micro exercise do not have any feedbacks to the macro model. Incorporating this type of feedback effects requires a much more complicated modeling structure and a near full-reconciliation of macro and micro data through 60 cross-entropy or other numerical techniques (see Robilliard et al, 2008, for an application and a discus- sion of advantages and limitations of such approaches). Furthermore, the simulation is implemented as a sequence of individual steps as described in the table below. Table C.2: Sequential nature of the micro model Step Action Households are re-weighted to match the new population structure given by the 1 macro model 2 Overall population growth is constrained to match that of the macro model The welfare of each household is scaled up or down depending on changes in the 3 total income of its corresponding representative household group (e.g., unskilled urban) in the macro model The welfare of each household from step 3 above is scaled up or down depending on changes in the consumption prices faced by its corresponding representative 4 household group (e.g., unskilled urban) in the macro model and the share of each good in each household’s consumption bundle. Overall change in real consumption per capita is constrained to match that of the 5 macro model 61 Annex D: Impact of discrete changes in independent variable(s) on the estimated probability The objective of the empirical strategy is to identify the characteristics that make an individual more likely to lose his or her job and, once unemployed, identify the characteristics that make him or her more likely to abandon this labor status. is the size of household h (either in terms of persons or adult equivalents) In non-linear models such as probit or logit, inferringcomposed of i-type households independent variable is the size of population group the impact of changes in an is the its mean is not straightforward. To see why this an i-type household for values away from total post-tax income (labor, capital, and transfers) ofis the case, take the probability of ersons or adult equivalents) outcome, of good c consumed by an i-type household observing the realization of is the price y=1, as a function of a latent variable y*: f i-type households is the size of household h (either in terms of persons or adult equivalents) old hof an i-type household is the adultof population group composed of i-type households ers) (either in terms of persons or size equivalents) ulation group composed of i-type households (labor, capital, and transfers) of an i-type household is the total post-tax income D.1 D.1 type household e (labor, capital, and transfers) ofis the price of good cof household han i-type householdpersons or adult equivalents) is the size an i-type household consumed by (either in terms of good c consumed by an i-type household is the size of population group composed of i-type households D.2 where D.1 is the cumulative distribution function (CDF). In the case of the probit model, the CDF is the total post-tax income (labor, capital, and transfers) of an i-type household D.1 is the normal cumulative which is defined by the following expression: household h (either in terms of persons or adult equivalents) of good c consumed by an i-type household is the price D.2 D.1 D.3 of population group composed of i-type households D.2 D.2 of household hcapital, and transfers) of an or adult equivalents) ncome (labor, (either in terms of persons i-type household D.2 D.1 ize of population group composedD.3i-type households h (either in terms of persons or adult equivalents) rice of good c consumed by an i-type is the size of household of household From equation (D.2) the marginal effect, i.e. the change in the probability of observing an outcome D.3 D.4 ax income (labor, capital, and transfers) of an i-typepopulation group composed of i-type households is the size of household given a small change in an independent variable keeping all other variables constant, of continuous D.2 D.3 e price of good c consumed bythe total post-tax income (labor, capital, and transfers) of an i-type household is an i-type household variable X is defined as: D.1 D.4 is the price of good c consumed by an i-type household D.4 D.3 D.5 D.3 D.2 D.1 D.4 D.1 where is the normal probability density function.D.2 clear from the above expression that the mar- It is D.5 D.3 D.4 D.5 D.2 E.1 ginal effect of X depends on the value of all independent variables and the parameters associated with them. To get the probability of transitioning to employment or unemployment at different values of D.5 D.3 E.2 years of schooling as reported, the K marginal effects –for the same number of independent variables in D.3 the model—were estimated (these results are reported in Table 7). The K marginal effects were multiplied E.1 D.4 by the mean value of the independent variables to get the expected probability of E.1 observing an outcome D.5 E.2 (given mean values of independent variables): E.1 D.4 E.2 E.3 D.4 E.2 D.5 D.4 E.1 E.3 D.5 E.2 E.3 D.5 E.1 E.4 E.3 E.2E.1 E.4 E.3 E.4 E.1 62 E.2 E.4 E.2 E.3 D.3 D.4 Finally, the expression above was used to compute the probability of observing an outcome at different values of years of schooling (S S’): D.5 D.5 E.1 E.2 E.3 E.4 63 D.1 D.1 is the price of good c consumed by an i-type household D.3 D.2 D.3 is the size of household h (either in terms of persons or adult equivalents) D.1 D.2 is the size of population group composed of i-type households D.2 D.2 D.1 terms of persons orincome (labor, capital, h (either inh of an in terms D.4 D.5 the equivalents) D.2 is the size of household terms of persons of adult or or persons is the size of household h (either inis the total post-tax isadultsize of household and transfers)(either i-type household equiva D.3 D.3 D.4 i-type households of the size of group composed of i-type householdsh is the size consumed by an i-type group composed of i-type is the size of population group composed of is the price of good cis population population household Annex E: Incorporating State Dependence D.3 D.3 D.2 is the total post-tax income (labor, capital, and transfers)the an i-type household (labor, capital, and transfers) of an i-type hou post-tax income is of total is the total post-tax income (labor, capital, and transfers) of an D.3 E.1 is the probability c consumed by “j� transits from of good c of good c consumed =e an i-type is the priceis the priceconsumed by an i-type household hous Define P =u as the price of good that individualan i-type householdemployment to unemployment and P by D.4 9. Annex E: Incorporating State Dependence as the probability that individual “i� transits from unemployment to employment. Let D.4 and be la-D.5 D.5 D.3 Define E.2 as determining that individual ―j‖D.4 tent linear functionsthe probability the transition probabilities as defined by the following expressions: transits from employment to unemployment andD.4 31 D.4D.1 D as the probability that individual ―i‖ transits from unemployment to employment. Let and D.5 be latent linear functions determining the transition probabilities as defined by the followingE.1 E.1 E.1 D.4 expressions:31 D.2 D.5 E.3 D.5 D.5 E.2 D.5 E.2 E.2 E.1 D.3 E.1 E.1 D.5 where and E.1 E.2 are vectors of personal and household characteristics determiningE.1 probability of the losing and finding a job, respectively; and are random terms distributedE.2 E.4 E.2 E.1 normally with mean zero E.3 and known variance. is the normal cumulative E.3 E.2 distribution function. Finally, and are vectors of D.4 E.2 parameters to be estimated. E.2 E.1 where and are vectors of personal and household characteristics determining the probability of E.3 E.2 We first estimate the following model of the probability of being employed (l = 1) at time t: losing and finding a job, respectively; and mean are random terms distributed normally with E.4 zero E.3 and known variance. E.3 is the normal cumulative distribution function. Finally, D.5 E.4 vectors andE.3 are of parameters to be estimated. E.3 E.1 E.4 We first estimate the following model of the probability of being employed (l = 1) at time t: E.3 E.3 E.2 E.1 E.4 E.4 E.4 E.3 Where, as before, X is a matrix with observable characteristics affecting E.2 probability of being em- E.4 the ployed. Notice that E.3 exploits the cross-sectional variation instead of the time-variation as in equa- E.4 Where, as before, X is a matrix with observable characteristics affecting the probability of being E.4 E.3 tions E.1 and E.2. As demonstrated by Card and Sullivan (1988), one of the most important variables employed. Notice that D.3 exploits the cross-sectional variation instead of the time-variation as in equations D.1 probability demonstrated by Card and Sullivan (1988), one history, which, according determining theand D.2. As of employment in t is the worker’s employmentof the most important the authors, can be the probability fi employment in t is the worker‘s employment history, which, tovariables determining simplified by a ofrst-order Markov process. Therefore, we could re-write (3) as: E.3 according to the authors, can be simplified by a first-order Markov process. Therefore, we could re-write (3) as: E.4E.4 E.4 Where denotes the employment status in the period and E.4 a parameter measuring the impor- is E.4 Where lt-1 denotes the employment status in the period t-1 and � is a parameter measuring the importance of of worker‘s employment history to to determine worker‘s contemporary employment status. tancethe the worker’s employment historydetermine the the worker’s contemporary employment status. Equation D.4 a more general form of probability of employment dynamics which includes, as special Equation E.4 isis a moregeneral form of probability of employment dynamics which includes, as special cases, D.1 and D.2: cases, E.1 and E.2: 31 A similar specification can be found in Mondragon-Velez et al (2010) 6431 A similar specification can be found in Mondragon-Velez et al (2010) 55 References Annabi, N. (2003). “Modeling Labor Markets in CGE Models: Endogenous Labor Supply, Unions, and Efficiency Wages,� Poverty and Economic Policy (PEP) Network Working Paper, May. Armington, P. (1969). “A Theory of Demand for Products Distinguished by Place of Production.� Staff Paper 16, International Monetary Fund. Aschauer, D. (1989). “Is Public Expenditure Productive?� Journal of Monetary Economics 23, 177-282. Barzel, Y., and R. McDonald (1973). “Assets, Subsistence, and the Supply Curve of Labor,� American Economic Review 63(4), pp. 621-633. Birdsall, N. and Nellis, J. (2002) “Winners and losers: assessing the distributional impacts of privatiza- tion�, Center for Global Development Working Paper No. 6 Bourguignon, F. and Sepúlveda, C. (2009) “Privatization and development: some lessons from experi- ence�, Policy Research Working Paper 5131, The World Bank, Washington, DC. Bourguignon, F., and F. H. G. Ferreira (2005). “Decomposing Changes in the Distribution of Household Incomes: Methodological Aspects� In F. Bourguignon, F. H. G. Ferreira, and N. Lustig, eds., The Microeconomics of Income Distribution Dynamics in East Asia and Latin America. New York: Oxford University Press. Bourguignon, F., M. Bussolo, and L. Pereira da Silva, eds. (2008). The Impact of Macroeconomic Policies on Poverty and Income Distribution: Macro-Micro Evaluation Techniques and Tools. New York: Palgrave Macmillan. Boxill, I., S. French, J. Talbot, and D. Ramjeesingh. 2006. Social Impact Assessment of the Likely Effects of Sugar Reform Options on Industry Workers in Jamaica, Report prepared for the PIOJ. Bussolo, M. and De Hoyos, R. (2009), Gender aspects of the trade and poverty nexus: a macro-micro approach, Palgrave Macmillan, London and Washington DC Bussolo, M., and D. Medvedev (2008). “Do Remittances Have a Flip Side? A General Equilibrium Analysis of Remittances, Labour Supply Responses and Policy Options for Jamaica� Journal of Economic Integration, 23 (3): 734-64. Bussolo, M., J. Lay, D. Medvedev, and D. van der Mensbrugghe (2008). “Trade Options for Latin America: A Poverty Assessment Using a Top-Down Macro-Micro Modeling Framework,� in Bourguignon, F., M. Bussolo, and L. Pereira da Silva, eds., The Impact of Macroeconomic Policies on Poverty and Income Distribution: Macro-Micro Evaluation Techniques and Tools. New York: Palgrave Macmillan. 65 Card, D. and Sullivan, D. (1988) “Measuring the effects of subsidized training programs on movements in and out of employment�, Econometrica, Econometric Society, Vol. 57(2), pg. 411-45. Coady, D. (2006) ‘Indirect Tax and Public Pricing Reforms’, Analyzing the Distributional Impacts of Reform. De Melo, J., and D. Tarr (1992). General Equilibrium Analysis of US Foreign Trade Policy, Cambridge, Mass.: MIT Press. Deaton, A. (1997). The Analysis of Household Surveys: A Microeconometric Approach to Development Policy, Baltimore and London: John Hopkins U. Press for World Bank. Edmiston, K. D. and Bird, R. M. (2004) Taxing Consumption in Jamaica, Joseph L. Rotman School of Management, University of Toronto, Canada. Ferreira, F., P. Leite, L. Pereira da Silva, and P. Picchetti (2008). “Can the Distributional Impacts of Macroeconomic Shocks Be Predicted? A Comparison of Top-Down Macro-Micro Models with Historical Data for Brazil,� in Bourguignon, F., M. Bussolo, and L. Pereira da Silva, eds., The Impact of Macroeconomic Policies on Poverty and Income Distribution: Macro-Micro Evaluation Techniques and Tools. New York: Palgrave Macmillan. Handa, S. and King, D. (1996) “Structural adjustment policies, income distribution and poverty: a review of the Jamaican experience�, Discussion Paper FC1996-4, Centre for International Studies, Toronto, Ontario Hauptmeier, S., Heipertz, M. and L. Schuknecht (2006). “Expenditure Reform in Industrialised Countries: A Case Study Approach� ECB Working Paper No. 634. Kim, N. (2007). “The Impact of Remittances on Labor Supply: The Case of Jamaica,� World Bank Policy Research Working Paper 4120. King, D. (2000) “The evolution of structural adjustment and stabilization policy in Jamaica�, Economic Reforms Series, 65, ECLAC, Santiago, Chile La Porta, R. and López-de-Silanes, F. (1999) “The benefits of privatization: evidence from Mexico�, The Quarterly Journal of Economics, November. McKenzie, D. and Mookherjee, D. (2003) “The distributive impact of privatization in Latin America: evidence from four countries�, Economia, Vol. 3, Spring. Mondragón-Vélez, C., Peña, X. and Wills, D. (2010) “Labor market rigidities and informality in Colombia�, Documentos CEDE 6717, Universidad de los Andes-CEDE 66 Ravallion, M. and M. Lokshin (2008). “Winners and Losers from Trade Reform in Morocco,� in Bourguignon, F., M. Bussolo, and L. Pereira da Silva, eds., The Impact of Macroeconomic Policies on Poverty and Income Distribution: Macro-Micro Evaluation Techniques and Tools. New York: Palgrave Macmillan. Robilliard, A.-S., F. Bourguignon, and S. Robinson (2008). “Examining the Social Impact of the Indonesian Financial Crisis Using a Macro-Micro Model,� in Bourguignon, F., M. Bussolo, and L. Pereira da Silva, eds., The Impact of Macroeconomic Policies on Poverty and Income Distribution: Macro-Micro Evaluation Techniques and Tools. New York: Palgrave Macmillan. Robinson, S. and D. Willenbockel (2009). 2005 Social Accounting Matrix for Jamaica, mimeo. Sheshinski, E. and López-Calva, L.F. (2003) “Privatization and its benefits: theory and evidence�, CESifo Economic Studies, Vol. 49, 429-459. Sterner, T. (2007) ‘Fuel Taxes: An Important Instrument for Climate Policy’, Energy Policy, 35, 3194–3202. Taubman, P. and M. L. Wachter (1986). “Segmented Labor Markets,� in Ashenfelter, O. and R. Layard, eds., Handbook of Labor Economics, Vol. II, 1183-1217. van der Mensbrugghe, D. (2005a). “Instructions for Running the Prototype CGE model: A User’s Manual for the Tunisia CGE model.� Technical report, World Bank. van der Mensbrugghe, D. (2005b). “Prototype Model for a Single Country: Real Computable General Equilibrium Model.� Technical report, World Bank. World Bank. 2007. “Jamaica - Poverty Assessment : Breaking the Cycle of Unemployment, Vulnerability and Crime.� World Bank, Washington, DC. World Bank. 2011. “Jamaica Country Economic Memorandum: Unlocking Growth.� World Bank, Washington, DC. 67 68 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-Mail: feedback@worldbank.org