CHAPTER 1 From Vicious to Virtuous Circles - World...

1

CHAPTER 1

From Vicious to Virtuous Circles

That raising income levels alleviates poverty, and that economic growth can be more or less effective in doing so, is wellknown and has received renewed attention in the search for pro-poor growth. Less well explored is the reverse channel: thatpoverty may, in fact, be part of the reason for a country’s poor growth performance. This more elaborated view of the devel-opment process opens the door to the existence of vicious circles in which low growth results in high poverty and high povertyin turn results in low growth. This report is about the existence of those vicious circles in Latin America and about theways and means to convert them into virtuous circles in which poverty reduction and high growth reinforce each other.

inequality, it would have been more pro-poor. Second, evenwhen inequality remains unchanged, economic growth isless effective in reducing poverty in countries with lessequal distributions of income: To attain the same reductionof poverty, unequal countries must grow more than moreequal ones. Given the region’s acute growth divergenceduring the lost decade of the 1980s and the slowdown from1998 to 2003, as well as lack of progress on the inequalityfront, it is not surprising that income poverty has been sopersistent since 1980 (figure 1.3). Though the report dis-cusses important caveats in traditional comparisons acrosscountries and across time, it remains true that, with theexception of Chile, there has been little poverty reductionbeyond the gains of the 1950–80 period, and in manycountries growth has not been especially pro-poor.

Poverty as a multidimensional anddynamic conceptThese conclusions broadly hold when a broader view ofpoverty and welfare is taken (chapter 2). As the literatureincreasingly stresses, poverty is a concept that spans a rangeof dimensions, such as health, mortality, and security, thatmay be uncorrelated with conventional measures of incomepoverty. Further, a complete concept of well-being needs to

LATIN AMERICA’S TWIN DISAPPOINTMENTS OF

relatively weak economic growth and persis-tent poverty and inequality are longstandingand intimately related. That raising incomelevels alleviates poverty, and that economic

growth can be more or less effective in doing so, is wellknown and has received significant attention in the searchfor pro-poor growth. Less well explored is the reverse chan-nel—poverty may, in fact, be part of the reason for a region’spoor growth performance, creating vicious circles where lowgrowth results in high poverty and high poverty in turnresults in low growth. This report is about finding ways ofconverting this negative cycle into a virtuous circle ofpoverty reduction, in which broad-based attacks on povertyfeed back into higher growth that in turn reduces poverty.

Latin America’s economic performance in the last 50 yearshas been disappointing. Growth lagged behind core coun-tries of the OECD (Organisation for Economic Co-operation and Development), at a time when East Asia andSpain, the madre patria on the periphery of Europe, werequickly catching up (figure 1.1). Income inequality hasremained very high in Latin America over the past 50 years(figure 1.2), posing a double impediment to poverty reduc-tion. First, had growth been accompanied by reduced

pove_001-020.qxd 2/2/06 12:08 PM

incorporate income movements across lifetimes or evengenerations, which means that issues of risk and mobilitythrough the income distribution must be examined. Ignor-ing these considerations leads to large distortions in theconcepts of poverty and inequality.

Although the limited existing data on these aspects ofpoverty do not permit the kind of global comparisons thatmeasures of income inequality and headcount povertynumbers do, the picture they sketch is only somewhat moreoptimistic. It is true that mortality rates have fallen farmore than income levels would predict and account forlarge improvements in welfare in those countries with little

growth. However, intergenerational mobility remains lowerin Latin America and the Caribbean than in the worst ofthe OECD countries. Recent evidence indicates that thechildren of poor families and of parents with low educationface a relatively high probability of achieving low educa-tional levels, obtaining lower returns for their education,and remaining poor (figures 1.4 and 1.5). The fact thatChile is one of the most mobile societies in the region sug-gests that the modernization of the country across the lastdecades has offered more opportunities to the less well-off.Finally, as documented in the World Bank’s Latin Ameri-can region flagship Securing Our Future in a Global Economy(de Ferranti and others 2000), the high economic volatilityin the region implies that the poor are subject to higherrisks than the poor in other regions. Although macroeco-nomic volatility was reduced in the 1990s after peaking inthe 1980s, it still remains exceptionally high, and labormarket volatility remains substantially higher than it is inthe United States, for example.

As later chapters show, all these dimensions not onlyprovide a more complete view of poverty, they also consti-tute channels back to growth.

The twin disappointments: Destiny or choice?Is there something intrinsic to the region that has left itwith relatively low growth and high levels of inequality

2

P O V E R T Y R E D U C T I O N A N D G R O W T H : V I R T U O U S A N D V I C I O U S C I R C L E S

Spain

LAC

East Asia0

0.2

1870

0.4

0.6

0.8

Ratio

1880

1890

1900

1913

1925

1929

1938

1950

1960

1970

1975

1980

1990

2000

Source: Authors’ calculations based on Prados de la Escosura (2005)and Maddison (2005). Note: LAC � Argentina, Brazil, Chile, Mexico, República Bolivarianade Venezuela, and Uruguay. East Asia � South Korea,Taiwan (China), Hong Kong (China), and Singapore.

FIGURE 1.1

Per capita income relative to the OECD, 1870–2000

0.401950 1960

0.45

0.50

0.55

0.60

1970 1980 20001990

Source: Authors’ calculations based on Altimir (1987) and Londoñoand Szekely (1997). Note: Based on data for Brazil, Chile, Mexico, and RepúblicaBolivariana de Venezuela.

FIGURE 1.2

Gini coefficient for Latin America, 1950–2000

01950 1960

10

30

50

70

Percent

60

40

20

1970 1980 20001990

Source: Authors’ calculations for 1950–1980; Gasparini, Guitierrez,and Tornarolli (2005) for 1990 and 2000. Note: We used a poverty line of US$2 a day; poverty rates for1950–1980 are estimated using a lognormal approximation.

FIGURE 1.3

Poverty rates in Latin America, 1950–2000

pove_001-020.qxd 2/2/06 12:08 PM

and poverty? The World Bank’s Latin American regionflagship Inequality in Latin America: Breaking with History?(de Ferranti and others 2004) argued that exclusionaryinstitutions set up during the European conquest to exploitexisting mineral wealth and indigenous populations, andthe particular crops suited to the region’s climate (such assugar plantations based on a slave workforce), led to highlyunequal access to land, education, and political power at

least until the late 1800s and thus had adverse conse-quences for growth and inequality for a long time.

In chapter 3, we show that indeed Latin America waswell behind the advanced economies in the mid-1800s,when the region’s per capita income levels representedabout 60 percent of the U.S. levels and 55 percent of thosein the broader OECD group. More important, we also showthat a significant part of the current development gap in

3

F R O M V I C I O U S T O V I R T U O U S C I R C L E S

0

0 to 6 6 to 11

4

12

20

Children’s years of education Average years of education of adult sons

Colombia Brazil

Mother’s years of education Father’s years of education

16

8

more than 11 0 to 4 5 to 8 9 to 11 more than 110

4

12

20

16

8

Source: Authors’ estimates based on household survey data. Note: Average years of school for adults aged 24–65 is determined by their parents’ years of school.

FIGURE 1.4

Low educational traps persist across generations among the poor and excluded

40% poorest 20% richest Whites Pretos Pardos

0.1Completeprimary

Somesecondary

0.3

0.7

1.1

1.7

Wages relative to level of education

Source: Authors’ estimates based on household survey data. Note: Average schooling returns for workers from families in the bottom and top quintiles of the income distribution; from Mincer earningsregressions, controlling for work experience, gender, and urban residence.

Chile

FIGURE 1.5

Although they stand to gain the most from education, poor people actually have low returns

1.5

1.3

0.9

0.5

Completesecondary

Someuniversity

Completeuniversity

Wages relative to level of education

Nicaragua

0.1

0.3

0.7

1.1

1.7

1.5

1.3

0.9

0.5

Completeprimary

Somesecondary

Completesecondary

Someuniversity

Completeuniversity

20% richest20% richest

20% poorest20% poorest

pove_001-020.qxd 2/2/06 12:08 PM

the region dates from the middle of the 20th century, whenother regions took more advantage of the rapid pace ofglobal expansion. Latin America’s relative retardation inthis period was in all likelihood related to the extremeinward-looking policies instituted then and to the lack ofmacroeconomic prudence that led to the devastating debtcrisis of the 1980s. Although policies are importantly con-ditioned by historical context, more promising roads werenot taken.

The same appears true in the realm of income distribu-tion. The report shows that as the 20th century began,France, Spain, the United Kingdom, and the United Statesall had high levels of income inequality. Yet they managedto lower income inequality dramatically during the centuryand over relatively short periods of time (two to threedecades). Such achievements appear related to the universalprovision of basic education and health services and theestablishment of highly redistributive welfare states.

Both Latin America’s loss in relative income position inthe last 50 years and the OECD’s ability to sharply reduceinequality are, perhaps counterintuitively, good news: ourhistory is not our destiny—choices of policies and institu-tions can lead to major improvements along both dimen-sions. Breaking with history is indeed difficult, but it is byno means impossible.

The link from growth and developmentto income-poverty reductionChapter 4 of the report concentrates on the effect of growthand changes in inequality on income-poverty reduction incountries with different characteristics. It shows thatachieving the greatest reduction in poverty may implyplacing differing relative emphasis on growth versus redis-tribution depending on the individual country’s initialconditions: poor countries (such as Bolivia, Haiti, andHonduras) and relatively equal countries that, bluntly put,have little to distribute, need first and foremost high andsustained growth, even at the expense of some increases ininequality; this might be called the China model. Incontrast, relatively richer and more unequal countries—most of Latin America, and especially Argentina, Brazil,Colombia, and Mexico—need both higher growth andsignificant redistribution if they want to make a fast andsignificant dent in poverty reduction (table 1.1).

Chapter 5 examines how different policies and differentsectoral patterns of growth affect income-poverty reduc-tion. It finds that sectoral composition matters: different

industries show large differences in labor intensity (agricul-ture and construction are generally more labor intensivethan manufacturing and services, and the latter are morelabor intensive than mining and utilities); and povertyreduction is stronger when growth has a labor-intensiveinclination. The chapter also finds that policies such asincreased access to education and infrastructure have haddirect positive impacts on growth, inequality, and povertyreduction, while others, such as trade opening, have hadpositive effects on growth but have tended to increaseinequality and even poverty in the short run. In the longrun, however, all pro-growth policies tend to reduce incomepoverty.

Chapter 5 also discusses the importance of transfers as ameans of sharing the fruits of growth by investing in thepoor. Bringing the historical discussion above into the pres-ent, the chapter shows that roughly half of the stark differ-ence in income inequality between Latin America andcontemporary OECD countries results from differences inreturns to factors of production—the result of the unequaldistribution of human and other capital in Latin America.But the other half results from the generally unprogressivenature of Latin America’s system of transfers. The coreOECD countries use transfers from the rich to the poor, andextensive pension schemes that distribute income from thethose working today to those retired tomorrow, to lower

4


TABLE 1.1

Growth rates needed to compensate for a 1-percentage-point

increase in inequality

Compensatory CompensatoryCountry growth rate Country growth rate

Argentina 2.5 Peru 1.6Chile 2.4 St. Lucia 1.5Brazil 2.3 Guatemala 1.5Mexico 2.1 Paraguay 1.5Costa Rica 2.1 El Salvador 1.4Colombia 2.1 Venezuela, 1.2Trinidad and Tobago 2.0 R.B. deDominican Republic 1.9 Ecuador 1.1Panama 1.9 Nicaragua 1.1Belize 1.8 Guyana 1.1Uruguay 1.8 Bolivia 1.0Jamaica 1.7 Honduras 0.8

Source: Authors’ calculations.Note: The table reports the growth rates that would leavepoverty unchanged when the Gini coefficient increases by 1percent. Higher values indicate that inequality plays a moreimportant role in poverty reduction.

pove_001-020.qxd 2/2/06 12:08 PM

the Gini (the standard measure of inequality) by about15 percentage points (from, for instance, 0.53 in theUnited Kingdom to 0.35).1 Transfers in a typical LatinAmerican country, in contrast, alter the Gini by 2 percent-age points or less, although there are a few exceptions suchas Chile, which managed to reduce the Gini by twice asmuch (figure 1.6).

Whether the pure transfers of the magnitudes discussedabove for Europe have been optimal from a growth point ofview is debatable, as is their wisdom or political feasibilityin Latin America. Arguably, for a variety of reasons, and inparticular to be consistent with growth objectives, redis-tributive policy probably should focus on equalizingopportunities through more equal access to assets, such ashuman capital, rather than on equalizing outcomes mea-sured as incomes per se. What is clear, however, is thatLatin America has not made the efforts to mobilize theresources to attack poverty that it could. First, the region’stax collections are below those in similar countries (whenbenchmarked by income per capita), with a few exceptionssuch as Brazil and Nicaragua, and collections for progres-sive taxes, such as personal income and property taxes, are

especially low. More important, although Latin Americanpublic expenditures underwrite large, progressive items(basic education and health), they also fund large regressiveitems (subsidies to pensions, tertiary education, andenergy), which offset the progressive spending. An encour-aging recent development is the introduction of successfulpolicies such as Progresa/Oportunidades in Mexico, Familiasen Acción in Colombia, and Bolsa Escola in Brazil, that com-bine fiscal transfers to the poor with incentives for them tobuild human capital through both health and educationinvestments from early childhood.

Closing the virtuous circle: The link from povertyto growthThe more novel thesis of the report is that Latin America’spersistent poverty may itself be impeding the achievementof higher growth rates—that there are reinforcing viciouscircles that keep families, regions, and countries poor andunable to contribute to national growth. The now-expansiveliterature on poverty traps has elaborated a large number ofchannels that may perpetuate poverty. The emphasis weplace on the multidimensionality of poverty and on lifetime

5


Source: Authors’ calculations.

Gini market incomes

FIGURE 1.6

Gini coefficients for market and disposable incomes

Gini disposable incomes

0.60

Latin AmericaIreland

Canada

Latin AmericaIreland

United Kingdom

Portugal

FinlandDenmark

Italy

Greece

EU15

United States

Spain

Belgium

Sweden

Germany

France

Luxembourg

Netherlands

Austria

0.55 0.45 0.35 0.200.250.300.400.50 0.20 0.25 0.35 0.45 0.600.550.500.400.30

pove_001-020.qxd 2/2/06 12:08 PM

and intergenerational considerations in welfare measure-ment further enriches the universe of channels throughwhich poverty impedes growth. To list just a few we discuss:

• Poor people often have limited access to financial mar-kets or other necessary complements to private invest-ment (such as property rights and infrastructure)essential to the accumulation of physical and knowl-edge capital and participation in the growth process.

• Poor people are often in poor health, which reducestheir productivity and impedes their ability to man-age and generate knowledge.

• Poor people attend low-quality schools and the lowand late returns to education and diminishedprospects for mobility deter the accumulation ofhuman capital essential for growth. Education enhancesearnings potential, expands labor mobility, promotesthe health of parents and children, and reduces fertil-ity and child mortality.

• Poor people may face more labor market risk, or maybe less able to hedge against it, and thus find returnsto investing in human capital adjusted for risk to beless attractive. Further, the inability to diversify riskprevents specialization in agriculture or movementsto off-farm activities, for example, that would lead togreater productivity. Since the poor are typicallymore risk averse than the rich because losses hurtthem more severely, in the absence of well-functioninginsurance and credit markets, the poor skip profitableinvestment opportunities that they deem too risky.Once again, societies with high poverty rates show atendency to underinvest.

• Poor regions and countries have fewer individualscapable of adopting, managing, and generating newtechnologies that would contribute to productivity.

• Poor regions may lack the infrastructure or humancapital that would make them attractive to extra-regional investment or the resources to develop themand that would facilitate sectoral and territorial labormobility in search of higher income opportunities.

• Poor countries with poor regions may find ethnic orracial tensions exacerbated by income disparities lead-ing to interregional tensions that make both regionsand the country as a whole riskier to invest in.

In each case, poverty in itself prevents taking actionsthat would facilitate the exit from poverty and results in

lower aggregate growth. Such vicious circles can lead to“convergence clubs”—richer and poorer countries, regions,or households tend to converge to different income or wel-fare levels even in the long run. Whether these are, in fact,poverty traps that cannot be escaped without intervention,or whether it simply takes much longer to transition tohigher-income states, is to us a distinction of secondaryimportance, particularly when political economy issues areconsidered. What we do argue is that smart investments inthe poor can lead to virtuous circles and that the issue of“pro-growth poverty reduction” should perhaps be asimportant a policy concern as traditional concerns with“pro-poor growth.” In other words, investing in the poor isgood business for society as a whole, not just for the poor.

Tracing these reinforcing circles implies necessarilymoving away from static concepts of poverty and studyingthe dynamics of poverty at every level, and this reportaspires to break new ground in this area. It provides evi-dence on the existence of convergence clubs at the house-hold, regional, and international level and in several casesshows that these appear to reveal the evidence of poverty-trap dynamics.

Global convergence clubsDo poorer countries grow less than richer countries? Theevidence presented in chapter 6 suggests that, with a fewnotable exceptions, they do. Panel a of figure 1.7 suggeststhat, apart from two short periods (one in the second half ofthe 1970s and another in the early 2000s), the typicaldeveloping country (and Latin America is not an exceptionhere) has always experienced lower growth rates than thetypical rich country. Over the 1963–2003 period, medianper capita growth in industrial countries outpaced mediangrowth in developing countries by an average of more than1 percent per year.

The difference in per capita growth rates between thedeveloped and developing countries has led to an expandinggap between rich and poor countries over time (figure 1.7,panel b). In the early 1960s the median Latin Americancountry had an income level that was slightly less than one-third the income of the median developed country; todaythat gap is less than 20 percent. Globally speaking, the typ-ical developing country had an income level about 12 per-cent that of the richer countries in 1960; and today it iscloser to 5 percent. There is little to support the conver-gence hypothesis that poorer countries will tend to catchup with the richer ones. Rather, as panel c of the figure

6


pove_001-020.qxd 2/2/06 12:08 PM

suggests, the poor stay poor, while the rich get richer. Thehistogram for the world in 1999 suggests a trimodal distri-bution, with a low peak at $1,100; a second at between$5,000 and $8,000, and a third peak around $35,000 form-ing poor, middle-income, and rich convergence clubs.(Chapter 7 shows that since 1960 there has been conver-gence within these clubs but divergence among them.)Panel d shows that Latin America as a region is unimodalwith its single peak at about $8,000 and belongs to themiddle cluster that is slowly separating both from the verypoor and, distressingly, from the very rich.

Convergence clubs at the cross-national level are alsoevident, though much less so, when nonincome dimensionsof welfare are considered. For example, figure 1.8 presentsthe cross-national life expectancy histograms for 1960 and2002. These histograms indicate the presence of a two-peaked pattern in both periods, but it is also evident thatthe mass of the low peak declines between 1960 and 2002,

whereas the mass of the high peak increases (worldwide lifeexpectancy has increased and is slowly converging).

Does poverty matter for growth?Are high poverty levels to blame for the disappointinggrowth performance of poorer countries? A bimodal distri-bution in income or life expectancy levels does not, initself, prove that poverty is a brake on growth, andchapter 6 finds only mixed evidence for the extreme case ofpoverty traps. However, the chapter does identify severalself-reinforcing mechanisms that may retard growth andcause poverty to persist, and these may be more relevantfrom a policy point of view. Looking across countries,poverty does appear to deter growth and investment (fig-ure 1.9), especially when the degree of financial develop-ment is limited. More specifically, we estimate in chapter 6that, for the average country, a 10-percentage-pointincrease in income poverty lowers the growth rate by about

7


�2

1963

�1

1

5

a. Growth rates b. Relative incomes

3

4

2

0

1960

0

0.05

0.15

0.35

0.25

0.30

0.20

0.10


FIGURE 1.7

Indicators for poor and rich countries

0

0.4

5

15

25

c. World

Per capita income, US$ thousands

Income relative to OECDPercent

Number of countriesNumber of countries

Per capita income, US$ thousands

d. Latin America

20

10

0.7 1.1 1.8 3 5 8 13 22 35 60 0.40

2

6

10

8

4

0.7 1.1 1.8 3 5 8 13 22

Latin AmericaDeveloping Developed Latin America Developing

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

pove_001-020.qxd 2/2/06 12:08 PM

1 percent, holding other determinants of growth constant.Further, we estimate that a 10-percentage-point increase inincome poverty reduces investment by 6–8 percentagepoints of gross domestic product (GDP) in countries withunderdeveloped financial systems. These results validatethe predictions of theory: that poverty may limit growthwhen financial sectors are imperfect because the poor, wholack access to credit and insurance, will not undertakemany socially profitable investments, thus depressing theaggregate level of investment and growth. The report alsofinds evidence that poverty limits the level of innovation (as

measured by research and development expenditures) andthe accumulation of human capital (see below), both ofwhich are additional channels through which poverty influ-ences aggregate growth.

Regional convergence clubsChapter 7 finds an unusual combination of convergingincome among subnational units, but increased spatial con-centration within countries. Modern spatial econometrictools show that within Brazil, Chile, and Mexico, there areclear convergence clubs of rich and poor regions, that appearto be drifting increasingly apart (figure 1.10). This findingis consistent with the New Economic Geography literaturethat has focused on how larger, already established regionsenjoy scale economies while lagging regions are less produc-tive and hence less attractive to factors of production.

These dynamics, and those discussed for national povertytraps in chapter 6, apply to national or subnational unitsequally. Two considerations are particular to the latter,however. The first is that within countries, labor can legallymove freely. In practice it does not, leaving large wage gapsof often 50 percent among regions. Evidence from Chileand Mexico suggests that this phenomenon is partly theresult of another poverty-trap dynamic—the poor cannotmuster the savings or liquidity to migrate and hence can-not leave. But other evidence suggests that this story maybe incomplete. Nonincome measures of poverty, such asmortality, show convergence within countries, much the

8


350

10

20

35

Number of countries


1960

FIGURE 1.8

Convergence clubs in life expectancy throughout the world

30

25

15

5

40 45 50 55 60 65 70 75 40 45 50 55 60 65 70 75

Number of countries

2002

Life expectancy at birth, years Life expectancy at birth, years

0

20

40

60

50

30

10

80 85

01

(poorest)2 3 4 5

World income ranking by decile

6 7 8 9 10(richest)

10

20

30

50

% of people living in poverty Rate of investment

12

14

16

18

20

2470

40

60 22


FIGURE 1.9

Poverty and investment throughout the world

Poverty (%)

Investment (% of GDP)

pove_001-020.qxd 2/2/06 12:08 PM

way they do internationally, suggesting that the welfaregap broadly considered may be less dramatic. Further, sim-ply asking people how poor they feel reveals some provoca-tive anomalies. The poorest group in the Bolivian altiplano(largely indigenous) self-rates as the least poor in Bolivia,while inhabitants of the rich province of Buenos Aires ratethemselves as the poorest in Argentina. These findings sug-gest that “congestion externalities”—the negative aspectsof living in concentrated urban areas—may be important,that relative income disparities may be more brutallyapparent in urban contexts, or simply that researchers aremissing key dimensions of well-being that are uncorrelatedwith income.

Second, laggard regions in general have low levels ofeducation and infrastructure that require special efforts tobring them toward the country average. However, to thedegree that agglomeration externalities—the economies ofscale that may arise from concentrating economic activity—dictate that poor regions have lower growth potential andlower returns to investment, governments may be con-fronted eventually with a trade-off between aggregategrowth and geographical equity.

Household-level poverty trapsThe fundamental building block underlying the interna-tional and regional analyses discussed above is the household.Addressing persistent poverty requires an understanding ofthe factors preventing poor families from moving out of

low-productivity economic activities. The poverty-trapsliterature emphasizes insufficient asset holdings (includinghuman capital), thresholds in the returns to those assets,fixed costs of productive transitions, and limited access tocredit or insurance among the poor as main determinants oftheir inability to take advantage of growth opportunities.Of particular importance is the ability of the poor to usetheir labor (their most abundant asset) in wage jobs, self-employment, or their own microenterprises. Labor earningsoften account for more than two-thirds of total householdincome of the Latin American poor. The pricing of laborreflects productivity differentials across workers and jobs,sector and regional supply-demand imbalances, and non-market factors. Low-earnings traps can arise from deficien-cies in the endowments that enhance the productivity(quality) of labor assets (such as human capital and infra-structure) and from earnings differentials unrelated toskills (such as ethnic discrimination and location) that arisefrom barriers to mobility in the labor market.

Chapter 8 examines some of the mechanisms that mayprevent the Latin American poor from participating in thegrowth process and lead to persistent poverty. Unfortu-nately, the limited long-span panel data prevent in-depthanalyses of the duration of poverty and its main determi-nants throughout Latin America. The chapter draws on thelimited, though highly consistent, evidence available onthese issues and reaches two main conclusions. First, lowlevels of productivity, rather than labor market segmenta-tion, is the overwhelming driver of low earnings. Mostpoverty is thus not generated directly by labor market fail-ures but by deficiencies in workers’ productive endow-ments, especially education, combined with the low levelsof overall productivity of their local economy. This effect isexacerbated by high volatility and the inability to insureagainst shocks, much more so than in developed countries.Second, detailed analyses of rural El Salvador and consistentevidence from other countries suggest that poverty trapssurrounding the accumulation of these productive assetsare a phenomenon of practical relevance in the region.

Chapter 9 then takes on one of the central channels thatcan support a two-way causality between poverty and eco-nomic growth: the accumulation of human capital. Humancapital, proxied by education or health levels, is generallybelieved to be one of the key determinants of long-termgrowth, while cross-country empirical evidence suggeststhat poverty may affect education levels (see chapter 8).Chapter 9 investigates the micromechanisms that could

9



Note: Figure shows relative state income distributions at time t and ten years later for the period 1955–2000. It suggests littlemovement in states’ relative positions and a persistent two humpeddistribution.

FIGURE 1.10

Regional income dynamics in Brazil: The persistence of twoconvergence clubs

0 0.5 1.0 1.5 2.0 2.5 3.0

0

1.0

2.0

3.00

0.5

1.0

1.5

2.0

2.5

Relative income

Country relative, period t � 10

Country relative,period t

pove_001-020.qxd 2/2/06 12:08 PM

support this double causality, so that specific actions toincrease the educational attainment of the poor couldignite a virtuous circle of faster growth and poverty reduc-tion in the region.

The chapter begins with a well-known fact: families withlittle education (specifically those with less than secondaryschooling) tend to be poor, and in turn they tend not to investenough in their and their children’s education to escapepoverty. The chapter documents several pieces of evidence onself-reinforcing mechanisms driving this vicious circle.

First, despite the region’s recent progress toward univer-sal primary enrollment, there is a clear and persistent educa-tional divide in educational attainment. The populationsorts into two groups: individuals with low-educationattainments (typically less than secondary education) andindividuals with secondary education and above (fig-ure 1.11). Rural residents and the poorest families, includ-ing disadvantaged ethnic groups, are predominantlytrapped in the low educational group. This divide continuesreplicating itself among the current cohort of students inhigh rates of repetition and dropout of these same groups.The smooth decline in enrollments during the secondary

cycle in most countries suggests that lack of school facili-ties is not the main driving factor, although in some coun-tries physical access constraints remain a problem.

Second, returns to schooling tend to increase with thelevel of education, a finding consistent with a skill bias inlabor demand caused by technological change in theregion, as detailed in the World Bank’s Latin Americanregion flagship Closing the Gap in Education and Technology(de Ferranti and others 2003). Schooling returns are flatduring the basic and secondary cycles and increase aftercompletion of secondary education; in some cases, the fullreturn materializes only after completion of tertiary educa-tion. That is, schooling returns become attractive just asthe opportunity cost, in terms of wages forgone by the stu-dent, becomes most acute for poor families. In addition, thechapter strikingly shows that in most countries poor fami-lies face below-average returns to tertiary (and sometimessecondary) education, plausibly due to low-quality schoolsas well as disadvantages arising from family background andattitudes toward education (see figure 1.4). Poor familieshave to juggle current subsistence needs against schoolinginvestments with a remote, uncertain, and less-attractive

10


00

10

20

35

40

45

Percent

Source: Authors’ estimates based on household survey data.Note: Distribution of the working-age population across schooling levels from families in the bottom and top quintiles of the incomeper capita distribution.

Argentina

FIGURE 1.11

The sharp educational divide between the poor and the rich in Latin America

30

25

15

5

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18+ 00

10

20

35

40

45

Percent

30

25

15

5

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18+

Brazil

Years of education Years of education

30% poorest 30% richest

pove_001-020.qxd 2/2/06 12:08 PM

payoff. The statistical evidence describing the low incen-tives and barriers to accumulating human capital is corrob-orated by the responses that poor children and youth givefor dropping out of school: high opportunity costs at olderages, perceived low benefits in the 1–12 grade schoolingcycle, and physical access constraints.

In sum, the completion of a secondary education neces-sary for poor families to move out of poverty remains out ofreach and children’s education remains strongly correlatedwith that of their parents. The educational divide is self-reinforcing across generations and is a critical underlyingdriver of the vicious circles of poverty observed at thehousehold, regional, and national levels.

Implications of the reportA number of implications emerge from the analysesdescribed above. We discuss them along two main dimen-sions: strategic and policy levels.

Strategic implicationsThe report uncovers several lessons that have implicationsfor the way we view poverty reduction.

1. Pro-poor growth and pro-growth poverty reduction. Theexistence of virtuous circles between growth andpoverty reduction enriches the debate on optimalpoverty reduction strategies in several ways. • First, the debate about whether strategies should

emphasize pro-growth or pro-poor policies nowappears somewhat less germane. Strategies that donot focus on growth forswear perhaps the mostpotent weapon for improving human well-being atour disposal, especially in light of the likely limitsof explicitly pro-poor policies discussed above. Yetfailing to take account of the constraints facing thepoor in participating in and contributing togrowth undermines its generation. For example,liquidity constraints, risk, and indivisibilities orlumpiness in human capital investments appear toprevent the poor from acquiring the education thatwould move them out of poverty and fuel growth.Redressing these constraints gives rise to an under-examined dimension of policy analysis that mightbe called pro-growth poverty reduction.

• Second, the bidirectional relationship betweengrowth and poverty reduction suggests that ide-ally consideration of policies should take into

11


account their direct and indirect effects on growthand poverty reduction. This awareness introducesnew but necessary levels of complexity in the eval-uation of policy options on both agendas. As asimple but important example, conditional cashtransfer programs have an impact on poverty thatgoes beyond the increased incomes for poorhouseholds provided by straight transfer policies.Conditional transfer programs also relieve creditconstraints on and provide a further incentive to theaccumulation of human capital that raises incomeboth at the household and, eventually, at the econo-mywide level.

• Third, pro-growth policies that have short-runadverse impacts on distribution and poverty, asappears to be the case with trade opening, mayactually create a drag on growth creation (seechapter 5). However, when combined with com-plementary policies such as improved access toeducation and infrastructure, the short-run adversepoverty effect can be mitigated, enhancing boththe direct and indirect effects on growth. Further,compensatory actions to offset some of these effects(for example, support to small farmers in noncom-petitive sectors during trade opening) gain a newrationale in increasing the efficiency of reformpolicies in addition to those justifications relatedto social protection.

• Finally, transfer programs should always seek todirectly stimulate the accumulation of assets thatwill advance the growth process, as programs likeOportunidades in Mexico, Bolsa Escola in Brazil, andFamilias en Acción in Colombia do.

2. Pro-poor growth vs. pro-poor government policy. The find-ing that at most half of the difference in inequalitybetween Latin America and OECD countries arisesfrom differences in the distribution of marketincomes implies two things. First, while efforts needto be made to improve both the endowments of thepoor and the returns to them offered by the market,there appear to be limits to what can be done. Forexample, Sweden, a country well known for its con-cerns with equity and human capital formation, has amarket distribution that is very similar to that ofmany countries in Latin America, suggesting thateven states that put equity high on their policyagenda may end up with high levels of inequality in

pove_001-020.qxd 2/2/06 12:08 PM

In short, policy makers need to consider morecomprehensive measures of poverty and inequalitynot only to get a more accurate view of the evolutionof societal well-being but to better understand andtake advantage of the channels back to growth.

4. Nonlinear thinking: Humps and black holes, agglomera-tion externalities, and complementarities. One criticalinsight of the poverty-traps literature is that theresponse to policy is nonlinear: it may vary depend-ing upon the magnitude and comprehensiveness ofthe effort.• There are thresholds (or humps) below which

effort may have no impact; in such cases policymakers are effectively throwing resources down ablack hole. For example, the fact that the returnsto secondary education often materialize onlyupon completion—or, worse, upon completion oftertiary education—implies that it is not worth itfor households to invest beyond primary school.Programs that seek to create incentives to investin education may have a greater impact on povertyif they are designed to get the student “over thehump”—through the end of secondary school andnot just to the next grade level.

• The literature suggests that the returns to assets,such as human capital, depend greatly on otherpublic assets that are complements, such as roads,communications systems, and credit markets.Major investment in education, for example, mayhave limited payoff if individuals cannot com-mute to a job that uses the higher level of skills. Inthe same way, a pro-growth policy of buildingroads in a region may have a greater impact if thepopulation has the human capital to work inemerging industries than if they are sick, illiter-ate, or constrained by language.

• Policies toward lagging regions may be complicatedby the fact that concentrations (agglomerations) ofeconomic activity are self reinforcing—that is,they are more economically dense. Richer areasmay have intrinsic dynamism and yield higherreturns to capital and labor than poorer areaswhere there is no natural equilibrating tendencytoward geographical equality over the long run.There seems to be ample scope for policies thatwould facilitate growth and labor mobility in

12


market incomes. Second, much of the heavy lifting ofequalizing incomes in the OECD countries appearsto have been done by their expansive transfer systemsthat dwarf anything found in the Latin Americanregion to date, although the report suggests that,here too, there are limits posed by political economyand efficiency. In short, policies designed to obtainequal opportunities for development of human capi-tal, and hence more equal market incomes, need to becomplemented with redistribution through taxes andtransfers.

3. Multiple dimensions of poverty, multiple channels togrowth. The narrowness of the traditional focus onincome poverty becomes increasingly unsatisfactoryin the context of tracing feedbacks to growth. Asexamples:• The strong gains in longevity in the region are

only weakly correlated with income growth. Insome countries where incomes have remained stag-nant, welfare has risen substantially because ofimprovements in health care and disease preven-tion. As noted above, health is linked to produc-tivity growth, and policies dedicated to redressingthis dimension of poverty are thus both pro-poorand pro-growth.

• The prospect of moving out of poverty or upwardin the income distribution is a major motivationfor the accumulation of human capital. However,the lower, late, and uncertain rates of return toeducation of the poor, for the reasons discussedabove, foreclose such mobility and discourageindividuals and their children from accumulatingthis capital. Clearly, one lesson is that redressingthese disincentives both improves social indicatorsthat more completely measure poverty and sti-mulates growth. But a second lesson is that anti-poverty policy must take a life-cycle view, withpolicies that look at the barriers to mobility in acomprehensive way.

• The risk associated with unanticipated mobility—high volatility in wages, for example—is also adisincentive to long-term investments in humancapital. Clearly, reducing the high macroeconomicvolatility of the region, as well as designing mech-anisms to mitigate the various types of risk—health or income, for example—reduces poverty inall its dimensions and has pro-growth impacts.

pove_001-020.qxd 2/2/06 12:08 PM

regions whose citizens have had particularly lowlevels of access to markets, education, and infra-structure. Yet, as discussed in the World Bank’sLatin American region flagship Beyond the City:The Rural Contribution to Development (de Ferrantiand others 2005), investing excessive state re-sources in some of these areas could lower overallaggregate growth, and thus governments mayeventually face a growth-equity dilemma. Even insuch cases, however, a smart combination of con-ditional cash transfers for the poor and paymentsfor environmental services can enhance both povertyreduction and long-term growth.

Policy implicationsThese considerations have important implications for spe-cific policies. The report does not offer universal recipes tobreak the vicious circle between low growth and poverty.For one thing, different countries will likely have differentpolicy priorities; policy makers in poorer and more equalcountries should focus mainly on growth, whereas those inricher and more unequal countries should try to balancegrowth-enhancing objectives with policies to reduceinequality. Nonetheless, the following examples emergefrom the report as illustrative.

Making growth more pro-poor There is no doubt that economic growth has to be at thecenter of the development strategies, and numerous studiesconducted by the Latin American Region of the WorldBank have explored constraints on growth that the regionfaces. For example, both the 2002 and 2003 World Bank’sLatin American region flagships (de Ferranti and others,2002, 2003) stressed the need to address the gaps in edu-cation (particularly secondary schooling) and innovation toget the most out of its existing endowments and to developdynamic new areas of comparative advantage. Similarly, theWorld Bank’s Latin American regional study The Limits ofStabilization: Infrastructure, Public Deficits, and Growth inLatin America (Easterly and Servén 2003) stressed how theregion’s wide gaps in infrastructure implied significant lostopportunities in growth and welfare.

This report offers suggestive evidence that investmentsin these areas have, in fact, been highly efficient in bothpromoting growth and allowing the poor to connect withthat process over the last 40 years, providing a classical“win-win” situation (see chapter 5). As Inequality in Latin

America: Breaking with History? (de Ferranti and others2004) showed, the poor were the primary beneficiaries ofefforts within the region in the 1990s to provide universalbasic education and health services and to expand somepublic services, such as access to safe water and electricity(that were already provided to rich and middle-incomegroups). Going forward, care must be taken to guaranteethat the poor continue to benefit from efforts to expandcoverage of secondary and tertiary education (which up tonow have benefited more middle- and high-income groups)and to improve educational quality. In the same vein,future investments in infrastructure must benefit laggardregions and increase the poor’s access to those serviceswhere past expansions primarily benefited rich and middle-income groups (telecommunications and access to theInternet, for example).

In addition, under a broad definition of poverty, twoother areas have the complementary potential to reducepoverty and promote growth. First, improvements inhealth have important impacts on welfare and demon-strated positive effects on growth. Second, the report pro-vides conceptual grounds for treating the income, health,and other risks that households face as a critical dimensionof poverty. The macroeconomic instability arising fromunsound policy therefore has a direct impact on the well-being of the poor and a documented adverse impact ongrowth.

There are, however, other pro-growth areas where LatinAmerica needs to make progress but where there may bepotential trade-offs with inequality and even with povertyreduction goals in the short run, according to the resultsdiscussed in chapter 5. Indeed, several previous studieshave found that trade openness (an area of particular rele-vance given potential liberalization efforts) may lead tohigher inequality through greater divergence of wageincomes.2 This result appears to be related to the verydesirable adoption of technologies that tend to be skillbiased and thus enhance the returns and the demand foreducation. This phenomenon, found globally, nonethelessleaves the poor, and often poor regions, behind in the shortrun. Chapter 5 argues that governments may need to takecomplementary policies behind the border—facilitatingaccess to education, expanding infrastructure to laggingareas with potential to tap into the benefits of liberaliza-tion, and providing conditional transfers for poor peasantswho may lose out in the transition. Such policies permit acountry to take full advantage of the opportunities brought

13


pove_001-020.qxd 2/2/06 12:08 PM

about by trade opening, and thus significantly mitigate theinequality effects and considerably enhance the growtheffects of trade liberalization. A parallel argument could bemade based on concerns that greater trade openness willincrease the risk that workers face. To date, little evidencehas emerged to suggest that this is true, but were it thecase, income support programs could mitigate the impacton poverty and the disincentive effects on human capitalaccumulation.

Although chapter 5 suggests that financial deepeningover the past 40 years appears to have had adverse impactson inequality and even on poverty in the short term, chap-ter 6 finds that it is precisely in countries with low access tofinancial services where poverty may become more of a dragfor investment and growth. Chapters 8 and 9 reinforce thisconclusion at the household level. Thus, even if past lim-ited advances in financial deepening in the region may haveleft most of the poor behind, it is essential that futureefforts guarantee that the poor gain access to both creditand insurance markets. Now that Latin America has appar-ently succeeded in achieving more resilient financial sectorsto avoid the costly crises of the past, extending access tocredit and insurance markets appears as a key policy agendato strengthen the virtuous circles between poverty reduc-tion and growth.

Another strand of the literature has explored the impacton poverty of the structure of growth. In particular this lit-erature argues that the higher the representation of sectorsthat use unskilled labor, the more the favorable effect onpoverty. Findings reported in chapter 5 give support to thisview. The potential conceptual conflict is that policies thatinduce a sectoral bias in growth may conflict in the longrun with pursuit of a country’s natural comparative advan-tage, leading to growth-impeding inefficiencies. Whilethis report does not delve deeply into the complex (country-specific) issues surrounding the sources of growth andinterlinkages across sectors or into the political economy ofgovernment intervention, the evidence provided here andin de Ferranti and others (2005) suggests that interventionsto induce strong sectoral biases are probably ill advised. Adifferent matter is to ensure that policy biases and ineffi-ciencies against rural development, for example, are liftedand that growth opportunities are enhanced by the efficientprovision of public goods and national and sectoral “inno-vation” policies. Incomes of the poor, including those fromagriculture and off-farm activities, thrive with higher tradeopenness, when public rural expenditures focus on the

provision of public goods (such as rural roads, health andeducation, research and development, and extension ser-vices) and when policy biases against labor mobility (suchas fiscal generosity for capital-intensive activities and stifflabor markets) are removed.

Nor does this report delve into policies to stimulatemore “labor intensity” within all sectors, apart from mak-ing sure that potential biases against labor use are removed.However, the previous discussion suggests that one wouldhave to carefully weigh the potential adverse effects on effi-ciency and growth of more “active” policies in this regardagainst potential short-term gains in poverty reduction.Given the potential short-term adverse effects of tradeopening on poverty and the negative effects of poverty ongrowth, an area of future research regards the desirability ofattempting to keep undervalued exchange rates in the earlyphases of trade opening, as long as inflationary pressures arekept at bay, as Chile did after 1984 and China is currentlypracticing.

Pro-poor government policyIn the end, the relatively young literature on pro-poorgrowth has not given us a feel for how much it is possibleto engineer growth in order to promote income distribu-tion. That the differences in the distributions of marketincomes between Latin American and OECD countriesexplain at most only 50 percent of differences in dispos-able incomes suggests the important complementary roleof taxes and public expenditures to ensure that the fruitsof growth are broadly distributed. Chapter 5 argues thatLatin America has made relatively modest use of thesetools. Although recent trends toward universal basic edu-cation and health and the introduction of targeted condi-tional transfers (among others) are likely to have had aprogressive impact on the distribution of income, manybig-ticket items continue to be highly regressive: the highsubsidies to pensions do not benefit the poor since they areseldom covered; since the poor seldom finish secondaryeducation, they do not benefit from subsidized universi-ties; gasoline, electricity, and other goods and servicessubsidized by the state are mostly consumed by the well-to-do.

Achieving a more redistributive and efficient pattern ofpublic expenditures similar to the OECD patterns wouldgreatly reduce poverty and inequality. However, given thecentrality of growth to the goal of poverty reduction, policymakers may wish to ensure that state efforts of such

14


pove_001-020.qxd 2/2/06 12:08 PM

magnitude have favorable effects on growth. Vehicles thatcondition cash transfers on the acquisition of human capitalcould be substantially expanded. The forthcoming WorldBank’s Latin American regional study The RedistributiveImpact of Transfers in Latin America and the Caribbean findsthat conditional cash transfers tend to be well targeted andmake a strong marginal contribution to social welfare, out-ranking not only social insurance schemes but also most ofthe existing social assistance programs. However, the cen-tral thesis of this report is that, in addition to conditionalcash transfers, there are numerous other areas where inter-ventions to aid the poor would also be pro-growth. Some ofthese interventions are reviewed in the next sections.

First, we should emphasize once more that the relativeweight of different instruments depends on initial condi-tions in individual countries. As mentioned above, poor(and more equal) countries should concentrate on achievingincreased growth, even at the expense of some increases ininequality, while middle-income countries with highinequality should aim for policies that achieve a better bal-ance of pro-growth and pro-poor effects (including redistri-bution through conditional transfers).

Pro-growth poverty reductionThe report presents some of the first empirical evidencethat poverty adversely affects growth at economywidelevels. As noted above, a central channel appears to workthrough underdeveloped financial sectors—more specifi-cally, through the poor’s lack of access to credit. This lackmay arise from institutional failures that make contractenforcement difficult and do not address the problems ofinformation asymmetries and the poor’s lack of collateraliz-able wealth. The search for efficient means and innovationsto overcome information asymmetries (including creditbureaus) and enforcement constraints and to convert thescarce wealth of the poor into collateralizable assets are keypriorities for policy and further research.

Addressing spatial concernsAll the concerns that could potentially lead to lower eco-nomic growth at the national level hold for low growth insubnational regions as well, and a case can be made for poli-cies analogous to those discussed above. Further, regionalinequalities correlated to ethnic, linguistic, or religious di-visions provide fertile ground for internal conflict that canundermine economywide growth. Yet in the world of the

New Economic Geography, the case for major reorientationof resources to disadvantaged zones becomes less clear, andthe literature to date has been very circumspect on policyprescriptions. Fundamentally, if the existing agglomera-tion externalities imply that those regions that are alreadymost advanced are also those with the highest potentialfor growth, concentrating all types of costly infrastructureinvestments on poor regions may decrease national growth.Unfortunately, the literature offers little guidance onwhether the externalities relative to agglomeration or thoseleading to dispersion of activity are more important, so wecannot know whether existing agglomerations are too bigor too small. However, as indicated in Beyond the City: TheRural Contribution to Development (de Ferranti and others2005), some policies targeted to rural areas, such asimproved rural education and access to communications,are clearly win-win solutions: they would increase produc-tivity in agriculture and other rural activities and at thesame time increase labor mobility toward more productiveactivities and toward richer areas with higher growthpotential.

A more subtle use of geographic information can atten-uate the potential trade-offs to some extent. In manycountries—the report looks specifically at Bolivia andBrazil—lagging regions frequently have the highestpoverty rates, but larger urban areas actually contain themost poor people. Therefore, the theoretical trade-offs, pro-viding existing agglomerations are not too large already,may be less important than initially thought: a large chunkof the poor are, in fact, in areas with potentially highergrowth. In addition to those advanced regions with nopoverty, three different spatial categories emerge thatimply distinct policies, some of which allow investment inpotential high-growth areas with large numbers of poorpeople.

• Areas with high poverty rates but low poverty density lackeconomies of scale arising from agglomeration exter-nalities and are unlikely to develop substantial eco-nomic dynamism. Policies thus need to focus moreon direct poverty alleviation and on programs thatwill impart skills useful in other more dynamicregions. Conditional cash transfer programs or othereducation and health initiatives, agricultural researchand development, and payments for environmentalservices would be most appropriate in these circum-stances (see de Ferranti and others 2005).

15


pove_001-020.qxd 2/2/06 12:08 PM

• In areas with low poverty rates but high poverty density,often urban or relatively dense rural areas whereagglomeration forces have already taken place, policiesaimed at fostering growth have a good chance of reach-ing the poor and translating into important povertyreductions. The major problem is to ensure thatwealthy groups do not capture the flow of resources.For this reason, self-targeting mechanisms, such asthose envisaged in the Argentine and Colombianworkfare programs, are particularly appropriate. Thatsaid, conditional cash transfer schemes, such as thosein Colombia and Mexico where targeting is quitegood, perform well in this type of situation.

• Areas with high poverty rates and high poverty densityhave the potential to take advantage of projects witheconomies of scale with low levels of leakage ofresources to the nonpoor. Infrastructure investmentssuch as rural roads may be a good example of the typeof projects for these kinds of areas.

From a practical point of view, the increasing use ofdetailed poverty maps to identify poor groups and targetpoverty policies may yield high dividends.

History suggests, however, that policy makers ofteneither judge that current agglomerations are too big orallow other considerations to lead them to resist abandoningentire regions to low levels of economic activity and exten-sive conditional cash transfer programs. In fact, as severalrecent World Bank reports have noted, Latin America hassubstantial experience with ambitious regional develop-ment programs that have met with mixed success. The nowvast OECD literature on the effects of public investmentpolicies generally finds a positive impact on growth andsometimes inequality, although, as the Spanish case sug-gests, they do not necessarily maximize national growth.The evidence for Latin America is thinner but generallyconcurs.

What should be emphasized, however, is that traditionalregional policy has not focused enough on the complemen-tary roles of human capital, knowledge transmission, inno-vation, and improved economic environments, all of whichconsistently emerge as correlated with differences inregional income.

Addressing household concernsCoordinated policies are needed to reverse the viciouscycles of poverty and low asset accumulation in the region.

One of the findings of the report in this area is that publicinvestments and policies in one area may have differentimpacts depending on the existing level of assets and otherinitial conditions affecting the poor. Ensuring that poorhouseholds have access to minimum bundles of assets (suchas education, health, or access to infrastructure) is essentialfor their capacity to exploit growth opportunities.

On the human capital front, demographic forces offermany countries in the region a unique opportunity totranslate the human capital accumulation of young cohortsinto a more productive labor force and faster reduction inpoverty. There is a need for integrated, long-term strate-gies for skills development that go beyond narrow educa-tional policies and exploit the synergies in the life-cyclehuman capital accumulation process in which bothfamilies and schools play a central role. This calls foractions to correct deficiencies in early-childhood develop-ment of poor children, strengthen degree completion andschooling transitions, upgrade education quality for thepoor, and improve the fluidity of labor markets. The mainspecific implications for human capital formation poli-cies are:

• Leveling the initial playing field for children at risk. Theunequalizing impact of deficiencies in early-childhood development and deficient parenting onpoor children’s educational attainment and returns toeducation as adults needs to be addressed. Almosthalf of the countries in the region are off track onmeeting the UN Millennium Development Goal ofhalving malnutrition by 2015. Early-childhoodinterventions and other policies that strengthen thecapacities of families to create early human capitalshould be given more attention. For example, condi-tional cash transfer programs should systematicallyincorporate health and nutritional components formothers and infants. The experience with the HeadStart program in the United States and similar inter-ventions elsewhere in the world may merit considera-tion for replication in the region.

• Strengthening the full option value of education for the poor.Education policies should aim to strengthen transi-tions to secondary school and enable opportunitiesfor tertiary education for the poor. While spendingand reform priorities must be set according to bind-ing constraints, acting at all levels of the educationsystem, even on a small scale, is crucial to signal

16


pove_001-020.qxd 2/2/06 12:08 PM

low-income families that their educational invest-ments have better chances of maturing in highergrades. Where returns are high and basic infrastruc-ture is deficient, the construction and upgrading ofschools and roads are of paramount importance. Thedevelopment of multigrade schools, learning frombest practices such as the Colombian Escuela Nuevaand the Chilean MECE Rural, can address supplyconstraints cost-effectively; when appropriate, public-private partnerships and other modalities such asdistance education should be considered. Schemes touse conditional cash transfers to the poor for encour-aging completion of full courses of education (basicor lower secondary) may hold promise to reducedropouts especially of children from poor familiesand parents with little education. Also needed arepolicies to promote the development of the tertiaryeducation market, including student loan programsand well-designed (means-tested and merit-based)university scholarships.

• Making education count for the poor. Increasing or level-ing the returns to educational investments of thepoor is key to encourage them to move up the educa-tion ladder. Well-informed actions to improve thescholastic performance of poor children are needed.These may include removing automatic promotionpolicies in early grades, offering special programs toaddress learning deficiencies resulting from a poorlearning environment at home, and addressing fail-ures in the instruction process such as inadequateteaching and large class sizes. Effective interventionsinclude decentralizing school management to getparents more involved in their children’s schoolprogress, offering incentives to encourage qualifiedteachers to work in disadvantaged schools, adaptinginnovations to improve learning environments indisadvantaged schools and communities, upgradingtextbooks and school aids, providing teacher train-ing, expanding computer education in secondaryschools, and consistently using international stan-dardized tests to assess performance progress. Sometargeted and performance-based increases in publicexpenditures, particularly at the secondary level,might be needed in some countries.

Finally, chapter 9 shows that the higher levels of labormarket risk found in the region have strong disincentive

effects on the accumulation of human capital that, in turn,slow down growth. Income security policies, such as unem-ployment insurance, workfare programs, or conditionalcash transfers as used in Colombia, therefore become bothpro-poor and pro-growth. Policies to improve access to jobsmay be needed that include enacting and enforcing antidis-crimination laws and establishing labor market intermedi-ation services that help well-educated ethnic and racialpopulations gain greater access to better-quality jobs.

Some of the best policies from a social cost-benefit calcu-lation, such as early-childhood interventions and overhaulsof the educational system, may be complex to implementfor reasons of political economy. However, considering thepositive spillovers on technology adoption, productivity,and growth from a labor force with a minimum level ofeducation, it is hard to overstate the critical importance ofovercoming political failures that prevent pushing “educa-tion for all” (see de Ferranti and others 2003). This is criti-cal to the region’s long-term human capital accumulationand prospects for sustained growth. In many countries, thedemographic window of opportunity is closing; the time toinvest is now.

Bridging the gaps in both the quantity and quality ofeducation and other productive characteristics of workerscan go a long way toward reducing the wide earnings dis-parities in the region, but it will not be enough to reducepoverty significantly. In most countries, low levels of laborproductivity are a chief constraint to earnings potential.Policies that promote an economic and institutional envi-ronment conducive to productivity growth are thus impor-tant to reduce the incidence of low-paid jobs and in turnmake investments in skills more attractive.

For example, rural investments seem to correlate posi-tively with rural household characteristics, indicating a needto increase access to markets through expansion of basicinfrastructure while simultaneously strengthening thecapacity of households to ensure a minimum level of wealthand education skills.

Rural development could be made more inclusive withsome minimum coordination of rural investments andprograms—such as education, the construction of roads tomarkets, the establishment of microcredit schemes, and theprovision of agricultural extension—to ensure that allthe potential returns to these investments are realized andthe conditions of the rural poor improved. A minimumcoordination of public interventions in poor areas can helpexploit synergies and overcome the associated potential

17


pove_001-020.qxd 2/2/06 12:08 PM

poverty traps that may affect households with a bundling ofunfavorable characteristics.

How are we going to pay for these interventions?This report offers a relatively large number of areas thatmay require additional attention if the vicious circlebetween growth and poverty is to be converted into a virtu-ous circle. For example, it urges that the levels of humancapital and public infrastructure in the region be expanded,in particular by increasing the poor’s access to quality edu-cation and infrastructure. Similarly, it argues that anexpansion of conditional cash transfer programs (especiallyin richer countries) would likely have a sustained impact onpoverty reduction and growth. But what are the real possi-bilities the region has for financing these interventions,which in some cases can be quite expensive?

It is crucial that policy makers step up efforts towardimproving the efficiency of the system and achieving bettertargeting before they increase public spending. For exam-ple, as noted in chapter 5, a number of big-ticket itemssuch as tertiary education are highly regressive. Moreover,many public transfer programs such as pensions or unem-ployment insurance are typically poorly targeted and donot reach many of the poor. Policy makers are likely to face

a trade-off between targeting and coverage: the greater thenumber of poor covered by a program, the more difficult itis to avoid leakages. A careful review of existing social pro-grams, however, can result in significant savings that maybe redirected to priority areas. Even more important,although they would require politically difficult reforms,highly regressive subsidies—of pensions for the well-to-do,of university students from wealthy families or who payback educational credits, and of the consumption of energyby the middle class and the rich—offer huge opportunitiesto reallocate expenditures.

Once these potential gains have been tapped, and onceefforts to curtail tax evasion have been stepped up, policymakers can consider increasing tax rates. In this regard,chapter 5 argues that most countries in the region (with afew exceptions such as Brazil and Nicaragua) have tax col-lections that are below what would be expected from theirper capita income (figure 1.12). This, too, is a window ofopportunity because bringing Latin America in line withthe international experience in tax collections would allowsome extra space to finance part of the expenditure priori-ties of the region. One related issue discussed in chapter 5is that countries aiming at increasing tax collections shouldavoid, to the extent that it is possible, tax structures with

18



Log, per capita GDP

FIGURE 1.12

Total tax revenue versus per capita income, throughout the world

United States

Spain

ItalyFrance

Uruguay

Estonia

Brazil

NicaraguaMexicoHonduras

Guatemala

El Salvador

Dominican Rep.

Costa Rica

Colombia Paraguay

Peru

Chile

Bolivia

Argentina

0

5

10

15

20

25

30

35

40

45

Total tax revenue (% GDP)

4.5 5.5 6.5 7.5 8.5 9.5 10.5 11.5

LAC Selected countries throughout the world

pove_001-020.qxd 2/2/06 12:08 PM

high efficiency costs. Latin American countries tend to haveespecially low levels of collections from personal incomeand property taxes—the very taxes that may have someredistributive effect without large costs to economicgrowth. Thus well-designed systems could increase tax col-lections while keeping the impact on growth low. Also, theregion’s value added and income tax productivity is signifi-cantly lower than it is in the OECD countries, and mostLatin American countries maintain a large set of exemp-tions that significantly reduce the tax base. Thus the elimi-nation of exemptions combined with additional efforts toenforce compliance would likely increase collections.

Converting the state into an agent that promotes equal-ity of opportunities and practices efficient redistribution is,perhaps, the most critical challenge Latin America faces inimplementing better policies that simultaneously stimu-late growth and reduce inequality and poverty.

Notes1. The Gini coefficient is a standard measure of inequality that

ranges between 0 and 1. A value of 0 would indicate a perfectly equaldistribution. As inequality increases, the Gini coefficient also tendsto increase.

2. See, for example, de Ferranti and others (2003); Lederman,Maloney, and Servén (2005); and World Bank (2005c).

19


pove_001-020.qxd 2/2/06 12:08 PM

pove_001-020.qxd 2/2/06 12:08 PM

CHAPTER 1 From Vicious to Virtuous Circles - World...

Documents

Transcript of CHAPTER 1 From Vicious to Virtuous Circles - World...