76979 THE WORLD BANK Inequality in Focus The State of Opportunities in South Africa: Inequality among Children and in the Labor Market Ambar Narayan and Sandeep Mahajan Figure 1 D-Indices and HOIs for Key S outh Africa, the African continent’s largest economy Opportunities of South African Children, 2010 by far, displays strikingly high and persistent inequal- ity for an upper middle-income country. While GDP growth has averaged a credible 3.2 percent a year since 1995 (1.6 percent per capita), it has been highly uneven in its distribution.1 In large part, inequality in South Africa is an enduring legacy of the apartheid system, which denied the non-whites the chance to accumulate capital in any form— land, finance, skills, education, or social networks. Even though social assistance grants have had a moderating influence on inequality, an income Gini of about 0.70 in 2008 makes South Africa one of the most unequal countries in the world.2 The high and persistent patterns of inequality predictably polarize the political and economic debate in the country, with consensus being especially hard to achieve on the questions of Note: IOO refers to inequality of opportunity, as measured by D-Index. what notions of equality (or equity) should guide policy, and Source: SAEU3, based on General Household Survey (2010). how? Consensus, however, is easier to reach on the need to promote equality of opportunity—the principle that predeter- children’s access to basic services in education, health care, and mined circumstances such as gender, ethnicity, and location of essential infrastructure, which provide an individual the op- birth or family origins should not play a role in determining an portunity to advance and reach his or her human potential. How individual’s chance of success in life. 3 The simple yet powerful universal the basic services are—and the extent to which avail- idea of “leveling the playing field for all� tends to find accep- ability is influenced by a child’s circumstances—are important tance across the ideological spectrum in most countries, and (if imperfect) predictors of future outcomes, including inequal- South Africa is no exception. ity in earnings and of economic mobility within and across Equality of opportunity is the lens through which the generations. Given the important role played by labor markets World Bank’s recent South Africa Economic Update Volume in driving income inequality, the report also examined how 3 (SAEU3) has examined inequality. This includes a focus on individual attributes influence inequality in employment status. This article is based on Section 2 of South Africa Economic Update Opportunities among Children: (Issue 3): Focus on Inequality of Opportunity, World Bank (July 2012). The report was produced by a team led by Sandeep Maha- A Story of Mixed Progress jan and Fernando Im, with Allen Dennis, Sailesh Tiwari, Alejandro Access to a set of basic goods and services is taken as a Hoyos Suarez, Shabana Mitra, Phindile Ngwenya, and Ambar Na- proxy for opportunities among South African children. rayan, who was a special guest coauthor for the focus topic. The metric chosen is the well-known Human Opportunity Poverty Reduction and Equity Department : : www.worldbank.org/poverty : : Volume 2, Number 1 : : April 2013 Index (HOI), which is the coverage rate of a good or service Figure 3 Contribution of Circumstances discounted by how equitably it is distributed among groups to Inequality of Opportunity, 2010 with different circumstances, such as gender, race, and family 100 50% background.4 The HOI runs from zero to 100; a higher number indicates greater opportunity.5 Embedded within HOI is a Marginal Contributions (%) 80 40% “dissimilarity� index (or D-Index) that is a metric for inequal- 60 30% D-Index ity of opportunity, with a higher value indicating greater inequality.6 40 20% For South Africa, the circumstances considered are per- 20 10% sonal and family-related: gender and ethnicity of the child; 0 0% household composition;7 education, gender, and age of the ECD exposure Finish primary Access to improved Have health insurance household head; orphan status (whether both parents are sanitation alive); and location of the household (urban townships and Gender Household composition Ethnicity Orphan status informal settlements, other urban areas, or rural areas). Op- Education of the household head Other household head characteristics Location D-Index (%) portunities include exposure to early childhood development Note: Results of Shapley decompositions of D-Index, shown only for (ECD) programs (ages 0–4 years), school enrollment (ages 6–11 opportunities where D-Index > 0.5. and 12–15), completion of primary school (ages 13–15), adequacy Source: World Bank staff calculations in SAEU3, based on General of school infrastructure and teachers (as reported by parents), Household Survey (2010). and having health insurance. Completion of primary school and parents’ perceptions of schools are proxies for education Some opportunities, like school attendance and access quality, used in lieu of measures of student learning achieve- to telecommunications, are nearly universal (HOI above 90 ment that are not available. Exposure to ECD programs is percent) among South African children. Other opportuni- a proxy for cognitive inputs early in a child’s development ties—health insurance, access to safe water and improved process.8 Having health insurance is an indirect measure of sanitation, adequate space without overcrowding, and finish- access to quality health services in the South African con- ing primary school— are inadequate and distributed with high inequality among children of different circumstances. text.9 Other opportunities include access to safe water on site Still other opportunities—access to ECD programs, neighbor- and improved sanitation, which are known to reduce the risk hood safety, and access to electricity—are below universal, of diseases that are the leading cause of undernourishment but have low to moderate inequality of opportunity (figure 1). in children.10 Access to electricity and telecommunications, Accordingly, South Africa fares well in international com- lack of overcrowding in the household, and the opportunity parisons on HOI for school attendance. But for completion to grow up in a safe neighborhood, for a number of different of primary school on time, which is related more closely to reasons, are likely to improve a child’s cognitive and academic quality of education, South Africa is surpassed by most of its development.11 Latin American peers (figure 2). On access to safe water and Figure 2 Finishing Primary School on Time, improved sanitation, South Africa, though ahead of other South Africa and Other Countries African countries, lags behind all except the poorest among 100 30 Latin American countries (e.g. El Salvador and Honduras). 25 South Africa made significant progress between 2002 and 80 2010 in providing access to telecommunications, and to a 20 lesser extent in sanitation, adequate infrastructure in school, D-Index (%) 60 HOI (%) 15 40 and electricity. Most of the gains in HOI were driven by a 10 general expansion of service coverage for children of all cir- 20 5 cumstances. But for improvements in sanitation and telecom- 0 0 munications, reduction in inequality also played a key role, with opportunities improving more than proportionately for B. (2 ) nt (2 ) m 20 ) ex (2 ) 0) ug (2 5) 2) El ra l (2 6) Ar Ch (20 8) n du (2 5) Pa ubl (200 ) in Ho vad (20 ) Co Ric (20 ) ua (2 8) N ny ia ( 08) bw 00 9) Co rag c (2 6) m (2 ) ) a 08) Ke iger (20 ) el n ( 9) 3) m ia 08 ) 01 ) R. a 08 lo a ( 08 ge ile 06 M ina 006 7 l a 8 8 Ja ico 008 11 N ana 007 (2 7 Zi mib (20 008 Ur Peru (200 ica n or 0 00 Re ra 00 de 00 01 e 6-0 do 00 ba (2 -0 ica zi 0 Sa gu 00 st ay 0 Ec ay 00 zu Pa bia 0 0- a, am 20 N Bra (20 0 aic 0 (2 (2 2 a a 2 Gh a ( a r ala p s underserved groups. ric bi i u u m m Af e Za a at h ut Gu In most cases where inequality of opportunity is moderate So ne om Ve or high, the most important circumstances are a child’s loca- D Finished primary on time HOI Finished primary on time D-Index Source: SAEU3, using General Household Surveys (2002 and 2010) for tion and the education of the head of the child’s household South Africa; national household surveys for Latin America and the (figure 3). Location is particularly important for opportunities Caribbean countries; and DHS for African countries. related to infrastructure; education of the household head 2 : : April 2013 : : Inequality in Focus Figure 4 Contrasting Opportunities of Three (Hypothetical) Children, 2010 Probability of access to the opportunity 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 School attendance (12-15 yr) School attendance (6-11 yr) Access to telecomunications Finish primary school on time ECD exposure Safe location Access to electricity No overcrowding Access to health insurance Access to improved sanitation Access to safe water on site Andries: white boy, in urban (non-township) area, in a household headed by someone with secondary education Thandiwe: black African girl, in a township, in a household headed by someone with no education Nothando: black African girl, in a rural area, in a household headed by someone with no education Source: SAEU3, based on General Household Survey (2010). is the most important for finishing primary school on time Both Thandiwe and Nothando’s chances of attending and having health insurance, underscoring how the family’s school up to the age of 15 are just as good as Andries’s. But socioeconomic background holds a lock on children’s future. compared with Andries, they have a lower chance of finishing Household size matters the most for the opportunity to grow primary school by age 15, having exposure to an ECD pro- up in a house with no overcrowding. Interestingly, the child’s gram, health insurance, and access to basic infrastructure, and gender contributes appreciably to inequality only for finishing living in a safe environment with adequate space. Thandiwe primary school on time. Ethnicity contributes to inequality, and Nothando’s opportunities also vary relative to each other. but does not rank among the top two contributors for any For access to safe water on site, improved sanitation, and elec- opportunity.12 This seems to suggest that the contributions of tricity, Thandiwe has a big advantage over Nothando, reflect- race and gender to inequality of opportunity among children ing the inadequacy of these services in rural areas. For primary are intertwined with that of family background and location. school completion and access to ECD and telecommunica- Thus many of the apparent racial and gender gaps in oppor- tions, the gaps between the two children are smaller. Living tunities for children in South Africa today can be narrowed if in a township, however, puts Thandiwe at a disadvantage in opportunities could be equalized across groups differentiated terms of a safe environment and adequate living space. by socioeconomic status and, above all, location. To what extent key circumstances matter, and for which Inequality in Access to Jobs, opportunities, can be illustrated through an example of three and How Circumstances Matter imaginary children with contrasting profiles and their likeli- Lack of access to employment is a driver of income inequal- hood of access to different basic services (figure 4). In this ity and an obstacle to economic mobility. This is especially hypothetical example, Thandiwe and Nothando are black true in South Africa, where the narrowest measure of unem- African girls, growing up in households headed by individuals ployment stood at 25 percent in 2012,13 with nearly 70 percent with no education, with the difference that Thandiwe lives of the bottom income quintile being unemployed in 2008. In- in an urban township and Nothando in a village. Andries is a equality in employment, and the extent to which it is related white boy, living with a household head who has secondary to circumstances, is thus an important concern. education, in an urban area that is not a township or informal Two definitions of employment were used in SAEU3: an settlement. adult of working age (ages 15–64) is considered to be “em- Inequality in Focus : : April 2013 : : 3 ployed� if s/he has any job (as opposed to being unemployed Figure 5 IAC and Inequality between Groups or discouraged from seeking employment), or “employed Differentiated by Circumstances, full-time� (as opposed to being unemployed, underemployed, Education, and Age or discouraged). The HOI methodology is used to estimate 100% 16% an “inequality-adjusted coverage rate� (IAC), which dis- counts average employment rate by inequality in employment 80% between groups of working-age adults differentiated by two 12% types of attributes: circumstances that a worker has little or no 60% D-Index control over (gender, ethnicity, and location) and characteris- IAC 8% tics (education level and age of the worker).14 The contribution 40% of circumstances to overall inequality indicates the extent of 4% inequality of opportunity in the labor market—the part of 20% inequality that is not explained by the individual’s education 0% 0% or experience, but by attributes the individual has no control Employed Employed full-time over. 2008 IAC 2012 IAC IAC for being employed fell between the first quarters of 2008 and 2012, reflecting lower availability of jobs due to the 2008 D-index 2012 D-index global crisis as well as a rise in inequality between groups Source: SAEU3, based on Quarterly Labour Force Surveys (2008q1 and (figure 5). Even as IAC for being employed full-time rose 2012q1). slightly over the period, so did inequality between groups. Figure 6 D-Index and Inequality-Adjusted Thus inequality between groups has risen in employment and Coverage by Employment, circa 2008 full-time employment, while the adverse impacts of the global 100% 15% crisis are seen more on part-time than full-time employment. 80% When compared with 17 middle-income countries around 10% 2008, South Africa—ranked 13th out of 18 countries by per 60% D-index IAC capita GDP – was ranked the lowest by IAC and highest by 40% 5% between-group inequality in having a job (figure 6). The low 20% IAC in South Africa is explained not only by too few jobs 0% 0% (high unemployment) but also by higher inequality in how Turkey Kazakhstan Chile Russia Ukraine Serbia Peru Dominican Rep Brazil Argentina Romania Ecuador South Africa Bulgaria Colombia Venezuela Belarus Mexico the available jobs are distributed among workers of different attributes. More than half the between-group inequality (in being IAC D-Index employed or employed full-time) in South Africa is driven by Note: Computed using similar but not identical definitions of differences in education and age. The remaining inequality circumstances across countries. is attributable to circumstances, of which location appears to Source: SAEU3, based on Latino Barometro (2008) for Latin America contribute the most, followed by ethnicity and gender (figure and the Caribbean countries; Life in Transition Surveys (2006) for Europe and Central Asian countries; and Quarterly Labour Force 7). The chances of being employed or employed full-time for Survey (2008q1) for South Africa. a township and informal settlement resident or especially a rural resident are much lower than those of a resident of other of inequality in employment status, has increased in impor- urban areas, and the gap is larger in 2012 than it was in 2008. tance from 2008 to 2012.15 While this is a trend toward a fairer Being a woman and non-white increases the likelihood of be- labor market, it also implies that disadvantages conferred by ing unemployed or underemployed significantly, even as these unequal opportunities in education earlier in life are increas- circumstances have declined in importance in the past four years. Importantly, circumstances related to parental socioeco- ingly important obstacles to an individual’s economic mobil- nomic background, likely to be correlated with circumstances ity. More calculations show that for formal-sector employ- such as location and ethnicity, are missing due to a lack of ment outside agriculture, a proxy for quality jobs, education data. Given this caveat, the contributions of race and location accounts for a much higher and increasing share of inequality. are best interpreted as reflecting socioeconomic factors (in- This appears to be consistent with other literature that finds cluding, but not limited to, race and location) that contribute high (and rising) skills premium to be a key driver of income to inequality in employment opportunities. inequality in South Africa.16 Some literature also suggest Education, which contributes between 15 and 20 percent sharp disparities in wage earnings among those with jobs, by 4 : : April 2013 : : Inequality in Focus Figure 7 How Circumstances analysis shows that inequality of opportunity—the part of Contribute to Inequality inequality attributable to gender, ethnicity and location—is 30% 15% much higher among workers ages 15-29 than among older workers.18 Thus not only do young workers face a disad- % contribution to D-Index vantage in the labor market, they also compete for jobs in a 20% 10% market seemingly more “unfair� in allocating opportunities D-Index among the young, relative to what is seen for older workers. 10% 5% Final Thoughts Equality of opportunity provides a powerful guiding princi- 0% 0% ple for policy on which political consensus is easier to achieve. 2008 2012 2008 2012 As with any objective, an intuitive and objective measure of Employed Employed full-time progress is crucial—a need that is in part filled by the human Gender Ethnicity Location D-Index opportunity index for South Africa. Analysis using this index Note: Each contribution is expressed as % of total value of D-Index. shows that extraneous circumstances that a South African Source: SAEU3, based on QLFS (2008q1 and 2012q1). child is born into (ethnicity, location, gender, and family race, gender, location, and union membership—an aspect background) variably affect the child’s access to basic oppor- of group-based inequality that our measures do not capture tunities. Some of these circumstances (such as location and because they are based on a binary measure of employment. ethnicity) are also important for inequality in employment op- While labor markets typically reward experience, the extent portunities later in life, raising the prospect of a persistent role to which age contributes to inequality in employment status of adverse circumstances in perpetuating inequalities over in South Africa is unusually large, relative to its middle- successive stages in life. A child born into disadvantageous income comparators.17 When computed separately for each circumstances has to work harder to overcome these in order age group, IAC increases and inequality declines progressively to attain her human potential; having done so, she may find with age of the workers (figure 8). The trends also seem to be that disadvantages reemerge when she enters the job market. against young workers—the 15-24 age group is the only group And the disadvantages do not necessarily end in a generation, for which IAC has fallen and inequality risen for both employ- but are likely to be carried over to the next. ment and full-time employment between 2008 and 2012. More While there are no simple, elegant policy solutions in the Figure 8 Inequality-Adjusted Coverage and D-Index for Workers of Different Age Groups, 2008 and 2012 100% 20% 80% 16% 60% 12% D-Index IAC 40% 8% 20% 4% 0% 0% 40-49 years 40-49 years 40-49 years 40-49 years 50-65 years 50-65 years 50-65 years 50-65 years 30-39 years 30-39 years 30-39 years 30-39 years 25-29 years 25-29 years 25-29 years 25-29 years 15-24 years 15-24 years 15-24 years 15-24 years 2008 2012 2008 2012 Employed Employed full-time IAC D-Index Note: Circumstances/characteristics are gender, ethnicity, education, location, and age. Source: SAEU3, based on Quarterly Labour Force Surveys (2008q1 and 2012q1). Inequality in Focus : : April 2013 : : 5 quest for equity, a few broad principles seem to be relevant for service in one society is determined to a greater extent by South Africa. To be sure, that would involve leveling the play- personal circumstances beyond their control. ing field in the quality of education children get and the em- 6 The relationship between HOI and D-Index can be ployment opportunities they face as young adults, irrespective expressed as HOI = C(1-D); where C: coverage rate of of location, gender, or ethnicity. It would also be important the good or service, and D: dissimilarity index. D is also to pay special attention to the water, sanitation, and health- equivalent to the ratio of “penalty� due to inequality to the care needs of rural areas and townships, and overcrowding in coverage rate. Intuitively, D measures the share of available townships. Academic research has found interventions that opportunities that needs to be reallocated across circum- equalize opportunities earlier in life to be much more cost- stance groups in order to achieve equality of opportunity. effective and successful than those later in life. Policy design 7 Household composition includes presence of the spouse of also needs to recognize that children of certain circumstances the household head in the household, total number of chil- are vulnerable to deprivations in multiple dimensions simul- dren ages 0–16 in the household, and whether both parents taneously. For example, black South Africans living in rural live in the household. areas, and with household heads lacking education, are much more likely to not complete primary school, and not to be cov- 8 See, for example, Chetty and others (2010) for evidence ered by an ECD program and health insurance. Policy inter- that early childhood education has substantial long-term ventions in different sectors thus need to coordinate closely in impacts, ranging from adult earnings to retirement savings. order to achieve better efficiency and results. 9 Ideally, opportunities would also include indicators of child health such as nutritional outcomes, access to preventive Notes care, and access to maternal care services, which are absent 1 In 2008, the top decile of the population accounted for in the dataset for this analysis. 58 percent of the country’s income while the bottom half 10 Child malnutrition has been shown to generate life-long accounted for less than 8 percent (Leibbrandt and others, learning difficulties, poor health, and lower productivity 2010). and earnings over a lifetime (Alderman and others 2001; 2 Noncontributory and means-tested (except for foster care) Hoddinott and others 2008). financial transfers from the budget account for more than 11 See, for example, Gove and others (1979) for the adverse 70 percent of the income of the bottom quintile, up from 15 percent in 1993 and 29 percent in 2000. In the absence of impacts of overcrowding. social assistance, estimated average incomes of the bottom 12 The limited role of gender in explaining inequality in ac- four deciles of the population would have actually fallen cessing infrastructure facilities is driven by the fact that between 1995 and 2005. Bhorat and van der Westhuizen access is measured at the household level (and not the (2011) find that in 2005 income Gini would increase from individual child’s). 0.72 (including social assistance grants) to 0.77 (without the 13 Unemployment rate is 25 percent in 2012 (quarter 1), and grant incomes). 34 percent when discouraged workers are included in a 3 While the concept of equality of opportunity—a subject broader measure (SAEU3, page 7). of growing research among academics and at the World 14 Unlike gender and ethnicity, location is somewhat debat- Bank—draws from a large volume of literature, the econo- able as a circumstance because working-age individuals mist John Roemer’s 1998 book Equality of Opportunity was have some control over where they live. In practice, how- the first to formalize an equality of opportunity principle. ever, spatial mobility is restricted by cost considerations, 4 Equality of opportunity was highlighted by the 2006 World (lack of) social networks, and cultural and family ties. In Development Report on Equity and Development. Recent South Africa particularly, mobility out of rural areas and work by researchers has provided various measures of townships and informal settlements is fairly restricted be- inequality of opportunity in countries (for example, see cause of historical reasons that affect the poorest more than Ferreira and Gignoux, 2011). The Human Opportunity others. Index (HOI) was developed by a group of World Bank staff 15 Inequality of opportunity in employment would be even and external researchers. First applied in Latin American higher than what is estimated here, if the indirect effect of countries in 2009, it has now been computed for a grow- circumstances on employment through education were to ing list of countries around the world. For a description of be taken into account. The estimate here of inequality pro- HOI, see Barros and others (2009, 2010, and 2012). duced in the labor market does not take into account the 5 Thus two societies with the same coverage rate for any unequal distribution of opportunities earlier in life (such as service can have different HOIs if citizens’ access to that timely completion of schooling or ECD exposure) that mat- 6 : : April 2013 : : Inequality in Focus ter for human capital formation and are also affected by the tions?� DPRU Working Paper 09/138. University of Cape same circumstances. Town, Development Policy Research Unit, Cape Town, 16 See, for example, Bhorat and others (2009). South Africa. 17 The average contribution of age to between-group inequal- Chetty, R., J. Friedman, N. Hilger, E. Saez, D. Schanzenbach, ity in employment was 42 percent in 2008, compared to and D. Yagan. 2010. “How Does Your Kindergarten Class- 19 percent for the 17 middle-income countries included in room Affect Your Earnings? Evidence from Project STAR.� figure 7. NBER Working Paper 16381, National Bureau of Economic Research, Cambridge, MA. 18 Circumstances contribute around 70 percent of the D-Indices in being employed or employed full-time for Ferreira, F., and J. Gignoux. 2011. “The Measurement of those between 15 and 29 years, compared to less than 60 Inequality of Opportunity: Theory and an Application to percent for those between 30 and 65 years of age. Latin America.� Review of Income and Wealth 57 (4): 622–57. Gove, W., M. Hughes, and O. Galle. 1979. “Overcrowding References in the Home: an Empirical Investigation of its Possible Alderman, H., J. Behrman, V. Lavy, and R. Menon. 2001. Pathological Consequences.� American Sociological Review “Child Health and School Enrollment: A Longitudinal 44 (February): 59–80. Analysis.� The Journal of Human Resources 36 (1): 185–205. Hoddinott, J., J. Maluccio, J. Behrman, R. Flores, and R. Mar- Barros, R, F. Ferreira, J. Molinas Vega, and J. Saavedra. 2009. torell. 2008. “The Impact of Nutrition during Early Child- Measuring Inequality of Opportunities in Latin American and hood on Income, Hours Worked, and Wages of Guatema- the Caribbean. Washington, DC: World Bank. lan Adults.� The Lancet 371 (February): 411–16. Barros, R., J. R. Molinas Vega, and J. Saavedra. 2010. “Measur- Leibbrandt, M., I. Woolard, H. McEwen, and C. Koep. 2010. ing Progress Toward Basic Opportunities for All.� Brazilian “Employment and Inequality Outcomes in South Africa.� Review of Econometrics 30 (2). Southern Africa Labour and Development Research Unit and School of Economics, University of Cape Town. Bhorat, H., and C. van der Westhuizen. 2011. “Pro-poor Growth and Social Protection in South Africa: Exploring Roemer, J. 1998. Equality of Opportunity. Cambridge, MA: Har- the Interactions.� Input paper prepared for the National vard University Press. Planning Commission. World Bank. 2005. World Development Report 2006: Equity and Bhorat, H., C. van der Westhuizen, and T. Jacobs. 2009. “In- Development. Washington, DC. come and Non-Income Inequality in Post-Apartheid South World Bank. 2012. South Africa Economic Update (Issue 3, July Africa: What are the Drivers and Possible Policy Interven- 2012): Focus on Inequality of Opportunity. Washington, DC. Climate Change and Inequality: What Can Current Patterns of Food Insecurity Tell Us? Nancy McCarthy, Leslie Lipper, and Maximillian Ashwill maintaining a livelihood. According to the Food and Agricul- M uch of the discussion about climate change ture Organization (FAO) of the UN (2012), in 2010-2012 nearly and inequality focuses on how the greenhouse 870 million people were chronically undernourished, with 850 gas emissions of wealthy nations contribute to million of those living in developing countries. That is almost 15 the negative climate impacts that poor nations percent of the population of poor countries. FAO (2008) shows face. The discussions center on how devel- that climate change is having an impact on all dimensions of oped countries like the United States and quickly developing food security, with changing food prices playing a major role, as countries like India and China can reduce their emissions, or highlighted in figure 1. at least compensate poorer countries that emit low amounts of This leads to important questions regarding climate change, greenhouse gas but face the most perilous impacts. food security, poverty, and inequality. How will climate change To a poor farmer in the highlands of Bolivia or a fisher in the affect the food security of poor households in rural and urban Maldives, these discussions are light years away from the real areas? How will climate-change impacts on poverty and food problem. These individuals must contend with the everyday security affect poverty and inequality? consequences of climate impacts, such as feeding a family and Evidence discussed below suggests that the food security of Inequality in Focus : : April 2013 : : 7 the urban poor will deteriorate as food prices rise. Because poor Figure 1 The Rising Food Price Index people in urban environments spend a large share of their bud- gets on food, they will be hit disproportionately by relatively higher food prices, leading to both higher poverty rates and increased inequality in urban areas. In rural areas, the impacts will be more complex. Some producers might benefit from higher food prices by maintain- ing or expanding output despite higher temperatures and more erratic climate. Rural laborers might also benefit if the overall agricultural economy expands. Other farmers, however, might not be able to expand production; in fact, output might well fall despite price increases. In this case, rural poverty will increase. Crucially, given thin or nonexistent insurance markets, farm- ers will need to be able to self-insure against increased climate risks. Because wealthier farmers are better placed to self-insure, inequality in rural areas might well increase. In developing countries, overall impacts on poverty and inequality will be largely driven by the rural sector because a large share of the Note: The Food Price Index consists of the average of five population lives in rural areas, as highlighted in figure 2. commodity group price indices (meat, dairy, cereals, oils and fats, and sugar) weighted with the average export shares of each of the Impacts of Climate Change groups for 2002–2004. Source: FAO 2013b. on Food Production and Prices Global weather patterns are changing. The planet is warming mate change, and competing demands for agricultural lands (for and as a consequence environmental change is happening more example, for biofuel production) have caused prices to reach quickly. According to the Intergovernmental Panel on Climate near all-time highs (see figure 1) (World Bank 2013). Change (IPCC), floods, droughts, and other weather-related Climate change is also affecting the ability of some lands shocks are predicted to become more severe and more frequent. to grow food. Higher temperatures, increased water scarcity, Certain parts of the globe are becoming wetter and others are be- floods, rising soil salinity, and extreme weather all can dimin- coming drier. In some cases, formerly productive lands are now barren. The Figure 2 Location of the World’s Rural and Agricultural Populations weather is becoming more variable and seasons are becoming more dif- ficult to predict (IPCC 2012). All of this can have negative impacts on the world’s food producers. In turn, climate change is hav- ing an impact on food security. All components of the global food system, including food markets and prices, stand to be affected. Food processing, production, storage, and transport can be affected through increased disruptions and costs associated with productive land, energy, and water. Since 1970, ris- ing incomes for the world’s poor, combined with falling food (real) prices have helped make much of the global population more food secure (Schmidhuber and Tubiello 2007). But population growth, cli- Source: FAO 2013a. 8 : : April 2013 : : Inequality in Focus are expected to affect availability (food supply), access (food The Four Dimensions of Food Security prices as well as agricultural incomes), and utilization (spoilage, Food availability: The availability of sufficient quantities human health). of food of appropriate quality, supplied through domestic Within developing countries, the impacts of climate change production or imports (including food aid). on food security are likely to be unequally distributed as well. Food access: Access by individuals to adequate resources The distribution of these impacts is largely determined by (entitlements) for acquiring appropriate foods for a nutritious two factors: first, by how much people depend on agricultural diet. Entitlements are defined as the set of all commodity production for their livelihoods and incomes, and second, by bundles over which a person can establish command given how much they depend on purchased food to maintain food the legal, political, economic, and social arrangements of the security. community in which they live (including traditional rights We have identified five distinct types of households that such as access to common resources). depend on these two factors in different ways. These types Utilization: Utilization of food through adequate diet, clean are: 1) net food-selling farm households, 2) self-sufficient farm water, sanitation and health care to reach a state of nu- households that neither purchase nor sell food to the market, tritional well-being where all physiological needs are met. 3) net food-purchasing farm households, 4) rural landless and This brings out the importance of non-food inputs in food non-farm rural households dependent on rural wages, and 5) security. poor urban households. These groups are presented in order Stability: To be food secure, a population, household, or indi- of those most likely to gain from higher food prices (as long as vidual must have access to adequate food at all times. They land productivity does not fall too much) to those most likely should not risk losing access to food as a consequence of to lose. sudden shocks (e.g. an economic or climatic crisis) or cyclical events (e.g. seasonal food insecurity). The concept of stabil- Food-Selling Farm Households ity can therefore refer to both the availability and access Results from an analysis of nine developing countries show dimensions of food security. that, on average, 23 percent of all households and 32 percent Source: FAO. of rural households were net food sellers (Aksoy and Isik- Dikmelik 2008). Like all farm households, net food sellers will ish land productivity. Some of this loss can be offset by the be vulnerable to reduced crop yields because of climate-change- increased use of previously unused or moderately productive induced reductions in farmland productivity, hence impacting lands, or by adopting different agronomic practices. Still, this availability. However, the net effects on access could be positive, is an alarming trend because climate change is projected to to the extent that food-selling households benefit from higher have a negative impact on agricultural systems located in poor, prices as long as negative impacts on productivity are not too low-latitude countries where agriculture is an important source severe (e.g. negative impacts on productivity outweighing posi- of livelihood (Fischer et al. 2002; Parry et al. 2004). For example, tive price effects). The degree to which households are integrat- Africa is projected to bear the largest loss in productive lands ed into markets and can reap the benefits of higher prices is an (Fischer et al. 2002). So, although crop production and yields important determinant of net effects. However, even if incomes continue to increase to meet global demand, these are growing remain constant, impacts on utilization could affect the food security of this group, though it is difficult to predict how cur- at a slower pace than in the past. According to FAO (2009), crop rent net sellers will alter their consumption basket in response yields over the next 50 years will grow at half the rate that they to changes in yields and agricultural prices. Food safety and have increased historically. food waste may become a more difficult issue because higher Impacts of Climate Change on Food Security temperatures and changing rainfall patterns require improved Generally, projections indicate that developing countries are on-farm storage (Vermueulen et al. 2012). And stability could be threatened by increased climate risk and uncertainty if the likely to suffer greater negative impacts on the various dimen- incentives to invest in farming are reduced (Dercon and Chris- sions of food security—in both the short and long term—al- tiaensen 2011; Hurley 2010). though some exceptions and considerable uncertainty still remain. In the short term, the increased frequency and inten- Self-Sufficient Farm Households sity of climate shocks are expected to directly affect stability, Self-sufficient households usually constitute a small share with indirect effects on the other three dimensions—avail- of the rural population, because most households engage in ability, access, and utilization (see box). In the long term, climate some form of market transactions (Karfakis et al. 2011). This change is projected to affect the entire food system. Changes in group includes subsistence farmers, herders, fishers, and forest- temperature, rainfall patterns, and pest and disease incidence dependent people who rely on their own production. As a Inequality in Focus : : April 2013 : : 9 result, they are vulnerable to production risks and the loss of change on the rural economy are negative, but as food prices natural resources (land, water, fish) from climate change. This increase, access for this group will significantly decline. can negatively affect availability, as well as access since their Reduced incomes can result in households switching to less own production is their sole means of access. The extent to nutritious and lower-quality foods, negatively affecting utiliza- which they are impacted is largely determined by the extent of tion. Their stability is vulnerable to the effects of climate shocks the climate impact on production levels. It is more likely that on rural incomes and food prices (FAO 2011). highly vulnerable households will have to reduce consump- Poor Urban Households tion or sell off assets in the event of a climate shock (Kazianga and Udry 2006; Skoufias and Quisumbing 2005). Wealthier Like the rural landless, the urban poor’s food security de- households will have a greater capacity to adapt and prepare for pends on relative changes in incomes and food prices. Gener- the impacts of climate change. Also, self-sufficient households ally, the livelihoods of urban families, and food availability, have limited access to markets, which can make them less able are less negatively affected by the weather than those of rural than others to purchase food, impacting access. Like other farm families. However, urban consumers are especially vulnerable households, self-sufficient producers may face higher costs to to changes in global food prices because they are more likely to improve storage to ensure food safety, affecting utilization. They consume staple foods derived from globally traded commodi- are also more likely than net sellers to be negatively affected by ties. This can have severe impacts on access. By contrast, rural the reduced nutritional value of their food consumption bas- populations are more dependent on local market surpluses ket. The fate of self-sufficient households is less influenced by and more traditional staple crops such as roots or tubers (FAO changing food prices than are net sellers or buyers of food, so 2008a). This makes both urban and rural groups susceptible to price volatility has less of an impact on food stability. However, price volatility caused by climate shocks, although their vulner- like net food sellers, self-sufficient households will experience ability depends on the relative volatility in their market source. more frequent climate shocks that will reduce incentives to Also, if urban incomes rise at a slower pace than food prices, make long-term investments in farming in the absence of policy the urban poor are less likely to consume nutritious and safe interventions to help manage risks. foods, or more likely to make dietary cuts. This negatively impacts utilization. The stability of urban food supplies would be Food-Purchasing Farm Households negatively impacted by food price volatility, but this could be Most farming households in developing countries are both buy- mitigated by greater access to non-local food markets or storage. ers and sellers of food. They use markets to supplement a lack of quantity or variety in domestic production or to bridge seasonal Summary food shortages. However, those that generally buy more than As we have seen, there are many different pathways through they sell make up the largest share of poor rural households (FAO which climate change might impact food security via impacts 2008b). These households are vulnerable to both price increases on the four dimensions of food security for the five household and production losses; as a result, they are vulnerable to negative groups. Net impacts for any particular group (the “sum� of impacts on availability from climate change. The double impacts impacts across the four dimensions) are generally ambiguous of reduced production and higher food expenditures constitute a and conditional on other factors. However, net food sellers are threat to the food security of this group. Food-purchasing house- the most likely to gain, and the urban poor are most likely to holds are less likely to have adequate food-storage facilities than lose. Self-sufficient and net food-purchasing households might food-selling households; this affects utilization by increasing their gain if their ability to benefit from higher prices outweighs exposure to spoiled or unsafe food (Brown et al. 2009). Stability for other negative impacts on utilization and stability, but lose other- these households will be vulnerable to both market and produc- wise. Rural non-farmers and the landless might gain if the rural tion volatility. economy expands and they see increased incomes, but the gain must be significant enough to offset the likely negative impacts Rural Landless and Non-Farm Rural Households on utilization and stability. In rural areas, there are many individuals who don’t partici- pate in the local agricultural economy, but still rely on it. These How to Minimize the Impact of Climate Change non-producers are typically landless but they are still involved on Poverty and Inequality in agriculture through wage labor, trading, services, or input For the urban poor, maintaining and improving food secu- sales. These individuals may benefit from increased food prices, rity will be directly related to their ability to access nutritious depending on the overall effects on agricultural production and foods. Minimizing the impact of climate change on poverty and incomes of local producers. Rising rural wages could thus be inequality depends on protecting the access and utilization di- associated with increases in access for this group, even if avail- mensions of food security. Broad-based economic growth and ability decreases locally. However, where net effects of climate development certainly is key. Additionally, well-functioning 10 : : April 2013 : : Inequality in Focus global food markets—and domestic trade regulations regulating Dercon, S., and L. Cristiaensen. 2011. “Consumption Risk, access to these markets—can help by dampening increases in Technology Adoption, and Poverty Traps: Evidence from locally driven food prices and ensuring a more stable supply of Ethiopia.� Policy Research Working Paper 4257. Washing- affordable, nutritious food. ton, DC: World Bank. In rural areas, the stability and availability dimensions of food FAO. 2008a. “Climate Change and Food Insecurity: A Frame- security will be particularly important in determining the ultimate work Document.� Rome: Food and Agriculture Organiza- impacts of climate change on poverty and inequality. Inequality in tion of the United Nations (FAO). rural areas is likely to be determined by the differences in agricul- tural households’ ability to take advantage of higher long-term FAO. 2008b. “Challenges for Sustainable Land management food prices to expand production and thus incomes. (SLM) for Food Security in Africa,� Twenty-fifth Regional Unfortunately, current wealth is likely to be a key determi- Conference for Africa. Nairobi, Kenya, June 16-20, 2008. nant in this differential ability. Relatively wealthy producers, Rome: FAO. who can self-insure or are better connected to individuals not FAO.2009. “Global Agriculture Towards 2050.� High Level directly dependent on the local rural economy (such as those Experts Forum, Office of the Director, Agricultural whose educated children have migrated to urban areas), will Development Economics Division Economic and Social be better able to manage increases in climate risks. They will Development Department, October. Rome: FAO. http:// also be less likely to reduce important expenditures on food, www.fao.org/fileadmin/templates/wsfs/docs/Issues_papers/ education, and health care. In addition, they may not need to HLEF2050_Global_Agriculture.pdf sell productive assets in response to climate shocks, thereby FAO. 2011. “The state of food insecurity in the world: How protecting the income of the next generation. does international price volatility affect domestic econo- Social safety nets and innovative insurance products can help mies and food security?� Rome: FAO. level the playing field, enabling poorer farmers to take advantage FAO, WFP, and IFAD. 2012. “The State of Food Insecurity in of higher prices now, and avoid having to sell assets, remove their the World 2012: Economic growth is necessary but not suf- children from school, and reduce nutritious consumption—all of ficient to accelerate reduction of hunger and malnutrition.� which would have negative impacts on future income. Rome: FAO. More general rural agricultural-development policies that reduce transaction costs can also enable more producers to FAO. 2013a. “FAOSTAT.� FAO Statistics Division. http://fao- benefit from relatively higher prices, thereby reducing poverty stat.fao.org/site/, accessed February 26, 2013. rates and inequality. Finally, policies and extension services FAO 2013b. “World Food Situation: FAO Food Price Index.� that promote wider adoption of sustainable land-management http://www.fao.org/worldfoodsituation/wfs-home/food- practices can also help minimize the negative impacts of climate pricesindex/en/, accessed April 11, 2013. extremes, and potentially lead to increases in yield. Fischer, G., M. Shah, and H. van Velthuizen. 2002. “Climate References Change and Agricultural Vulnerability.� Special report prepared as a contribution to the World Summit on Sus- Aksoy, A., and A. Isik-Dikmelik. 2008. “Are Low Food Prices tainable Development, International Institute for Applied Pro-Poor? Net Food Buyers and Sellers in Low-Income Countries.� Policy Research Working Paper 4642. Wash- Systems Analysis, Laxenburg, Austria. ington, DC: World Bank. Hurley, T. 2010. “A Review of Agricultural Production Risk Brown, M., B. Hinterman, and N. Higgins. 2009. “Markets, in the Developing World.� Harvest Choice Working Paper Climate Change, and Food Security in West Africa,� Envi- 11. Washington, DC: International Food Policy Research ronmental Science and Technology, 43: 8016–8020. Institute. Darwin, R., M. Tsigas, J. Lewandrowski, and A. Raneses. 1995. ILO. 2007. “Employment by sector.� Key indicators of the “World Agriculture and Climate Change: Economic Adap- labour market (KILM), 5th edition, Chapter 4. tations.� Agricultural Economic Report 703. Washington, Ivanic, M., and W. Martin. 2008. “Implications of Higher DC: Department of Agriculture. Global Food Prices for Prices and Poverty in Low-Income Daviron, B., M. Aubert, N, Bricas, H. David-Benz, S. Dury, J. Countries.� Policy Research Working Paper 4594. Washing- Egg, F. Lancon, and V. Meuriot. 2008. “Les mechanismes ton, DC: World Bank. de Transmission de la Hausse des prix Internationaux des IPCC. 1996. “Impacts, Adaptations and Mitigation of Climate Produits Agricoles dans les Pays Africans.� Paris: Fon- Change: Scientific-Technical Analysis.� 427–467. Authors: dation pour l’Agriculture et la Ruralité dans le Monde J. Reilly, W. Baethgen, F.E. Chege, S.C. van de Geikn, L. (FARM), CIRAD. Erda, A. Iglesias, G. Kenny, D. Petterson, J. Rogasik, R. Inequality in Focus : : April 2013 : : 11 Rotter, et al. Editors: R.T. Watson, M.C. Zinyowera, R.H. Parry, M.L., C. Rosenzweig, A. Iglesias, M. Livermore, G. Moss. Cambridge: Cambridge University Press. Fischer. 2004. “Effects of Climate Change on Global Food IPCC. 2007. “Climate Change 2007: The Physical Science Production.� Global Environmental Change 14: 53– 67. Basis.� Contribution of Working Group I to the Fourth Schmidhuber, J., and F. Tubiello, 2007. “Global food security Assessment Report of Intergovernmental Panel on Climate under climate change,� Proceedings of the National Acad- Change. Cambridge: Cambridge University Press. emy of Science (PNAS) 104 (50): 19,703–19,708. IPCC. 2012. “Managing the Risks of Extreme Events and Di- Skoufias, E., and A. Quisumbing. 2005. “Consumption Insur- ance and Vulnerability to Poverty: A Synthesis of Evidence sasters to Advance Climate Change Adaptation (SREX),� from Bangladesh, Ethiopia, Mali, Mexico and Russia.� The Special Report of the Intergovernmental Panel on Climate European Journal of Development Research 17: 24–58. Change. Cambridge: Cambridge University Press. Vermuelen, S.J., B.M. Campbell, J.S.I. Ingram. 2012. “Climate Karfakis, P., M. Knowles, M. Smulders, and J. Capaldo. 2011. Change and Food Systems.� Annual Review of Environ- “Effects of Global Warming on Vulnerability to Food Inse- mental Resources 37: 195–222. curity in Rural Nicaragua.� ESA-FAO WP Series 11–18. World Bank. 2013. “Food Price Watch.� PREM Network. Kazianga, H., and C. Udry. 2006. “Consumption Smoothing? http://siteresources.worldbank.org/EXTPOVERTY/Re- Livestock, Insurance and Drought in Rural Burkina Faso.� sources/336991-1311966520397/Food-Price-Watch-Novem- Journal of Development Economics 79: 413–446. ber-2012.htm, accessed on February 27, 2013. The Inequality in Focus series aims at informing the public debate on equity, inequality of opportunity, and socioeconomic mobility. It features articles written by World Bank staff, as well as researchers and policy makers from the broad development community. The views and interpretations in the articles are those of the authors and do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent. The Inequality in Focus series is not copyrighted and may be reproduced with appropriate source attribution. Editorial Committee: Pedro Olinto (managing editor), Jaime Saavedra, Francisco Ferreira, Luis-Felipe Lopez-Calva, John Newman, Gabriel Demombynes, and Anna Reva Editor: Mary Anne Mulligan THE WORLD BANK Poverty Reduction and Equity Department Poverty Reduction and Economic Management Network (PREM)