85760 v2 January 2014 When Prosperity is not Shared The Weak Links between Growth and Equity in the Dominican Republic When Prosperity is not Shared: The Weak Links between Growth and Equity in the Dominican Republic World Bank January 2014 Acknowledgements This study was led by Javier E. Baez and Luis Felipe Lopez-Calva. The core team included Andres Castañeda and Ali Sharman. The team thanks Louise Cord, Miguel Sanchez, Christine Richaud, Leonardo Lucchetti, Jose Cuesta, Anna Fruttero, Maria Ana Lugo, Juan Baron, Aline Coudouel, Carolina Rendon, Macdonald Benjamin, Magdalena Lizardo and Antonio Morillo for providing relevant material and sharing thoughtful insights. The team also thanks participants at consultations held at the World Bank office in Santo Domingo, the Dominican Republic, and at headquarters in Washington, DC, USA. The work was conducted under the general guidance of Louise Cord (Sec- tor Manager, LCSPP). The views and interpretations expressed here are the sole responsibility of the authors and should not be attributed to the World Bank, the Board of Executive Directors or the countries that they represent. Summary Despite strong economic growth over the past decade, large inequities persist in Dominican society and are declining more slowly than expected. GDP per capita rose almost 50 percent from 2000 to 2011, yet many of the country’s 10 million people missed out on the benefits. Moderate poverty has fallen by only half of the dramatic spike that followed the decade’s only growth setback, a 2003-2004 economic crisis. Chronic poverty—in which people endure long spells of being poor—remains high. Of greater concern, almost one third of the population is poor despite having the skills and assets to generate higher income. The Dominican Republic also has low economic mobility, with less than 2 percent of its people climbing to a higher income group during the decade, compared to an average 41 percent in the Latin America and Caribbean region as a whole. Despite improving access to basic goods and services such as water and education, coverage and quality remain uneven, thus limiting the economic opportunities of many disadvantaged people. This re- flects their inability to influence the system to their benefit, a manifestation of weak political agency. This report identifies three broad goals for addressing the underlying causes of economic inequity in the Domin- ican Republic. 1. Promote equitable, efficient and sustainable fiscal policy, 2. Build fair, transparent and efficient institutions that will improve the provision and quality of public goods and services, expand economic opportunities, increase upward mobility and better protect economically vulnerable Dominicans, and 3. Strengthen access of the poor to labor markets and increase the demand for their labor, so as to make efficient use of human capital and allow the poor to benefit from economic growth. Contents Overview 9 Growth and equity in the Dominican Republic: The puzzle 9 A conceptual framework for equity analysis 10 The weak links between economic growth and poverty reduction in the Dominican Republic 10 Escaping poverty and staying away from it: Income dynamics and the lack of upward economic mobility 12 Multidimensional poverty and its dynamics: The persistent and increasing risk of becoming chronically poor 13 Fairness: Is there equality of opportunities for all? 15 Policy options to strengthen the links between economic growth and equity 16 Conclusions 22 1. Growth and equity in the Dominican Republic: The puzzle 24 2. A Conceptual framework for equity analysis 26 3. The weak links between economic growth and poverty reduction 28 3.1 The macro context: A top performer in economic growth 28 3.2 A slowly shifting economic landscape 29 3.3 Poverty trends: Recovering slowly from the economic crisis 31 3.4 Income inequality trends: Moderate improvements, mostly for rural households 37 3.5 Unpacking the changes in poverty and inequality 37 3.5.1 The role of income growth and distribution 37 3.5.2 The role of different sources of income 38 3.5.3 The role of population shifts between urban and rural areas 40 3.4.4 The role of population shifts between formal and informal jobs 41 4. Escaping poverty and staying away from it: Income dynamics and the lack of upward economic mobility 44 4.1 Conceptualizing and measuring economic mobility 44 4.2 Assessing economic mobility in the Dominican Republic or is it economic immobility and economic insecurity? 45 4.2.1 Overall long-term mobility 45 4.2.2 Understanding upward mobility out of poverty and vulnerability 48 4.3 Initial conditions do matter for economic mobility 51 5. Identifying chronic and transient poverty using a multidimensional approach 52 5.1 Non-monetary poverty: Improvement in key aspects of human welfare 52 5.2 The Dominican Republic has persistent chronic and increasing transient poverty 54 6. Equality of opportunities and life chances 60 6.1 Measuring opportunities for all 60 6.2 Have human opportunities for children improved? 61 6.3 Unpacking the changes in human opportunities 62 6.4 “Circumstances” constraining equality of opportunity and economic mobility across generations 62 7. Policy options to strengthen the links between economic growth and equity 66 Conclusions 72 Annexes Annex A. Total number of poor people by area (2000-2011) 75 Annex B. Poverty decomposition by income source (2000–2011) 76 Annex C. Poverty decomposition by income source for moderate and extreme poverty (2000–2011) 77 Annex D. Coverage and incidence of remittances as a percentage (selected years) 78 Annex E. Decomposition of Gini coefficient by income source (2000-2011) 79 Annex F. Synthetic panels using cross-sectional data to predict economic mobility 80 Annex G. Defining vulnerable and middle-class groups 81 Annex H. Constructing the Human Opportunity Index 82 Annex I. Poverty and Inequality in the Dominican Republic compared to Central America 83 Figures Figure 1: Moderate poverty rate: Dominican Republic v. LAC 11 Figure 2: Gini coefficient (2000–2011) 12 Figure 3: Economic mobility across groups and the size of the middle class (2000–2011) 13 Figure 4: Initial characteristics and economic class in 2011 14 Figure 5: Matrix of multidimensional and income poverty in DR, 2000–2011 (selected years) 15 Figure 6: Contribution of “circumstances” to inequality of opportunity, 2011 17 Figure 7: Real GDP growth: Dominican Republic and LAC, 1990–2011 (%) 29 Figure 8: GDP per capita: Dominican Republic and LAC, 1990–2011 (PPP constant 2005 $) 29 Figure 9: Percentage share of economic sectors in value added 30 Figure 10: Sector contribution to value added growth (%) 30 Figure 11: Share of segments in services value added (%) 30 Figure 12: Contribution to change in employment 31 Figure 13: Labor productivity and real earnings 31 Figure 14: Employment to population ratio, 15+ (%) 31 Figure 15: Unemployment rate (%) 31 Figure 16: Moderate and extreme poverty rates (2000–2011) 32 Figure 17: Moderate poverty rates in Dominican Republic, Panama and LAC (2000–2011) 33 Figure 18: GDP growth and changes in poverty 34 Figure 19: Growth rate of income per capita using macro and micro data 35 Figure 20: Gini coefficient (2000–2011) 36 Figure 21: Income per capita growth incidence curves, 2000–2011 38 Figure 22: Poverty decomposition by income source (2000–2011) 40 Figure 23: Gini coefficient decomposition by income source (2000–2011) 41 Figure 24: Decomposition of changes in poverty into intra- and inter-regional population shifts 41 Figure 25: Access to services and documentation 2012 42 Figure 26: Share of jobs by sector 2012 43 Figure 27: Educational attainment 2012 (% completed) 43 Figure 28: Decomposition of changes in poverty into intra- and inter-sectoral shifts 43 Figure 29: Sliders, climbers and stayers in economic status 47 Figure 30: People joining the middle class: DR vs. LAC 48 Figure 31: Upward mobility out of poverty and vulnerability: Origin and destination, 2000–2011 49 Figure 32: Anonymous and non-anonymous growth incidence curves, 2002–2004 and 2004–2011 49 Figure 33: Initial characteristics and economic class in 2011 50 Figure 34: Initial characteristics and upward economic mobility 50 Figure 35: Share of population deprived for each dimension 53 Figure 36: Multidimensional poverty for different values of k 54 Figure 37: Share of population deprived for each dimension by income poverty status 55 Figure 38: Multidimensional headcount by poverty status 57 Figure 39: Matrix of multidimensional and income poverty, 2000–2011 (selected years) 58 Figure 40: Chronic and transient poverty by household characteristics, 2011 58 Figure 41: Matrix of multidimensional and extreme income poverty, the Dominican Republic and LAC (2003 and 2011) 59 Figure 42: Contribution of each circumstance to inequality of opportunity, 2000 and 2011 63 Figure 43: Impact of parental background on children’s education gap at age 15 in LAC, 1995–2009 63 Figure 44: Correlates of teenage pregnancy 64 Figure 45: Moderate Poverty Rates in the Dominican Republic and Central America (2000–2011) 83 Figure 46: Gini coefficient in the Dominican Republic and Central America (2000–2011) 83 Tables Table 1: Moderate and extreme poverty rates (2000–2011) 12 Table 2: Gini coefficient (2000–2011) 12 Table 3: HOI for education, safe water and sanitation, housing and assets ownership 16 Table 4: Real GDP per capita growth: Dominican Republic and LAC, 1990–2011 (%) 29 Table 5: Moderate and extreme poverty rates (2000–2011) 34 Table 6: General and extreme poverty gap (2000–2011) 35 Table 7: Ratio of mean household income to macro indicators 36 Table 8: Gini coefficient (2000–2011) 37 Table 9: Poverty decomposition into growth and redistribution effects 39 Table 10: Intra-generational mobility in the Dominican Republic–Percentage of population (2000–2011) 46 Table 11: Intra-generational mobility in the Dominican Republic, by median income change 46 Table 12: Intra-generational mobility in the Dominican Republic, by percentage of median income change (2000–2011) 46 Table 13: Intra-generational mobility in LAC—Percentage of population (circa 1995–2010) 46 Table 14: Selected indicators and deprivation criteria 53 Table 15: HOI, coverage rate and penalties in DR, 2000–2011 (selected years) 61 Table 16: HOI for education, safe water and sanitation, housing and assets ownership 61 Table 17: Contribution of the “composition” and “coverage” effects to changes in the HOI in DR 2000–2008 62 Table 18: Contribution of the “scale” and “equalization” effects to the “coverage” effect in DR 2000–2008 62 Table 19: Contribution of each circumstance to inequality of opportunity in DR, 2011 63 Boxes Box 1: The Poverty committee experience in the Dominican Republic 32 Box 2: A profile of the Haitian immigrants 42 Box 3: Teenage pregnancy as a proxy of agency in the Dominican Republic 64 » Overview When Prosperity is not Shared: The Weak Links between Growth and Equity in the Dominican Republic Growth and equity in the Dominican Republic: Looking at types of poverty in the Dominican Republic The puzzle reveals that two thirds of income-poor Dominicans, in principal, have the skills and assets needed to gener- C ontrary to the overall experience of the Latin ate higher incomes for themselves but have been un- America and Caribbean region (LAC), strong eco- able to do so. The other portion of the poor consists of nomic growth in the Dominican Republic over the people caught in chronic poverty—long, in some cases past decade has not been accompanied by strong life-long, spells of deprivation. This remains a critical is- improvement across a number of equity dimensions. sue as these are likely the most disadvantaged members These dimensions include the right to be free from abso- of society. Indeed, the Dominican Republic has very low lute poverty, fairness in access to economic opportunities economic mobility with less than 2 percent of the popula- and the ability of individuals to make effective choices and tion moving to a higher economic group over the past de- transform those choices into outcomes. This disconnection cade compared to the regional average of 41 percent. In between growth and shared prosperity in the Dominican fact, over 19 percent of Dominicans actually experienced Republic signals weak fundamentals of equity. From 2000 a worsening in economic status from 2000 to 2011. The to 2011, GDP per capita in the Dominican Republic grew Dominican Republic is also underperforming compared at an annual rate of 3.8 percent compared to a LAC annual to LAC in regards to promoting equitable access to basic average of 2.9 percent. A small contraction of the Domin- goods and services for children. This limits the economic ican Republic’s economy during the 2003-2004 financial opportunities of disadvantaged people. crisis led to a dramatic increase in poverty, with nearly half the population engulfed by it. By 2011, the poverty In the Dominican Republic, following a pattern observed rate had fallen to 40.4 percent, which is higher than the in other countries in the region, the middle class tends When Prosperity is not Shared LAC average and remains higher than the Dominican Re- to opt out of the social contract by demanding private public’s own level in 2000 (32 percent). This trend is more services and refusing to contribute to public goods. This marked in the Dominican Republic’s urban areas, where generates a vicious cycle of low tax compliance, low pub- there are now twice as many poor people as there were in lic services quality and exclusion of the poor (Sanchez and 2000. In these areas, levels of income inequality have bare- Senderowitsch 2012; Ferreira et al. 2012). Strengthening ly changed, while a moderate reduction has occurred in the capacity of institutions to provide quality services rural areas. Overall, total income inequality in the Domini- could reverse this process and reinforce a more cohesive can Republic has been falling at a slower rate than in LAC. social contract. 9 The fundamentals of equity, clearly weaker in the Do- Growth and equity are key outcomes to sustain a robust minican Republic than in other countries of the region, process of shared prosperity. While economic growth is can be strengthened by focusing on three broad policy important for increasing welfare, how equitable a society goals to tie growth to equity and generate a positive cy- is also plays a primary role in driving progress. Moreover, cle of shared prosperity. an equity-driven policy agenda may enhance the capacity to grow in a sustained manner. Enabling people who are 1. Promote equitable, efficient and sustainable fiscal policy, currently marginalized to improve their conditions will un- leash their inherent economic potential, increasing overall 2. Build fair, transparent and efficient institutions that will productivity and thus spurring growth. Beyond normative improve the provision and quality of public goods and reasons, increasing equity has an important instrumental services, expand economic opportunities, increase up- value. It contributes to poverty reduction, increases effi- ward mobility and better protect the poor and vulnerable cient utilization of human and physical resources and po- from economic shocks and tentially increases the sustainability of the social contract by making institutions more responsive to all members of 3. Strengthen access of the poor to labor markets and in- society. This report argues that designing policy options in crease the demand for their labor, so as to make efficient the Dominican Republic to address both equity and growth use of human capital and allow the poor to benefit from as interconnected outcomes will maximize economic economic growth. progress and enhance social and economic inclusion. A conceptual framework for equity analysis The weak links between economic growth and poverty reduction in the Dominican Republic This report uses a comprehensive definition of “equi- ty” which entails that citizens must have equal access Recently, the Dominican Republic has experienced tre- to opportunities, be able to live in dignity and have mendous growth in comparison to LAC, closing the gap the autonomy and voice to participate fully in their with the region. GDP per capita in the Dominican Repub- communities and decide on life plans that they have lic grew by almost 50 percent from 2000 to 2011. The fig- reason to value. The conceptual framework is based on ure for the Latin America and Caribbean (LAC) region as a a tripartite definition of equity, an equity triangle. The whole was 26 percent. This gap has been a recurrent pat- first dimension of equity, fairness—or independence tern in the last few decades: GDP growth in the Dominican from original circumstances—lies in the notion that Republic exceeded the region’s in the 1980s and 1990s initial background characteristics of individuals that as well. This has led to strong, sustained convergence in are out of their control, such as gender or area of res- average incomes between the Dominican Republic and idence, should not limit the set of opportunities avail- LAC. In 1990, the average income in the region was 90 able to them and dictate their achievement. The second percent higher than in the Dominican Republic; by 2011, dimension, the elimination of absolute poverty, entails it was just 24 percent higher. While the country was hit by that all members of a society are guaranteed a well-de- a severe domestic economic crisis in 2003 and 2004 that fined minimum standard to live with dignity. All must contracted its GDP, the growth rate quickly recovered. All have sufficient access to goods and services that they in all, the economy has been remarkably resilient to the are able to achieve well-defined outcomes. The third recent global economic crises. dimension is process freedom, which is about strength- When Prosperity is not Shared ening individuals’ capacity to set goals, make choices Given this level of growth, poverty has not declined as and transform those choices into desired actions and much as would be expected. An extreme increase in pov- outcomes (represented by the notion of agency). This erty due to the economic crisis and slow recovery there- framework allows for a cohesive characterization of in- after shifted poverty levels from below those in the LAC tra- and inter-generational economic mobility, chronic region at the start of the decade to above by the end. In and transient poverty, and between-group inequities 2000, 32 percent of Dominicans were poor, more than 10 that potentially prevent certain vulnerable populations percentage points below the overall LAC figure. The eco- from fully participating and benefiting from the devel- nomic crisis in the Dominican Republic caused the figure 10 opment process. there to shoot up by 17 percentage points, peaking at 50 percent in 2004 to surpass the LAC average of 41 percent. Figure 1: Moderate poverty rate: Dominican Republic v. LAC In the years that followed, the Dominican Republic did re- duce poverty, but levels failed to decline at the same pace 50 as in LAC and the gap widened. By 2011, poverty in the 40 Dominican Republic was at 40.4 percent, dropping only Poverty headcount (%) 9 of the 17 percentage points that it gained during the 30 crisis and remaining above the level at the start of the de- cade.1 In contrast, poverty fell substantially in LAC during 20 the 2000s, reaching 27 percent. Nearly 70 million people emerged from poverty. 10 0 A particular characteristic of poverty in the Dominican 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Republic is its asymmetric response to business cycles. Dominican Republic LAC Compared with other countries in the region, the poverty Note: Poverty line used for the calculation is the LAC level of $4 PPP per capita per day. rate in the Dominican Republic is especially sensitive to Source: Study team’s own estimates using SEDLAC (Socio-Economic Database for Latin economic shocks. Though GDP declined by just 0.3 per- American and the Caribbean) cent in 2003, the start of the crisis, poverty shot up by 52 percent. It took much larger declines in GDP to bring 2000 and 2011. While the Gini, like poverty, increased in about commensurate poverty increases in Mexico in 1995 2003 with the advent of the crisis, it peaked during the re- and Argentina in 2001. Conversely, the Dominican Repub- covery at 0.52 in 2006, before falling in subsequent years. lic’s recovery period, which saw economic growth soar to Thus, while the Gini did fall over the decade and remained around 10 percent in 2005 and 2006, reversed only about lower than the LAC average (0.53 in 2011), the difference half of the increase in poverty suffered in 2003 and 2004. between the Dominican Republic and LAC shrank by So, while the crisis in the Dominican Republic in 2003 and 0.02 points (see Figure 2a). Disaggregation by area sug- 2004 affected the overall economy relatively less than gests that most of this income inequality reduction was other country crises in the region, it exposed the vulnera- attained in the rural parts of the country. While income in- bility to poverty of a large portion of the Dominican pop- equality in urban areas fluctuated between 0.49 and 0.52 ulation with lasting effects. during the period of analysis, and was just one point low- er in 2011 than in 2000, rural inequality fell by four points Poverty in the Dominican Republic is becoming more (nearly 9 percent) over the period, from 0.45 to 0.41. urban. While it is still higher in rural areas than in ur- ban areas, the gap between the two has lessened over Rising monetary labor incomes have been the most im- the decade. This was primarily due to a net increase in portant factor behind the modest poverty reduction in urban poverty following the economic crisis. Though pov- the Dominican Republic. While the incomes of the poor erty declined in urban areas post-crisis by 7.6 percentage did not show any positive growth between 2000 and points, this was slower than the 11.5 percentage point 2004, they grew moderately during the post-crisis peri- reduction in rural poverty, leaving urban areas with a od: around 19 percent in total between 2004 and 2011. poverty headcount which was 54 percent higher in 2011 Labor income, the most important income source of than in 2000 while rural poverty returned to 2000 levels. households, has thus been the largest contributor to pov- Technical analysis confirms that this “urbanization” of pov- erty reduction. This effect is larger for rural households. When Prosperity is not Shared erty is primarily the result of new poor in urban areas as Non-labor income such as public transfers also boosted opposed to rural poor migrating to urban areas. family incomes and helped eliminate poverty, partic- ularly in the second half of the decade. This is probably Income inequality fell moderately during the 2000s, explained by the implementation of Solidaridad, a target- mostly driven by an improvement in income distribu- ed conditional cash transfer scheme linked to education tion in rural areas. The Gini coefficient, a standard mea- and health. This program was created in September 2005 sure of income inequality, fell from 0.51 to 0.48 between to enhance a previously minimal and fragmented social safety net. In 2012, this program expanded into Progre- 1  Extreme poverty followed a similar trend as moderate poverty in sando con Solidaridad. Now covering 80-90 percent of 11 the Dominican Republic. Table 1: Moderate and extreme poverty rates (2000–2011) Poverty Area 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 line Moderate Total 32.0 32.8 32.7 41.5 49.8 47.8 44.2 43.6 44.2 42.1 41.6 40.4 Rural 47.3 47.5 47.2 55.4 59.9 57.0 54.0 51.8 55.2 50.8 50.4 48.4 Urban 23.7 24.8 24.9 33.9 44.1 42.8 38.8 39.1 38.8 37.9 37.3 36.5 Extreme Total 8.1 7.8 8.7 12.0 15.5 16.6 13.7 13.2 13.4 11.8 11.4 10.2 Rural 14.7 13.2 16.1 19.4 22.9 23.7 19.8 18.4 20.6 17.9 16.9 15.3 Urban 4.6 4.9 4.8 7.9 11.4 12.6 10.4 10.4 9.9 8.8 8.7 7.7 Source: Study team’s own estimates based on ENFT 2000-2011 Table 2: Gini coefficient (2000–2011) Area 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Total 0.510 0.497 0.494 0.511 0.506 0.506 0.524 0.493 0.487 0.493 0.478 0.480 Rural 0.448 0.417 0.418 0.431 0.415 0.455 0.484 0.434 0.408 0.413 0.412 0.411 Urban 0.498 0.489 0.483 0.516 0.515 0.507 0.520 0.500 0.489 0.500 0.482 0.486 Source: Study team’s own estimates based on ENFT 2000-2011 Figure 2: Gini Coefficient (2000-2011) Panel A: DR v. LAC Panel B: Urban v. Rural 0.59 0.58 0.51 0.50 0.49 0.48 0.54 0.53 0.46 0.45 Gini Coe cient Gini Coe cient 0.52 0.51 0.41 0.49 0.48 0.41 0.44 0.36 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Gini - LAC Gini - DR Rural Urbano Note: Gini coe cient for Dominican Republic and LAC is based on income per capita. Source: Study team’s own estimates based on ENFT 2000-2011 and SEDLAC (Socio-Economic Database for Latin American and the Caribbean) eligible households in extreme and moderate poverty, it uals, families or other groups of people to improve their represents a commitment by the government to alleviate economic and social status—either by individuals over poverty in the short term and shield vulnerable popula- time (intra-generational) or by families across generations When Prosperity is not Shared tions from future large shocks. The prominence of labor (inter-generational). The analysis presented in this study income and non-labor incomes in reducing poverty is analyzes mobility within generations by measuring direc- consistent with the pattern observed in LAC as a whole. tional income movement, i.e., the net upward or downward movement in individual incomes over time. To do so, it de- Escaping poverty and staying away from it: Income fines three economic groups: (1) the poor, those below the dynamics and the lack of upward economic mobility Dominican Republic’s moderate monetary poverty line of $4.70 per day 2005 purchasing power parity in urban ar- Economic mobility is a key element of economic devel- eas and $4.20 in rural areas, (2) the vulnerable, people with 12 opment. This type of mobility means the ability of individ- between $4.70 a day and $9 in urban areas and $4.20 and Figure 3: Economic mobility across groups and the size of the middle class (2000–2011) Panel A: Sliders, climbers and stayers Panel B: People in the middle class 90% 50 78.9% 80% 70% 40 57.1% Headcount (%) 60% 30 50% 41.4% 40% 20 30% 19.3% 20% 10 10% 1.8% 1.5% 0% 0 DR LAC 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Stayers Climbers Sliders Dominican Republic LAC Note: The numbers in Panel A cover the period 2000-2011 for the Dominican Republic and circa 1995-2010 for LAC. Indicators of economic mobility (“sliders,” “climbers” and “stayers”) are measured in percentages on the left-hand Y-axis. “Middle class” consists of individuals with a per capita income per day higher than $10 in LAC and $9 and $7.35 in the Dominican Republic for urban and rural households, respectively. Both values are expressed in 2005 $ purchasing power parity (PPP). Source: Study team’s own estimates based on ENFT 2000-11, World Bank (2013) with data from SEDLAC (Socio-Economic Database for Latin American and the Caribbean) and WDI. $7.35 in rural areas, incomes levels that place them out of ber of people in the middle class exceeds the number of poverty but with a 10 percent or greater chance of falling people who are poor. Such an inflection point remains far back into it and (3) the middle class, individuals who have away for the Dominican Republic. daily incomes above $9 (urban) and $7.35 (rural) and an upper income threshold of $25. Unsurprisingly, initial endowments, assets and eco- nomic opportunities matter for economic mobility. In addition to limited poverty reduction, the Domini- Households that are today in the middle or better-off can Republic has fostered very little upward economic classes are headed by individuals with higher initial mobility over the past decade, resulting in a stagnant school attainment than heads of households that are in middle class. While the 2003-2004 crisis brought an in- poverty or vulnerability. In fact, middle class household crease in downward economic mobility, the great eco- heads in 2011 had on average four years more education nomic growth during the pre- and post-crisis periods did 11 years earlier than heads of households that were poor not substantially raise people out of poverty or econom- (9.7 compared to 5.7) (Figure 4). Likewise, half of house- ic vulnerability, meaning limited economic mobility and holds in the middle class and three quarters in the well few entrants into the middle class. From 2000 to 2011, just off in 2011 had heads who were employed in the formal under 2 percent of the population in the Dominican Re- sector in 2000. This contrasts with the less than 30 percent public experienced upward mobility (e.g., escaping pov- of household heads initially employed in the formal sec- erty to move into the vulnerable group or moving from tor among typical poor families in 2011. A similar trend vulnerable to the middle class), while 19 percent of the emerges when analyzing location: Current middle class population slid down in economic status, and the over- and well off household heads were more likely to have whelming majority, 79 percent, had no change in eco- resided in urban areas in 2000. nomic status. A miniscule 3.2 percent of Dominicans who were initially poor in 2000 had escaped poverty by 2011, Multidimensional poverty and its dynamics: When Prosperity is not Shared but all of them remained vulnerable to falling back into The persistent and increasing risk of becoming poverty. As a result of these trends, the share of people chronically poor in the middle class remained fixed over the long term. By contrast, in LAC, 41 percent of people entered a higher Widening the lens of poverty measurement to include economic group from 1995 to 2010, and of the people non-monetary indicators allows for a more comprehen- initially in poverty, almost half had climbed out by 2010. sive characterization of disadvantaged people, aiding Furthermore, in the last decade, the middle class in the re- policy design and targeting. Identifying the chronic poor gion increased by almost half, from 21 to 30 percent. Now, requires information about the same individuals over for the first time in the recorded history of LAC, the num- time, which is generally not available in the Dominican 13 Figure 4: Initial characteristics and economic class in 2011 A. School attainment B. Proportion of formal 14 13.4 80% 73.4% 12 60% 10 9.7 50.9% 7.5 42.9% 8 40% 6 5.7 29.3% 4 20% Poor Vulnerable Middle Class Better o Poor Vulnerable Middle Class Better o C. Proportion of urban D. Male head of household 100% 84% 83.3% 87.0% 85% 80.2% 76.4% 80% 71.4% 78.7% 70% 75.7% 76% 55% 51.1% 40% 72% Poor Vulnerable Middle Class Better o Poor Vulnerable Middle Class Better o Note: Initial characteristics calculated in 2000 include: (a) household head’s years of education (b) portion of household heads whose main job was in the formal sector, (c) portions of households residing in urban areas and (d) portions of households headed by a male. Source: Study team’s own estimates based on ENFT 2000–2011 Republic. However, non-monetary indicators of poverty monetary poverty. The proportion of the population who tend to be associated with its persistence. These indica- have incomes below the moderate poverty line but are tors include access to services such as electricity, clean not multi-dimensionally poor are the transiently poor, water, sanitation, quality housing and education, as well while the remaining members of the population, those as asset ownership. When a person is deprived across who are neither multi-dimensionally poor nor income multiple indicators, he or she is classified as “multi-di- poor, are part of the better off. mensionally” poor. Combining this measure of multi-di- When Prosperity is not Shared mensional poverty with income-based measures creates Despite the progress the Dominican Republic has made four groups with which to categorize the population. The in improving access to a spectrum of basic services, the first two groups are people who are multi-dimensionally country has experienced a persistent problem of “chron- poor: the chronic poor are multi-dimensionally poor but ic” poverty over the last decade. While the incidence of also poor in a monetary sense, while the not-income poor multi-dimensional poverty is closely related to that of but deprived group is comprised of the multidimensional monetary poverty in the Dominican Republic, the gap in poor whose incomes are above the poverty line. Among access to basic services between the monetary poor and the chronic poor, those in extreme monetary poverty are non-poor has narrowed as a result of a consistent decline 14 identified as the severe poor given the intensity of their in multi-dimensional poverty. This decline was precipitat- Figure 5: Matrix of multidimensional and income poverty in Dominican Republic, 2000-2011 (selected years) 2000 2011 Not poor but deprived Better o Not poor but deprived Better o 11.9% 55.5% 6.6% 51.2% 3,238 3,238 Income poor Income poor Chronic poor Transient poor Chronic poor Transient poor 1,458 1,458 17.6% 15.0% 13.2% 29.0% 8 7 6 5 4 3 2 1 0 8 7 6 5 4 3 2 1 0 Multi-dimensionally poor Multi-dimensionally poor Extreme poverty line, incomes below: 6.0% of the pop. Extreme poverty line, incomes below: 5.0% of the pop. Note: Dotted line shows the monetary extreme poverty line, $2.12 for urban residents and $2.03 for rural. Severe poor (i.e. people below the extreme poverty line) and MPI poor in 2000 = 5.9 percentage points and in 2011 = 4.9 percentage points. X-axis measures the number of household deprivations; Y-axis is the household per capita income (R$). Source: Study team’s own estimates based on ENFT 2000 and 2011 ed by the Dominican Republic’s success over the decade people became income-poor. Transient poverty remains in providing greater access to a range of services, particu- high seven years after the crisis, suggesting that though larly schooling and sanitation, and improving the figures the Dominican Republic has endowed people who are on house quality and overcrowding. In fact, the propor- transiently poor with physical and human capital (as they tion of Dominicans classified as poor in both a monetary are not multi-dimensionally poor by classification), there and a multi-dimensional sense (the chronically poor) fell are conditions, particularly in the labor markets and en- from 17.6 to 13.2 percent between 2000 and 2011. How- terprise development, that impede them from translating ever, the severe poor fell by only 1 percentage point (from these human and physical assets into higher earnings and 6 to 5 percent), a decline which pales in comparison to better standards of living. In addition to the possibility the achievements of other countries in the region, such of being hit by a serious economic shock, there is also a as Brazil, which managed to reduce severe multi-dimen- chance that some of their productive assets, for instance, sional poverty by almost five and a half percentage points job skills, will depreciate over time, raising the risk that (from 7.1 to 1.5 percent) between 1999 and 2011. Behind the transiently poor will become chronically poor. this limited decline in severe poverty over the decade was its rise to 9.2 percent in 2004, outpacing the growth in Fairness: Is there equality of opportunities for all? chronic poverty in general. This suggests that the slow re- covery from the economic crisis may be due to the depth The notion of fairness as equality of “life chances” in- of poverty into which people were thrown. Those who did volves equitable access to key goods and services that not have access to basic services sank ever-deeper, there- allow a person to progress in life, independently of that by making it harder to escape during times of growth. person’s origin or circumstances. The chances of enjoy- ing a fully productive life are largely influenced by access In addition to the persistence of multi-dimensional to—and quality of—basic goods and services (denoted poverty among the extreme poor, the growth of the here as opportunities) such as food, school enrollment, When Prosperity is not Shared percentage of people who are transiently poor pos- clean water, sanitation and electricity, as well as by the es another important challenge to the country. Even ownership of a minimum set of key assets. The concept of though there were significant drops in the chronic poor fairness followed in this report refers to the idea that per- above the extreme monetary poverty line and the group sonal circumstances that are out of the control of children, that is not income-poor but is deprived , the proportion such as birthplace, gender, race, wealth, parental educa- of people who are classified as transiently poor in the Do- tion and area of residence, should not determine their hu- minican Republic almost doubled over the past decade, man opportunities. For example, all children should enjoy from 15 to 29 percent. This too is largely the result of the access to safe water irrespective of whether they live in an economic crisis, with the middle class shrinking as more urban or rural location. 15 Table 3: HOI for education, safe water and sanitation, housing and assets ownership Year Education Safe water & sanitation Housing Assets Enrollment Completing 6th grade on time Water Sanitation House with hard floor Asset ownership 2000 97 43 61 37 87 62 2004 98 56 64 44 92 59 2011 96 68 61 55 95 62 Note: Values are expressed in percentages and denote coverage rates adjusted by how equitably they are distributed. Source: Study team’s own estimates based on ENFT 2000, 2002, 2004 and 2011 Equitable access to services has expanded in the last for almost 90 percent of the inequality in opportunities. decade, pointing in the direction of enhanced equali- In isolation, parental education and family income each ty of opportunities. However, the rate of improvement explain a quarter of the uneven distribution of opportuni- has been sufficiently slow that it would take more than ties measured in the HOI. Parental background continues a generation to level the playing field for children in to be a strong determinant of the human capital of chil- the Dominican Republic. The Human Opportunity In- dren and thus of their ability to progress in life. Not only dex (HOI) is a measure of the coverage rate of opportu- do children of more-educated parents complete more nities that is adjusted according to how equitably they years of education but there are also substantial differ- are distributed amongst different subgroups based on ences in the quality of schooling that they receive. In sum, circumstances. It shows that access for children to key ba- limited human capital, parents’ inability to earn higher in- sic services and goods has increased over time and that come and the gender of the child are key elements that some opportunities, such as school enrollment, are close appear to limit the child’s opportunities. This restrains so- to universal. Increases in access for circumstance-specific cial inter-generational mobility. disadvantaged groups (those with below-average cover- age rates for services) has fostered equitable allocation The third element of the equity triangle, the notion of opportunities, helping improve the country’s HOI. The of agency, is associated with gaps in service provision Conditional Cash Transfer program and a parallel effort to and differences in service quality. Different groups close supply gaps in basic health and education services have different capacity to influence the system (em- has also contributed to this achievement. Interestingly, powerment), which results in institutions offering dif- the economic crisis that began in 2003 did not seem to ferentiated responses to these groups (UNDP 2008). impede improvements in the HOI—the index grew on Scant service provision is a very basic indicator of weak average by approximately 1 percent per year between agency. Indeed, the political economy of service pro- 2000 and 2011. At this pace, however, it would take the vision, the patterns of exclusion and the overall sense Dominican Republic just under 30 years to universalize of “fairness” in the system become particularly import- the opportunities captured in the HOI. This is similar to ant in the analysis of mobility and middle classes.2 the amount of time it would take for Central America, 36 years, while the LAC region as a whole is projected to Policy options to strengthen the links between achieve this objective earlier, in approximately 24 years. economic growth and equity Factors such as place of residence (urban or rural), ed- Setting out a policy agenda aimed at ensuring that the When Prosperity is not Shared ucation level of parents, family income and gender of gains from economic growth and prosperity are more the child still constrain equality of opportunity and eco- evenly distributed requires an understanding of the nomic mobility across generations. These circumstances links between growth and equity. Equity and growth as help explain why the Dominican Republic has been less a mutually reinforcing virtuous cycle constitute shared successful in providing children with equitable access prosperity (see diagram below). When all members of so- to key services (clean water and sanitation) and goods ciety have the opportunity and skills to generate income, linked with better quality of life (refrigerator, telephone then the society is more equitable. At the same time, with and clean cooking fuel stove) as well as the opportunity 16 to complete 6th grade on time. The four factors account 2  See Ferreira et al. (2012), Chapter 6. more people contributing to the economy, the economy Figure 6: Contribution of “circumstances” grows and everyone, whether poor or non-poor, comes to inequality of opportunity, 2011 out ahead. In this process, the virtuous cycle enhances societal welfare. Equity entails freedom from absolute 30% poverty. There is equality of opportunities regardless of 26% 26% 25% original circumstances and the ability to make autono- 20% mous decisions regarding important life choices. Within 20% the framework of shared prosperity, there are four main 15% 15% channels through which growth and equity reinforce 10% 7% each other: (1) equitable, efficient and sustainable fiscal policy, (2) fair, transparent institutions and effective provi- 5% 3% 2% sion of public goods, (3) well-functioning and accessible 0% markets, and (4) comprehensive and efficient risk man- Area Education of Family income Both parents Gender of household head Number of sibling household head Gender of child agement. A number of policy instruments can strengthen these channels. Improving fiscal policy and the effective- ness of institutions will, for instance, facilitate equity by increasing the opportunity set and economic mobility of Source: Study team’s own estimates based on ENFT 2000 and 2011 the poor and vulnerable. Making markets function better will facilitate the efficient use of the skills and assets of tors, even for workers with high skills, and real earnings the poor. Finally, better risk management can reduce the have fallen across skill groups and sectors. Furthermore, exposure to and impact of shocks on the poor and vul- too few Dominicans have quality education and job train- nerable which, otherwise, could force them to engage in ing and many still lack access to even basic services, di- negative risk coping behaviors. minishing the potential of the Dominican Republic’s pool of human capital. These factors have made it difficult to The virtuous cycle of shared prosperity framework is connect people to economic growth and improve equity, useful in understanding the challenges in the Domini- resulting in limited mobility and inequality of opportuni- can Republic and the entry points for effective policy in- ties. While the Dominican Republic is improving risk-man- terventions. This report has shown that despite econom- agement systems to protect vulnerable populations from ic growth, the Dominican society remains highly unequal. such dramatic increases in poverty as were seen in the For instance, the fastest-growing sectors of the economy 2003-2004 crisis, these social protection programs were have not significantly increased employment. Instead, as of 2011 still leaving over 40 percent of the population employment has increased most in low-productivity sec- in poverty. Strikingly, the number of people who are en- The Virtuous Circle of Shared Prosperity I. Fiscal policy II. Institutions & stability & public goods Fairness When Prosperity is not Shared Growth SHARED Equity PROSPERITY Equity Triangle Freedom Process from absolute freedom III. Well-functioning IV. Risk poverty markets management Source: SEDLAC and The World Bank. 17 dowed with human capital and assets but are unable to tax reforms have not raised the progressivity of the escape poverty has almost doubled over the decade. tax system, particularly by failing to impose more direct taxes. Micro-simulations of the 2012 tax re- The fundamentals of equity can be strengthened by fo- form indicate that, as expected, the increase in in- cusing on the policy areas to be discussed below, thus direct taxes (for instance, an increase in the VAT tying growth to equity and generating a positive cycle rate) will be regressive (Valderrama et al. 2013).3 of shared prosperity. The links between socioeconom- ic outcomes relate not only to economic processes, but • Budget rigidities limit the ability to increase and also to political and social dynamics. Inequality begets redirect allocations to key social sectors so as to inequality, for example, through a distorted allocation of provide more and better public goods and services, resources due to the differing capacity to lobby of groups including systems to protect the poor and vulnera- that have been historically excluded, such as woman and ble from negative shocks. The already limited fiscal minorities, even in the context of a well-functioning de- space is further constricted by structural factors on mocracy (Esteban and Ray 2005). If institutional failures the expenditure side, notably use of public resources such as corruption and weak accountability mechanisms to fund a large electricity deficit. In 2008, that deficit also exist, the distortions are even larger and typically re- alone accounted for 2.7 percent of GDP. The figure has gressive (Gupta et al. 2002). These realities should be con- since decreased but remained substantial at 1.8 per- sidered in policy design. cent of GDP in 2012. All in all, the Dominican Repub- lic remains well below LAC averages in social alloca- Policy area 1 tions. In 2011, the country channeled just 2.3 percent of GDP in public funds towards education, one of the Objective: Promote equitable, efficient and sustainable lowest levels in the LAC region. Steps in the right di- fiscal policy. rection have been taken to raise education spending to 4 percent of GDP in 2013 but capacity constraints Problem to address: The current structure of fiscal pol- may undermine the quality and efficiency of that icy, both on the revenue and expenditure sides, limits spending. Health expenditures have also increased the ability of the Dominican Republic to provide suffi- in recent years, but as of 2011 they still amounted cient and quality public goods and services and places to just half the LAC average. More than a third of the too much of the burden on the poor. In particular, Dominican Republic’s people have no access to clean water; nearly half have no access to sanitation ser- • The fiscal system is hampered by low revenue col- vices. Despite a substantial increase in resources for lection. While the average tax burden in LAC is 20 social assistance, almost half of these expenditures percent of GDP, public revenues in the Dominican Re- are still not allocated on the basis of formal targeting public averaged 13.7 percent of GDP over the past de- criteria. Targeting has become especially important cade, only slightly higher than the level of tax revenue because the economic shock of 2003-2004 caused of Guatemala, the lowest in the region. Unfortunately, some groups to slip into poverty and others to fall the tax reform measure that was passed in November even further into it. 2012 missed the opportunity to address the low tax burden—it is expected to raise fiscal revenues by only Policy options: 1.4 percent of GDP. When Prosperity is not Shared On the revenue side, adjusting the fiscal system can • The tax system relies heavily on indirect taxes, strengthen its redistributive capacity and ensure that limiting progressivity. Nearly one third of total tax the Dominican Republic can afford to uphold a more collection comes from value-added taxes (VAT), on ambitious social contract and improve economic which income has no bearing, though basic goods growth. Specific policy actions to achieve this include: from the consumption basket are exempted. Fur- thermore, despite the exemptions of basic goods, about 50 percent of tax exemptions and incentives 3  These micro-simulations do not take into account increases in property tax or tax on motor vehicles, both of which would likely 18 benefit the more affluent members of society. Past make the results slightly more progressive. • Making the tax system more progressive by replac- used for the Progresando con Solidaridad program, ing some of the current indirect taxation with direct non-contributory health insurance and electricity and taxation (e.g., personal and corporate income taxes) gas subsidies, which helped reduce poverty after the and ensuring exemption of basic goods; 2003-2004 economic crisis. This will help assure that help goes only to people who need it; • Revising existing tax exemptions (now 5.9 percent of GDP) and incentives to make them more progres- • Consolidating small existing social programs, partic- sive; and ularly in the social protection sector, to avoid duplica- tion and fragmentation of services and help rational- • Strengthening the ability of tax collection mecha- ize expenditures; and nisms to detect and limit evasion, for instance, by increasing the capacity of the Tax Administration Au- • Addressing the structural deficit in the electricity thority to conduct tax audits and enforce control. Tax sector, which diverts a substantial amount of public evasion is likely to make actual taxation less progressive. resources away from the social sector. On the expenditure side, more public resources should Policy area 2 be allocated to education, health, water and sanitation, social protection and risk management systems, with Policy objective: Build fair, transparent and efficient in- steps taken to maximize the efficiency of resources. stitutions that will improve the provision and quality of Specific policy actions to achieve this include: public goods and services, expand economic opportu- nities, increase upward mobility and better protect the • Investing more resources in water and sanitation poor and vulnerable from economic shocks. and quality education and health to endow the poor with the skills and assets required to take full advan- Problem to address: The quality of public services is low tage of their productive capacity and live better lives; and, despite significant improvement, inequities in ac- cess to basic public services remain, particularly to the • Strengthening current social assistance programs poor. This constrains their set of human “opportunities” and disaster risk management systems, including to progress in life.4 In particular, safety nets to protect the poor and vulnerable from the effects of major shocks. This will ultimately lessen the • The low quality of governance affects service de- impact of shocks on the overall economy and acceler- livery in key sectors. There have been significant im- ate recovery to combat the asymmetric response of provements in service delivery, for instance, efforts poverty in the Dominican Republic to business cycles; to adopt transparent targeting mechanisms, publish budget data and open lines for queries and com- • Analyzing the differences between monetary and plaints. Yet many Dominicans believe that inefficient non-monetary deprivations to better distinguish the bureaucracy, lack of transparency, anti-reform coa- chronic from the transient poor and to design and litions and low trust in governance systems remain target strategies and programs that address each important obstacles to raising competitiveness and group’s needs, such as better infrastructure and ed- improving service delivery in the social sectors.5 ucation services for the chronic poor and enhanced When Prosperity is not Shared safety nets for the transient poor. This type of targeting • Despite remarkable increases in school enrollment, may have particular impact in urban areas, where there inefficient use of budget allocations and low qual- are more likely to be both types of poverty (and thus ity continue to hamper the education system. The a need to distinguish between the two). Overall, the effect will be to stanch the “urbanization of poverty;” 4  The Transparency International Corruption Perceptions Index (TI-CPI) perceives the Dominican Republic as one of the most corrupt countries in LAC—the TI-CPI ranks the Dominican Republic as 118th • Further linking public social expenditures to current among 174 countries evaluated. targeting mechanisms, such as the SIUBEN (Siste- 5  According to Kaufmann et al. (2012), the the Dominican Repub- lic has had historically weak rule of law, and government effective- ma de Identificacion Unica de Beneficiarios), currently ness has been low. 19 system is characterized by operation in double shifts, • Further developing and strengthening incentive, high teacher/pupil ratios, high teacher absenteeism monitoring and accountability structures between and little use of performance data to manage for re- service providers, clients and policy makers, as well sults. Not surprisingly, more than 41 percent of third as procurement and financial management reforms graders lack basic math skills, while over 31 percent (e.g. the introduction of framework agreements) to cannot meet the lowest defined level of the SERCE fight corruption and promote good governance, reading test. Students in the Dominican Republic quality service delivery, and accountability. The IDEC show some of the lowest academic performance in initiative, which brings public and private sector and the LAC region, and this lack of efficiency in education civil society actors together to improve the quality of affects the poor disproportionally. While the better education, and the community scorecards piloted by off study for 12.3 years and complete 10.1 grades, the the Progresando con Solidaridad program are good poor study for 11.1 years and complete 7.8 grades. examples. As the analysis shows, households headed by people with higher education are more likely to experience • Building on existing improvements in budget trans- upward economic mobility and their children enjoy parency and strengthening efforts to better link more opportunities to enhance their economic poten- planning and budget formulation and to further tial. Thus, education is a major tool in promoting both adopt performance-based budgeting and manage- inter- and intra-generational economic mobility and ment instruments to link increased efficiency and ensuring sustainable returns to social expenditure.6 effectiveness according to established standards of quality. A crucial aspect of such instruments is the • Health coverage and quality are patchy, ranking generation and use of performance data (for instance, behind those of countries that spend at similar lev- linking student test scores to teacher evaluations); els or less. Maternal mortality rates in DR are around and 150 per 100,000 live births, over a third higher than the regional average, and infant mortality, at 22.3 per • Making efficiency analyses (such as cost-benefit, 1,000 live births, is also above the LAC average. Like- cost-effectiveness and economic rate of return) wise, immunization coverage, though it has increased, and monitoring and evaluation systems far more remains below regional levels. Coverage through full common to economically value the return on public health insurance has increased considerably but still investments, monitor the achievement of results and leaves half of the poor uninsured, resulting in out-of- learn what works and what does not work. pocket expenditures financing a high share of health services. Combining quality of education and health Policy area 3 services, the World Economic Forum ranked the Do- minican Republic 107th out of 139 countries in 2010. Policy objective: Strengthen access of the poor to labor markets and increase the demand for their labor, so as Policy options: to make efficient use of human capital and allow the poor to benefit from economic growth. Developing new mechanisms and strengthening exist- ing ones to monitor the quality and provision of public Problem to address: Low labor force participation and goods and services with a goal of increasing account- dearth of formal, well-paying jobs, particularly among When Prosperity is not Shared ability and ensuring that expenditures give maximum the poor and vulnerable, youth and women, at a time value to society. In addition to increasing spending in the when real earnings have been systematically falling social sector and further linking it to effective targeting across most economic sectors. In particular, mechanisms, policy actions to increase accountability, cit- izen engagement and service quality could include: • The sectors in which labor productivity has in- creased, such as manufacturing, the wholesale and 6  From 2006 to 2012 the level of satisfaction in local government retail trades, communications and financial services, services regarding education and tax fairness at municipal levels de- have not contributed much to job creation. For ex- creased almost 18 percent, according to Latinobarometro, for LAPOP 20 (2012) ample, manufacturing had the second-highest annu- al average of contribution to the gross value added working nor studying and women, as a group, have growth rate from 2001–2011. Yet it was the sector that even lower labor force participation. lost the most jobs over the same period, reducing its share of total employment by 6 percentage points • An inadequately educated workforce and a skills (Abdullaev and Estevão 2013). This kind of growth mismatch. According to the Global Competitive- without job creation may explain much of the lack of ness Report (2011-2012), one of the biggest prob- poverty reduction in spite of economic growth during lems in doing business in the Dominican Republic the second half of the 2000s. is an inadequately educated workforce. The Domin- ican Republic ranks in the bottom third of the 142 • The majority of job creation, instead, has occurred countries analyzed in terms of higher education and in low-skill, low-productivity sectors, which sug- training. As this report shows, another reason why gests that many of the jobs created are of low quali- the number of transient poor doubled (from 15 to ty. Three in every four jobs created between 2004 and 30 percent) could be that though some people are 2011 were in the informal sector. Increased employ- endowed with certain human capital to generate in- ment during this period did lift some people out of come, they are unable to do so. This is due to a skills poverty; however, the low quality of jobs may explain mismatch: the low relevance and quality of their ed- why the country failed to reduce poverty to at least ucation and training do not match what employers pre-crisis levels. As shown in the analysis, nearly 30 are looking for. percent of the population was poor in 2011 despite having basic education and access to services and/ • Several factors reduce the ability of the private sec- or assets, signifying that some people who are en- tor to create more and better jobs. The enclave type dowed to generate adequate income are unable to of development that has characterized two of the do so. Moreover, the Dominican Republic had the sec- leading industries, special economic zones (Zonas ond-highest growth in the share of low-skill services Francas) and tourism, restrains employment gener- as a percentage of employment between the late ation and business linkages to the overall economy. 1990s and late 2000s out of a sample of selected LAC The business environment is further undermined by countries (Aedo and Walker 2012). complex regulatory processes, including labor rules, and by weak institutions, lack of transparency, inef- • Real earnings have been falling uniformly along the ficient government bureaucracy, partial access to fi- earnings distribution, and high-skilled workers are nance, underdeveloped capital markets and poor la- being continuously absorbed into low-skill jobs. bor skills and infrastructure. On average, real earnings per hour both of self-em- ployed and private sector wage workers were about Policy options: 27 percent lower in 2011 than in 2000. Even workers employed in high-productivity sectors or who have Improving the relevance and quality of education. Spe- tertiary education have not seen their incomes rise. cific policy actions to achieve this include: These trends are consistent with the low upward mobility and high downward mobility documented • Increasing investments in secondary and tertiary in this study. The fact that real earnings are also flat education, changing the content of education and or declining in sectors that have strong productivity training and creating the right incentive structure When Prosperity is not Shared growth and make the largest contribution to the over- within key institutions in the education sector to all output is puzzling but may be explained by the ensure that all Dominicans have the necessary skills high percentage of people in need of jobs following to work in high-productivity jobs. Bearing in mind the crisis. that rising labor income was the biggest engine of poverty reduction in the last decade, provision both • Low labor force participation. Labor force partici- of the endowments and avenues for people to par- pation in the Dominican Republic is 10 percentage ticipate in the labor market will raise standards of points lower than the regional average. More than a living and increase the global competitiveness of the third of people between 18 and 29 years are neither country. 21 Increasing labor market information flow and training key economic sectors and further opening up markets linked to job opportunities to raise the employability to local and foreign investment, including the finan- and productivity of the labor force. Creating targeted ac- cial sector, and tive labor market programs (ALMPs) designed to enhance opportunities for high-productivity job creation and for- • Creating and targeting incentives for entrepreneur- mality can raise the employability of the labor force and ship and innovation in sectors with the capacity to facilitate young people’s transition from school to work, generate large numbers of jobs that are likely to have without creating major market distortions. Further bene- significant positive spill-overs and income distribu- fit for training and intermediation will come from consol- tion effects for the rest of the economy. idating the existing ALMPs, the Dominican Republic such as the national training institute INFOTEP, the intermedi- Conclusions ations services of SENAE, the labor observatory OMLAD and the youth training program PJyE. Specific policy ac- Sustained economic and social progress requires a vir- tions to achieve this objective include: tuous cycle of growth and equity as the fundamental policy goal. If growth is the result of an equitable process • Further developing temporary employment pro- of income generation, societies can claim to be on a path grams to allow ALMPs to provide a counter-cyclical of shared prosperity. Based on these notions, this report response in times of economic downturns and em- has applied a variety of empirical methodologies and data ployment crisis and to activate the more disadvantage to assess whether the strong economic growth that the segments of the labor force; Dominican Republic has enjoyed during most of the last decade benefitted the more disadvantaged groups in an • Updating and expanding training and retraining economic and social sense. In doing so, the report exam- programs that couple educational attainment with ined the conditions in the Dominican Republic upholding productive skills to improve the quality of the labor the key pillars of what constitutes an equitable society, force and widen access to job opportunities; namely equality of opportunities regardless of original circumstances, the capacity of the system to bring peo- • Strengthening intermediation services to provide ple out and keep them out of absolute poverty, and the job-search assistance and reduce information asym- existence of agency—the ability of people to make deci- metries between labor supply and demand; sions and convert them into actions—for all. This involved the analysis of the patterns that characterize aggregate • Providing entrepreneurship training and grants to monetary poverty and income inequality trends, house- equip individuals, including the poor and vulnerable, holds’ income mobility, and non-monetary dimensions of with the skills and capital to start and sustain busi- welfare that portray the deprivations experienced by the nesses; and poor, as well as the distribution of opportunities for chil- dren and proxies for the amount of agency that different • Consolidating existing institutions and scaling up groups in society have. effective interventions to ensure coordination, in- tegration and efficacy of existing ALMPs as a means Despite strong economic growth over the past decade, to develop and implement a country-wide system to large inequities still exist in Dominican society and are promote employment. declining at a slower rate than expected. Strong growth When Prosperity is not Shared in the Dominican Republic occurred in every year of the Improving the business environment to foster better last decade except 2003 and 2004 when the economy competition, investment climate, entrepreneurship contracted due to a banking crisis. In total, the Dominican and job creation. Many of the policies proposed above Republic grew its GDP per capita by almost 50 percent are expected to contribute to this goal. Additional policy from 2000 to 2011. But despite this growth, the funda- actions include: mentals of equity remain low in the Dominican Republic. For instance, though strong growth resumed after the cri- • Improving competition policy by eliminating an- sis, the country has been slow in decreasing poverty that 22 ti-competitive practices through market regulation in soared by 17 percentage points to reach nearly half the population. The rate remains at 40.4 percent, higher than progressivity. The narrow fiscal space and institutional the 32 percent level of 2000. Chronic poverty—people en- weaknesses, in turn, constrain the size and effectiveness during long spells of poverty—remains an issue with only of social expenditures, limiting access to crucial goods a one percentage point decrease since 2000, from 6 per- and services, particularly for the poor and vulnerable. Of cent to 5 percent of the population as of 2011. Even more great concern is that even if they are endowed with the concerning is that about one third of the poor are in this means to progress in life, many individuals are employed state despite having the skills to generate higher income. in informal, low-paying jobs and have high vulnerability This group, the transient poor, has almost doubled since to economic shocks. Growth, especially in the last decade, the level in 2000. In addition, the country shows very low is becoming more and more concentrated in sectors such economic mobility with less than 2 percent of the popula- as financial services, transportation and communications tion rising to a higher economic group. In fact, more than and tourism) that have either little employment creation 19 percent of the population experienced a worsening or low-paying jobs. in economic status from 2000 to 2011. Furthermore, de- spite improving access to services, coverage and quality To address these issues, this study identifies three poli- remains uneven across population groups, thus limiting cy actions to address the underlying factors of inequity. the economic opportunities of disadvantaged people. These include: (1) adjusting the structure of fiscal policy, both on the expenditure and revenue sides, to make it Compared to LAC, growth in the Dominican Republic more equitable, efficient and sustainable; (2) developing is stronger but the country is falling behind the wider and strengthening monitoring, social accountability and region in a number of equity dimensions. Largely as a incentive mechanisms to increase the quality and provi- result of the 2003-2004 crisis, poverty rates in the Domin- sion of public goods; and (3) strengthening the access of ican Republic, lower than LAC’s overall rates in 2000, now the poor and other disadvantaged groups to labor mar- exceed the region’s average. Furthermore, though pov- kets and increasing the demand for their labor to make erty began to decline after the crisis, the rate of decline efficient use of human capital, allowing the poor to maxi- has been slower than LAC’s over the same time period. mize returns on their endowments and to ultimately ben- In terms of reducing income inequality, the Dominican efit from economic growth. Republic continues to perform better than the region. However, as improvements have been modest over the Finally, policy design aimed at promoting a more bal- decade, LAC is catching up to the Dominican Republic. anced development path could be effectively informed In terms of inter-generational income mobility, a striking by further analytical work. Particularly salient work difference remains between the Dominican Republic and could be performed in the areas of fiscal policy, social LAC—while 41.4 percent of people rose to a higher eco- sectors and labor markets. This would help to further nomic group in LAC, only 1.8 percent did so in the Domin- uncover the underlying factors that inhibit the gains of ican Republic. The country is also underperforming LAC in growth from being more evenly shared across the popu- terms of increasing access to basic goods and services for lation. A relevant concern in this analysis is discrepancies children. At the rate of improvement of the past decade, between the Dominican Republic’s national accounts DR would take longer to reach universal access than the and household survey data. Serious analytical efforts LAC average. While investigating the underlying causes should be devoted to understanding the apparent dis- behind disappointing progress in equity is beyond the connection between macro and micro data that hinders scope of this report, the study postulates some hypothe- the ability of national statistics to accurately reflect mac- When Prosperity is not Shared ses as to why, compared to LAC, the fundamentals of eq- roeconomic and social progress. uity are weak in the Dominican Republic. This report identifies areas of priority for policy to ad- dress underlying factors of inequity in the Dominican Republic, including fiscal policy, institutional effective- ness and the performance of labor markets. In short, the Dominican Republic has a weak fiscal capacity, as it raises low revenues and does so in a manner that harms 23 » Chapter 1 Growth and equity in the Dominican Republic: The puzzle C ontrary to the overall experience of the Latin Looking at types of poverty in the Dominican Republic America and Caribbean region (LAC), strong reveals that two thirds of income-poor Dominicans, in economic growth in the Dominican Republic principal, have the skills and assets needed to gener- over the past decade has not been accompa- ate higher incomes for themselves but have been un- nied by strong improvement across a number of eq- able to do so. The other portion of the poor consists of uity dimensions. These dimensions include the right people caught in chronic poverty—long, in some cases to be free from absolute poverty, fairness in access to life-long, spells of deprivation. This remains a critical is- economic opportunities and the ability of individuals sue as these are likely the most disadvantaged members to make effective choices and transform those choic- of society. Indeed, the Dominican Republic has very low es into outcomes. This disconnection between growth economic mobility with less than 2 percent of the popula- and shared prosperity in the Dominican Republic sig- tion moving to a higher economic group over the past de- nals weak fundamentals of equity. From 2000 to 2011, cade compared to the regional average of 41 percent. In GDP per capita in the Dominican Republic grew at an fact, over 19 percent of Dominicans actually experienced annual rate of 3.8 percent compared to a LAC annual a worsening in economic status from 2000 to 2011. The average of 2.9 percent. A small contraction of the Do- Dominican Republic is also underperforming compared minican Republic’s economy during the 2003-2004 fi- to LAC in regards to promoting equitable access to basic nancial crisis led to a dramatic increase in poverty, with goods and services for children. This limits the economic nearly half the population engulfed by it. By 2011, the opportunities of disadvantaged people. poverty rate had fallen to 40.4 percent, which is high- When Prosperity is not Shared er than the LAC average and remains higher than the In DR, following a pattern observed in other countries in Dominican Republic’s own level in 2000 (32 percent). the region, the middle class tends to opt out of the social This trend is more marked in the Dominican Republic’s contract by demanding private services and refusing to urban areas, where there are now twice as many poor contribute to public goods. This generates a vicious cy- people as there were in 2000. In these areas, levels of cle of low tax compliance, low public services quality and income inequality have barely changed, while a moder- exclusion of the poor (Sanchez and Senderowitsch 2012; ate reduction has occurred in rural areas. Overall, total Ferreira et al. 2012). Strengthening the capacity of institu- income inequality in the Dominican Republic has been tions to provide quality services could reverse this process 24 falling at a slower rate than in LAC. and reinforce a more cohesive social contract. The fundamentals of equity, clearly weaker in the Do- minican Republic than in other countries of the region, can be strengthened by focusing on three broad policy goals to tie growth to equity and generate a positive cy- cle of shared prosperity. 1. Promote equitable, efficient and sustainable fiscal policy, 2. Build fair, transparent and efficient institutions that will improve the provision and quality of public goods and services, expand economic opportunities, increase up- ward mobility and better protect the poor and vulnerable from economic shocks and 3. Strengthen access of the poor to labor markets and in- crease the demand for their labor, so as to make efficient use of human capital and allow the poor to benefit from economic growth. When Prosperity is not Shared 25 » Chapter 2 A Conceptual framework for equity analysis T his report uses a comprehensive definition of for equity analysis is based on a tripartite definition of “equity,” based on ideas from welfare econom- “equity,” encapsulated by the equity triangle depicted be- ics and political philosophy. There is a vast liter- low: ature that ponders the nature of equity and how to achieve it. According to Bourguignon et al. (2002) and The first dimension of the equity triangle, fairness, is World Bank (2006), a social state is equitable when it satis- based on the notion that individuals’ initial circumstanc- fies two conditions: (1) there is “equality of opportunity,”7 es and background characteristics should not limit the implying that achievement and life chances are not as- set of opportunities available to them. To be equitable, sociated with an individual’s origin or circumstances and a society must allocate resources and opportunities (2) absolute poverty has been eliminated. If we include such that no group is limited in their advancements and a “comprehensive outcomes” perspective, in which the achievements due to certain characteristics out of their process to achieve a result is an essential part of our eval- control, such as place of birth, family’s economic status, uation of the final state (Sen, 1998, 2010), then the no- race or gender. For instance, the fact that members of a tion of equity also involve the process aspects of equity. country’s indigenous groups, whose minority status is a In this view, the processes by which individuals achieve characteristic beyond their control, would have lower ac- socioeconomic advancement, effectively opting among cess to clean water is unfair, and thus, inequitable as de- life options they have reasons to value, become a funda- fined in this report. The second dimension of the equity mental part of the evaluation of the social state (Cord and triangle, the right to be free from absolute poverty, entails Lopez-Calva 2013). This relates to individual agency and that all members of a society are guaranteed access to When Prosperity is not Shared autonomy for specific groups of society. goods and services such that they are able to achieve a well-defined minimum standard to live in dignity. In oth- Thus, for a society to be “equitable,” its citizens must er words, if a society is to be considered equitable, no have equal access to opportunities, be able to live in member should be living in extreme poverty—defined dignity and have the autonomy and voice to partici- in the Dominican Republic as life on less than $2.12 a day pate fully in their communities, choosing life plans that in urban areas and $2.03 in rural ones—and all individu- they have reasons to value. This conceptual framework als ideally should be safe from falling into that condition. The third dimension of the equity triangle, process free- 26 7  See Rawls (1971), Roemer (1998) and Cohen (2001) dom, requires that individuals have the social, economic and political means to make effective choices and trans- Dominican Republic in the near future to fully assess the form those choices into outcomes that they have reasons equity situation and devise responsive policies. to value. This involves the concept of agency, a “capacity to do by themselves.” The constitutive elements of agen- The interaction among these three constitutive ele- cy are: (1) aspects of the individual, including objective ments of equity is determined by context, markets and ones (for instance, a person’s good health or lack of it) policy. Context—the functioning of specific markets, the and subjective ones (such as aspirations and self-drive); exposure to risks and the country’s institutional envi- (2) contextual elements, such as social norms, culture ronment—exercises deep influence over the equity pic- and formal institutions; and (3) power, defined as the ture. Markets play a great role too because members of capacity to align the actions of others to one’s own in- the labor force need the skills and education in order to terest (Cord and Lopez-Calva 2013). This report will look compete in them. Policies, in turn, affect the dynamics in mostly into conditions related to the first two corners of the search for better outcomes, establishing, for example, the triangle (fairness and absence of absolute poverty). specific fiscal structures, expenditure plans and interven- As discussed in the concluding section, agency-related tions to overcome inequities embedded in the system. issues, including those related to economic empower- Thus, equity assessment can inform policy makers on ment (particularly for women), institutional capacity and ways to decrease poverty, inequality and barriers to agen- aspirational gaps, should be part of the agenda in the cy through key channels. Equity Triangle Fairness MARKETS POLICIES Freedom from absolute poverty Process freedom CONTEXT Market structure Institutions Norms Exposure to risk When Prosperity is not Shared 27 » Chapter 3 The weak links between economic growth and poverty reduction 3.1 The macro context: A top performer creased by 51 percent in the 2000s to $5,785 and by 45 in economic growth percent in the following decade to reach $8,387 in 2010. This rise of 2.2 times between 1990 and 2010 is impres- S trong and sustained economic growth in the last sive when judged by regional standards. For instance, in- two decades has made the Dominican Republic come per capita in LAC increased annually in the 2000s by one of the fastest-growing economies in the re- 2.2 percent on average, but the corresponding value for gion. On average, the economy grew annually in the Dominican Republic was 3.8 percent. The difference real terms by 5.3 percent between 2001 and 2011, out- was even more marked for the period between 2005 and performing the Latin America and Caribbean region (3.4 2011, 2.9 percent for LAC and 5.3 percent for the Domini- percent) by almost two percentage points. Gross domes- can Republic (Table 4). Had the GDP per capita of the Do- tic product in the Dominican Republic expanded by 72 minican Republic grown in the 2001–2011 period at LAC’s percent compared to 43 percent in the region over this average growth rate, it would have been 12 percent lower time period. This gap has been a recurrent pattern decade than what it actually was in 2011. The Dominican Repub- after decade. Indeed, the Dominican Republic exceeded lic’s relatively strong performance has raised average in- the region’s GDP performance in the 1990s by the same comes, converging with levels of the region. In 1990, the margin of 2 percentage points and by more than 1 per- average income in the region was 85 percent higher than centage point in the 1980s. Overall, the real GDP growth in the Dominican Republic; by 2011, the gap between the rate of the Dominican Republic exceeded LAC’s in 16 out two had fallen to 21 percent (Figure 8). Assuming that go- of the 22 years spanning the period 1990-2011 (Figure 7). ing forward the GDP per capita of the Dominican Repub- This long-term trend is largely the result of economic re- lic and LAC will grow at the average rates seen in the last When Prosperity is not Shared forms of the early 1990s that sought to introduce sound five years, the gap will vanish by 2020. macroeconomic policies and attract foreign capital. Large flows of foreign direct investment and remittances have A domestic banking crisis in the first half of the 2000s helped drive the economic expansion. contracted the Dominican Republic economy; by con- trast, the country was remarkably resilient in the face The Dominican Republic’s fast pace of wealth creation of global economic crises. The country enjoyed a period raised average per capita incomes substantially. The av- of rapid and stable economic growth during most of the erage annual income of Dominicans in 1990 was $3,833 1990s and early 2000s. But in 2003, with rapid currency 28 (measured in purchasing power parity, or PPP). This in- depreciation and inflation in the wake of a domestic fi- Table 4: Real GDP per capita growth: Dominican Republic and LAC, 1990–2011 (%) 1990-95 1996-2000 1990-2000 2001-05 2006-2011 2000-2011 DR 1.5 5.2 3.2 2.0 5.2 3.8 LAC 1.1 1.6 1.3 1.3 2.9 2.2 Source: World Development Indicators Figure 7: Real GDP growth: Dominican Republic and LAC, Figure 8: GDP per capita: Dominican Republic and LAC, 1990–2011 (%) 1990–2011 (PPP constant 2005 $) 12 12000 10 8 10000 6 8000 4 2 6000 0 -2 4000 -4 2000 -6 -8 0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Dominican Republic LAC Dominican Republic LAC Source: World Development Indicators Source: World Development Indicators nancial crises precipitated by a major bank failure, the public was contributed by the service sector, compared economy contracted, by 0.3 percent. After slow growth in to 52.3 percent in 2000. Accordingly, shares of other 2004, it recovered in 2005 and grew at even faster rates sectors have fallen—manufacturing declined the most than those seen in the pre-crisis period. While the global with a drop of 5.6 percentage points. This switch to ser- economic crisis, particularly the recession in the United vices can likely be explained by an initial decline in con- States, slowed down the Dominican economy in 2008 struction’s share due to the economic crisis in 2003 and and 2009,8 growth remained positive in both years (5.2 2004 and a shift from manufacturing towards services percent in 2008 and 3.4 percent in 2009) and well above throughout the decade. the levels in the LAC region as a whole.9 The resilience to crisis was not intrinsic to the Dominican economy; rather, Growth in the services sector has consistently driven the government implemented a large expansionary fiscal total value added growth. A simple average of the con- policy to protect against shock. While the policy served tribution to value added growth reveals that the services its purpose, such actions translate into higher debt which industry contributed, on average, 3.2 points per year out can limit both future governmental capacity and the sus- of a total annual average growth of 4.4 points from 2000 tainability of high levels of economic growth. to 2011. This represents 73.7 percent of annual growth. The services sector is also the only sector in which growth When Prosperity is not Shared 3.2 A slowly shifting economic landscape occurred every year. Construction, by contrast, declined in six of the 12 years considered and was a major cause of The Dominican Republic is shifting towards a more the decline in GDP during the crisis. service-based economy, but slowly. For instance, in 2011, 61 percent of value added in the Dominican Re- The composition of the service sector changed over time. For instance, communications increased its share of 8  The economic slowdown was mostly driven by falls in remittanc- the total services value added from 9.3 percent in 2000 to es, exports and tourism. nearly 30 percent in 2011. In contrast, the share of every 9  The real GDP of the region grew by 4.1 percent in 2008 and con- tracted by 1.7 percent in 2009. other segment fell over the time period. This fall was most 29 Figure 9: Percentage share of economic sectors in value added 100 90 80 Percent of value added GDP 52.3 53.8 54.5 55.7 55.9 56.4 57.3 58.9 60.7 61.7 61.8 61.0 70 60 50 40 30 30.6 29.1 28.8 28.6 29.0 28.6 27.1 26.1 25.5 24.4 24.4 25.0 20 10 6.7 6.2 6.1 5.0 4.8 4.9 5.6 5.4 5.1 4.8 5.0 4.8 1.0 0.8 0.8 0.9 0.9 0.9 0.9 0.8 0.5 0.3 0.2 0.4 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011* Agriculture Mining Construction Manufacturing Services Source: Central Bank of the Dominican Republic. Note: * preliminary gures Figure 10: Sector contribution to value added growth (%) Figure 11: Share of segments in services value added (%) 12 100 90 13.3 11.1 10.1 10 3.6 2.9 2.3 Percent of value added GDP 80 2.2 2.1 1.8 2.4 2.4 1.8 8 70 12.4 11.5 9.0 6 60 9.3 21.4 29.2 50 13.3 4 40 10.7 14.1 9.0 2 30 14.2 11.0 20 0 20.7 16.1 16.1 10 3.7 2.7 2.4 -2 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011* 2000 2005 2011* Energy & water Trade Hotels, bars and restaurants Services FTZ manufacture Construction Agriculture Communications Education Transportation and stocking Mining GDP growth Value added Health Housing rents Pub. admin., defense, SS Other activities Source: Central Bank of the Dominican Republic. Note: * preliminary gures Source: Central Bank of the Dominican Republic. Note: * preliminary gures pronounced in trade and transportation, which were down most to the relatively low employment growth from 1996 4.5 and 4.2 percentage points, respectively (Figure 11). to 2011 (4.1 percentage points), notably “other services,” which includes community, social, health, education and The sectors which have contributed the most to the private households with employed persons, fell as a share strong economic growth in the Dominican Republic of total value added and tend to consist of lower-skill ac- have not created a substantial number of jobs. Abdul- tivities (Figure 12 and Figure 13). When Prosperity is not Shared laev and Estevao (2013) find that the sectors which drove economic growth in the Dominican Republic over the Real earnings have fallen across the income distribu- past two decades, which include manufacturing, trans- tion. From 2000 to 2011, real earnings fell by 27 percent portation and communication and financial services, (Abdullaev and Estevao 2013). This decrease in wages has have either declined or have not changed as a share of affected people from all earnings levels and was slightly total employment. Thus, rising labor market productivity stronger for the top earners. The steep dip in hourly earn- in these sectors rather than a substantial increase in em- ings during the 2003-2004 crisis partially explains this ployment contributed to GDP growth in the Dominican dramatic reduction in wages, and there was an upward 30 Republic. By contrast, the sectors that contributed the trend in earnings in the post-crisis period. However, im- Figure 12: Contribution to change in employment rate Figure 13: Labor productivity and real earnings by sector (2011-1996 percentage points) 5.20 Agriculture 5.00 Mining and quarrying Manufacturing 4.80 Electricity, gas and water 4.60 Construction 4.40 Wholesale and retail trade Hotels, bars and restaurants 4.20 Transport and communication 4.00 Public administration and defence 3.80 Other services 1996 1998 2000 2002 2004 2006 2008 2010 -5.00 -3.00 -1.00 1.00 3.00 5.00 Labor productivity (output per hour) Real earnings per hour Note: Labor productivy and real earnings shown in logarithm. Source: Abdullaev and Estevao 2013. Source: Abdullaev and Estevao 2013 Figure 14: Employment to population ratio, 15+(%) Figure 15: Unemployment rate (%) 63 20 61 18 59 16 57 14 55 12 53 51 10 49 6 47 4 45 2 1991 1994 1997 2000 2003 2006 2009 2000 2002 2004 2006 2008 2010 Latin America and Caribbean Dominican Republic Broad Open Source: Abdullaev and Estevao 2013. Source: Abdullaev and Estevao 2013. provements are coming more slowly than would be ex- differences introduced over time in the collection of pected given the growth during this time. In fact, even data constrain the validity of long series to track mon- workers in high-productivity sectors have faced a decline etary and non-monetary indicators of household wel- in earnings over the period as they were not able to re- fare. For example, the semi-annual labor force survey cover to 2000 levels following the crisis. There is also low Encuesta Nacional de Fuerza de Trabajo (ENFT), the most labor force participation compared to the region, which regular source of information to measure and charac- When Prosperity is not Shared may be explained by the scarcity of quality work opportu- terize the living conditions of Dominicans, including nities and high unemployment, especially among women family incomes, is only comparable for the period 2000- (Figure 14 and Figure 15). 2011. Additionally, until 2012, the Dominican Republic lacked an official and commonly accepted methodolo- 3.3 Poverty trends: Recovering slowly from the gy to measure and monitor poverty. This report utilizes economic crisis official poverty estimates derived from the 2012 launch of an official poverty methodology produced by an in- Assessing long-term poverty trends in the Dominican ter-institutional technical poverty committee created Republic poses serious challenges. Methodological for that purpose. 31 Box 1: The Poverty committee experience in the Dominican Republic The effect of economic growth and income redistribution on poverty can only be measured correctly if a country has well-defined methodologies for the construction of welfare aggregate and poverty lines. Before 2012, the Dominican Republic lacked these, relying instead on multiple poverty estimates that were computed by international institutions such as the World Bank, the IADB and ECLAC based on international poverty lines. Consequently, the Government gathered a team of experts on poverty measurement from different institutions, including The World Bank, in order to devise official methodologies for poverty calculations. The committee agreed on several technical and theoretical points such as the most appropriate welfare income and its components; the conformation and estimation of the basic bundle of needs including a food basket for extreme poverty; and current expenditure for the estimation of the Engel coefficient used in moderate poverty. The committee also agreed on the best way to calculate the monetary value of poverty lines, the frequency of updates and the years of comparability over time. On July 31, 2012, the Dominican Republic launched this official poverty methodology and issued official poverty numbers for the 2000-2011 period. Figure 16: Moderate and extreme poverty rates percentage points annually. As of 2011, 40.4 percent of Dominicans were characterized as poor. The trend of the (2000–2011) extreme poverty rate—the proportion of people living 50 49.8% on $2.12 a day in urban areas and $2.03 a day in rural 44.2% 44.2% 47.8% 41.6% areas, making them unable to cover basic food needs— 40 43.6% 42.1% 41.5% 40.4% mirrored movements of the moderate poverty rate both 32.8% at the height of the economic crisis and during the sec- Headcount (%) 30 30.0% 32.7% ond half of the 2000s. The extreme poverty rate declined 20 during the recovery period but remained higher at 10 15.5% 13.7% 13.4% 11.4% percent in 2011 than it was prior to the crisis. 8.1% 8.7% 16.6% 10 12% 13.2% 11.8% 10.2% 7.8% While GDP has risen faster in the Dominican Republic 0 relative to LAC over the decade, this growth has been 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 less inclusive of the poor than that of the region. Pover- Moderate Extreme ty in the Dominican Republic has risen, shifting poverty Source: Study team’s own estimates based on ENFT 2000-2011 levels from below those in the LAC region at the start of the decade to above by the end. Using a moderate pover- Data from the official methodology show that there ty line comparable across countries in the region ($4 PPP was no poverty reduction at the beginning of the 2000s per day), 32 percent of Dominicans were poor in 2000, and poverty spiked during the banking crisis in 2003- more than 10 percentage points below the percentage of 2004. As Figure 16 shows, moderate poverty (defined as people who were poor in the LAC region as a whole. In the the number of people with insufficient incomes to afford years that followed, while poverty in LAC declined, pover- a basic basket of goods and services) remained flat be- ty rose dramatically in the Dominican Republic, peaking tween 2000 and 2002 but increased dramatically during at almost 50 percent in 2004 following the country’s eco- the 2003-2004 crisis. Poverty went up from 32 percent in nomic crisis. By 2011, poverty in the Dominican Republic 2002 to 41 percent in 2003 and peaked in 2004, when the had declined by half of the increase brought on by the When Prosperity is not Shared incomes of half of the population fell below the poverty crisis, remaining above the level at the start of the decade line. Extreme poverty followed a similar trend, remaining and the LAC level (Figure 17, Panel a). In contrast, pover- flat pre-crisis and spiking during 2005 when it reached ty fell substantially in LAC during the 2000s, resulting in 16.6 percent, more than double the rate in 2000. nearly 70 million fewer people living in poverty. Post-crisis, poverty declined, but at a slow rate in spite Poverty reduction in the Dominican Republic is also of strong economic growth during the recovery period. slow when compared with the performance of Panama, From the economic recovery in 2005 onwards, moderate a close neighbor country with a similar level of eco- 32 poverty started to gradually decrease, on average by 1.3 nomic growth. When compared to the average poverty Figure 17: Moderate poverty rates in Dominican Republic, Panama and LAC (2000–2011) Panel A Panel B 50 50 40 40 Poverty headcount (%) Poverty headcount (%) 30 30 20 20 10 10 0 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Dominican Republic LAC Dominican Republic Panama Note: Poverty line used for the calculation is $4 PPP per capita per day. Source: Study team’s own estimates using SEDLAC (Socio-Economic Database for Latin American and the Caribbean) reduction in Central America, the Dominican Republic re- global crisis (on average by 1.9 percent), yet poverty rates tained a lower poverty headcount, even during the crisis, remained almost intact. and performed similarly in trend thereafter (see Annex I, Figure 45). Given the substantial economic growth of the The same asymmetric response is seen in the Dominican Dominican Republic during the 2000s, a more compa- Republic during times of robust economic growth. For in- rable Central American country may be Panama. In fact, stance, seemingly no poverty reduction occurred in 1997- among all the countries in Central America, Panama at- 2000 and 2002, even though the economy expanded tained the highest GDP growth rates, climbing on aver- without interruption at annual rates of nearly 6 percent age by 4.6 percent per year. In the Dominican Republic, it or higher.10 Moreover, sustained average real GDP annu- increased on average by 3.8 percent. However, in contrast al growth of 7 percent between 2005 and 2011 helped to the trend in the Dominican Republic, Panama’s story eliminate only half (9 percentage points) of the increase of substantial economic growth has been accompanied in poverty that occurred in 2003 and 2004. Specific years by impressive and fairly inclusive poverty reduction. Us- of rapid growth in the Dominican Republic (for exam- ing the $4 line, overall poverty decreased by a startling ple, near or above 10 percent in 2005 and 2006) brought 19 percentage points between 2000 and 2011, from 40 to poverty down on average by only 2.5 percentage points 21 percent. Comparing just the post-crisis period, Panama (Figure 18, Panel B). Again, this trend diverges from the also achieved a faster rate of poverty reduction than did pattern found in LAC as a whole, where poverty fell by the Dominican Republic. (Figure 17, Panel b). a similar magnitude but following much slower average real annual GDP growth (7 percent in the Dominican Re- A particular feature in the Dominican Republic is the public vs. 4.2 percent in LAC). asymmetric responsiveness of poverty to business cycles: big jumps during economic crises but slow re- Due to the moderate impact of rapid economic growth duction in times of strong economic growth. As noted on poverty in the Dominican Republic, the number of When Prosperity is not Shared above, moderate poverty in the Dominican Republic in- poor people in 2011 remains at levels similar to those at creased by 18 percentage points during the domestic the peak of the 2003-2004 economic crisis. The number financial crisis of 2003-2004, more than doubling, even of poor in 2011 (just over 4 million) is only slightly lower though GDP contracted by only 0.3 percent in 2003 and than the level reached in 2004 and 50 percent higher than growth resumed in 2004 at a 1.3 percent rate. By compar- the level observed in 2000. This trend is more marked in ison, it took much larger contractions of GDP in Mexico in urban areas, where there are now twice as many poor 1995 and Argentina in 2001 to generate similar increases in poverty there (Figure 18, Panel A). Similarly, GDP fell 10  See “Dominican Republic Poverty Assessment: Achieving More significantly in the whole LAC region in 2009 during the Pro-Poor Growth,” World Bank, Washington, October 30, 2006. 33 Table 5: Moderate and extreme poverty rates (2000–2011) Poverty Area 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 line Moderate rural 47.3 47.5 47.2 55.4 59.9 57.0 54.0 51.8 55.2 50.8 50.4 48.4 urban 23.7 24.8 24.9 33.9 44.1 42.8 38.8 39.1 38.8 37.9 37.3 36.5 total 32.0 32.8 32.7 41.5 49.8 47.8 44.2 43.6 44.2 42.1 41.6 40.4 Extreme rural 14.7 13.2 16.1 19.4 22.9 23.7 19.8 18.4 20.6 17.9 16.9 15.3 urban 4.6 4.9 4.8 7.9 11.4 12.6 10.4 10.4 9.9 8.8 8.7 7.7 total 8.1 7.8 8.7 12.0 15.5 16.6 13.7 13.2 13.4 11.8 11.4 10.2 Source: Study team’s own estimates based on ENFT 2000-2011 Figure 18: GDP growth and changes in poverty A. GDP contraction and poverty changes in LAC B. GDP growth and poverty changes in Dominican Republic 20 10 Mex (95) 2003 DR (03) 8 2004 Arg (02) Par (02) 15 6 Change in poverty (pp) Change in poverty (pp) 10 4 2 2002 Uru (02) 5 Per (98) 0 2005 -2 0 2006 -4 -5 -6 -12 -10 -8 -6 -4 -2 0 -5 0 5 10 15 Real change of GDP Real change of GDP (2000 - 2010) Note: Panel A includes a selected sample of countries in LAC. Changes in poverty are shown in percentage points. Source: Study team’s own estimates based on WDI and SEDLAC (Socio-Economic Database for Latin American and the Caribbean) people as there were in 2000 (2.4 vs. 1.2 million). The whereas rural poverty returned to the pre-crisis level in number of people in extreme poverty decreased relative- 2011, the poverty headcount in urban areas, though it ly faster after the economy recovered in 2005, but it is still declined during the recovery, was still 54 percent higher 50 percent higher than in 2000 (Annex A). in 2011 than in 2000. An analogous pattern is observed for the extreme poverty rate. In sum, urban centers now Poverty has become a more urban phenomenon. account for 60 percent of the total number of poor, up The gap in the incidence of poverty between rural and from 48 percent in 2000. urban areas has narrowed in the last decade. In 2000, the rural poverty rate (47 percent) was almost twice as The evolution of the intensity of poverty mirrors the high as the urban rate (24 percent). Eleven years later, trends of the poverty headcount. The intensity of pov- When Prosperity is not Shared the headcount in rural areas is less than a third higher erty is captured by the poverty gap (PG), an index that than in urban areas [see Table 1]. The absolute differ- measures the average distance between the income of ence between one and the other has fallen from 23 to the poor and the poverty line for the whole population 12 percentage points. However, the decline in the gap in poverty – either moderate or extreme. The PG in the is not the result of reduction in rural poverty but is in- Dominican Republic was flat at around 11-12 percent of stead driven by a significant increase in urban pover- the poverty line in 2000, jumped to 20 percent in 2004 ty, most of it linked to the 2003-2004 economic crisis. and 2005 and started falling over the second half of the Poverty went up by 86 percent in urban centers in that decade to hit 15.1 percent in 2011 (Table 6). While the PG 34 time, compared to a 26 percent rise in rural areas. And in rural areas has recurrently been higher than in urban Table 6: General and extreme poverty gap (2000–2011) Poverty Area 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 line Extreme rural 4.7 3.7 4.6 5.7 6.9 8.1 6.6 5.8 6.5 5.3 4.8 4.4 urban 1.4 1.5 1.5 2.5 3.2 4.0 3.2 3.2 3.0 2.5 2.2 2.1 total 2.6 2.2 2.6 3.6 4.5 5.4 4.4 4.1 4.1 3.4 3.1 2.8 Moderate rural 18.4 17.7 18.8 22.7 25.5 25.4 23.1 21.5 23.3 20.5 20.1 19.2 urban 8.1 8.5 8.7 12.5 17.0 17.3 15.0 15.2 14.9 14.0 13.8 13.0 total 11.7 11.7 12.2 16.1 20.0 20.1 17.9 17.4 17.6 16.1 15.9 15.1 Source: Study team’s own estimates based on ENFT 2000-2011 Figure 19: Growth rate of income per capita using macro and micro data Latin America (circa 2000–2010) Dominican Republic (2000–2011) 7% 15% 6% 10% 5% 3.8% 4% 5% 3% 2.3% 2% 0% 1% -5% 0% -1% -1.5% 2.0% -10% -2% -15% Brazil El Salvador Argentina Honduras Venezuela, Rep Costa Rica Guatemala Nicaragua Panama Peru Dominican Rep. Ecuador Uruguay Chile Colombia LAC Bolivia Paraguay Mexico -20% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 GDP per capita growth Mean income growth (ENFT) Growth rate GDP per capita Growth rate mean income Source: Study team’s own estimates based on WDI, SEDLAC (Socio-Economic Database for Latin American and the Caribbean) and ENFT 2000–2011 areas, the difference has narrowed over time. As of 2011, of living, i.e. income growth, is similar in both types of the PG in rural the Dominican Republic was 19.2 percent data is to compare the annual growth rate of GDP or ag- compared to 13 percent in urban areas (Table 6). Overall, gregate consumption per capita (calculated with macro direct transfers of a total amount of nearly 1.2 percent of data) against the annual growth rate of the mean income the GDP would be necessary to bring the income of all (calculated with micro data) (Figure 19). Comparing the poor households up to the poverty line.11 This compares differences between the Dominican Republic’s national to an almost identical value of 1.3 percent to achieve the accounts (macro) and the ENFT labor force survey (micro) same objective in 2000. data with macro-micro gaps in 17 other countries in LAC shows that the gap in the Dominican Republic is among A serious disconnection exists between macro and mi- the largest in the region. A deeper inspection of the data When Prosperity is not Shared cro data in the Dominican Republic. National accounts in the Dominican Republic shows that (1) the growth rate (an example of macro data) generally serve as the under- of the mean income is usually lower than the growth rate lying data for measures of aggregate economic activity of GDP per capita and (2) the gap broadened during the (notably GDP) and growth while household surveys (mi- years of the economic crisis (2003-2004) (Figure 19, Panel cro data) provide the figures for poverty estimates. A stan- b). The discrepancies between survey estimates and na- dard way to check whether the change in the standards tional accounts continue when the annual growth rate of the mean income is compared against the annual growth 11  This assumes that there are no administrative costs associated rates of consumption per capita and gross national prod- with these transfers and that they can be made in a perfectly effi- cient manner (e.g. lump sum transfers). uct, both taken from national accounts. 35 Table 7: Ratio of mean household income to macro indicators Ratio of income to: 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 GDP 78.3 75.6 70.1 57.6 51.5 50.7 51.5 46.6 45.3 45.6 43.6 43.3 GNP 81.8 79.1 73.3 61.6 56.2 53.7 54.3 49.1 47.2 47.4 45.0 45.0 Consumption 101 97.3 89.9 73.6 66.0 61.6 62.4 56.4 51.6 53.5 50.0 50.0 Source: Study team’s own estimates based on WDI and ENFT 2000–2011 Figure 20: Gini coefficient (2000–2011) Panel A: Dominican Republic v. LAC Panel B: Urban v. Rural 0.59 0.58 0.51 0.50 0.49 0.48 0.54 0.53 0.46 0.45 Gini Coe cient Gini Coe cient 0.51 0.51 0.41 0.49 0.48 0.41 0.44 0.36 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Gini RD Gini LAC Rural Urban Note: Gini coe cient for Dominican Republic and LAC based on income per capita. Source: Study team’s own estimates based on ENFT 2000–2011 and SEDLAC (Socio-Econo- mic Database for Latin American and the Caribbean) Research using cross-country data over time indicates to answer a survey actually completes it. In many cases, that overstating growth and understating poverty re- the probability of response is negatively or positively cor- duction is a fairly common issue worldwide. The dis- related with socioeconomic characteristics. For example, crepancy between macro and micro data is not isolated in surveys better-off households are generally less likely to to the Dominican Republic. Using 557 survey-based esti- respond, leaving top-income earners less represented in mates of mean consumption or income per head for 127 the results. Another potential cause of the difference be- countries covering the period 1979–2000, Deaton (2005) tween sources is the use of price and price indices, which shows that income per head measured from household are also subject to their own sampling and non-sampling surveys often grows less rapidly than GDP per head mea- errors. Finally, discrepancies may be caused by variations sured in national accounts. He further shows that the ratio in survey design, such as coverage, recall period, survey of the former to the latter is less than 60 percent, tending responder and non-market set prices (Deaton 2005). Ad- to fall over time as real income in macro data increases. ditional hypotheses usually put forward in the case of the This is consistent with the trend observed for the Domini- Dominican Republic—but not formally tested yet—point can Republic (Table 7). to economic growth being driven by capital gains or ac- When Prosperity is not Shared cruing to top income earners and foreign investors, all of The difference between the two sources of data may which are not well captured in household surveys. have a number of causes. The definition, and success in meeting the definition, of consumption and income may In spite of the mismatch between the two sources of vary between survey data (e.g., excluded services, rent data, there are reasons to believe that the Dominican and financial services imputations, etc.) and national ac- Republic exhibits low poverty-growth elasticity in times counts. The disparity between consumption or income of strong economic growth. Disconnections between data and national accounts could also be tied to survey macro and micro data are common, and while the gap in 36 non-response rates, because not everyone who is asked the Dominican Republic is particularly large compared to Table 8: Gini coefficient (2000–2011) Area 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Rural 0.448 0.417 0.418 0.431 0.415 0.455 0.484 0.434 0.408 0.413 0.412 0.411 Urban 0.498 0.489 0.483 0.516 0.515 0.507 0.520 0.500 0.489 0.500 0.482 0.486 Total 0.510 0.497 0.494 0.511 0.506 0.506 0.524 0.493 0.487 0.493 0.478 0.480 Source: Study team’s own estimates based on ENFT 2000-2011 that of other countries in LAC, it does not preclude the context, inequality in the Dominican Republic (0.48) by viability of certain inferences about the Dominican Re- 2011 was lower than in the pooled LAC region (0.53) and public’s poverty and growth trends. The direction of the followed a similar trend, but at a smaller magnitude, as change in the poverty headcount derived from the Do- countries in Central America (see Annex I, Figure 46). minican Republic’s micro data corresponds to the growth trends derived from the Dominican Republic’s macro data However, most of the improvement in income distribu- and is consistent with the countercyclical nature of pov- tion is concentrated in rural areas. Disaggregation by erty, namely that it tends to decrease when the overall area suggests that most of the income inequality reduc- economy is growing and vice versa. Thus, the response of tion was attained in the rural parts of the country. While poverty to the state of the Dominican Republic economy income inequality in urban areas fluctuated between 0.49 has moved in the expected direction. Even if one were to and 0.52 during the period of analysis, and was just one cut growth rates by half—i.e., to a growth rate similar to point lower in 2011 than in 2000, rural inequality fell by the average performance of LAC—the reduction in pov- four points (nearly 9 percent) over the period, from 0.45 to erty would be small relative to decreases in poverty seen 0.41, a 7 point drop from the high of 0.48 reached in 2006. in other countries in the region. In other words, taking As discussed in more detail below, the larger reduction in into account the limitations of the data, the results still rural income inequality contributed to the relatively great- show that the Dominican Republic is underperforming in er reduction in poverty accomplished in rural areas, where terms of poverty reduction considering the positive mac- a full recovery from the economic crisis was realized. roeconomic conditions (growth, stability) that the coun- try enjoyed through most of the 2000s. 3.5 Unpacking the changes in poverty and inequality 3.4 Income inequality trends: Moderate improvements, mostly for rural households 3.5.1 The role of income growth and distribution Income inequality fell moderately during the 2000s. The By raising household incomes, economic growth is ex- Gini coefficient, a standard measure of income inequal- pected to contribute to reducing poverty. Growth inci- ity,12 fell from 0.51 to 0.48 between 2000 and 2011. The dence curves (GIC) are an instrument to separately assess fluctuations in the coefficient follow the economic trends the impact of economic growth on the poor, the middle during the pre-crisis, crisis and post-crisis periods, falling class and the wealthy. To do so, GICs plot the growth rate by 2 points from 2000 to 2002, increasing by 3 points be- between two points in time of the welfare measure (in- tween 2002 and 2006—partly coinciding with the timing come or consumption) for each percentile of the baseline When Prosperity is not Shared of the economic crisis—and then declining by 4 points distribution (for instance, using household surveys), thus through the second half of the decade. Other measures examining how the gains of economic growth—or the to gauge the dispersion of the income distribution (for losses of recessions—are distributed across the popula- instance, the ratio of the average income per capita of tion. GICs performed for the Dominican Republic uncover the 90th to the 10th percentile or the 75th to the 25th two marked trends. percentile) exhibit a similar evolution. In an international The incomes of the poor did not show any positive growth between 2000 and 2004. Whereas macro data 12  The Gini coefficient varies between 0 and 1, where 0 means complete equality and 1 means complete inequality. show that the economy expanded between 2000 and 37 Figure 21: Income per capita growth incidence curves, 2000–2011 2000-2002 2004-2011 0% 17% -2% 16% 15% -4% 14% -6% 13% -8% 12% -10% 11% -12% 10% 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Source: Study team’s own estimates based on ENFT 2000–2011 2002, income growth for the same period derived from composition using the ENFT vary across the periods of the micro data (ENFT) indicates that real incomes fell. Un- analysis. In the pre-crisis period (2000-2002), income surprisingly, the decline in household income was sub- contraction and a decline in inequality operated in op- stantially larger during the 2003-2004 crisis, at around 23 posite directions, thus keeping poverty levels almost flat, percent on average for families below the poverty line. particularly in rural areas. At the time of the crisis (2003- Moreover, GICs by area indicate that urban households 2004), both income contraction and increased inequality bore the largest burden in terms of income contraction. contributed to the substantial increase in poverty that For instance, the incomes of the urban poor decreased al- occurred. Nevertheless, there are marked differences in most twice as much (28 percent) as incomes of the rural the influence of each factor in explaining the spike in poor (15 percent). poverty. The effect of negative income growth was near- ly seven times stronger than the effect brought about by In contrast, strong economic growth during the post-cri- the change in welfare distribution. In contrast to their ef- sis period benefited the poor but just moderately. Poor fects during the crisis period, income growth and a more rural households enjoyed the largest real income growth, equitable income distribution brought about an average although the magnitude was not impressive: around 19 7.65 percentage point reduction in poverty from 2004 percent in total between 2004 and 2011. Likewise, the onwards, contributing 5.2 and 2.5 percentage points incomes of the urban poor grew relatively faster than respectively, while in rural areas, the contribution of in- those of the better off. The relatively slow increase in the come growth relative to the redistribution component incomes of people below and slightly above the poverty was substantially higher (Table 9). line is consistent with a period of robust economic growth that has brought only modest gains in poverty reduction. 3.5.2 The role of different sources of income Alternative analyses show that economic growth and Rising monetary labor incomes have been the most changes in inequality have contributed in different important factor behind the modest poverty re- When Prosperity is not Shared degrees to increases and reductions in poverty. Anoth- duction in the Dominican Republic. Unpacking the er way to shed light on the factors driving the trends in changes in household income by source can further welfare is to decompose changes in poverty into chang- illuminate the factors driving changes in poverty. Re- es due to balanced income growth and changes due to shifting welfare distribution in the absence of income sence of changes in inequality—a change in the mean income while growth (Datt and Ravallion 1992).13 Results from the de- holding the Lorenz curve (i.e. income distribution) fixed at some lev- el, and (2) a change in inequality in absence of economic growth—a change in the Lorenz curve while keeping the mean income con- 13  This decomposition is based on the notion that the level of pov- stant at some reference level. For more details on the methodology, 38 erty could change due to two factors: (1) economic growth in ab- see Datt and Ravallion, 1992. Table 9: Poverty decomposition into growth and redistribution effects. Total and by area 2000–2011 Period Area Distribution Growth Total change 2000-2002 Rural -5.21 4.77 -0.43 Urban -2.05 3.22 1.17 Total -3 3.6 0.6 2002-2004 Rural 1.5 10.87 12.36 Urban 4.21 14.52 18.73 Total 2.17 14.43 16.6 2004-2011 Rural -1.87 -10.2 -12.07 Urban -1.72 -3.09 -4.81 Total -2.49 -5.16 -7.65 2000-2011 Rural -6.29 6.15 -0.14 Urban 0.01 15.08 15.09 Total -3.17 12.73 9.55 Note: Negative values show poverty reduction effects of each component. The official poverty rate for each year is the average of the poverty rates of the first and second semester. However, the total poverty rate changes presented in this exercise are calculated using the second semester of each year and thereby do not match with official numbers. Source: Study team’s own estimates based on ENFT 2000-2011 sults from these decompositions indicate that labor treme and moderate poverty, respectively.14 The average income, the most important income source of house- transfer was $30.53 per household.15 As evidenced by the holds, was the largest contributor to poverty reduction dramatic rise in poverty during the economic crisis, the so- throughout the 2000s. This effect was larger for rural cial safety net at that time was not adequate to protect the households (see Annex B). Unsurprisingly, the poverty vulnerable from falling into poverty during shocks. The es- reduction role of labor income has been stronger in the tablishment of Solidaridad, the precursor to Progresando post-crisis period but was also present in pre-crisis and con Solidaridad, in 2005 represented a shift in policy to crisis times. The prominence of labor income in reduc- lessen the effects of social risks and protect the vulnerable ing poverty is consistent with the trends observed for from future shocks, helping to decrease the elasticity of LAC as a whole, where labor market income account- poverty to economic contractions. ed for 55 percent of the reduction of poverty in recent years (World Bank 2011). Remittances account for a high share of national in- come but their evolution over time has not contrib- Non-labor income, probably public transfers, also boost- uted to poverty reduction. According to macro data ed family incomes and helped reduce poverty. However, (i.e. national accounts), remittances flowing into the the contribution of non-labor incomes is only evident in Dominican Republic represented around 8 percent of the second half of the decade and is equally important the GDP between 2000 and 2011. This share peaked at in both urban and rural households. Non-labor income around 11 percent during the 2003-2004 economic cri- explains approximately 18 percent of the reduction in sis but started to fall gradually from then to 6 percent poverty achieved between 2004 and 2011, similar to the in 2011. Micro data from the ENFT shows a similar trend When Prosperity is not Shared portion (20 percent) estimated for the LAC region in the but that the ratio of remittances to total household in- 2000s. The poverty reduction effect of non-labor income is come is between two and three times higher.16 The de- larger for households in extreme poverty than households compositions of poverty by income source indicate that below the moderate poverty line. The increasing influence of non-labor income is probably driven to a large extent 14  Numbers based on records from the Technical Division of the by the implementation and expansion of Progresando con Social Policy Cabinet. 15  Department of Planning, Progresando con Solidaridad Solidaridad, a targeted conditional cash transfer scheme 16  This result is expected because the GDP captures more com- that by 2012 covered 655,394 households, representing ponents of income such as public expenditures, inversion and net exports while the labor survey only captures earned and unearned around 90 and 80 percent of the eligible households in ex- household income. 39 Figure 22: Poverty decomposition by income source (2000–2011) Panel A Panel B 6 10 4.44.0 4.1 4.9 4.1 4 6.0 6.3 1.8 1.1 1.9 4.8 2 1.1 0.8 5 3.6 3.8 3.1 3.6 0.7 0.5 1.3 1.5 0 -0.3 -0.2 0 -2 -1.8 -0.4 -2.3 -4 -5 -6 -4.9 -8 -10 -9.3 -10 -10.5 -12.4 -12 -15 2000-2002 2002-2004 2004-2011 2000-2011 2000-2011 Moderate poverty Moderate poverty Extreme poverty Remittances In-kind non-labor income Non-labor income In kind labor income Labor income Income from rent Note: Panel A presents results from a decomposition of changes in moderate poverty across three time periods. Panel B presents results from decompositions of changes in extreme and moderate poverty for the period 2000-2011. The o cial poverty rate for each year is the average of the poverty rates of the rst and second semester. However, the total poverty rate changes presented in this exercise are calculated using the second semester of each year and thereby do not match with o cial numbers. Source: Study team’s own estimates based on ENFT 2000–2011. changes over time in the flow of remittances—as well and in-kind non-labor income lowered the Gini coeffi- as labor and non-labor in-kind income—increased the cient—the rest contributed to making the income dis- incidence of poverty (Figure 22). A simple incidence tribution more unequal, particularly remittances. In the analysis using the ENFT confirms this. As a proportion after-crisis period, labor and non-labor income reduced of total incomes, remittances are between 5 to 10 per- income inequality, more than offsetting the opposite centage points higher among households in the upper influence on the Gini coefficient of in-kind (labor and half of the distribution. Moreover, while the first decile of non-labor) income (Figure 23 and Annex E). the pre-remittance income distribution received around 1 percent of the total amount of remittances reported in 3.5.3 The role of population shifts between urban and the ENFT, nearly 40 percent of the remittances accrued rural areas to the top decile. Remittances were even more regres- sive during the 2003-2004 economic crisis, a period in Migration between rural and urban areas explains which over half of the total remittances were sent to very few of the changes in poverty, whether increases households in the top income decile (Annex D). None- or reductions. Changes in poverty can also be decom- theless, remittances are an important source of income posed by changes in income and its distribution that for households at the bottom of the distribution. A sim- took place inside each area (the “intra-regional” effect) ple simulation that subtracts remittances from house- and by changes in these two factors driven by the in- hold income without taking into account any behavioral crease in the share of people in urban areas migrating responses would increase poverty from 40.4 to 46.3 per- from rural areas (the “inter-regional” effect). The decom- cent in 2011. position shows that the intra-regional effect is the vastly dominant effect, accounting for more than 95 percent of When Prosperity is not Shared Similar to their influence on poverty reduction, labor it across the different economic periods (pre-crisis, crisis and non-labor income have improved the distribution and post-crisis). And while the inter-regional (migration) of household incomes. Labor incomes have systemati- effect did increase during the period of rapid economic cally contributed to reducing the Gini coefficient while growth (2004-2011), it is too small at only 6.1 percent to the effect of other sources of income on the Gini coeffi- have been a major driver of changes in poverty. In other cient has been variable, in degree and in direction, across words, the emergence of new poor in the urban areas of time. Between 2000 and 2002, all income sources moved the Dominican Republic created the observed “urbaniza- in the same direction, making the income distribution tion” of poverty rather than the arrival of poor migrants 40 more equal. During the crisis period, only labor income from rural areas. Figure 23: Gini coefficient decomposition by income source (2000–2011) 0,048 0,05 0,04 0,026 0,03 0,02 0,015 0,01 0,00 -0,01 -0,006 -0,004 -0,02 -0,008 -0,013 -0,03 -0,04 -0,029 -0,05 -0,044 -0,06 -0,07 -0,066 -0,08 2000-2002 2002-2004 2004-2011 2000-2011 Remittances In-kind non-labor income Non-labor income In kind labor income Labor income Income from rent Total change Source: Study team’s own estimates based on ENFT 2000-2011 Figure 24: Decomposition of changes in poverty into intra- and inter-regional population shifts 2000-2002 2002-2004 2004-2011 120% 120% 120% 101.4% 99.4% 96.7% 100% 100% 100% 80% 80% 80% 60% 60% 60% 40% 40% 40% 20% 20% 20% 6.1% 0.1% 0.8% 0% 0% 0% -1.5% -0.2% -2.7% -20% -20% -20% Intra-regional Inter-regional Interaction e ect Intra-regional Inter-regional Interaction e ect Intra-regional Inter-regional Interaction e ect Note: The decomposition presented in the table follows the methodology in Ravallion and Huppi (1991). The two regions considered for the decomposition are urban and rural. Source: Study team’s own estimates based on ENFT 2000-2011 3.4.4 The role of population shifts between formal and Though job informality is generally linked with low- informal jobs er pay and quality, a decomposition of the changes in poverty in the Dominican Republic during the 2000s The income losses or gains that affected poverty are shows that the growth during the crisis as well as the not explained by shifts of workers between the for- decline post-crisis of poverty cannot likely be explained When Prosperity is not Shared mal and informal sectors. Many of the new jobs cre- by changes in informality (the “inter-sectoral” effect) ated in the Dominican Republic in recent years are in (Figure 28). Rather, the overwhelming driver of poverty the informal sector, with low pay, keeping the level of changes in the analysis was the change in the real wag- job informality high (Abudallaev and Estevao, 2013).17 es and earnings of workers within their corresponding sectors (“intra-sectoral” effect). In fact, real wages and 17  This is based on the “productivity-based” measure of informal earnings in both the formal and informal sectors have employment, which characterizes workers as informal if they are em- been falling over time. ployed in enterprises with fewer than five employees or are employ- ers or self-employed non-professionals in a low productivity field of work including farmers, operators and drivers, artisans and workers, merchants, sellers and unskilled workers. 41 Box 2: A profile of the Haitian immigrants In 2012, the Dominican Republic’s National Statistics Office (ONE) performed the nation’s first survey specifically targeted at the nation’s immigrant population. The Encuesta Nacional de Inmigrantes (ENI-2012) found many differences between native Dominicans and immigrants, who account for 7.9 percent of the Dominican Republic’s population. Furthermore, household and demographic differences between Haitian immigrants, who account for 87.3 percent of the Dominican Republic’s immigrant population (excluding descendants), and immigrants from other nations were found. While the majority of immigrants live in urban areas (68.2 percent), Haitian immigrants are more likely to live in rural areas than the rest of the immigrant population (34.9 percent compared to 10.4 percent). Furthermore, average household size is slightly lower for Haitian immigrants (2.3) than non-Haitian immigrants (2.7). Consistent with a higher likelihood of Haitian immigrant households having a male head (83.1 percent compared to 77.9 percent of non-Haitian immigrant households), the share of males in the Haitian immigrant population is larger (65.4 percent) than among non-Haitian immigrants (57.5 percent). There are also large differences in terms of access to certain basic services and important personal documentation (see Figure 32). Immigrants from countries other than Haiti are more likely to be literate, have access to utilities and health insurance and to possess documents such as passports, birth certificates and national ID cards. Although first-generation Dominicans are more likely to be literate and have access to electricity than Haitian immigrants, they are less likely to have documentation. This creates problems with access to financial institutions and services since this documentation is required by many providers. The primary reason which led people to migrate to the Dominican Republic was to find work (45.2 percent).18 Thus, it is not surprising that, in comparison to the native Dominican population, immigrants are much more likely to be of working age. There are also differences in the types of work performed by Haitian and non-Haitian immigrants (see Figure 26). While the majority of Haitian immigrants work in agriculture or construction, the majority of immigrants from other countries work in business and other services. The contrasts between these two immigrant groups may be partially explained by differences in their levels of educational attainment (see Figure 27). While over three quarters of other immigrants have completed their secondary education with 46.9 percent having completed their tertiary education, only 21.6 percent of Haitian immigrants have completed secondary education and less than 6 percent have completed tertiary education. Though more likely than Haitian immigrants to complete primary school, first-generation Dominicans also lag behind on educational attainment, with only 24 percent finishing secondary school. Many immigrants do not plan to remain in the Dominican Republic but intend to return to their home country or to move on to another country. Less than half (47.5 percent) of immigrants plan on staying in the Dominican Republic for the next five years, and the percentage of Haitians planning on returning to their home country is 8.1 percent higher than immigrants from other countries. Figure 25: Access to services and documentation 2012 100% 80% 60% When Prosperity is not Shared 40% 20% 0% Haititan Immigrants Other Immigrants All Immigrants First Generation Water Inside House Have Health Insurance Have Passport Have National ID Have Birth Certi cate Access to Electricity Literacy Rate Source: ENI-2012 42 18  The second-largest reason for immigration (25.7 percent) was to improve quality of life Figure 26: Share of jobs by sector 2012 Non-Haitian immigrants Haitian immigrants 2,8% 0,1% 1,2% 10,4% 10,5% 0,3% 27,1% 0,8% 1,9% 3,8% 3,0% 35,7% 16,5% 1,5% 19,6% 9,1% 0,2% 3,8% 10,8% 12,8% 25,9% 0,1% Other Services Public and Defense Administration Finance Transportation and Storage Hotels and Restaurants Agriculture Buisiness Construction Utilities Manufacturing Mining Source: ENI-2012 Figure 27: Educational attainment 2012 (% completed) 100% 5.9% 11.1% 90% 5.8% 80% 21.6% 18.2% 46.9% 22.5% 70% 60% 50% 47.8% 40% 44.2% 59.9% 28.8% 30% 20% 10% 22.3% 19.1% 19.6% 0.9% 10.3% 0% Haitian Immigrants Other Immigrants All Immigrants First Generation Tertiary Secondary Primary None Source: ENI-2012 Figure 28: Decomposition of changes in poverty into intra- and inter-sectoral shifts 2000-2002 2002-2004 2004-2011 120% 120% 120% 100.0% 102,9% 102,9% 100% 100% 100% 80% 80% 80% When Prosperity is not Shared 60% 60% 60% 40% 40% 40% 20% 20% 20% 0,9% 0,9% 0% 0% 0% 0.0% 0.0% -3.8% -3.8% -20% -20% -20% Intra-sectoral inter-sectoral Interaction Intra-sectoral inter-sectoral Interaction Intra-sectoral inter-sectoral Interaction e ect e ect e ect e ect e ect e ect e ect e ect e ect Note: The decomposition presented in the table follows the methodology in Ravallion and Huppi, (1991) The decomposition is applied only to households with heads who are employed. The two sectors of work considered are formal and informal. Source: Study team’s own estimates based on ENFT 2000-2011 43 » Chapter 4 Escaping poverty and staying away from it: Income dynamics and the lack of upward economic mobility 4.1 Conceptualizing and measuring so cannot be used to investigate long-term trends. Even economic mobility when panels do exist, they usually have limited coverage and, hence, are not representative of the entire popula- E conomic mobility is a key element of economic tion of the country. Furthermore, issues of nonrandom development and equity. It represents the ability attrition and measurement error in panel data may pose of individuals, families or other groups of people additional methodological problems. To overcome the to improve their economic and social status—ei- shortage of proper panel data in the Dominican Republic, ther across the same individuals over time (intra-gener- the analysis of mobility carried out here applies an exten- ational) or across the same families across generations sion of the “small-area estimation” imputation methodol- (inter-generational). This study analyzes mobility within ogy used for poverty mapping (Elbers et al. 2002, 2003) to generations by measuring directional income movement, construct “synthetic panels” with repeated cross-sectional i.e., the extent of income growth as the net upward or data. Essentially, the method creates longitudinal data of downward movement in individual incomes over time. To households or individuals from cross-sectional data (two do so, the analysis compares the average of the growth or more rounds) by predicting income for the same units rates in individual incomes based on an initial and final of analysis in the future or in the past (Dang et al. 2011). vector of incomes, generating transition matrices into See Annex F for an overview of the methodology. The and out of poverty and other social groups such as the methodology applied for the Dominican Republic uses vulnerable, middle class and well off—results which can selected rounds of the ENFT for the 2000–2011 period. be used to assess the extent of mobility of individuals within a generation. Income thresholds are used to define three economic When Prosperity is not Shared groups for the analysis of economic mobility. They are Due to the absence of longitudinal data, the study of (1) the poor, those below the Dominican Republic’s mod- intra-generational economic mobility in the Domini- erate monetary poverty line of $4.70 per day 2005 pur- can Republic is based on the construction of “synthet- chasing power parity in urban areas and $4.20 in rural ic panels.” Ideally, analyses of economic mobility within areas, (2) the vulnerable, people with between $4.70 a day generations rely on panel data that follow individuals or and $9 in urban areas and $4.20 and $7.35 in rural areas, households over a long period of time. In reality, however, incomes levels that place them out of poverty but with long panels are rarely available, particularly in Latin Amer- a 10 percent or greater chance of falling back into it and 44 ica. Often panel data span just a short period of time and (3) the middle class, individuals who have daily incomes above $9 (urban) and $7.35 (rural) and an upper income come change was negative (-$1.16 PPP per day per head) threshold of $25. According to the ENFT labor force sur- between 2000 and 2011. This represents a fall in the medi- vey, less than one percent of individuals in the Domini- an income of 17.4 percent. In fact, the net income change can Republic have incomes above the middle class upper was positive for only two subgroups: those that escaped threshold. See Annex G for an overview of technical de- poverty and joined the vulnerable ($0.86 PPP increase per tails followed on the choice of the thresholds to define day per head) and the vulnerable that entered the middle the vulnerable and middle-class groups. class ($1.54 PPP per day per head). For all other groups in the income distribution, even in those whose economic 4.2 Assessing economic mobility in the Dominican status did not change, net income fell. Republic or is it economic immobility and economic insecurity? The results for the Dominican Republic starkly contrast with the stronger economic mobility observed in the 4.2.1 Overall long-term mobility Latin America and Caribbean region (LAC) as a whole over the past 15 years. On average, 41 percent of people Dominican Republic experienced very little upward in the region advanced to a higher group, compared to economic mobility during the 2000s. Table 10 presents 1.8 percent in the Dominican Republic. The difference is the transition matrix across classes (poor, vulnerable and also remarkable for mobility out of poverty. Almost half middle class) for the 2000–2011 period. The measure of of the poor abandoned that condition across the region intra-generational upward mobility is the share of the to- during the period, whereas, in the Dominican Republic, tal population that moved up across classes, namely, from only 3.2 percent climbed out of poverty. Similarly, over poor to vulnerable and from vulnerable to middle-class. 50 percent of people in the condition of vulnerability in Overall, the analysis reveals that there has been very nar- LAC joined the middle class while less than 3 percent did row upward economic mobility: between 2000 and 2011, in the Dominican Republic. And while the net income less than 2 percent of the population (1.1 percent of them change in the Dominican Republic in the 2000s was neg- in the poor group, 0.7 percent in the vulnerable) rose in ative, the incomes in LAC grew, on average, by $3.30 PPP economic status. per day per capita. In contrast, downward mobility was substantially high- Upward economic mobility in the Dominican Repub- er. Around 19 percent of the population in the Domini- lic has been low even in periods of robust economic can Republic shifted to a lower status (into poverty or out growth. The economy of the Dominican Republic con- of the middle class). Another particular finding is that a tracted in 2003 in the wake of a major domestic financial substantial part of the population (79 percent) was eco- crisis, then started recovering in 2005. In fact, between nomically immobile in either direction. Almost all of the 2005 and 2011, real GPD expanded annually on average Dominicans who were below the poverty line in 2000 and by 7.1 percent—nearly three percentage points higher two thirds of the Dominicans who were in the condition than the growth rate recorded for LAC. Despite this ex- of vulnerability remained in that condition in 2011. Of the traordinary performance, there was very little economic people who did not stay in vulnerability in 2011, only 2.2 mobility during this period. On average, 12.5 percent of percent moved to the middle class while the rest slid into Dominicans improved their economic status. But close to poverty (Table 10). 60 percent of total upward mobility was driven by tran- sitions from poverty to vulnerability. Indeed, basically all When Prosperity is not Shared Analyzing aggregate mobility, defined as the sum of of the people who managed to escape poverty between all income changes, confirms that there has been little 2004 and 2011 remained economically insecure, meaning improvement in economic status. Totaling the changes that another economic shock could send some of these in income (in levels or percentages) for the entire dis- climbers back down into poverty. Similarly, the pre-crisis tribution provides a sense of the magnitude of overall period, in which the economy of the Dominican Republic intra-generational mobility over the time period. Rein- grew at a more than satisfactory pace (a real annual av- forcing the findings that a greater proportion of people erage of 4.4 percent between 2000 and 2002), also had shifted down in economic status, Table 11 shows that to- low mobility (overall and upward), while the years of the tal mobility for the Dominican Republic in terms of net in- economic crisis (2003–2004) led to larger than expected 45 Table 10: Intra-generational mobility in the Dominican Republic–Percentage of population (2000–2011) Destination (2011) Poor Vulnerable Middle class Total Origin (2000) Poor 34.1 1.1 - 35.2 Vulnerable 10.3 20.6 0.7 31.6 Middle class 0.1 8.9 24.2 33.2 Total 44.5 30.6 24.9 100.0 Note: “Poor” means individuals with a per capita income per day lower than the official poverty line, $4.70 in urban areas and $4.20 in rural. “Vulnerable” means individuals with a per capita income per day between the official poverty line and the middle class lower threshold ($9 and $7.35 for urban and rural household respectively). “Middle class” means individuals with a per capita income per day higher than their corresponding middle class threshold and below $25. Poverty lines and incomes are expressed in 2005 $ purchasing power parity (PPP). The table shows lower-bound mobility estimates using the Dang et al. (2011) technique. Source: Study team’s own estimates based on ENFT 2000 and 2011 Table 11: Intra-generational mobility in the Dominican Republic, by median income change $ PPP per capita per day (2000–2011) Destination (2011) Poor Vulnerable Middle class Total Origin (2000) Poor -0.53 0.86 - -0.54 Vulnerable -1.28 -1.09 1.54 -0.98 Middle class -5.34 -2.57 -3.32 -2.84 Total -0.62 -1.42 -3.24 -1.16 Note: “Poor” means individuals with a per capita income per day lower than the official poverty line, $4.70 in urban areas and $4.20 in rural. “Vulnerable” means individuals with a per capita income per day between the official poverty line and the middle class lower threshold ($9 and $7.35 for urban and rural household respectively). “Middle class” means individuals with a per capita income per day higher than their corresponding middle class threshold and below $25. Poverty lines and incomes are expressed in 2005 $ purchasing power parity (PPP). The table shows lower-bound mobility estimates using the Dang et al. (2011) technique. Source: Study team’s own estimates based on ENFT 2000 and 2011 Table 12: Intra-generational mobility in the Dominican Republic, by percentage of median income change (2000–2011) Destination (2011) Poor Vulnerable Middle class Total Origin (2000) Poor -18.2 21.1 - -18.3 Vulnerable -24.6 -16.1 19.9 -16.0 Middle class -57.4 -25.9 -19.5 -19.1 Total -18.4 -19.1 -19.2 -17.4 Note: “Poor” means individuals with a per capita income per day lower than the official poverty line, $4.70 in urban areas and $4.20 in rural. “Vulnerable” means individuals with a per capita income per day between the official poverty line and the middle class lower threshold ($9 and $7.35 for urban and rural household respectively). “Middle class” means individuals with a per capita income per day higher than their corresponding middle class threshold and below $25. Poverty lines and incomes are expressed in 2005 $ purchasing power parity (PPP). The table shows lower-bound mobility estimates using the Dang et al. (2011) technique. Source: Study team’s own estimates based on ENFT 2000 and 2011 Table 13: Intra-generational mobility in LAC—Percentage of population (circa 1995–2010) Destination (2011) When Prosperity is not Shared Poor Vulnerable Middle class Total Origin (2000) Poor 22.5 21.0 2.2 45.7 Vulnerable 0.9 14.3 18.2 33.4 Middle class 0.1 0.5 20.3 20.9 Total 23.4 35.9 40.7 100.0 Note: “Poor” means individuals with a per capita income per day lower than the official poverty line, $4.70 in urban areas and $4.20 in rural. “Vulnerable” means individuals with a per capita income per day between the official poverty line and the middle class lower threshold ($9 and $7.35 for urban and rural household respectively). “Middle class” means individuals with a per capita income per day higher than their corresponding middle class threshold and below $25. Poverty lines and incomes are expressed in 2005 $ purchasing power parity (PPP). 46 The table shows lower-bound mobility estimates using the Dang et al. (2011) technique. Source: Study team’s own estimates based on ENFT 2000 and 2011 Figure 29: Sliders, climbers and stayers in economic status A. Dominican Republic vs. LAC B. Economic groups and GDP growth 90% 100% 7.1 8 78.9% 90% 80% 7 70% 80% 6 60% 57.1% 70% 60% 4.4 5 50% 41.4% 50% 4 40% 40% 3 30% 30% 19.3% 2 20% 20% 0.5 10% 10% 1 1.8% 1.5% 0% 0% 0 DR LAC 2000-02 2002-04 2004-11 Sliders Climbers Stayers Sliders Climbers Stayers Growth rate Note: The numbers in Panel A cover the period 2000-2011 for the Dominican Republic and circa 1995-2010 for LAC. Indicators of economic mobility (“sliders”, “climbers” and “stayers”) measured in percentages on the left-hand Y-axis in both panels. Average annual rate of GDP growth measured on the right-hand Y-axis in Panel B. Source: Study team’s own estimates based on ENFT 2000-11, World Bank (2013) with data from SEDLAC (Socio-Economic Database for Latin America and the Caribbean) and WDI. downward mobility, primarily driven by the worsening that returned in the second half of the 2000s helped par- economic status of almost one fourth of the population. tially reverse these trends. Between 2004 and 2011, over 5 percent of the population entered the middle class, Overall, the middle class did not expand in the Do- roughly 7 percent of the poor exited poverty, and less minican Republic over the past decade. Considering than 1 percent worsened their economic status. How- the narrow upward economic mobility in the Domini- ever, this level of upward mobility was still substantially can Republic, it is not surprising that the share of peo- lower than in LAC. While 20 percent of the population in ple in the middle class has remained essentially fixed the region (26 percent of the poor and vulnerable) joined over the long term. The mobility which did occur was the middle class between 1995 and 2010, only 5.4 per- largely in the wrong direction. About 21 percent of the cent of Dominicans entered this economic group in the population switched to another economic group, but of second half of the 2000s. those, only about two percent rose in economic status while the rest fell. Individuals who were initially vulner- Overall, compared to the region, there has been ex- able comprised the majority of the low inflow of people tremely high downward economic mobility in the Do- into the middle class. In fact, whereas the probability minican Republic. Estimates suggest that 19.3 percent of for a person who was poor in 2000 to exit poverty and Dominicans who were not poor in the baseline year 2011 enter the vulnerable group by 2011 was 3.1 percent, the entered poverty or moved into the vulnerable group probability that that same person would join the middle from the middle class by 2011. This puts the Dominican class was virtually zero. Republic at the top end of countries in the region in terms of downward intra-generational mobility: the Dominican The stagnation of the middle class is the result of oppo- Republic ranked third—after Bolivarian Republic of Ven- site mobility trends in the first and second half of the ezuela and Paraguay—among 18 countries ordered by decade. The periods of positive growth (2000-2002) and the magnitude of the population originally not poor that When Prosperity is not Shared economic contraction (2002-2004) translated together became poor during the period circa 1995-2010 (World into very little upward intra-generational mobility into Bank 2013). Furthermore, the probability of sliding from the middle class. Instead, a substantial share of the pop- vulnerability into poverty in the Dominican Republic is ulation worsened their economic status in both of these one of the highest in LAC, around 10.3 percent based periods. During the economic crisis, close to 25 percent on these lower bound estimates. Clearly this has impli- of the population suffered a fall in economic class from cations for the design of policies oriented to protect the their situation in 2000. Eleven percent had belonged welfare of households that have high risk of falling into to the middle class before the economic shock pushed poverty (for instance, social protection and risk manage- them down to vulnerability. Modest upward mobility ment strategies). 47 Figure 30: People joining the middle class: Dominican Republic vs. LAC 25% 50 20.4% 20% 40 Headcount (%) 15% 30 10% 20 5.4% 5% 10 1.0% 0.2% 0% 0 2000-02 (DR) 2002-04 (DR) 2004-11 (DR) 1999-2010 (LAC) 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Dominican Republic LAC Note: The initial year in each time interval corresponds to the baseline year for the mobility calculations. The indicator for LAC was calculated for 18 countries circa 1995-2010. Numbers are based on lower-bound mobility estimates using the Dang et al. (2011) technique. Results are weighted using country-speci c population estimates of the last available period. Source: Study team’s own estimates based on ENFT 2000-2011 and World Bank (2012) 4.2.2 Understanding upward mobility out of poverty and class threshold. Income growth among households that vulnerability exited poverty was not high enough to get them into the middle class. Instead, households that managed to enter The few poor people who escaped poverty tended to the middle class began with incomes above the poverty have incomes relatively close to the poverty line. An ex- line but still in the condition of vulnerability. With an ini- amination of the initial and final income distribution for tial median income of $7.80 PPP (Panel C), most of these those who exited poverty reveals that economic mobility households clustered just below the middle class thresh- in the Dominican Republic varied across different parts of old ($9.00 PPP for urban households and $7.35 PPP for ru- the income distribution. In Figure 31, Panel A shows that ral households) in 2000. Representing 2.2 percent of the the 3 percent of poor households that exited poverty be- initially vulnerable, these households managed to surpass tween 2000 and 2011 had baseline incomes just below the middle class threshold due to a 20 percent change in the poverty line. The median of this group among rural average net income among the vulnerable population households in 2000, for instance, was $4.10 per head per over the decade. However, these newly risen households day, at a time when the rural poverty line was at $4.20 PPP. still remained bunched close to the middle class line (Pan- By contrast, almost none of the households that Panel A el D), albeit on the other side, with a median income of shows emerging from poverty had incomes below the $9.30 PPP in 2011. Restricting the analysis to the post-cri- 2000 extreme poverty lines of $2.03 for rural and $2.12 for sis period (2004–2011) of strong and sustained economic urban. In other words, upward mobility out of extreme growth yields similar findings. poverty was virtually zero. Panel B shows that the poor families that abandoned poverty clustered right above Despite very little poverty reduction and income mo- the poverty line in 2011, with a median income of $4.90 bility in the Dominican Republic, incomes grew faster PPP per day, well below the middle-class thresholds ($4.70 among the poor than among households in upper parts PPP for urban and $4.20 PPP for rural households). This of the distribution. Using non-anonymous GICs (a plot of When Prosperity is not Shared indicates that people who escaped poverty only experi- income growth rates for the same households followed in enced a small upward shift in income, rather than a large a synthetic panel) to assess household welfare trajectories rise, and remained vulnerable to falling back into poverty. in the long-term provides a progressive picture. When in- These findings confirm that economic mobility was very comes fell during the 2002-2004 period, they did so rela- narrow both in terms of number of people who improved tively less among the poor than among other households, their economic status and the net income change. and when incomes rose (2004-2011), they did so relatively faster for the poor. The findings of the anonymous GICs Similarly, most of the people who joined the middle (a plot of income growth rates for specific parts of the 48 class had initial incomes only slightly below the middle income distribution) that are presented in section 3.5.1 Figure 31: Upward mobility out of poverty and vulnerability: Origin and destination, 2000–2011 Panel A: Initial income distribution: poor in 2000 who are nonpoor in 2011 Panel B: Final income distribution: poor in 2000 who are nonpoor in 2011 3.00 2.00 1.50 2.00 1.00 1.00 0.50 0.00 0.00 0 1 2 3 4 5 0 2 4 6 8 10 Panel C: Initial income distribution: Panel D: Final income distribution: poor and vulnerable in 2000 who entered middle class in 2011 poor and vulnerable in 2000 who entered middle class in 2011 1.50 3.00 1.00 2.00 0.50 1.00 0.00 0.00 0 2 4 6 8 10 0 2 4 6 8 10 12 Per capita income (2005us$ PPP / day - year 2000) Per capita income (2005us$ PPP / day - year 2000) Note: The gure shows estimates of lower-bound households using the Dang et al. (2011) technique. Panels A and B show the initial and nal income distributions of those originally poor who exited poverty. Panels C and D show the initial and nal income distributions of those originally poor or vulnerable who joined the middle class. Poverty and middle class lines in 2005 $ PPP per head per day as follows: urban extreme $2.10; rural extreme $2; urban moderate $4.70; rural moderate $4.20; urban middle class $9; rural middle class $7.35. Source: Study team’s own estimates based on ENFT 2000–2011. Figure 32: Anonymous and non-anonymous growth incidence curves, 2002–2004 and 2004–2011 Panel A Panel B -17% 17% -18% 16% -19% 15% 14% -20% When Prosperity is not Shared 13% -21% 12% -22% 11% -23% 10% -24% 9% 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 10 2002 - 2004 2004 - 2011 Note: Anonymous growth incidence curves (GICs) are shown in solid lines and non-anonymous GICc in dotted lines. For non-anonymous GICs, rst-round incomes are the actual incomes reported in the survey and second-round incomes correspond to lower-bound estimates using the Dang et al. (2011) technique. For anonymous GICs, rst- and second-round incomes are based on the cross-sectional nature of the ENFT. Left and right vertical lines show the proportion of poor and vulnerable people in 2004 (Panel A) and 2011 (Panel B). The X-axis shows deciles of income per 49 capita. Source: Study team’s own estimates based on ENFT 2000–2011 Figure 33: Initial characteristics and economic class in 2011 a. School attainment b. Proportion of formal 14 13.4 80% 73.4% 12 60% 10 9.7 50.9% 7.5 42.9% 8 40% 6 5.7 29.3% 4 20% Poor Vulnerable Middle Class Better o Poor Vulnerable Middle Class Better o c. Proportion of urban d. Male head of household 100% 84% 83.3% 87.0% 85% 80.2% 76.4% 80% 71.4% 78.7% 70% 75.7% 76% 55% 51.1% 40% 72% Poor Vulnerable Middle Class Better o Poor Vulnerable Middle Class Better o Note: Initial characteristics calculated in 2000 include: (a) portion of households headed by a male, (b) portion of household heads whose main job is in the formal sector, (c) portion of households residing in urban areas and (d) household head’s school attainment in years of education. Source: Study team’s own estimates based on ENFT 2000–2011 Figure 34: Initial characteristics and upward economic mobility A. Out of poverty B. Into middle class 8% 4% 7% 6,2% 6,3% 6% 3% 2,7% 5% 4,1% 4,0% 4% 3,5% 3,3% 2% 1,4% 1,5% 3% 1,3% When Prosperity is not Shared 2,3% 2,4% 2,2% 1,2% 2,0% 0,9% 0,9% 0,9% 2% 1% 0,8% 1% 0,4% 0,3% 0,0% 0% 0% Informal Informal Female Male Primary Secondary Tertiary Formal Rural Urban Urban Rural Female Male Primary Secondary Tertiary Formal Rural Urban Urban Rural Gender Education Sector Area Migrant Gender Education Sector Area Migrant Note: Initial characteristics calculated in 2000 include: (1) gender of the household head, (2) education level of the household head, (3) sector of work of the household head, (4) area of residence and (5) migration status. Source: Study team’s own estimates based on ENFT 2000–2011 50 (“The role of income growth and distribution”) show sim- resided in urban areas in 2000. Finally, there also seems ilar results. However, as noted before, the relatively bet- to be an association, although less strong, between the ter (or less bad) performance among the first four or five gender of the household head at the beginning of the deciles of the income distribution was not large enough period and the household’s economic status today, with to have a substantial contribution to poverty reduction households headed by males in 2000 less likely to be in and, more broadly speaking, upward economic mobility poverty in 2011. Analogous patterns across similar indi- within this group of individuals. cators have also been documented for many other coun- tries in LAC. 4.3 Initial conditions do matter for economic mobility Initial endowments, assets and economic opportunities are also associated with upward mobility. Highlighting Certain factors seem to be associated with the ability of the percentage of individuals who improved their eco- households to improve their economic status. One way nomic status grouped by initial characteristics, Figure 34 to understand the underlying dynamics of economic mo- offers some potential policy implications. The households bility in order to enhance policy design is to identify attri- most likely to have experienced upward economic mobil- butes that could be linked to more or less mobility. In the ity (either out of poverty or into the middle class) include absence of actual longitudinal data, the synthetic panel households initially: (1) headed by a male, (2) located in methodology limits the assessment to determining only an urban area—either permanently or having migrated whether the initial characteristics of households (i.e., in there from a rural area, (3) headed by an person with a 2000) are correlated with economic mobility.19 To do so, secondary or tertiary education, and/or (4) headed by a the analysis compares the 2000 characteristics across the person employed in the formal sector. four 2011 income-based groups (poor, vulnerable, mid- dle class and well off ) in terms of human capital (school attainment of household head), sector of work (formal employment of household head), area of residence (ur- ban or rural) and household structure (gender of house- hold head). Unsurprisingly, the results show that initial conditions do have implications for earnings in later periods. Households that are today in the middle or well off class- es have been headed by individuals with higher initial human capital (proxied by the average years of school- ing) compared to households in the current condition of poverty or vulnerability. Heads of households that be- longed to the middle class in 2011 had four years more of education in 2000 than heads of households that were poor in 2011 (Figure 33). Likewise, half of households from the middle class and three quarters from the well off in 2011 had heads who were employed in the formal When Prosperity is not Shared sector in 2000. This contrasts with the less than 30 per- cent of household heads initially employed in the for- mal sector among typical poor families in 2011. A similar trend emerges when analyzing location; current middle class and well off households were more likely to have 19  Panel data allows examining the extent to which changes in these characteristics, as opposed to initial levels, drive upward eco- nomic mobility. 51 » Chapter 5 Identifying chronic and transient poverty using a multidimensional approach 5.1 Non-monetary poverty: Improvement in key sured are constrained by the data available in the ENFTs, aspects of human welfare they were chosen to capture important objectives set out in the Millennium Development Goals and in some I ncome-based measures of human welfare are un- flagship national anti-poverty and social protection pro- likely to fully capture the multiple aspects of pover- grams. The seven indicators analyzed—all with identical ty. In addition to having low incomes, poor people of- weights—are listed in Table 14. ten face multiple deprivations such as low standards of living, lack of education (whether in quantity and qual- The proportion of Dominicans deprived has dropped ity), inadequate jobs, poor health, limited and inefficient systematically across most of the dimensions during risk-coping mechanisms, social exclusion and lack of voice. the 2000s. As Figure 35 shows, the largest improvements, Income- or consumption-based measurements alone can in relative terms, occurred in school attainment, access to fail to capture deprivations of core human wellbeing and sanitation, quality of housing and house overcrowding. functioning. Measures that account for the type and in- The indicators show that child school attendance, one tensity of non-monetary deprivations experienced by the of the most important dimensions, has one of the lowest poor are thus a valuable complement to assessments of rates of deprivation. Similarly, most houses are found to human welfare based on monetary poverty measures. have hard floors, which signals better quality of housing More importantly, multidimensional measures of poverty and possibly better health and cognitive development can inform policies intended to relieve these deprivations outcomes for children.20 But there are still high levels of as well as those intended to address poverty more broadly. deprivation in access to water and sanitation and asset ownership. For instance, as of 2011, 25.1 and 31.1 percent When Prosperity is not Shared This section presents findings from a multi-dimensional of the people lived in dwellings that are not connected analysis of poverty in the Dominican Republic for the pe- to safe water and sanitation services, respectively, while riod of 2000-2011. The exercise draws from the ENFTs that more than 25 percent of Dominicans own only one or were used for most of the analysis discussed so far. The none of the assets defined in the assets ownership indi- dimensions of poverty selected constitute deprivations cator (refrigerator, phone and clean-cooking fuel stove). in areas with which poor households in the Dominican Republic typically contend, such as education, access to 20  Evidence from Mexico suggests that replacing dirt floors with basic services, housing characteristics, living conditions cement floors significantly improves the health of young children, reducing parasitic infestations, diarrhea and anemia, and helps im- 52 and assets ownership. Even though the deprivations mea- prove their cognitive development (Cattaneo et al. 2009). Table 14: Selected indicators and deprivation criteria Indicator A household is considered deprived if: Child school attendance Any school-aged (7-17) child in the household is regularly out of school School attainment None of the household’s members has eight years of education or more Safe water The dwelling has no access to piped water provided by the general network of distribution, well or spring Sanitation The dwelling has no access to the general network or septic tank Quality of housing The dwelling has dirt floors House overcrowding The dwelling has three or more permanent members per room Asset ownership The household does not own at least two of the following: (1) refrigerator/freezer, (2) telephone/mobile, (3) clean cooking fuel stove (gas or electric cooker) Figure 35: Share of population deprived for each dimension. Total and by area (2000, 2005 and 2011) Total population 44,9% 39,7% 39,6% 33,8% 30,8% 29,3% 31,1% 28,6% 29,2% 26,9% 25,1% 20,1% 16,2% 13,5% 11,0% 8,0% 6,4% 6,0% 6,1% 3,3% 2,4% 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 School attainment Assets Children in School Floor Overcrowding Water Sanitation 79,4% Urban Rural 26,7% 26,3% 25,6% 67,6% 24,2% 63,7% 60,5% 21,7% 20,0% 52,0% 51,1% 50,4% 18,7% 48,8% 48,2% 16,9% 45,0% 16,2% 43,5% 16,0% 15,1% 14,1% 13,9% 12,0% 30,5% 10,2% 20,1% 16,1% 6,1% 13,7% 5,7% 12,5% 5,2% 11,4% 7,8% 7,5% 7,0% 2,1% 5,8% 1,3% 0,7% 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 2000 2005 2011 When Prosperity is not Shared Source: Study team’s own estimates based on ENFT 2000, 2005 and 2011 There are marked differences in the incidence of the ship and school attainment. While 43.5 and 60.5 percent deprivations across areas. In most of the indicators, the of people in rural areas are deprived of regular access to portion of people who are deprived fell both in urban and water and sanitation, the corresponding numbers in ur- rural areas throughout the 2000s. But rural households ban areas, aided by the general distribution network, are still exhibit higher rates of deprivation in all indicators 16.2 and 16.9 percent. Moreover, the deprivations in these than do urban households (Figure 35). The largest differ- four indicators have continuously been the highest of all ences are in access to water and sanitation, asset owner- the dimensions assessed. 53 Figure 36: Multidimensional poverty headcount for different values of k k=2 k=3 k=4 2000 45,5% 55,5% 2000 29,5% 70,5% 2000 16,2% 83,8% 2002 43,7% 56,3% 2002 29,6% 70,4% 2002 15,9% 84,1% 2004 41,9% 58,1% 2004 26,4% 73,6% 2004 14,2% 85,8% 2011 33,7% 66,3% 2011 19,8% 80,2% 2011 9,2% 90,8% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% Note: The dark blue part of the bar shows the proportion of people poor in a multi-dimensional sense. Source: Study team’s own estimates based on ENFT 2000, 2005 and 2011 The proportion of the population which is poor, as 5.2 The Dominican Republic has persistent chronic defined by the non-monetary, multidimensional ap- and increasing transient poverty proach, has fallen, particularly between the mid- and late 2000s. The intensity of non-monetary poverty—how The overlay of the monetary and non-monetary pover- many deprivations are experienced at the same time— ty data reveals interesting patterns as to how income can be measured by aggregating the different dimen- poverty and multidimensional poverty relate to each sions considered in the analysis using identical weights.21 other. On one hand, the two approaches could be cap- This results in a multidimensional headcount (MH) that turing similar trends. On the other hand, non-monetary shows the percentage of people that are deprived in k or and monetary measures may diverge for some groups more dimensions. The main conclusion of the analysis for of people, because they reflect different conceptions of the Dominican Republic is that the non-monetary pover- poverty. By combining the two measures, a more compre- ty rate, regardless of the number of k, steadily declined hensive vocabulary of poverty can be used to categorize during the 2000s. For instance, whereas 44.5 percent of the population into welfare groups, which is important in Dominicans were poor in two dimensions (k=2) in 2000, defining both what kinds of services are most needed and the headcount fell to33.7 percent in 2011; reductions of targeting these services to the correct people. 9.7 and 7.0 percentage points were also observed for the same period of time for poverty thresholds of k=3 and There is a strong association between the incidence of k=4, respectively. Adjusting the MH to account for the non-monetary poverty and monetary poverty in the Do- breadth of poverty confirms the declining trend, as the minican Republic. Not surprisingly, for each of the seven higher the value of k in the analysis, the larger the results indicators, the share of people who are deprived is sub- of poverty reduction.22 In other words, the Dominican stantially larger for those who are below the income pov- Republic has made the greatest progress in reducing erty line than for those above the line. For instance, close the most intense forms of multidimensional poverty. The to 50 percent of the income poor live in dwellings which When Prosperity is not Shared drop in the MH is evident in urban and rural areas, but it is lack access to sanitation services compared to a 20 percent much larger in the latter. rate among the non-poor. Differences of similar magni- tude are seen for other dimensions such as water access, asset ownership and house overcrowding and are found both in urban and rural areas. Nevertheless, the gap in the incidence of deprivations between the two groups (mon- 21  Identical weights, rather than empirical, are used to avoid is- etary and non-monetary poor) has narrowed over time, es- sues such as factor or discriminant analysis. pecially in the education-based indicators of child school 22  This adjusted measure takes into consideration the (weighted) 54 share of deprivations experienced by the poor. participation and educational attainment (Figure 37). 2000 53,4% 2000 21,5% 2000 30.2% 2005 37,2% 2005 18,5% 2005 24.0% 2011 30,1% 2011 12,5% 2011 17.5% 2000 38,6% 2000 10,7% 2000 18.2% 2005 34,4% 2005 12,0% 2005 18.6% 2011 34,3% 2011 16,2% 2011 21.3% 2000 9,1% 2000 4,9% 2000 6.1% (2000, 2005 and 2011) 2005 5,6% 2005 3,4% 2005 4.1% 2011 4,6% 2011 3,7% 2011 4.0% 2000 7,0% 2000 0,7% 2000 2.4% School attainment 2005 3,3% 2005 0,5% 2005 1.3% 2011 2,9% 2011 0,3% 2011 1.1% Non poor Non poor Non poor 2000 10,7% 2000 9,1% 2000 9.5% 2005 8,8% 2005 6,2% 2005 7.0% 2011 6,3% 2011 5,3% 2011 5.6% Assets 2000 44,1% 2000 11,4% 2000 20.3% 2005 36,9% 2005 12,0% 2005 19.3% 2011 35,5% 2011 10,1% 2011 17.3% Source: Study team’s own estimates based on ENFT 2000, 2005 and 2011 2000 68,8% 2000 18,6% 2000 32.2% 2005 52,4% 2005 12,8% 2005 24.4% 2011 47,0% 2011 8,8% 2011 19.6% RURAL URBAN Children in School TOTAL POPULATION 2000 75,1% 2000 43,5% 2000 59.9% Floor 2005 57,4% 2005 35,1% 2005 44.6% 2011 30,9% 2011 19,6% 2011 21.1% 2000 66,9% 2000 33,3% 2000 50.7% 2005 62,5% 2005 30,8% 2005 44.2% 2011 56,3% 2011 31,3% 2011 41.1% 2000 14,0% 2000 9,9% 2000 12.1% 2005 9,5% 2005 8,7% 2005 9.0% Overcrowding 2011 10,5% 2011 7,8% 2011 8.9% 2000 21,2% 2000 6,3% 2000 14.0% 2005 9,8% 2005 2,4% 2005 5.5% Poor Poor Poor 2011 8,8% 2011 1,4% 2011 4.3% 2000 30,5% 2000 30,0% 2000 30.3% Water 2005 21,6% 2005 19,8% 2005 20.6% 2011 19,0% 2011 18,8% 2011 18.9% 2000 58,9% 2000 21,8% 2000 41.1% Figure 37: Share of population deprived for each dimension by income poverty status, total and by area 2005 56,8% 2005 27,6% 2005 39.9% 2011 51,9% 2011 26,8% 2011 36.7% 2000 91,2% 2000 50,9% 2000 71.8% 2005 79,1% 2005 39,4% 2005 56.2% Sanitation 2011 75,0% 2011 30,8% 2011 48.1% 55 When Prosperity is not Shared The intensity of deprivation is also correlated with mon- The country has been experiencing a persistent prob- etary poverty. As just discussed, the income poor are lem of “chronic” poverty over the last decade, but has more likely to be deprived in any one of the dimensions made some progress. After a decade of strong economic than the non-poor. In addition, people with incomes be- growth, excluding the crisis period, 13.2 percent of the low the poverty line are more likely to suffer from multidi- poor were chronically poor in 2011, down from 17.6 per- mensional poverty—that is, two or more deprivations (k cent in 2000. Focusing on this group, the proportion of ≥ 2). For k=3, for example, one third of the income poor people in severe poverty (multi-dimensionally poor and are also multidimensionally poor, compared to 16 percent extreme poor in monetary sense) has remained virtually among the non-poor. The gaps also exist for higher values constant since 2000. At the beginning of the decade, 6 of k but have narrowed since the early 2000s. percent of Dominicans were unable to afford a basic food basket (i.e. they were below the extreme poverty line) and The intensity of non-monetary poverty did not increase were deprived in at least three core, non-monetary mea- during the 2002-04 economic crisis. As noted before, the sures of human welfare (i.e. k=3). As of 2011, this number economic crisis led to a substantial fall in family incomes had fallen by one percentage point. By way of compar- that resulted in a remarkable increase in monetary pov- ison, Brazil managed to reduce severe poverty between erty. In contrast, the non-monetary poverty headcount 1999 and 2011 from 7.1 of the population to 1.5 percent. did not go up even at the height of the crisis, irrespec- The main explanation for the persistence of chronic pov- tive of the value of k. Moreover, the whole decade saw a erty throughout the decade is the spike in severe poverty sustained fall in the MH, though the reduction was more during the 2003-2004 economic crisis (it rose from 6.0 to marked in the second half, a period in which the economy 9.2 percent). Comparing the 4.2 percentage point fall of enjoyed stronger economic growth and stability. severe poverty between 2004 and 2011, then, reveals that the Dominican Republic’s recovery from the crisis on this The combination of multidimensional and monetary front was substantial, but additional mechanisms need to poverty indicators allows for a more comprehensive be in place to prevent people from sliding back into this classification of people into poverty groups based on depth of poverty in the future. varying degrees of exposure to overlapping monetary and non-monetary deprivations. To accomplish this, the Notably, the percentage of Dominicans who are “tran- people who are multidimensionally poor are broken down siently poor” almost doubled. In addition to the per- into groups according to their income poverty status: ex- sistence of severe poverty, the proportion of people treme poor, moderate poor and non-poor. This produces who are classified as transiently poor in the Dominican four different sub-sets of the population. The chronic poor Republic increased from 15 percent to 29 percent in the refer to the multidimensional poor who are also poor in a 2000-2011 period (Figure 39). This too stems from the monetary sense. The fact that they are deprived in one or increase in the monetary poverty headcount during more of the human welfare dimensions could make them the economic crisis, suggesting that there is a group of less likely to depart the condition of poverty. Among the people who are not protected from economic shocks chronic poor, those in extreme poverty can be identified despite having access to services and/or assets. There as the severe poor given the intensity of the monetary is also a possibility that some of their productive assets, poverty.23 The multidimensional poor who are between for instance job skills, depreciate over time, raising fur- the extreme and moderate poverty lines are the moder- ther the risk that the transiently poor become chronical- ate poor. The not-income poor but deprived group is com- ly poor. Moreover, the slow reduction in transient pover- When Prosperity is not Shared prised of the multidimensional poor whose incomes are ty since the crisis (30.6 percent in 2004 to 29 percent in above the poverty line. The transient poor include people 2011) suggests that labor markets have taken unusually who are not deprived in any of the non-monetary dimen- long to stabilize and highlights the importance of this sions despite being income poor. Finally, people who are area for policy focus. not considered poor by either approach are referred as to the better off. In contrast, the proportion of people who are “mod- erately” poor and “not income-poor but deprived” 23  We use “severe” in order to make a distinction from the people has continuously fallen. Moderate poverty (chronical- who are “extreme” poor, a classification that throughout the section 56 is based on income. ly poor above the extreme poverty line and below the Figure 38: Multidimensional poverty headcount by income poverty status 2000 73,1% 60,7% 53,3% MPI K=2 36,4% 35,3% 31,9% 27,6% 2000 2002 2004 2011 2000 2002 2004 2011 Not poor Poor 57,5% 52,0% 41,5% 33,5% MPI K=3 22,7% 21,7% 20,0% 16,2% 2000 2002 2004 2011 2000 2002 2004 2011 Not poor Poor 36,2% 30,3% 23,0% MPI K=4 16,6% 11,5% 10,4% 10,7% 8,0% 2000 2002 2004 2011 2000 2002 2004 2011 Not poor Poor When Prosperity is not Shared Source: Study team’s own estimates based on ENFT 2000, 2002, 2004 and 2011 moderate poverty line) decreased from 11.6 percent in off) decreased between 2000 and 2011 from 55.5 per- 2000 to 8.2 percent in 2011. Similarly, the percentage cent to 51.2 percent. The decline of the better off corrob- of the population who are not-income poor but deprived orates the results of the almost null economic mobility dropped significantly from 11.9 percent to 6.6 percent in the Dominican Republic that was discussed previous- over the same period of time. In contrast, the propor- ly (Figure 39) but hides the progress made during the tion of people who are neither income poor nor de- recovery of an 8.2 percentage point growth of the mid- prived in any of the non-monetary dimensions (better dle class since 2004. 57 Figure 39: Matrix of multidimensional and moderate income poverty, 2000–2011 (selected years) 2000 2002 Not poor but deprived Better o Not poor but deprived Better o 11.9% 55.5% 12% 54.8% 3,238 3,238 Income poor Income poor Chronic poor Transiently poor Chronic poor Transiently poor 1,458 1,458 17.6% 15.0% 17.6% 15.6% 8 7 6 5 4 3 2 1 0 8 7 6 5 4 3 2 1 0 Multi-dimensionally poor Multi-dimensionally poor Extreme poverty line, incomes below: 6.0% of the pop. Extreme poverty line, incomes below: 6.7% of the pop. 2004 2011 Not poor but deprived Better o Not poor but deprived Better o 7.2% 43.0% 6.6% 51.2% 3,238 3,238 Income poor Income poor Chronic poor Transiently poor Chronic poor Transiently poor 1,458 1,458 19,2% 30.6% 13.2% 29.0% 8 7 6 5 4 3 2 1 0 8 7 6 5 4 3 2 1 0 Multi-dimensionally poor Multi-dimensionally poor Extreme poverty line, incomes below: 9.2% of the pop. Extreme poverty line, incomes below: 5.0% of the pop. Note: Dotted line shows the monetary extreme poverty line. Severe poor (i.e. people below the extreme poverty line) and MPI poor in 2000 = 6.0 percentage points, in 2002 = 6.7 percentage points, in 2004 = 9.2 percentage points and in 2011 = 5.0 percentage points. X-axis measures the number of household deprivations; Y-axis is household per capita income (Dominican pesos). Source: Study team’s own estimates based on ENFT 2000, 2002, 2004, and 2011 Figure 40: Chronic and transient poverty by household characteristics, 2011 A. Chronic Poverty Rates B. Transient Poverty Rates 30% 40% 34% 25% 24% 35% 31% 31% 31% 30% 29% 19% 26% 20% 19% 24% 25% 22% 15% 14% 20% 12% 15% 15% When Prosperity is not Shared 10% 8% 6% 7% 10% 5% 1% 5% 0% 0% Rural Rural Informal Informal Male Female None or primary Secondary Tertiary Formal Urban Male Female None or primary Secondary Tertiary Formal Urban Gender of School attainment Sector of work Area Gender of School attainment Sector of work Area household head of household head household head of household head 58 Source: Study team’s own estimates based on ENFT 2000, 2002, 2004, and 2011 Figure 41: Matrix of multidimensional and extreme income poverty, the Dominican Republic and LAC (2003 and 2011) DR 2003 DR 2011 Not poor but deprived Better o Not poor but deprived Better o 9% 69% 9% 77% $2.5 $2.5 Income poor Income poor Chronic poor Transiently poor Chronic poor Transiently poor 7% 15% 4% 10% 7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0 Multi-dimensionally poor Multi-dimensionally poor LAC 2003 LAC 2011 Not poor but deprived Better o Not poor but deprived Better o 5% 72% 4% 83% $2.5 $2.5 Income poor Income poor Chronic poor Transiently poor Chronic poor Transiently poor 6% 16% 3% 10% 7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0 Multi-dimensionally poor Multi-dimensionally poor Note: International extreme poverty line of $2.50 is used. X-axis measures the number of household deprivations; Y-axis is household per capita income (R$). Source: Study team’s own estimates based on SEDLAC data Factors related to human capital, household structure, families headed by women are more likely to be transient- type of job and area of residence raise the probability ly poor (Figure 40, Panel B). of being chronic or transiently poor. For instance, house- holds headed by individuals with little or no education, In terms of extreme income and multidimensional pov- an informal job and living in a rural area are much more erty, the Dominican Republic performed similarly to LAC. likely to be in a state of chronic poverty in a monetary Using the international poverty line of $2.50 per day to and non-monetary sense. In addition to the notable ef- compare with LAC reveals that chronic poverty decreased fect of school attainment on reducing the chances of be- by about one half and the transiently poor decreased by ing chronically poor, the area of residence also plays an about one third from 2003 to 2011 in both the Domini- important role. Without conditioning on anything else, can Republic and LAC. By contrast, the not poor but de- When Prosperity is not Shared a household in a rural area is three times more likely to prived group did not change in the Dominican Republic be chronically poor than an urban household (Figure 40, and only decreased by one percentage point in LAC. In Panel A). These differences are less marked in the case of fact, the Dominican Republic in 2011the Dominican Re- the transient poor. Remarkably, households whose heads public exhibited similar poverty rates to LAC in all of the have no or little education have similar chances of being groups except for the not poor but deprived, which was income poor but not multi-dimensionally poor, signaling about double the LAC average in the Dominican Republic an inability of individuals to convert higher human capital (9 percent compared to 4 percent). This is also reflected in into more earnings and higher standards of living. In con- the lower share in the better off group in the Dominican trast to the profile of households that are chronically poor, Republic (77 percent) compared to LAC (83 percent). 59 » Chapter 6 Equality of opportunities and life chances 6.1 Measuring opportunities for all HOI takes into account both the total amount of goods and services available to the population and how fairly T he notion of fairness in development opportu- they are distributed (see Annex H with more details on nities underpins assessments of the equitable the methodology).25 Furthermore, the HOI can provide availability of key goods and services that are insight into inter-generational mobility by analyzing the necessary to progress in life. A vast amount of em- influence of the circumstances which have direct rela- pirical literature has shown that the chances of enjoying a tionships with mobility across generations (e.g., parental fully productive life are largely influenced by access to— education and income) on inequality of opportunities. and quality of—basic goods and services such as food, The index can be improved by changing people’s circum- school enrollment, water, sanitation and electricity, as well stances (“composition effect”), raising coverage to all peo- by the ownership of a minimum set of key assets. These ple (“scale effect”), or distributing the coverage in a more key goods and services are denoted here as opportunities. equitable manner (“equalization effect”). The concept of fairness followed in this report refers to the idea that personal circumstances that are out of the control This report calculated the HOI for the Dominican Re- of individuals (particularly children), such as birthplace, public to see the evolution of the equitable availability gender, race, wealth, parental education and area of res- of opportunities during the last decade. To do so, it esti- idence, should not determine their human opportunities. mated the HOI at three points in time: 2000 (“pre-crisis”), For example, all children should enjoy access to safe water 2004 (“crisis”) and 2011 (“post-crisis”). The opportunities whether they live in an urban or rural location. which the analysis considered consist of (1) school enroll- ment; (2) timely completion of the sixth grade (a proxy When Prosperity is not Shared The Human Opportunity Index (HOI) measures both to quality of education); access to (3) safe water and (4) coverage and equity in access to key goods and services sanitation; (5) quality of housing, namely whether the that enable people to expand their productive poten- dwelling has a hard floor; and (6) family ownership of at tial. The HOI measures the coverage rate of opportunities least two out of three key assets.26 To assess inequality of and adjusts or “penalizes” this measurement according to opportunities, the analysis focused on children, defined how equitably the opportunities are distributed among different subgroups based on circumstances. 24 Thus the 25  The index does not capture, however, differences in the quality of goods and services. Quality itself could hinder equity despite in- creases in equitable access to opportunities. 24  The HOI runs from 0 to 100, with 100 denoting universal cov- 26  Refrigerator/freezer, telephone/mobile, clean cooking fuel 60 erage. stove (gas or electric cooker) as individuals between 0 and 16 years old, as they have to universalize the opportunities included in the HOI. This less control over their current situations, i.e., their cir- is similar to the amount of time it would take for Central cumstances are less likely to have been determined by America (36 years), while the LAC Region is projected to effort and choice. The circumstances considered consist achieve this objective earlier, in approximately 24 years of: (1) parental education; (2) family income; (3) gender (Table 15). The HOI grew at a slower pace than expected of household head; (4) number of siblings; (5) presence of considering the strong economic growth over the past both parents in the household; and (6) area of residence. decade, especially in comparison to other countries in the region. For example, Brazil had an average annual 6.2 Have human opportunities for children GDP growth rate of 3.2 percent from 1995 to 2010, during improved? which time the HOI increased by 18.7 points to reach 75.7 in 2010. The Dominican Republic, meanwhile, grew its GDP Opportunities for children in the Dominican Republic by an annual average of 5.8 percent over the same time have expanded in the last decade. As noted previously, period, yet the HOI only increased by 8.9, falling behind access to basic services and goods and asset ownership Brazil at 72.7 in 2010. Similarly, Mexico, Ecuador, Peru, Co- in the Dominican Republic have increased over time, in- lombia, Nicaragua, Costa Rica and Paraguay all increased cluding among the poor. The HOI also indicates that there the HOI by more than the Dominican Republic from 1995 was sustained improvement in opportunities for children to 2010 yet had lower average annual growth rates. during the 2000–2011 period, with the index increasing from 65 in 2000 to 69 in 2004 and 73 in 2011. A decrease The distribution of human opportunities varies across from 7.2 to 5.6 of the penalty for unfair distribution con- the different categories considered. On one hand, as tributed slightly to the rise of the index over the period. Table 16 shows, opportunities for children to access ed- As for comparison with the region, the HOI in the Domin- ucation services, i.e., enrollment, and to live in a house ican Republic in 2011 is identical to the average HOI of a of decent quality are close to universal, at indices of 96 group of 19 countries in the region across similar oppor- and 95 in 2011, respectively. On the other hand, the Do- tunities and circumstances (World Bank 2012) (Table 16). minican Republic has been less successful in providing children with equitable access to key services (water and The rate of expansion, however, has been slow—at cur- sanitation) and goods linked with better quality of life rent rates it would take more than a generation to level (refrigerator, telephone and clean cooking fuel stove), as the playing field for children. The equity-adjusted provi- well as the opportunity to progress in school on time. For sion of human opportunities captured by the index grew instance, although there has been substantial progress in on average by about 1 percent per year between 2000 raising coverage in sanitation (its HOI increased from 38 and 2011. At this pace, and assuming a linear expansion, to 56), that coverage remains low and unevenly allocated, it would take the Dominican Republic just under 30 years while the HOI for access to safe water, remaining low at Table 15: HOI, coverage rate and penalties in Dominican Republic, 2000–2011 (selected years) HOI HOI HOI Annual Coverage rate Coverage rate Penalty Penalty (2000) (2004) (2011) change (2000) (2011) (2000) (2011) 65 69 73 1.09 72 79 7.2 5.6 Note: values are expressed in percentages and denote coverage rates adjusted by how equitably they are distributed. Source: Study team’s own estimates based on ENFT 2000, 2002, 2004 and 2011 When Prosperity is not Shared Table 16: HOI, coverage rate and penalties in Dominican Republic, 2000–2011 (selected years) Year Education Safe water and sanitation Housing Assets Enrollment 6th grade on time Water Sanitation House with hard floor Asset ownership 2000 97 43 61 37 87 62 2004 98 56 64 44 92 59 2011 96 68 61 55 95 62 Source: Study team’s own estimates based on ENFT 2000, 2002, 2004 and 2011 61 61, has shown no improvement since 2000. Of note, the proportional increase in coverage across all circumstance increase in the availability and equitable distribution of groups, and (2) the equalization effect, at the heart of the opportunities was not interrupted during the 2003-2004 concept of equality of opportunities, which captures economic crisis, with the exception of asset ownership, relative changes in coverage that are specific to circum- which fell and has shown no progress over the long-term. stance groups with below-average coverage rates. Esti- mates for both the Dominican Republic and the LAC re- 6.3 Unpacking the changes in human opportunities gion show that around 30 percent of the overall change in the coverage effect of the HOI is due to the equaliza- Most of the expansion of human opportunities achieved tion effect, namely increased equality of opportunity. In in the Dominican Republic is explained by increased the three dimensions presented in Table 18, the scale coverage across the circumstance groups. Changes in effect dominates the equalization effect, results that are the HOI can be decomposed into (1) changes in the cover- consistent with the evidence for LAC. These results sug- age rates of specific circumstance groups (e.g., an increase gest that one policy method of increasing equitable allo- in school enrollment for children living in rural areas), also cation of opportunities going forward could be to target called the coverage effect, and (2) changes in the share of expansion in the provision of opportunities to the most the population among specific circumstance groups (e.g., disadvantaged groups. families migrating from rural to urban areas), called the composition effect. The analysis for the Dominican Repub- 6.4 “Circumstances” constraining equality lic shows that the coverage effect explains nearly three of opportunity and economic mobility across quarters of the improvement in the HOI between 2000 generations and 2011. In other words, the increase in the probability that a child from, say, a rural family with low income goes Distributions of human opportunities across sub-groups to school is the type of change that has chiefly driven the based on area of residence (urban or rural), parental ed- improvement in the HOI, rather than there simply being ucation, family income and the gender of the child are fewer children living in rural areas due to urban migration the most inequitable. For policy purposes, it is important (World Bank 2012) (Table 17). The decomposition for the to identify the contribution of each circumstance to the LAC region delivers similar results. overall inequality of opportunity. The analysis shows that, among the set of circumstances considered for the calcu- The relatively larger expansion of access to opportuni- lation of the HOI, the place where children live, the level of ties for the vulnerable circumstance groups accounted education of their parents and the income of their house- for one third of the total increase in coverage of key hold are the main factors affecting equality of opportunity services. The coverage effect can be further decomposed for access to basic goods and services. Differences in ac- into two effects: (1) the scale effect, which measures the cess to key public goods and services between groups de- Table 17: Contribution of the “composition” and “coverage” effects to changes in the HOI in Dominican Republic 2000–2008 Finished 6th grade on time Access to safe water Access to sanitation Total change “Comp” Effect “Cov” Effect Total change “Comp” Effect “Cov” Effect Total change “Comp” Effect “Cov” Effect 1.9 0.4 1.5 0.9 0.3 0.6 1.4 0.2 1.2 Note: “Comp” denotes the composition effect and “Cov” the coverage effect. Decompositions are calculated for the 2000–2008 period and presented in percentage points. When Prosperity is not Shared Source: World Bank (2012) Table 18: Contribution of the “scale” and “equalization” effects to the “coverage” effect in Dominican Republic 2000–2008 Finished 6th grade on time Access to safe water Access to sanitation Total coverage Scale effect “Equal” effect Total coverage Scale effect “Equal” effect Total coverage Scale effect “Equal” effect effect effect effect 1.5 1.1 0.4 0.6 0.4 0.2 1.3 0.9 0.4 62 Note: “Equal” denotes the equalization effect. Decompositions are calculated for the 2000–2008 period and presented in percentage points. Source: World Bank (2012) Figure 42: Contribution of each circumstance to inequality of opportunity, 2000 and 2011 2000 2011 35% 30% 30% 26% 26% 30% 28% 25% 26% 25% 20% 20% 20% 15% 15% 15% 10% 7% 10% 6% 5% 4% 3% 2% 5% 3% 2% 0% 0% Area Area Family income Gender of Family income Gender of household head Gender of child Number of sibling Both parents household head household head Gender of child Number of sibling Both parents household head Education of Education of Source: Study team’s own estimates based on ENFT 2000 and 2011 Table 19: Contribution of each circumstance to inequality of opportunity in Dominican Republic, 2011 Education Safe water and sanitation Housing Assets Circumstance Enrollment 6th grade on time Water Sanitation House with hard floor Asset ownership Head’s education 14 23 29 26 22 41 Family income 4 26 32 21 28 11 Number of siblings 2 5 9 6 6 11 Child gender 54 1 1 1 1 31 Both parents 10 3 1 3 3 0 Head’s gender 1 3 2 4 4 0 Note: Values are in percentage points. Source: Study team’s own estimates based on ENFT 2000, 2002, 2004 and 2011 Figure 43: Impact of parental background on children’s education gap at age 15 in LAC, 1995–2009 0.8 0.6 0.4 0.2 -0.0 -0.2 -0.4 -0.6 When Prosperity is not Shared -0.8 -1.0 -1.2 Brazil El Salvador Chile Costa Rica Honduras Nicaragua Ecuador Colombia Mexico Panama Dom. Rep. Argentina Peru Paraguay Venezuela Uruguay Bolivia C.1995 C.2009 C.2009 - C.1995 Note: “Educational gap” is de ned as the di erence between potential years of education at age 15 and the years of completed education circa 1995 and circa 2009. The blue and red bars represent the expected reduction in the schooling gap associated with one standard deviation in parental education in 1995 and 2009, respectively. The green bar is the di erence between the two. Other covariates in the regression are children’s gender, residence in an urban area and country xed e ects. The estimated e ect of parental education on the educational gap is always statistically signi cantly di erent from zero as are the di erences between 1995 and 1999. Source: World Bank using data from DESLAC (Socioeconomic Database for Latin America and the Caribbean) 63 Box 3: Teenage pregnancy as a proxy of agency in the Dominican Republic Agency, or the freedom to pursue processes by which individuals achieve socioeconomic advancement, is a fundamental aspect of equity (Cord and Lopez-Calva 2013). Agency entails that people have the social, economic and political means to pursue life options that they have reasons to value. In other words, agency is the ability of people, regardless of background, to make decisions and convert them into desired actions and outcomes. The constitutive elements of agency are: (1) individual aspects, including objective conditions (e.g. good health or lack of it) and subjective elements (aspirations and self-drive); (2) contextual elements, such as social norms, culture and formal institutions; and (3) power, defined as the capacity to align the actions of others to one’s own interest (Cord and Lopez-Calva 2013). Agency involves many aspects of life, whether it is ability to rise in the workplace, have political voice, or take control of personal health and family decisions. One possible manifestation of lack of agency is teenage pregnancy. The adolescent fertility rate (AFR) is defined as the number of births per 1,000 women aged 15-19. In the LAC region, the AFR decreased between 2000 and 2010 for all countries considered. Yet, the numbers for the Dominican Republic are not encouraging—the country had the second-highest AFR (after Nicaragua) among 21 countries. Indeed, during this period, the Dominican Republic also had the lowest reduction of AFR, only -4 percent. By contrast, Colombia achieved the highest reduction of AFR in the region,-25 percent.27 The adolescent fertility rate for the Dominican Republic is remarkably higher than in otherwise comparable countries. For instance, countries with similar GDP per capita such as Ecuador, Peru and Jamaica have AFRs which are two thirds lower than the Dominican Republic’s. In the case of countries with similar extreme poverty rates ($2.50 a day poverty line), the AFR of the Dominican Republic is again quite higher than theirs. For example, the extreme poverty rates in 2009 for Brazil and the Dominican Republic were 15.1 percent and 16.4 percent, respectively. However, the AFR for Brazil was 75.8 whereas it was 105.65 for the Dominican Republic, almost 40 percent higher than the Dominican Republic. Finally, using the Gini coefficient as a point of reference, the Dominican Republic is once again way higher than its counterparts such as Ecuador and Peru on the AFR. However, the Dominican Republic is not alone in this regard—Nicaragua and Guatemala, countries with high AFRs as well, are also quite far from their own comparable countries in terms of inequality. Figure 44: Correlates of teenage pregnancy 120 120 120 nic dom dom nic nic dom gtm gtm gtm 100 100 100 Adolescent fertility rate hnd ven ven hnd ven hnd ecu pan pan ecu slv ecu pan 80 bol bra 80 bra slv bol jam 80 bol jam slv mex jam col ury cri mexcol pry mex pry bra col pry ury ury cri 60 cri 60 chl 60 chl chlarg arg arg per per per 40 40 40 0 5000 10000 15000 0 10 20 30 40 40 45 50 55 60 GDP per capita - PPP 2005 prices Poverty rate usd 2.5 Gini coe cient Source: World Bank (2012), authors’ calculations. Years: Argentina (2010-11), Bolivia (2007), Chile (2009), Colombia (2007), Dominican Republic (2009), Ecuador (2009), El Salvador (2008), Guatemala (2006), Honduras (2009), Jamaica (2002), Mexico (2008), Nicaragua (2005), Panama (2009), Paraguay (2009), Peru (2009), Uruguay (2008), Venezuela (2006) When Prosperity is not Shared fined on the basis of these three circumstances accounted rental education and family income matter the most for for 87 percent of the inequality27in opportunities in 2011, whether or not a child lives in a household which owns almost identical to their share in 2000. In isolation, paren- assets and has access to safe water and sanitation (Table tal education and family income each explain a quarter of 19). In sum, limited human capital, the parents’ inability to the uneven distribution of opportunities measured in the earn higher income and the gender of the child are key el- HOI. A breakdown across opportunities indicates that pa- ements that appear to limit the child’s opportunities, thus serving as circumstances which restrain social inter-gener- 64 27  The World Bank (2012). Using WDI indicators ational mobility. Other analyses confirm that parental background con- tinues to be a strong determinant of the human capital of children and thus of their ability to progress in life. Correlations in school attainment across generations in- dicate that LAC is the least educationally mobile region in the world (Hertz et al. 2007). The high persistence of educational attainment is also evident in the Dominican Republic. The association between parental background and the educational gap of children at age 15 (the differ- ence between potential years of education at that age and the actual years of education completed) shows that the gap is larger among children raised in households with lower parental background, be it fewer years of edu- cation or lower income. In 2009, an increase of a standard deviation in parental education among households in the Dominican Republic reduces the schooling gap by nearly 0.74 years of education, to bring it close to the LAC region average. Even though the impact on children of extra pa- rental education was higher in 1995 (0.87 year of educa- tion), suggesting that educational persistence has fallen, the reduction was small. Another factor that further limits equality of opportunity and mobility across generations is the fact that not only do children of more educated par- ents complete more years of education, there are also sub- stantial differences in the quality of schooling that they receive. In fact, analyses, controlling for potential biases, of the relationship between parental background and the performance of students in standardized test scores show that children who have parents with tertiary education do substantially better than children with parents who attained only primary education (World Bank 2013). When Prosperity is not Shared 65 » Chapter 7 Policy options to strengthen the links between economic growth and equity S etting out a policy agenda aimed at ensuring skill set. Finally, risk management can reduce the expo- that the gains from economic growth and pros- sure to and impact of shocks for the poor and vulnerable perity are more evenly distributed requires an which, otherwise, could force them to engage in negative understanding of the links between growth and coping mechanisms. equity. Equity and growth as a self-reinforcing cycle con- stitutes shared prosperity (see diagram below). When This framework is useful to understand the challenges all members of society have the opportunity and skills in the Dominican Republic, as well as the entry points to generate income, then the society is more equitable. for effective policy interventions. This report has shown At the same time, with more people contributing to the that despite economic growth the society remains high- economy, the economy grows and everyone, both poor ly unequal. Though GDP per capita increased by over 50 and non-poor, is comes out ahead. In this process, the vir- percent over the decade, the nature of growth has not tuous cycle that enhances societal welfare, equity entails been inclusive. For instance, the fastest growing sectors freedom from absolute poverty, equality of opportuni- have not significantly increased employment. Instead, ties regardless of original circumstances, and the ability weak linkages in the market have caused employment to to make autonomous decisions regarding important life increase most in low productivity sectors, even for workers choices. Within the framework of shared prosperity, there with high skills, and real earnings to fall across skill groups are four main channels through which growth and equity and sectors. Furthermore, not enough Dominicans have reinforce each other: 1) equitable, efficient and sustain- quality education and job training and many still lack ac- able fiscal policy, 2) fair, transparent institutions and ef- cess to even basic services diminishing the potential of the fective provision of public goods, 3) well-functioning and Dominican Republic’s pool of human capital. These factors When Prosperity is not Shared accessible markets, and 4) comprehensive and efficient have made it difficult to connect people to economic risk management. A number of policy instruments can be growth and improve equity resulting in limited mobili- used to strengthen these channels. Actions on improving ty and inequality of opportunities. While the Dominican fiscal policy and the effectiveness of institutions will, for Republic is improving systems to protect vulnerable pop- instance, facilitate equity by increasing the opportunity ulations from risks to prevent such dramatic increases in set and economic mobility of the poor and vulnerable poverty as seen in the 2003-2004 crisis, social protection and other disadvantaged groups. Well-functioning mar- programs, by 2011, were still leaving over 40 percent of kets facilitate the efficient use of the skills and assets of the population without adequate means to live free from 66 the most vulnerable and allow the poor to maximize their poverty. Strikingly, the number of people endowed with The Virtuous Circle of Shared Prosperity I. Fiscal policy II. Institutions & stability & public goods Fairness Growth SHARED Equity PROSPERITY Equity Triangle Freedom Process from absolute freedom III. Well-functioning IV. Risk poverty markets management Source: SEDLAC and The World Bank. human capital and assets but who are unable to escape and quality public goods and services and places too poverty has almost doubled over the decade. much of the burden on the poor. In particular, Inequalities also reflect a lack of responsiveness of the • The fiscal system is hampered by low revenue col- institutional system to the needs of specific circum- lection. While the average tax burden in LAC is 20 stance groups. As discussed in other studies, inequality percent of GDP, public revenues in the Dominican Re- in service provision in the Dominican Republic is also public averaged 13.7 percent of GDP over the past de- associated to lack of empowerment for certain groups cade, only slightly higher than the level of tax revenue and illustrates weaknesses of the political system (UNDP, of Guatemala, the lowest in the region. Unfortunately, 2008). The heterogeneity of influence and voice results the tax reform measure that was passed in November in an unequal distribution of resources –this is so even in 2012 missed the opportunity to address the low tax well-functioning democratic systems, though it might be burden—it is expected to raise fiscal revenues by only exacerbated by capture and corruption (Esteban and Ray 1.4 percent of GDP. 2006). Strengthening the accountability, transparency and responsiveness of the system to all groups in society • The tax system relies heavily on indirect taxes, limit- will also feed into a positive cycle towards a more cohe- ing progressivity. Nearly one third of total tax collec- sive social contract. tion comes from value-added taxes (VAT), on which income has no bearing, though basic goods from the The fundamentals of equity, clearly weaker in the Domin- consumption basket are exempted. Furthermore, de- ican Republic, can be strengthened by focusing on the spite the exemptions of basic goods, about 50 per- policy areas to be discussed below, thus tying growth to cent of tax exemptions and incentives benefit the equity and generating a positive cycle of shared prosperity. more affluent members of society. Past tax reforms When Prosperity is not Shared have not raised the progressivity of the tax system, Policy area 1 particularly by failing to impose more direct taxes. Micro-simulations of the 2012 tax reform indicate Objective: Promote equitable, efficient and sustainable that, as expected, the increase in indirect taxes (for fiscal policy. instance, an increase in the VAT rate) will be regres- sive (Valderrama et al. 2013).28 Problem to address: The current structure of fiscal poli- cy, both on the revenue and expenditure sides, limits the 28  These micro-simulations do not take into account increases in property tax or tax on motor vehicles, both of which would likely ability of the Dominican Republic to provide sufficient make the results slightly more progressive. 67 • Budget rigidities limit the ability to increase and increasing the capacity of the Tax Administration redirect allocations to key social sectors so as to Authority to conduct tax audits and enforce control. provide more and better public goods and services, Tax evasion is likely to make actual taxation less pro- including systems to protect the poor and vulnera- gressive. ble from negative shocks. The already limited fiscal space is further constricted by structural factors on On the expenditure side, more public resources should the expenditure side, notably use of public resources be allocated to education, health, water and sanitation, to fund a large electricity deficit. In 2008, that deficit social protection and risk management systems, with alone accounted for 2.7 percent of GDP. The figure has steps taken to maximize the efficiency of resources. since decreased, but remained substantial at 1.8 per- Specific policy actions to achieve this include: cent of GDP in 2012. All in all, the Dominican Republic remains well below LAC averages in social allocations. • Investing more resources in water and sanitation In 2011, the country channeled just 2.3 percent of GDP and quality education and health to endow the poor in public funds towards education, one of the lowest with the skills and assets required to take full advan- levels in the LAC region. Steps in the right direction tage of their productive capacity and live better lives; have been taken to raise education spending to 4 percent of GDP in 2013 but capacity constraints may • Strengthening current social assistance programs undermine the quality and efficiency of that spend- and disaster risk management systems, including ing. Health expenditures have also increased in recent safety nets to protect the poor and vulnerable from the years, but as of 2011 they still amounted to just half effects of major shocks. This will ultimately lessen the the LAC average. More than a third of the Dominican impact of shocks on the overall economy and acceler- Republic’s people have no access to clean water; near- ate recovery to combat the asymmetric response of ly half have no access to sanitation services. Despite a poverty in the Dominican Republic to business cycles; substantial increase in resources for social assistance, almost half of these expenditures are still not allocat- • Analyzing the differences between monetary and ed on the basis of formal targeting criteria. Targeting non-monetary deprivations to better distinguish has become especially important because the eco- the chronic from the transient poor and to design nomic shock of 2003-2004 caused some groups to slip and target strategies and programs that address into poverty and others to fall even further into it. each group’s needs, such as better infrastructure and education services for the chronic poor and enhanced Policy options: safety nets for the transient poor. This type of targeting may have particular impact in urban areas, where there On the revenue side, adjusting the fiscal system can are more likely to be both types of poverty (and thus strengthen its redistributive capacity and ensure that a need to distinguish between the two). Overall, the the Dominican Republic can afford to uphold a more effect will be to stanch the “urbanization of poverty;” ambitious social contract and improve economic growth. Specific policy actions to achieve this include: • Further linking public social expenditures to current targeting mechanisms, such as the SIUBEN (Siste- • Making the tax system more progressive by replac- ma de Identificacion Unica de Beneficiarios), currently ing some of the current indirect taxation with direct used for the Progresando con Solidaridad program, When Prosperity is not Shared taxation (e.g., personal and corporate income taxes) non-contributory health insurance and electricity and and ensuring exemption of basic goods; gas subsidies, which helped reduce poverty after the 2003-2004 economic crisis. This will help assure that • Revising existing tax exemptions (now 5.9 percent help goes only to people who need it; of GDP) and incentives to make them more progres- sive; and • Consolidating small existing social programs, partic- ularly in the social protection sector, to avoid duplica- • Strengthening the ability of tax collection mecha- tion and fragmentation of services and help rational- 68 nisms to detect and limit evasion, for instance, by ize expenditures; and • Addressing the structural deficit in the electricity poor study for 11.1 years and complete 7.8 grades. sector, which diverts a substantial amount of public As the analysis shows, households headed by people resources away from the social sector. with higher education are more likely to experience upward economic mobility and their children enjoy Policy area 2 more opportunities to enhance their economic poten- tial. Thus, education is a major tool in promoting both Policy objective: Build fair, transparent and efficient in- inter- and intra-generational economic mobility and stitutions that will improve the provision and quality of ensuring sustainable returns to social expenditure.31 public goods and services, expand economic opportu- nities, increase upward mobility and better protect the • Health coverage and quality are patchy, ranking poor and vulnerable from economic shocks. behind those of countries that spend at similar lev- els or less. Maternal mortality rates in DR are around Problem to address: The quality of public services is low 150 per 100,000 live births, over a third higher than and, despite significant improvement, inequities in ac- the regional average, and infant mortality, at 22.3 per cess to basic public services remain, particularly to the 1,000 live births, is also above the LAC average. Like- poor. This constrains their set of human “opportunities” wise, immunization coverage, though it has increased, to progress in life.29 In particular, remains below regional levels. Coverage through full health insurance has increased considerably but still • The low quality of governance affects service de- leaves half of the poor uninsured, resulting in out-of- livery in key sectors. There have been significant im- pocket expenditures financing a high share of health provements in service delivery, for instance, efforts services. Combining quality of education and health to adopt transparent targeting mechanisms, publish services, the World Economic Forum ranked the Do- budget data and open lines for queries and com- minican Republic 107th out of 139 countries in 2010. plaints. Yet many Dominicans believe that inefficient bureaucracy, lack of transparency, anti-reform coa- Policy options: litions and low trust in governance systems remain important obstacles to raising competitiveness and Developing new mechanisms and strengthening exist- improving service delivery in the social sectors.30 ing ones to monitor the quality and provision of public goods and services with a goal of increasing account- • Despite remarkable increases in school enrollment, ability and ensuring that expenditures give maximum inefficient use of budget allocations and low qual- value to society. In addition to increasing spending in the ity continue to hamper the education system. The social sector and further linking it to effective targeting system is characterized by operation in double shifts, mechanisms, policy actions to increase accountability, cit- high teacher/pupil ratios, high teacher absenteeism izen engagement and service quality could include: and little use of performance data to manage for re- sults. Not surprisingly, more than 41 percent of third • Further developing and strengthening incentive, graders lack basic math skills, while over 31 percent monitoring and accountability structures between cannot meet the lowest defined level of the SERCE service providers, clients and policy makers, as well reading test. Students in the Dominican Republic as procurement and financial management reforms show some of the lowest academic performance in (e.g. the introduction of framework agreements) to When Prosperity is not Shared the LAC region, and this lack of efficiency in education fight corruption and promote good governance, affects the poor disproportionally. While the better quality service delivery, and accountability. The IDEC off study for 12.3 years and complete 10.1 grades, the initiative, which brings public and private sector and civil society actors together to improve the quality of 29  The Transparency International Corruption Perceptions Index education, and the community scorecards piloted by (TI-CPI) perceives the Dominican Republic as one of the most corrupt countries in LAC—the TI-CPI ranks the Dominican Republic as 118th among 174 countries evaluated. 31  From 2006 to 2012 the level of satisfaction in local government 30  According to Kaufmann et al. (2012), the Dominican Republic services regarding education and tax fairness at municipal levels de- has had historically weak rule of law, and government effectiveness creased almost 18 percent, according to Latinobarometro, for LAPOP has been low. (2012) 69 the Progresando con Solidaridad program are good which suggests that many of the jobs created examples. are of low quality. Three in every four jobs creat- ed between 2004 and 2011 were in the informal • Building on existing improvements in budget trans- sector. Increased employment during this period parency and strengthening efforts to better link did lift some people out of poverty; however, the planning and budget formulation and to further low quality of jobs may explain why the country adopt performance-based budgeting and manage- failed to reduce poverty to at least pre-crisis lev- ment instruments to link increased efficiency and els. As shown in the analysis, nearly 30 percent of effectiveness according to established standards the population was poor in 2011 despite having of quality. A crucial aspect of such instruments is the basic education and access to services and/or as- generation and use of performance data (for instance, sets, signifying that some people who are endowed linking student test scores to teacher evaluations); and to generate adequate income are unable to do so. Moreover, the Dominican Republic had the sec- • Making efficiency analyses (such as cost-benefit, ond-highest growth in the share of low-skill ser- cost-effectiveness and economic rate of return) vices as a percentage of employment between the and monitoring and evaluation systems far more late 1990s and late 2000s out of a sample of select- common to economically value the return on public ed LAC countries (Aedo and Walker 2012). investments, monitor the achievement of results and learn what works and what does not work. • Real earnings have been falling uniformly along the earnings distribution, and high-skilled workers are Policy area 3 being continuously absorbed into low-skill jobs. On average, real earnings per hour both of self-em- Policy objective: Strengthen access of the poor to labor ployed and private sector wage workers were about markets and increase the demand for their labor, so as 27 percent lower in 2011 than in 2000. Even workers to make efficient use of human capital and allow the employed in high-productivity sectors or who have poor to benefit from economic growth. tertiary education have not seen their incomes rise. These trends are consistent with the low upward Problem to address: Low labor force participation and mobility and high downward mobility documented dearth of formal, well-paying jobs, particularly among in this study. The fact that real earnings are also flat the poor and vulnerable, youth and women, at a time or declining in sectors that have strong productivi- when real earnings have been systematically falling ty growth and make the largest contribution to the across most economic sectors. In particular, overall output is puzzling but may be explained by the high percentage of people in need of jobs follow- • The sectors in which labor productivity has in- ing the crisis. creased, such as manufacturing, the wholesale and retail trades, communications and financial services, • Low labor force participation. Labor force partici- have not contributed much to job creation. For ex- pation in the Dominican Republic is 10 percentage ample, manufacturing had the second-highest annu- points lower than the regional average. More than a al average of contribution to the gross value added third of people between 18 and 29 years are neither growth rate from 2001–2011. Yet it was the sector that working nor studying and women, as a group, have When Prosperity is not Shared lost the most jobs over the same period, reducing its even lower labor force participation. share of total employment by 6 percentage points (Abdullaev and Estevão 2013). This kind of growth • An inadequately educated workforce and a skills without job creation may explain much of the lack of mismatch. According to the Global Competitive- poverty reduction in spite of economic growth during ness Report (2011-2012), one of the biggest prob- the second half of the 2000s. lems in doing business in the Dominican Republic is an inadequately educated workforce. The Domin- • The majority of job creation, instead, has oc- ican Republic ranks in the bottom third of the 142 70 curred in low-skill, low-productivity sectors, countries analyzed in terms of higher education and training. As this report shows, another reason why idating the existing ALMPs, such as the national training the number of transient poor doubled (from 15 to institute INFOTEP, the intermediations services of SENAE, 30 percent) could be that though some people are the labor observatory OMLAD and the youth training pro- endowed with certain human capital to generate in- gram PJyE. Specific policy actions to achieve this objec- come, they are unable to do so. This is due to a skills tive include: mismatch: the low relevance and quality of their ed- ucation and training do not match what employers • Further developing temporary employment pro- are looking for. grams to allow ALMPs to provide a counter-cyclical response in times of economic downturns and em- • Several factors reduce the ability of the private sec- ployment crisis and to activate the more disadvantage tor to create more and better jobs. The enclave type segments of the labor force; of development that has characterized two of the leading industries, special economic zones (Zonas • Updating and expanding training and retraining Francas) and tourism, restrains employment gener- programs that couple educational attainment with ation and business linkages to the overall economy. productive skills to improve the quality of the labor The business environment is further undermined by force and widen access to job opportunities; complex regulatory processes, including labor rules, and by weak institutions, lack of transparency, inef- • Strengthening intermediation services to provide ficient government bureaucracy, partial access to fi- job-search assistance and reduce information asym- nance, underdeveloped capital markets and poor la- metries between labor supply and demand; bor skills and infrastructure. • Providing entrepreneurship training and grants to Policy options: equip individuals, including the poor and vulnerable, with the skills and capital to start and sustain busi- Improving the relevance and quality of education. Spe- nesses; and cific policy actions to achieve this include: • Consolidating existing institutions and scaling up • Increasing investments in secondary and tertiary effective interventions to ensure coordination, in- education, changing the content of education and tegration and efficacy of existing ALMPs as a means training and creating the right incentive structure to develop and implement a country-wide system to within key institutions in the education sector to promote employment. ensure that all Dominicans have the necessary skills to work in high-productivity jobs. Bearing in Improving the business environment to foster better mind that rising labor income was the biggest en- competition, investment climate, entrepreneurship gine of poverty reduction in the last decade, pro- and job creation. Many of the policies proposed above vision both of the endowments and avenues for are expected to contribute to this goal. Additional policy people to participate in the labor market will raise actions include: standards of living and increase the global competi- tiveness of the country. • Improving competition policy by eliminating an- ti-competitive practices through market regulation in When Prosperity is not Shared Increasing labor market information flow and training key economic sectors and further opening up markets linked to job opportunities to raise the employability to local and foreign investment, including the finan- and productivity of the labor force. Creating targeted ac- cial sector, and tive labor market programs (ALMPs) designed to enhance opportunities for high-productivity job creation and for- • Creating and targeting incentives for entrepreneur- mality can raise the employability of the labor force and ship and innovation in sectors with the capacity to facilitate young people’s transition from school to work, generate large numbers of jobs that are likely to have without creating major market distortions. Further bene- significant positive spill-overs and income distribu- fit for training and intermediation will come from consol- tion effects for the rest of the economy. 71 Conclusions economic mobility with less than 2 percent of the popula- tion rising to a higher economic group. In fact, more than Sustained economic and social progress requires a vir- 19 percent of the population experienced a worsening tuous cycle of growth and equity as the fundamental in economic status from 2000 to 2011. Furthermore, de- policy goal. If growth is the result of an equitable process spite improving access to services, coverage and quality of income generation, societies can claim to be on a path remains uneven across population groups, thus limiting of shared prosperity. Based on these notions, this report the economic opportunities of disadvantaged people. has applied a variety of empirical methodologies and data to assess whether the strong economic growth that the Compared to LAC, growth in the Dominican Republic Dominican Republic has enjoyed during most of the last is stronger but the country is falling behind the wider decade benefitted the more disadvantaged groups in an region in a number of equity dimensions. Largely as a economic and social sense. In doing so, the report exam- result of the 2003-2004 crisis, poverty rates in the Domin- ined the conditions in the Dominican Republic upholding ican Republic, lower than LAC’s overall rates in 2000, now the key pillars of what constitutes an equitable society, exceed the region’s average. Furthermore, though pov- namely equality of opportunities regardless of original erty began to decline after the crisis, the rate of decline circumstances, the capacity of the system to bring peo- has been slower than LAC’s over the same time period. ple out and keep them out of absolute poverty, and the In terms of reducing income inequality, the Dominican existence of agency—the ability of people to make deci- Republic continues to perform better than the region. sions and convert them into actions—for all. This involved However, as improvements have been modest over the the analysis of the patterns that characterize aggregate decade, LAC is catching up to the Dominican Republic. monetary poverty and income inequality trends, house- In terms of inter-generational income mobility, a striking holds’ income mobility, and non-monetary dimensions of difference remains between the Dominican Republic and welfare that portray the deprivations experienced by the LAC—while 41.4 percent of people rose to a higher eco- poor, as well as the distribution of opportunities for chil- nomic group in LAC, only 1.8 percent did so in the Domin- dren and proxies for the amount of agency that different ican Republic. The country is also underperforming LAC in groups in society have. terms of increasing access to basic goods and services for children. At the rate of improvement of the past decade, Despite strong economic growth over the past decade, Dominican Republic would take longer to reach universal large inequities still exist in Dominican society and are access than the LAC average. While investigating the un- declining at a slower rate than expected. Strong growth derlying causes behind disappointing progress in equity in the Dominican Republic occurred in every year of the is beyond the scope of this report, the study postulates last decade except 2003 and 2004 when the economy some hypotheses as to why, compared to LAC, the fun- contracted due to a banking crisis. In total, the Dominican damentals of equity are weak in the Dominican Republic. Republic grew its GDP per capita by almost 50 percent from 2000 to 2011. But despite this growth, the funda- This report identifies areas of priority for policy to ad- mentals of equity remain low in the Dominican Republic. dress underlying factors of inequity in the Dominican For instance, though strong growth resumed after the cri- Republic, including fiscal policy, institutional effective- sis, the country has been slow in decreasing poverty that ness and the performance of labor markets. In short, soared by 17 percentage points to reach nearly half the the Dominican Republic has a weak fiscal capacity, as it population. The rate remains at 40.4 percent, higher than raises low revenues and does so in a manner that harms When Prosperity is not Shared the 32 percent level of 2000. Chronic poverty—people en- progressivity. The narrow fiscal space and institutional during long spells of poverty—remains an issue with only weaknesses, in turn, constrain the size and effectiveness a one percentage point decrease since 2000, from 6 per- of social expenditures, limiting access to crucial goods cent to 5 percent of the population as of 2011. Even more and services, particularly for the poor and vulnerable. Of concerning is that about one third of the poor are in this great concern is that even if they are endowed with the state despite having the skills to generate higher income. means to progress in life, many individuals are employed This group, the transient poor, has almost doubled since in informal, low-paying jobs and have high vulnerability the level in 2000. In addition, the country shows very low to economic shocks. Growth, especially in the last decade, 72 is becoming more and more concentrated in sectors such as financial services, transportation and communications and tourism) that have either little employment creation or low-paying jobs. To address these issues, this study identifies three poli- cy actions to address the underlying factors of inequity. These include: (1) adjusting the structure of fiscal policy, both on the expenditure and revenue sides, to make it more equitable, efficient and sustainable; (2) developing and strengthening monitoring, social accountability and incentive mechanisms to increase the quality and provi- sion of public goods; and (3) strengthening the access of the poor and other disadvantaged groups to labor mar- kets and increasing the demand for their labor to make efficient use of human capital, allowing the poor to maxi- mize returns on their endowments and to ultimately ben- efit from economic growth. Finally, policy design aimed at promoting a more bal- anced development path could be effectively informed by further analytical work. Particularly salient work could be performed in the areas of fiscal policy, social sectors and labor markets. This would help to further uncover the underlying factors that inhibit the gains of growth from being more evenly shared across the population. A rele- vant concern in this analysis is discrepancies between the Dominican Republic’s national accounts and household survey data. Serious analytical efforts should be devoted to understanding the apparent disconnection between macro and micro data that hinders the ability of national statistics to accurately reflect macroeconomic and social progress. When Prosperity is not Shared 73 References Abdullaev, U. and M. Estevão (2013). “Growth and Employment Lopez-Calva, L.F. and E. Ortiz-Juarez (2013). “A Vulnerability in the Dominican Republic: Options for a Job-Rich Growth.” Approach to the Definition of the Middle Class.” Journal of International Monetary Fund Working Paper WP/1340. Wash- Economic Inequality, January. ington D.C. Sanchez, M. and R. Senderowitsch (2012). “The Political Econ- Aedo, C. and I. Walker (2012). “Skills for the 21st Century in omy of the Middle Class in the Dominican Republic.” World Latin America and the Caribbean.” World Bank, Directions in Bank Policy Research Working Paper 6049. Washington D.C. Development—Human Development. Washington D.C. UNDP (2010). “Política Social: Capacidades y Derechos - Análi- Cattaneo, M.D., S. Galiani, P.J. Gertler, S. Martinez and R. Tit- sis y Propuestas de Políticas Sociales en República Domini- iunik 2009. “Housing, Health, and Happiness.”American Eco- cana.” Volume I, UNDP and Ministry of Economy, Planning and nomic Journal: Economic Policy, 1(1): 75-105. Development of Dominican Republic. Cord, L. and L.F. Lopez-Calva (2013). “The Equity Challenges in UNESCO (2008). “Student achievement in Latin America and Latin America: Opportunities, Markets and Institutions.” mim- the Caribbean—Results of the Second Regional Comparative eo, The World Bank. and Explanatory Study (SERCE).” Prepared by UNESCO and LLECE, Santiago, Chile. Cord, L., L.F. Lopez-Calva and C. Rodriguez-Castelan (2013). “Shared Prosperity: Challenges for Latin America.” mimeo, The Valderrama, D., M. Francos, M. Jiménez, M. Lizardo and A. World Bank. Morillo (2013). “El Efecto de un Programa de Transferencias Condicionadas sobre la Pobreza y Desigualdad en el Corto Deaton, A. (2005). “Measuring Poverty in a Growing World (or Plazo: El caso de Progresando con Solidaridad en República Measuring Growth in a Poor World).” The Review of Economics Dominicana.” Mimeo. National Office of Statistics and Min- and Statistics, MIT Press, vol. 87(1), pages 1-19, 04. istry of Economy, Planning and Development of Dominican Republic. Esteban, J.M. and D. Ray (2006). “Inequality, Lobbying, and Re- source Allocation.” American Economic Review. World Bank (2011). “On the Edge of Uncertainty: Poverty Re- duction in Latin America and the Caribbean during the Great Gupta, S., H. Davoodi and R. Alonso-Terme (2002). “Does Cor- Recession and Beyond.” Poverty, Gender and Equity Unit, ruption Affect Income Inequality and Poverty.” Economics of Poverty Reduction and Economic Management Team, Latin Governance, 3:23-45. America and Caribbean Region, Washington D.C. Guzmán, R. (2011). “Composición Económica Dominicana. El World Bank (2012). “Do Our Children Have a Chance? A Hu- Estrato de Ingresos Medios en el Umbral del Siglo XXI.” Mimeo. man Opportunity Report for Latin America and the Caribbe- When Prosperity is not Shared Document prepared for the Ministry of Economy, Planning an.” Directions in Development—Poverty. Washington D.C. and Development of Dominican Republic. World Bank (2012). “Poverty Dynamics in Brazil: Patterns, As- Huppi, M. and M. Ravallion (1991). “The Sectoral Structure of sociated Factors and Policy Challenges.” Policy draft. Washing- Poverty during an Adjustment Period: Evidence for Indone- ton D.C. sia in the mid-1980s.” World Development, Elsevier, vol. 19(12), pages 1653-1678, December. World Bank (2013). “Economic Mobility and the Rise of the Latin American Middle Class.” World Bank Latin American and Kaufmann, D., A. Gallina and R. Senderowitsch (2012). “Building Caribbean Studies. Washington D.C. Coalition for Reforms: The Case of the Participatory Anti-Cor- 74 ruption Coalition (IPAC) in the Dominican Republic.” mimeo. Annexes Annex A. Total number of poor people in Dominican Republic by area (2000-2011) Year total urban rural 2000 2,627,149 1,265,419 1,361,731 2001 2,740,586 1,350,338 1,390,248 2002 2,787,959 1,381,575 1,406,384 2003 3,595,442 1,900,253 1,695,189 2004 4,389,454 2,508,199 1,881,255 2005 4,279,834 2,471,572 1,808,263 2006 4,033,833 2,288,838 1,744,995 2007 4,055,352 2,352,567 1,702,786 2008 4,197,401 2,483,934 1,713,468 2009 4,076,939 2,470,707 1,606,233 2010 4,073,841 2,460,464 1,613,377 2011 4,009,289 2,439,498 1,569,791 Source: Study team’s own estimates based on ENFT 2000-2011 When Prosperity is not Shared 75 Annex B. Poverty decomposition by income source (2000–2011) Period Income source Total Urban Rural 2000-2002 Total 0.59 1.14 -0.42 Foreign remittances 1.10 1.23 1.20 In-kind non-labor income 0.80 0.77 0.75 Monetary non-labor income -0.29 -0.34 0.01 In-kind labor income 0.74 0.63 0.77 Monetary labor income -2.30 -1.40 -3.25 Imputed rents 0.54 0.24 0.10 2002-2004 Total 16.64 18.83 12.32 Foreign remittances 4.40 4.15 5.02 In-kind non-labor income 3.96 3.45 3.60 Monetary non-labor income 4.12 4.15 4.74 In-kind labor income 4.90 4.50 5.28 Monetary labor income -4.88 -1.81 -10.44 Imputed rents 4.15 4.38 4.11 2004-2011 Total -7.57 -4.71 -12.01 Foreign remittances 1.83 1.93 2.33 In-kind non-labor income 1.11 2.17 0.24 Monetary non-labor income -1.75 -1.74 -1.59 In-kind labor income 1.92 2.01 2.09 Monetary labor income -10.45 -9.04 -15.38 Imputed rents -0.22 -0.03 0.31 Source: Study team’s own estimates based on ENFT 2000-2011 When Prosperity is not Shared 76 Annex C. Poverty decomposition by income source for moderate and extreme poverty (2000–2011) Area Income source Extreme poverty Moderate poverty Total population Total 2.37 9.66 Foreign remittances 3.78 6.04 In-kind non-labor income 3.13 4.76 Monetary non-labor income -0.37 1.34 In-kind labor income 3.64 6.30 Monetary labor income -9.28 -12.36 Imputed rents 1.48 3.58 Urban population Total 3.68 15.25 Foreign remittances 3.10 6.03 In-kind non-labor income 3.12 5.81 Monetary non-labor income -0.67 1.69 In-kind labor income 2.88 6.35 Monetary labor income -6.29 -8.50 Imputed rents 1.54 3.87 Rural population Total 0.42 -0.11 Foreign remittances 5.13 6.25 In-kind non-labor income 3.39 3.30 Monetary non-labor income 0.55 1.09 In-kind labor income 5.53 6.57 Monetary labor income -16.13 -20.36 Imputed rents 1.95 3.03 Source: Study team’s own estimates based on ENFT 2000-2011 When Prosperity is not Shared 77 Annex D. Coverage and incidence of remittances as a percentage (selected years) Proportion of households receiving Percentage of total remittances Decile 2000 2002 2004 2011 2000 2002 2004 2011 1 13.0 5.8 9.4 10.5 0.9 0.6 0.6 1.3 2 22.6 18.5 11.8 14.0 2.9 3.2 1.3 3.0 3 17.1 7.9 16.8 10.6 2.9 1.9 2.5 3.0 4 21.0 15.0 21.7 14.3 4.5 4.6 4.0 5.0 5 20.4 15.7 25.0 11.7 5.5 6.0 5.7 4.9 6 13.0 18.8 21.0 16.6 4.4 8.8 5.9 8.5 7 19.8 22.3 15.5 18.2 8.5 13.1 5.4 11.8 8 17.7 16.3 21.2 19.3 9.8 12.4 9.5 16.2 9 25.7 17.8 22.0 21.3 20.2 18.8 13.9 24.5 10 20.2 11.9 30.2 7.9 40.3 30.6 51.2 21.7 Source: Study team’s own estimates based on ENFT 2000-2011 When Prosperity is not Shared 78 Annex E. Decomposition of Gini coefficient by income source (2000-2011) Period Income Source Total Urban Rural 2000-2002 All Income Sources -0.016 -0.015 -0.030 Foreign remittances -0.004 -0.004 -0.004 In-kind non-labor income -0.002 -0.001 -0.004 Monetary non-labor income -0.004 -0.005 -0.004 In-kind labor income -0.002 -0.001 -0.003 Monetary labor income -0.003 -0.004 -0.012 Imputed rents -0.001 -0.001 -0.004 2002-2004 All Income Sources 0.012 0.032 -0.003 Foreign remittances 0.024 0.022 0.020 In-kind non-labor income -0.001 -0.001 -0.004 Monetary non-labor income 0.006 0.008 0.000 In-kind labor income 0.003 0.004 -0.002 Monetary labor income -0.022 -0.006 -0.016 Imputed rents 0.002 0.005 -0.002 2004-2011 All Income Sources -0.025 -0.029 -0.005 Foreign remittances -0.009 -0.023 -0.004 In-kind non-labor income 0.001 0.004 0.000 Monetary non-labor income -0.012 -0.010 -0.015 In-kind labor income 0.005 0.005 0.005 Monetary labor income -0.007 -0.004 0.011 Imputed rents -0.003 -0.001 -0.002 2000-2011 All Income Sources -0.030 -0.012 -0.037 Foreign remittances 0.013 0.007 0.015 In-kind non-labor income -0.001 0.004 -0.006 Monetary non-labor income -0.015 -0.012 -0.022 In-kind labor income 0.008 0.007 0.004 When Prosperity is not Shared Monetary labor income -0.033 -0.016 -0.020 -0.016 -0.015 -0.030 Source: Study team’s own estimates based on ENFT 2000-2011 79 Annex F. Using synthetic panels with cross-sectional data to predict economic mobility ^ To estimate intra-generational mobility in the absence the first round (β1), where the superscripts denote which of panel data, this report utilizes the synthetic panel ap- survey the observations of households are taken from. proach developed by Dang, Lanjouw, Luoto and McK- enzie (2011). Synthetic panels are created using two or Given that the correlation between the error terms of the more rounds of cross-sectional data by modeling the re- two rounds is unknown but is likely to be non-negative, lationship between actual income (or consumption) and lower- and upper-bound estimates of mobility are ob- time-invariant household characteristics in one period tained by assuming either perfect correlation between and applying the model to observations of households the error terms in the former or zero correlation in the lat- from the other period to estimate income based on the ter. If zero correlation of the error terms is assumed, the same characteristics. By establishing a lower and upper upper bound of income or consumption in the first round bound for the estimated income based on assumptions for the households from the second round is predicted about the residuals, movements in and out of poverty by taking a random draw with replacement from the em- ~ over the period for different cohorts (e.g., poor to non- pirical distribution of first round estimated residuals (ε2i1, poor, always poor) can be placed within a range. drawn from the residuals of the OLS regression where ^ ε1i1= y1i1- β1’x1i1) as follows: Assume that there are two rounds of cross-sectional sur- = β1‘x2i1 ) + ~ ^2U ^ veys. Let x_it be characteristics which are observed in y i1 ε2i1. (F1.c) both rounds of household i in round t. These character- istics should be things that do not change between the By using the actual income from round 2 and the pre- two surveys such as gender, place of birth and parental dicted income of round 1 (for the same households), the education. These can also include characteristics in round fraction of households moving from one income group 2 referring to round 1, such as questions in the round 2 to another can be computed, for instance, for those who survey about whether the household head was employed escaped poverty by the following: in round 1 or whether the household had a TV in round ^ 1. Additionally, characteristics, regardless of variance, at Pr ( y2Ui1< p | y2Ui2 > p ). (F1.d) a more aggregate level can also be supplied, for example, unemployment rate, population of the working age or Repeating the procedure R times, due to the random variables from GIS data. drawing, and taking the average of the equation (F1.d) yields upper bound estimates of the movements in and Projecting round t consumption or income onto x_it out of poverty. To obtain the lower bound, perfect posi- (where ε_it is an error term) as follows: tive correlation of the error terms allows the estimates of the residuals from the second round ( ^ ε2i2 ) to be used di- yit = βt’ x it+ εit t = 1,2 (F1.a) rectly to predict income or consumption in the first round: ^ can yield information about the movement of household y ^2L i1 ε2i2 . = β1’x2i1 + ^ (F1.e) i across income groups, such as: The poverty dynamic of interest can then be calculated When Prosperity is not Shared Pr ( yi1< p | yi2 > p ) (F1.b) ^ using the lower bound estimate ( y2Li1 ). where p is the poverty line. To obtain both yi1 and yi2 giv- Using the lower bound provides a more conservative es- en that the household is only observed in one period, ob- timate of mobility (upwards and downwards) and is not servations from the second round can be used to predict prone to classical measurement error as the same dis- ^ income or consumption in the first round ( y2i1 ), by re- turbance term applies to both income (or consumption) gressing the sample of households in round 1 (y1i1 on x1i1) measures across the two cross sections. For this reason, to obtain the ordinary least squares (OLS) estimator from lower bounds estimates are presented in this report. 80 The synthetic panel approach for estimating income sults show that the true mobility generally lies within mobility has been validated in a recent paper by Cruces the upper and lower bounds of the synthetic panel es- et al. (2011) in the context of Latin America. The paper timates using two cross sections and that these results compares mobility estimates using synthetic panels are robust to further checks such as short and long term against actual panel estimates in three countries (Chile, estimates, focusing on particular sub-groups and sensi- Nicaragua and Peru) where panel data is available. Re- tivity tests. Annex G. Defining vulnerable and middle-class groups To define the group of people who are vulnerable to fall- related to vulnerability and the specific context in which ing into poverty, this report relies on the methodology for vulnerability is being defined. setting the lower threshold of the middle class developed by López-Calva and Ortiz-Juarez (2012). The people who López-Calva and Ortiz-Juarez apply the $10 per day have incomes between the LAC moderate poverty line of threshold found in the panel data to cross-sectional data $4 per day and the lower threshold of the middle class are to analyze income groups across time and find that the considered vulnerable—those who are neither poor nor groups increase and decrease with economic cycles in the economically secure. intuitive direction (the vulnerable group grows during economic recessions and contracts during growth, while To set the lower threshold of the middle class, López-Cal- the reverse occurs for the middle class). Using lower va and Ortiz-Juarez use panel data from Chile, Mexico and thresholds for the middle class labor market resources Peru to estimate the probability of being poor in the sec- and shocks affecting households (Xi). ond period of data and the incomes associated with these probabilities in the first period of data by incorporating pi = E ( poori | Xi ) = F ( Xi ∙βi ) (E1) the same independent variables in both models. Based on empirical evidence in Cruces et al. (2011) which finds Then, in order to capture the permanent component of that in a period of 15 years, 10 percent of Latin Americans household incomes, the same independent variables are fell into poverty every year, they define the lower thresh- used in the logit model to estimate an income equation, old of the middle class as the estimated income associ- where lnY_i is household per capita income. ated with a 10 percent probability of falling into poverty (~$10 per day in all three countries). lnYi = α + Xi ∙ βi + εi (E2) In the absence of panel data, this report relies on The average of the independent variables for each prob- cross-sectional data to provide the dependent variable in ability in Equation 1 is used with the coefficients from both models (the probability of being poor and income). Equation 2 to estimate the incomes associated with an This alters the concept of vulnerability slightly, from who array of probabilities of being poor. Finally, the estimated is likely to be poor in a future time period to who is likely probability of being poor is mapped against the estimat- When Prosperity is not Shared to be poor in the current period. With this exception, the ed income for each household. Alternatively, as done in report performs the same steps as López-Calva and Or- this study, the methodology can be applied using the ob- tiz-Juarez in defining the middle-class threshold. served income rather than the estimated income. Results using either alternative are, however, comparable. First, a logistic model is estimated to analyze the cor- relation between the probability of being in poverty As in López-Calva and Ortiz-Juarez, this report uses the (pi, where 1 = income of < $4 per day and 0 otherwise) predicted income of households with a 10 percent prob- and demographic indicators, vulnerable and it is relative ability of being poor ($9 and $7.35 for urban and rural in terms of the extent to which these characteristics are household respectively) to define the threshold between 81 the vulnerable and the middle class. Though this per- Ravallion (2010) of $2-10 and $2-13, respectively, seems centage is somewhat arbitrary, its relation to falling into to convolute these trends. Absolute standards, which poverty as found by Cruces (2011) provides the underly- may be more relevant in low-income countries, are made ing concept of economic security on which the definition over-inclusive by including people who are vulnerable of vulnerability, as applied here, is based. In other con- or poor. Making the distinction between people who are texts, this percentage may not be appropriate if house- not poor, yet vulnerable, and people who have achieved holds are more or less prone to falling into poverty. Thus, economic security can help policy makers better analyze this method of defining vulnerability is absolute in the the evolution of the middle class as well as develop tar- sense of establishing characteristics which cause house- geting mechanisms to more efficiently serve the poor holds to be proposed by Banerjee and Duflo (2008) and and the vulnerable. Annex H. Constructing the Human Opportunity Index First published in 2008 by Barros, Molinas Vega and Saave- Thus, the HOI decreases when differences between the dra Chanduvi in the context of LAC, the Human Opportu- coverage rates of circumstance groups increase on the nity Index (HOI) is a measure of the equitable coverage of principle that access to a basic good or service should be basic goods and services critical for enabling economic uncorrelated with personal characteristics, in which case progress in life. The HOI takes into account both overall the average coverage rates would be the same across access rates of basic services which provide opportuni- groups and the penalty would be 0. ties (e.g., access to safe water and school enrollment) and obtained by: differences in the coverage rate of circumstance groups n _ n ∑ wi pi and D = , 2C ^ ^ 1 ^ based on characteristics which are out of the control of in- C= ∑ i =1 ) wi | pi - C | (E4) 1 dividuals (e.g., birthplace, gender and race). Because chil- dren are less likely to have control of these characteristics, where wi = N_ or another sampling weight. An index of dif- 1 the HOI is calculated using individuals aged between zero ferences in access across groups, the “Dissimilarity Index” and 16 years old. or D-index, is the weighted average of the gap between circumstance group coverage rates and the overall cov- To account for both coverage and equity of distribution erage rate. In other words, the D-index is the number of in one number which ranges from 0 (no coverage) to 100 people in better-off groups whose access to a basic ser- (total coverage), the HOI penalizes P the overall coverage vice would need to be reassigned to worse-off groups for rate C by the amount of inequality of opportunity for a the distribution of coverage to be equitable. Accordingly, particular good or service. 1-D is the percentage of people with access to a service under equitable allocation. The D-index can be obtained HOI = C - P (E1) by dividing P by C, whereby: While C is simply the average access rate of the popu- HOI = C ( 1 - D ) (E3) When Prosperity is not Shared _ of the gaps between the number of lation, P is the total people required M for each group to achieve the average Thus, the HOI can be interpreted as the number of oppor- access rate of the population and the actual number of tunities in a society that have been distributed equitably. people covered M for each group for all groups k with below average coverage rates (opportunity-vulnerable The index is Pareto consistent in that it will rise if any groups) as a percentage of the total population N. person, regardless of circumstance, comes out ahead _ at no cost to anyone else. The HOI will also rise, holding _ ∑v (M -M ) P= 1 (E2) C constant, given an increase in the equality of distribu- N k =1 k k 82 tion. This entails that if a fixed number of opportunities is shifted from better-off to worse-off groups, the HOI as a function of the child’s circumstances. Next, the prob- ^ will increase as more weight is given to the coverage of ability of access p for each child i is calculated given the worse-off circumstance groups. If coverage rates increase coefficient estimates from the logistic regression. The HOI proportionally for all groups, the HOI will increase by the can then be computed by plugging into formula (E3) the ^ same proportion as both C and P would increase propor- overall coverage C and the D-index D tionally. Furthermore, an equal increase in coverage rates across groups would result in an equal increase in the HOI To construct a composite HOI of all opportunities, this re- as C rises and P remains unchanged. port used the simple average of the HOI of each opportu- nity analyzed (enrollment, 6th grade on time, water, sani- To empirically calculate the HOI, first a logistic model is tation, hard-floor dwellings and asset ownership). estimated on whether a child had access to a basic service Annex I. Poverty and Inequality in the Dominican Republic compared to Central America Figure 45: Moderate Poverty Rates in the Dominican Figure 46: Gini coefficient in the Dominican Republic Republic and Central America (2000–2011) and Central America (2000–2011) 60 0.65 50 0.60 Poverty headcount (%) 40 Gini Coe cient 0.55 30 0.50 20 10 0.45 0 0.40 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Central America Dominican Republic Costa Rica Dominican Rep El Salvador Guatemala Nicaragua Panama Honduras Note: The poverty rates for Central America are the weighted average by the population of Note: Dashed lines for Guatemala and Nicaragua represent linear interpolation for those years each country. Linear interpolation and extrapolation applied for those years in which the in which the country does not have survey data. Markers indicate years with data. Source: country does not have survey data. Source: Own estimates using data from SEDLAC Own estimates using data from SEDLAC (Socio-Economic Database for Latin American and the (Socio-Economic Database for Latin American and the Caribbean). Caribbean). When Prosperity is not Shared 83 The World Bank 1818 H Street, NW, Washington, DC 20433, USA. www.worldbank.org