WPS7782 Policy Research Working Paper 7782 Inequality of Opportunity in Sub-Saharan Africa Paolo Brunori Flaviana Palmisano Vito Peragine Africa Region Office of the Chief Economist August 2016 Policy Research Working Paper 7782 Abstract In the last decades, inequality of opportunity has been exten- the control of individuals control—is between 30 percent sively studied by economists on the assumption that, in and 40 percent in the countries considered. The results also addition to being normatively undesirable, it can be related indicate a positive association between total consumption to low potential for growth. This paper evaluates inequality of inequality and inequality of opportunity. Finally, this paper opportunity and the different sources of unequal opportuni- addresses a number of methodological issues that typically ties in 11 Sub-Saharan Africa countries. The results indicate arise when measuring inequality of opportunity with imper- that the portion of total inequality that can be attributed to fect data, which is the typical case in developing countries. exogenous circumstances—that is, circumstances outside This paper is a product of the Office of the Chief Economist, Africa Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at peragine.vito@gmail.com. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Inequality of Opportunity in Sub-Saharan Africa Paolo Brunori∗ Flaviana Palmisano† Vito Peragine‡ JEL classification: D63, E24, O15, O40. Keywords: Consumption inequality, Equality of opportunity, Sub-Saharan Africa. ∗ University of Bari. † University of Rome LUMSA. ‡ University of Bari. 1 Introduction Sub-Saharan African countries are especially known for their high levels of economic inequality and poverty (see, for instance, Ellis 2012; Moradi and Baten 2005; Thorbecke 2013). However, the specific features of these inequalities remain largely understudied. Yet, the understanding of the different sources of inequality is a necessary step toward the implementation of policies that may foster a sustained and shared growth in these coun- tries. There is, in fact, a rooted consensus around the argument that not all inequalities are the same: in particular, it has been convincingly argued (see Ferreira et al. 2014; Marrero and Rodriguez 2013; World Bank 2006) that the degree of inequality caused by differences at birth (such as gender, ethnicity, or parental background) or, more generally, by factors beyond the control of individuals may be related to low growth, more so than other effort- based inequalities. The idea is that, when exogenous circumstances play a strong role in determining individual outcome, there is a suboptimal allocation of resources and lower potential for growth. To put it differently, the existence of inequality traps, which system- atically exclude some groups of the population from participation in economic activity, is harmful to growth because it discourages effort and investment by individuals, provokes a loss of productive potential, and contributes to social and institutional instability. The arguments above suggest that analyzing the specific horizontal dimensions of inequality is particularly important in both developing and underdeveloped countries. One way to assess these kinds of inequalities is to implement the Equality of Oppor- tunity (EOp) framework (see Fleurbaey 2008; Roemer 1998), which provides a model to distinguish between that part of inequality caused by exogenous circumstances outside the individual’s responsibility, considered to be objectionable and therefore deserving a compensatory intervention, and the part of inequality generated by individual choices and effort, which is, in contrast, considered to be fair and should not be eliminated. The EOp theory has spurred a huge amount of theoretical and empirical work focusing on the mea- surement of inequality of opportunity (see the recent surveys by Ferreira and Peragine 2015; Ramos and Van de gaer 2015; Roemer and Trannoy 2015). However, most of the literature has been concerned with inequality of opportunity (IOp) in Western developed countries, with only a small set of studies dedicated to developing countries.1 One reason for this is that measuring IOp is not an easy task: its informational requirements are quite high if compared with the standard measurement of income or consumption inequality.2 Therefore, these are more commonly met in surveys and databases in wealthier countries. Hence, as argued above, such analysis would be particularly needed in developing countries. This paper is a contribution in this direction. Specifically, it is the first attempt to evaluate IOp in a large set of Sub-Saharan African countries by using 13 different surveys that contain information about individual circumstances and outcomes. 1 e-Somps (2008) In particular, only two contributions exist in the literature, namely, Cogneau and Mespl´ and Piraino (2015), that propose an analysis of IOp for African countries. 2 See Hassine (2011). 2 Our contribution to the literature is twofold. First, we contribute to the understanding of economic inequality in 11 Sub-Saharan African countries (i) by showing the portion of consumption inequality that can be attributed to IOp and (ii) by identifying the most disadvantaged groups of the population in each country. This analysis can help in un- derstanding the social and economic mechanisms that generate inequalities and can help in identifying priorities in antipoverty policies in different countries. Second, this paper offers a methodological contribution to the literature on the measurement of inequality of opportunity by addressing a number of methodological issues that typically arise in the realization of this task in the presence of imperfect data, which is the typical case in developing countries. Our analysis is made possible through the availability of large-sample surveys built upon a common methodology and providing information on the socioeconomic background of adult individuals. We use a set of 13 surveys that were implemented during a period ranging from 2000 to 2013 and covering the following countries: the Comoros, the Democratic Republic of Congo, Ghana, Guinea, Madagascar, Malawi, Niger, Nigeria (two waves), Rwanda, Tanzania (two waves), and Uganda (two waves). Our estimates use a subsample of the original data, adult household members with observable relevant characteristics; nevertheless, they uncover a dramatic picture. Total consumption inequality is remarkable in all the countries, although quite variable across them: the Gini coefficient ranges from 0.55 for the Comoros to 0.31 for Niger, but, in general, the Gini coefficient is around 0.4 in all countries considered. The entire region of Sub-Saharan Africa is confirmed as one of the most unequal regions in the world. More- over, for the three countries for which two waves are available (Nigeria, Tanzania, and Uganda), the results show an increase in inequality in recent years. As far as inequality of opportunity is concerned, our estimates show that the impact of exogenous circumstances is noticeable in every country, although this impact is quite variable across them: the por- tion of total inequality that can be attributed to (the observable) exogenous circumstances is between 30 percent and 40 percent for the countries considered. This is a striking result, particularly if one considers that the computed measures are lower bound estimates of the inequality of opportunity level in each country. We also look at the association between total consumption inequality and inequality of opportunity: although some re-rankings do exist, the data show a positive relationship between the two kinds of inequalities. The sources of unequal opportunities also differ across countries. For example, in the Comoros and Niger, birthplace plays the strongest role in determining IOp, while in Congo ethnicity is clearly the dominating circumstance. The ranking of countries in terms of inequality of opportunity is robust with respect to the inequality measure used, but our estimates are sensitive with respect to the estima- tion approach and to the choice of the exogenous circumstances. In this paper we address this issue by exploring two different estimation approaches (parametric and nonparamet- ric) and by proposing an adjusted inequality of opportunity measure, which takes into account the differences between countries in the number of the circumstances variables. 3 This methodology should make the cross-country comparison more reliable. Our results differ substantially from the only previous contribution that has focused on cross-country comparisons of inequality of opportunity in Sub-Saharan Africa:3 Cogneau and Mespl´ e-Somps (2008) analyzed five Sub-Saharan African countries (Cˆ ote d’Ivoire, Ghana, Guinea, Madagascar, and Uganda) by using data collected between 1985 and 1994. They use a coarse set of circumstances (parental background), and, in fact, their results show a much lower level of inequality of opportunity: with some variation between coun- tries, their estimates show that the portion of inequality attributed to exogenous circum- stances is between 10 percent and 20 percent. Unlike Cogneau and Mespl´ e-Somps (2008), we extend the analysis to a larger set of countries and a bigger set of circumstances for each country; moreover, we provide a more data-extended and methodologically intensive analysis. The paper is organized as follows. Section 2 briefly reviews the concept of opportunity inequality and discusses some measurement issues. Section 3 describes the data, and the nonparametric and parametric analyzes of inequality of opportunity for the periods and countries considered are presented in Sections 4 and 5, respectively. Section 6 provides a summary of the current findings and concludes with suggestions for further research on IOp in Sub-Saharan African countries. 2 Methodology 2.1 A model of equality of opportunity The canonical model of EOp assumes that the outcome of an individual, y , is entirely determined by two classes of variables: circumstances and effort (see Peragine 2002; Roe- mer 1998; Van de gaer 1993). For simplicity, we refer here to the individual outcome as income, but any other interpretation of outcome, such as consumption, would in principle be possible. Circumstances are denoted by c and belong to a finite set Ω: examples are gender, age, ethnicity, region of birth, or parental background. These are factors beyond an individual’s control, but nonetheless exogenously affect income. Effort is denoted by e and belongs to the set Θ, and it may be treated either as a continuous or a discrete variable. This is a factor that endogenously affects the individual income since it is the result of one’s own choices. The different forms of luck that may affect the individual income can be classified either as circumstances or as responsibility characteristics. Individual income can then be expressed as follows: y = g (c, e) (1) The production function g : Ω×Θ → R+ is assumed to be monotonic in both arguments, 3 See also Piraino (2015) for a study of IOp in South Africa. 4 while circumstances and effort are assumed to be orthogonal.4 This is a reduced form model in which neither the opportunities themselves nor the individual decision process to exert a given level of effort are explicitly modeled. The model builds on the argument that (non-observable) individual opportunities can be inferred by observing joint distributions of circumstances, effort, and income, which fully characterize a population of individuals. For simplicity, let us treat effort, as well as each element of the vector of circumstances, as discrete variables. This allows the population to be partitioned in two ways: into types in which all individuals share the same circumstances and into tranches in which everyone shares the same degree of effort. Roughly, the source of unfairness in this model is given by the effect circumstance vari- ables (which lie beyond individual responsibility) have on individual outcomes. However, there are different ways to measure this effect. The measurement exercise can be thought of as a two-step procedure. First, the actual distribution is transformed into a counterfactual distribution that reflects only and fully the unfair inequality, while all the fair inequality is removed. In the second step, a measure of inequality is applied to this counterfactual distribution. The construction of the counterfactual distribution should reflect two distinct and independent principles: the reward principle, which is concerned with the apportion of outcome to effort and, in some of its formulations, requires that one respects the out- come inequalities arising from effort and the compensation principle, according to which all outcome inequalities associated to exogenous circumstances are unfair and should be compensated for by society. In particular, the existing literature has developed two main versions of the compensation principle and two consequent approaches to the measurement of IOp, namely the ex ante and the ex post approach. According to the ex ante approach, there is EOp if the set of opportunities is the same for all individuals, regardless of their circumstances. Hence, in the ex ante version, the compensation principle is formulated with respect to individual opportunity sets: it requires that one reduce the inequality between these opportunity sets. In the model introduced above, the income distribution of a given type is interpreted as the opportunity set of all individuals with the same set of circumstances. Hence, the focus is on the inequality between types: the counterfactual distribution should eliminate the inequality within the types (reward) and reflect the inequality between the types (ex ante compensation). Let us underline here a dual interpretation of the types in the EOp model: on one hand, the type is a component of a model that, starting from a multivariate distribution of income and circumstances, allows us to obtain a distribution of (the value of) opportunity sets enjoyed by each individual in the population. On the other hand, given the nature of the circumstances typically observed and used in empirical applications, the partition into types may be of interest per se: they can often identify well-defined socioeconomic groups, possibly deserving special attention by policy makers. 4 This assumption is motivated by the theoretical argument that it would hardly be sustainable to hold people responsible for the factor e in a situation in which it is dependent on exogenous characteristics. 5 Alternatively, according to the ex post approach, there is EOp if and only if all those who exert the same effort end up with the same outcome. The compensation principle, in the ex post version, is thus defined with respect to individuals with the same effort but different outcomes. This means that IOp within this approach is measured as inequality within the tranches. Hence, the corresponding counterfactual distribution should reflect the inequality within the tranches (ex post compensation), but should eliminate the inequality between the tranches (reward). Different measures that are consistent with either the ex ante or the ex post approaches have been proposed in the literature (see Ferreira and Peragine, 2015; Ramos and Van de gaer, 2015): they express different and sometimes conflicting views on EOp, and, in fact, the rankings they generate may be different. In addition, their informational requirements are quite different: while, for the ex ante approach, one needs to observe the individual outcome and the set of circumstances, a measure of individual effort is required for the ex post approach. Therefore, in addition to normative considerations, the choice of which methodology to adopt should also reflect data availability. In our case, the database we use does not contain a satisfactory measure of effort. For this reason, we focus on the ex ante approach, and, among the various measures coherent with this approach, we use the between-types inequality measure, which was proposed, among others, by Checchi and Peragine (2010), Ferreira and Gignoux (2011), and Peragine (2002). It relies on a counterfactual distribution, which is obtained by replacing each individual’s income by the average income of the type an individual belongs to, independently of the level of effort exerted.5 This smoothing transformation, intended to remove all inequality within types, can be performed by using either a parametric or a nonparametric method. These methods are discussed in the following section. 2.2 Parametric and nonparametric approaches Given a distribution of income Y of size N, with n types indexed by i = 1, ..., n, for each type i, the population size will be denoted by mi , its population share by qi , and its mean income by µi (y ). According to the between-types inequality measure, the counterfactual distribution Ys is obtained by replacing each individual income with the value of the opportunity set of that individual, that is, the mean income of the type to which the individual belongs. Hence, by ordering the types on the basis of their mean such that µ1 (y ) ≤ ... ≤ µj (y )... ≤ µn (y ), the counterfactual distribution corresponding to Y is defined as Ys = (µ1 (y )11 , ..., µi (y )1i , ..., µn (y )1n ), where 1i is the unit vector of size mi . For a given measure of inequality I : R+ N → R , the part of inequality arising from initial + 5 The use of the average of the type for the smoothing transformation is justified, from a normative point of view, in light of the utilitarian reward principle, according to which society should express full neutrality with respect to inequalities due to effort. See Ferreira and Peragine (2015) for a discussion of the different formulations of the reward principle proposed in the literature and Lefranc, Pistolesi, and Trannoy (2009) and Peragine and Serlenga (2008) for empirical analyses based on different versions of the reward principle. 6 circumstances will be given by I (Ys ) or, in relative terms, by: I (Ys ) IOp = (2) I (Y ) Equation (2) measures the portion of overall inequality that can be attributed to un- equal opportunities. In most empirical analyses, I (Y ) is represented by the mean logarith- mic deviation (MLD), because it is perfectly decomposable into between- and within- types inequality. However, the MLD has some undesirable properties: in particular, it tends to be more sensitive to extreme values and is not bounded above; therefore, if inequality is measured on a distribution of type means from which extreme values have been removed by the smoothing operation, it tends to be underestimated by the MLD. For these rea- sons, in this paper, we follow Aaberge, Mogstad, and Peragine (2011) and use the Gini index, which has well-known desirable characteristics, although it is not perfectly decom- posable into between- and within- types inequality whenever the type income distributions overlap.6 Therefore, in general: Gini(Y ) = Gini(Ywithin ) + Gini(Ys ) + K (3) K is a residual greater than zero if there is overlapping between the type distributions. Given a set of selected circumstances defined on the basis of normative grounds and observability constraints, any within-type variation in individual outcome is attributed to personal effort. However, the vector of observed circumstances is likely to be a subvector of the theoretical (true) vector of all possible circumstances that determine a person’s outcome. Hence, as in any other empirical analysis of this kind, we face the issue of omitted circumstance variables. This problem is often addressed by the argument that the IOp estimates should be interpreted as lower-bound estimators of the true IOp, that is, the inequality that would be captured by observing the full vector of circumstances. It can be shown, in fact, that increasing the number of observed circumstances increases IOp (see Ferreira and Gignoux 2011; Luongo 2011). However, this interpretation renders IOp estimates barely comparable across studies, particularly if comparing, for instance, the IOp of a country with a large number of observ- able circumstances to the IOp of another country with only a few observable circumstances. Moreover, the error made in comparing these two quantities might not be random, but might be correlated with data quality. Elbers et al. (2008) discuss this issue in a more general setting concerning any estimate of between-group inequality. They claim that, if decomposing total inequality into a between component and a within component, the estimate of between-group inequality might be artificially too low because it compares between-group inequalities with the inequality measured in a counterfactual population in which each individual is a group. To overcome this problem, they propose an adjusted 6 In the appendix, for a robustness check, we also compute the mean logarithmic deviation. See Lambert and Aronson (1993) for an insightful discussion on the Gini decomposition. 7 measure of between-group inequality, which is equivalent to the actual between-group in- equality normalized by the maximum possible between-group inequality that could be reached in the population, given the number of groups. The latter is defined as the extent of between-group inequality in a counterfactual distribution (Ya ) obtained by ranking out- comes from the lowest to the richest and then partitioning the distribution in such a way that the groups have the same population share as the actual group. Hence, adjusted IOp (Adj-IOp) can be expressed as follows: I (Ys ) Adj-IOp = (4) I (Ya ) Although the problem they are looking at does not exactly correspond to our problem of partial observability, their solution can be usefully applied to this context. This adjusted measure is appealing because it accounts for the number of types and their relative weights. Adj-IOp solves, at least in part, the problem of comparing IOp estimates based on different numbers of observable characteristics. Therefore, in the following, we propose estimates of both IOp and Adj-IOp. The nonparametric approach discussed so far is data-intensive: as the partition into types becomes finer, the population size of each type decreases, bringing about a decline in the precision of the estimates of the type mean, consequently giving rise to a bias in the estimation of IOp. In countries such as those considered in this paper, where data limitations on circumstances might seriously hamper the analysis, an alternative, parametric approach to the estimation of IOp that economizes on the data requirement could be explored. This is based on the assumption that a simple linear relationship characterizes equation (1), given that circumstances are exogenous by definition, and they may also influence effort (see Bourguignon, Ferreira, and Menendez 2007; Ferreira and Gignoux 2011; Ferreira, Gignoux, and Aran 2011). Therefore, equation (1) could be re- expressed in reduced form as y = φ(c, ), and a linearized version of this equation would lead to: y = βc + (5) The estimated βˆ coefficient of the ordinary least squares (OLS) estimation of equation (5) will incorporate both the direct effect of the circumstances on the outcome and the ˆ estimated through indirect effect through effort. Clearly, this will be true only if the βs OLS are unbiased estimates of the real effect of circumstances. IOp will then be obtained ˆ from the by applying an index of inequality to the distribution of the predicted values y OLS estimation of equation (5), that is 7 y) IOp = I (ˆ (6) 7 See Ferreira and Gignoux (2011). 8 Relative IOp will be equal to y) I (ˆ IOp = (7) I (y ) The parametric approach is fully consistent with the ex ante utilitarian assumption used in the nonparametric modeling of IOp. Here, the only difference is that the expected outcome, given observable circumstances, is obtained using the predicted values from a linearized OLS model. Assuming a linear effect of circumstances, we no longer need to construct types to predict this outcome, and we can exploit all information contained in the variables describing circumstances, that is, all values assumed by each circumstance. The literature has also recognized that the parametric approach has some limitations, however. First, the approach indirectly imposes a precise functional form linking circum- stances and outcome. Moreover, the OLS estimation of equation (5) requires that one control for a number of dummy variables. In fact, the set of circumstances generally used in empirical analyses typically includes parental education, parental occupation, area of birth, and ethnicity. Such variables are not cardinal, and, to make equation (5) opera- tional, each one needs to be transformed into a number of dummy variables equal to the number of values it assumes. If no cardinal circumstance is observable, estimating equation (5) through an OLS regression brings to the estimation of a shift in the regression intercept associated with each category of every circumstance, for instance, having white-collar parents or being a first-generation immigrant. This implies a severe restriction in the construction of the counterfactual distribution because it imposes a fixed effect for each circumstance. For example, it could be the case that being a first-generation immigrant has a completely different meaning depending on whether one’s parents are university professors or con- struction workers. On the other hand, in a parametric approach, this effect is defined to be the same. To take into account the interaction between circumstances, one needs to interact dummies. However, once all dummies have interacted, one intercept is estimated for each type, and our OLS estimate becomes equivalent to the nonparametric approach. Thus, on one side, the motivations for the use of a parametric approach appear to be clear if cardinal measures of circumstances are available (such as parental income). How- ever, they are less convincing if all circumstances can only be modeled through dummies. On the other side, if we assume a fixed effect of each circumstance, the output of the parametric approach can be easily used to estimate the partial effect of each circumstance on outcome: for example, following Wendelspiess and Soloaga (2014), we could implement a Shapley-Shorrocks decomposition based on the average marginal effect of each circumstance over all the possible permutations. This method leads to a path-independent identification of the contributions of each circumstance. The discussion above suggests the adoption of both the parametric and nonparametric approaches. In fact, as underlined by Ferreira and Gignoux (2011), the two approaches may be considered complementary. 9 3 Data Our analysis is based on the following surveys: - Enquˆete Int´ es des M´ egrale aupr` enages (EIM) for the Comoros, carried out by the Statistical Office of the Ministry of Land Planning and Settlement; - Enquˆete sur la Consommation des M´ enages (ECM) for the Democratic republic of Congo (year 2010), carried out by the National Institute of Statistics (Ministry of Planning); - Ghana Living Standards Survey (GLSS) for Ghana (year 2013), carried out by the Ghana Statistical Service - National Data Archive (GSS); - Enquˆ e de Base pour l’Evaluation de la Pauvret´ ete Integr´ e (EIBEP) for Guinea (year 2003), carried out by the National Directorate of Statistics (Ministry of Economics and Finance) - Enquˆ eriodique aupr` ete P´ es des M´ enages (EPM) for Madagascar (year 2005), carried out by the National Institute of Statistics (INSTAT); - Third Integrated Household Survey (IHS3) for Malawi (year 2010), carried out by the National Statistical Office of Malawi; - National Survey on Household Living Conditions and Agriculture (ECVM) for Niger (year 2011-12), carried out by the National Institute of Statistics of Niger; - General Household Survey (GHS) for Nigeria (years 2010-11 and 2012-13), carried out by the National Bureau of Statistics of Nigeria; - Enquˆ egrale sur les Conditions de Vie des M´ ete Int´ enages (EICV) for Rwanda (year 2000), carried out by the National Institute of Statistics of Rwanda (NISR); - National Panel Survey (NPS) for Tanzania (years 2009-10 and 2010-22), carried out by the National Bureau of Statistics of Tanzania; - Uganda National Panel Survey (UNPS) for Uganda (years 2009-10 and 2010-11), carried out by the Uganda Bureau of Statistics. They are all representative at a national level and cover both urban and rural areas. Table 1 lists the surveys used, the year they refer to, their original sample size, and a link to the documentation. Our analysis is based on a subsample of the original data, obtained by considering only individuals aged 15 years or more on whom information about circumstances beyond an individual’s control are available. The outcome considered is per capita consumption, which encompasses consumption for both food and non-food 10 Table 1: Data sources Country Survey Year Sample size Documentation Comoros EIM 2004 18,373 IHSN Congo, Dem. Rep. ECM 2010 110,529 IHSN Ghana GLSS 2013 39,826 GSS Guinea EIBEP 2003 25,319 INSG Madagascar EPM 2005 30,271 INSTAT Malawi IHS3 2010 30,137 World Bank Niger ECVM 2011-12 12,118 World Bank Nigeria GHS 2010-11 14,916 World Bank Nigeria GHS 2012-13 14,560 World Bank Rwanda EICV 2000 17,69 INSR Tanzania NPS 2009-10 9,175 World Bank Tanzania NPS 2010-11 11,394 World Bank Uganda UNPS 2009-10 8,268 World Bank Uganda UNPS 2010-11 7,509 World Bank goods, that is, we assume a proportional intrahousehold distribution of consumption and zero economies of scale in consumption. Although we use different surveys, the results are comparable across countries because the consumption variable has been adjusted for inflation and translated into 2011 purchasing power parity (PPP) international dollars (World Bank, 2015). A fundamental step in the measurement of IOp is the identification of the vector of observable circumstances. This is a normative choice, subject to the constraint of data availability. Our data contain information on a small set of basic circumstances that are important. For each country, in fact, we can observe a subset of the following: ethnicity, parental education and occupation, birthplace (see Table ?? for details). Parental education and occupation are widely used as circumstances in the empirical literature on IOp that deals with developed countries. The importance of socioeconomic origin is emphasized also by the sociological literature on social stratification and social mobility, which focuses on occupation-based social classes. A vast amount of evidence has been produced on the effect of socioeconomic background on children’s outcomes dur- ing adulthood. This literature is, however, traditionally Western-centric and has rarely concentrated on Sub-Saharan African countries. Nevertheless, there is also evidence sup- porting the argument that parental education and occupation act as circumstances in individual outcomes in the specific Sub-Saharan African context. For instance, it has been shown that, in these countries, the nutritional status of a child is strongly correlated with parental occupation with obvious, although indirect, consequences on the outcome in the future (Madise, Matthews, and Margetts 1999). Parental education, meanwhile, has been 11 Table 2: Circumstances observed by country Country Circumstances birthplace parental education parental occupation ethnicity √ √ √ Comoros √ √ √ Congo, Dem. Rep. √ √ Ghana √ √ √ Guinea √ √ √ Madagascar √ √ √ Malawi √ √ Niger √ √ Nigeria √ √ √ Rwanda √ √ Tanzania √ √ Uganda Source : Surveys listed in Table 1. Note : Ethnicity for the Democratic Republic of Congo is observable, but the documentation to decode it is missing, thus rendering impossible the construction of the partition in the types necessary for the nonparametric estimates of IOp. In Malawi, mother tongue is used as a proxy for ethnicity. shown to be an important factor in determining whether or not a child is currently at- tending school; whereas, school improvements in parental education have been shown to increase the schooling of children, which, in addition to improving their health and reducing the status of extreme poverty, has direct effects on the outcome prospects of these children (see Glick and Sahn 2000; Lassibille and Tan 2005; Lloyd and Blanc 1996; Schultz 2004). Ethnicity and birthplace are variables of paramount importance in Sub-Saharan Africa, historically characterized by civil and ethnic conflicts, which arrest or even reverse the growth and development process of the Sub-Saharan part of the African continent. Even today, Sub-Saharan African countries face impressive challenges to peace and stability and have fallen prey to continuous armed ethnic conflicts. Between 1946 and 2002, not less than 1.37 million battle-related deaths occurred in 47 civil wars in Sub-Saharan Africa (Lacina and Gleditsch 2005). In 2011, for instance, Sub-Saharan Africa suffered 91 instances of this type of conflict, compared with 89 in 2010 (see Brautigam and Knack 2014; De Ree and Nillesen 2009). Moreover, previous studies have shown that high levels of ethnic diversity are strongly linked to high informal market premiums, poor financial development, low provision of infrastructure, and low levels of education. Ethnicity has a strong influence on inequality in Africa, where ethnic fractionalisation has given rise to a political economy of unequal subsidies and discrimination (Easterly and Levine 1997; Milanovic 2003). The area is also characterized by regional disparities in access to opportunities. Hence, it appears natural to treat ethnicity and birthplace as circumstances in the context of our analysis. 12 Cross-country comparisons of IOp must be interpreted while bearing in mind that the subset of circumstances used may vary across countries because different surveys usually collect different information on circumstances.8 To provide meaningful nonparametric estimates of IOp, the circumstances observed for each country require some additional treatment. While the parametric approach, assuming a linear effect of circumstances on outcome, can exploit all the information contained in the variables that describe circumstances, the nonparametric approach is forced to aggregate some of this information. Thus, to estimate the mean of each type with a sufficient degree of confidence, the sample size of each type should not be too small. Circumstances are therefore aggregated to reduce the number of types and to increase the size. Tables 7 to 16 in the appendix contain the details of the partition in types used for the nonparametric estimates in each country. These tables represents the opportunity profile (Ferreira and Gignoux 2011), a country-specific list of types, the rank, and the value of the opportunity set. These profiles are interesting per se because they identify the most deprived groups in each society. 4 Results: the nonparametric approach 4.1 Consumption inequality and inequality of opportunity Table 3: Inequality and IOp, nonparametric estimates Country Sample Consumption Types Inequality IOp IOp % Max between-group Adj-IOp % per capita Gini Gini Gini Gini Gini Comoros 5,936 2,975 36 0.5532 0.1657 29.95 0.5489 30.18 Ghana 42,519 1,838 24 0.4143 0.1457 35.16 0.3960 36.80 Guinea 24,866 1,000 32 0.4275 0.1594 37.28 0.4257 37.44 Madagascar 28,951 415 30 0.3701 0.1026 27.71 0.3680 27.87 Malawi 30,137 855 64 0.4739 0.2071 43.71 0.4734 43.75 Niger 11,774 1,071 48 0.3106 0.1022 32.91 0.3087 33.11 Nigeria 2010-11 14,916 1,298 20 0.3885 0.1459 37.54 0.3792 38.47 Nigeria 2012-13 14,560 1,601 20 0.3897 0.1429 36.66 0.3795 37.65 Rwanda 14,112 641 24 0.4436 0.1149 25.90 0.4385 26.20 Tanzania 2009-10 9,119 1,133 52 0.3935 0.1687 42.88 0.3930 42.93 Tanzania 2010-11 11,391 1,112 52 0.3966 0.1609 40.57 0.3961 40.62 Uganda 2009-10 8,194 1,157 24 0.4523 0.1785 39.46 0.4470 39.93 Uganda 2010-11 7,454 1,039 24 0.4748 0.1885 39.71 0.4691 40.19 Source : Calculations based on the surveys listed in Table 1. Note : Per capita consumption is expressed in 2011 PPP U.S. dollars. Table 3 reports, for each country and wave, the estimates of total inequality, IOp, and 8 All individuals with missing information on circumstances are dropped from the analysis. 13 the IOp ratio (all computed using the Gini coefficient). Moreover, the first three columns contain information about the sample size, the average per capita consumption, and the number of types into which each country is partitioned. We do not report any estimate for the Democratic Republic of Congo because of the already mentioned impossibility of aggregating ethnic groups to obtain types. Total inequality is remarkable in all the countries, although quite variable across them: the Gini coefficient ranges from 0.55 for the Comoros to 0.31 for Niger, but, in general, the Gini is around 0.40. The entire region is confirmed as one of the most unequal regions in the world.9 For the three countries on which observations for more than one year are available (Nigeria, Tanzania, and Uganda), the results bear witness to an increase in inequality; hence, the recent dynamics, where available, show a regressive pattern. The ranking of countries according to the level of inequality seems to be robust to the choice of the inequality measure (whether the Gini or the MLD index, the latter reported in appendix III table 16); there is, in fact, only one instance of re-ranking between Tanzania and Niger. For IOp, the estimates show an equally dramatic albeit different picture. The share of inequality that can be attributed to different exogenous factors is extremely high and variable across all countries: it ranges between 26 percent for Rwanda and 44 percent for Malawi, and is more generally between 30 percent and 40 percent for the other Sub- Saharan African countries. In other words, according to the observed circumstances, more than one-third of the observed inequalities in consumption can be attributed to exogenous factors, that is, to IOp. This is a striking result, particularly if one considers that the computed measures are only lower-bound estimates of the IOp in each country. The association between total consumption inequality and IOp is depicted in Figure 1. This figure could be interpreted as a generalization of the so-called “Great Gatsby” curve (Corak 2013) showing a negative relationship between income inequality and social mobil- ity. Our results demonstrate that, although countries with higher consumption inequality are characterized by a higher level (portion) of IOp, there is also considerable re-ranking between countries taking place in passing from total inequality to IOp. Notable here is the case of the Comoros, which has the highest total inequality, but it has the second to lowest IOp of all countries examined here. In sum, our estimates allow the division of the 10 Sub-Saharan African countries under analysis into three main groups. The first group is represented by the three countries with the highest share of IOp, namely, Tanzania, Malawi, and Uganda; Malawi and Uganda also have the highest level of total inequality. The second group is represented by the three countries with a lower share of IOp, namely, Rwanda, Madagascar, and the Comoros, that nevertheless exhibit a comparatively high level of consumption inequality. The third 9 For instance, the most recent estimates of the Gini coefficient are around 0.30 for advanced countries, such as the EU-28, Australia, or the United States. They are around 0.35 for Eastern Europe, and about 0.40 for East Asia. Only the Latin American countries are characterized by an average Gini coefficient above 0.50, which is higher than the Gini for Sub-Saharan African countries (see Tsounta and Osueke 2014). 14 Figure 1: Total inequality and IOp 0.5500 Comoros 0.5000 Uganda 2010-11 Malawi 0.4500 Uganda 2009-10 Rwanda Gini Guinea Ghana 0.4000 Tanzania 2010-11 Nigeria 2012-13 Tanzania 2009-10 Nigeria 2010-11 Madagascar 0.3500 0.3000 Niger 0.1000 0.1200 0.1400 0.1600 0.1800 0.2000 IOp (absolute Gini) Source : Calculations based on the surveys listed in Table 1. is represented by all the other countries having relatively middle shares of IOp (that is, Ghana, Guinea, Niger, and Nigeria). 4.2 Adjusted inequality of opportunity The last two columns of the third part of Table 3 report, respectively, the adjusted IOp according to the Gini index and its share in total consumption inequality. As discussed above, the normalization of inequality with respect to the number of types is particularly relevant in the present context because we are comparing IOp in countries in which the specific consumption distribution is partitioned into a different number of types: from a minimum of 20 in Nigeria to a maximum of 64 in Malawi. Figure 2 plots the difference between IOp and Adj-IOp as a percentage of IOp against the number of types. Figure 2 also shows a clear pattern for this correction (approximated with a fractional polynomial curve), approaching zero as the number of types increases. The figure makes clear that the adjustment procedure does not add particularly relevant information in our context. The correction is never above 5 percent, and it is smaller than 2 percent in countries with a number of types above 40. Hence, the higher the number of types, the lower the impact of the adjustment, and this result is rather general. To grasp this drawback, consider Figure 3, which plots the difference between the total Gini (twice the area between the black Lorenz curve and the diagonal) and the maximum 15 Figure 2: Adj-IOp correction and number of types 5 Ghana 4 correction (%) 3 Nigeria 2012-13 Nigeria 2010-11 2 Rwanda Uganda 2010-11 Uganda 2009-10 1 Comoros Madagascar Niger Guinea Tanzania 2010-11 Malawi Tanzania 2009-10 0 20 30 40 50 60 number of types Source : Calculations based on the surveys listed in Table 1. between-group Gini, twice the area between the blue broken line, for three hypothetical group partitions: 1 group, 5 groups, 10 groups. The difference between the two possible denominators of IOp will depend on the shape of the original Lorenz curve; the example clarifies that this difference approaches zero quickly as the number of types increases. Therefore, the adjustment proposed by Elbers et al. (2008) loses relevance whenever the number of types is in the order of 10s. 5 The parametric approach Table 4 reports, for each country and wave considered, the results of the parametric estimates of IOp. The first part of the table contains information about sample size, mean per capita consumption, and the number of regressors (all dummies) used to assess the share of total inequality explained by circumstances. The number of regressors is given by the number of observable circumstances, multiplied by the number of values that each circumstance can take.10 The second part of the table contains the estimates of total inequality and IOp in absolute terms and as a share of total inequality, using the Gini 10 The analytical results of the OLS regression for each country are available from the authors upon request. 16 Figure 3: Lorenz curve and maximum between-type inequality Lorenz curve 2 types 5 types 10 types Note : Lorenz curves for the maximum between-group inequality (light blue) are drawn assuming a population partitioned into equally sized types. coefficient.11 Table 4: Inequality and IOp, parametric estimates Country Sample Consumption Number of Total inequality IOp IOp (%) per capita regressors Gini Gini Gini Comoros 5,936 2,975 91 0.5532 0.2305 41.67 Congo DR 39,578 1,535 402 0.3634 0.1739 47.84 Ghana 42,519 1,838 125 0.4143 0.2304 55.61 Guinea 24,866 1,000 96 0.4275 0.1504 35.18 Madagascar 28,951 415 445 0.3701 0.2080 56.22 Malawi 30,137 855 71 0.4739 0.2637 55.64 Niger 11,774 1,071 50 0.3106 0.1249 40.22 Nigeria 2010-11 14,916 1,298 40 0.3885 0.1640 42.21 Nigeria 2012-13 14,560 1,601 40 0.3897 0.1661 42.62 Rwanda 14,112 641 76 0.4436 0.1851 41.74 Tanzania 2009-10 9,119 1,133 41 0.3935 0.1911 48.57 Tanzania 2010-11 11,391 1,112 40 0.3966 0.1820 45.90 Uganda 2009-10 8,194 1,157 100 0.4523 0.2159 47.74 Uganda 2010-11 7,454 1,039 102 0.4748 0.2497 52.58 Source : Calculations based on the surveys listed in Table 1. Note : Per capita consumption is expressed in 2011 PPP U.S. dollars. 11 See appendix III for a parametric estimate of IOp using MLD. 17 In general, nonparametric estimates tend to be lower than their parametric version; however, this is not necessarily the case. Recall that parametric and nonparametric ap- proaches differ in two aspects: the former imposes a linear relationship between circum- stances and outcome; the latter aggregates some information contained in variables beyond an individual’s control. Setting aside the problem of partial observability, both constraints imply that IOp is a downward bias estimate of the real IOp under general conditions. Imposing linearity reduces the variability that can be explained by circumstances in all cases except if y is a linear function of c. Similarly, ignoring some of the variability of the circumstances decreases the ability of these variables to explain total inequality, unless the inequality between the groups aggregated is zero. Thus, if the bias implied by the assumption of linearity is smaller than the bias intro- duced by aggregating circumstances, the parametric IOp is larger than the nonparametric IOp. However, there can be cases in which the linearity assumption implies a larger dis- tortion than the aggregation of circumstances: in such cases, the nonparametric IOp will be larger. Figure 4 shows the discrepancy between the two approaches. Parametric estimates are reported on the vertical axis, and the nonparametric estimates on the horizontal axis. The first feature that stands out is that, with the exception of Guinea, parametric estimates are always larger than nonparametric ones. With the considerable exception of Guinea, Ghana, and Madagascar, there is a clear positive relationship between the rankings generated by the two approaches. The discrepancy between the two approaches seems to be driven by the high number of regressors used to estimate equation (5) and the rather low number of types used to construct the counterfactual distribution for the nonparametric estimates. An extreme case is Madagascar, in which the number of regressors is the highest, 462, while the number of types is 30, one of the lowest. Moreover, Madagascar jumps from being one of the least unequal countries when IOp is parametrically estimated to being one of the most unequal when IOp is nonparametrically estimated. Such a difference should be expected whenever the number of regressors (which by definition increases the total variability explained) is much larger than the number of types. However, the high number of regressors in Madagascar is mainly driven by the high number of possible birthplaces, that is, 397 dummies, far more than the six provinces in which Madagascar was divided at the time of the survey (now 22 regions), and also more than three times the 111 districts of the country. Birthplaces in this survey are cities (communes urbaines). The coefficients for the dummies of such a detailed subdivision of the territory are generally not statistically significant. It therefore seems unreasonable to include all the possible birthplaces among the controls of the OLS estimation of equation (5) because the estimates of their effect on circumstances would not be reliable. A viable solution consists of aggregating birthplaces into districts or provinces. Indeed, this is exactly what we do with the nonparametric approach: we trade some of the variability of our regressors to gain statistical significance. Therefore, in cases such as Madagascar, with 18 Figure 4: IOp parametric and nonparametric estimates 60.00 Madagascar Ghana Malawi Uganda 2010-11 50.00 Tanzania 2009-10 parametric IOp % Uganda 2009-10 Tanzania 2010-11 Nigeria 2012-13 Rwanda Nigeria 2010-11 40.00 Niger Comoros Guinea 30.00 20.00 30.00 40.00 50.00 60.00 non parametric IOp % Source : Calculations based on the surveys listed in Table 1. few observable qualitative characteristics that can take a large number of values, it would be more advisable to follow a nonparametric approach, which has the additional quality of not imposing linearity. This issue is examined for all countries in Figure 5, where we determine whether the difference between parametric and nonparametric estimates is really associated to the dif- ference between the number of types and the number of regressors. The vertical axis reports the ratio between the two estimates (nonparametric over parametric), and the horizontal axis reports the ratio between the number of types and the number of regressors. The pos- itive correlation between the two ratios suggests that the number of regressors does play a role in making parametric estimates. The correlation is far from perfect, and Guinea is an interesting case. Although, for this country, we have 113 regressors and 32 types, the parametric estimate of IOp is smaller than the nonparametric one. The case of Guinea provides an example of how assuming a linear effect of circumstances on outcome actually provokes a downward bias in our IOp estimates, which is larger than the bias induced by aggregating circumstances in using the nonparametric approach. The literature has traditionally judged the assumption of linearity to be less important in determining the magnitude of IOp than the issue related to the number of circumstances. However, the case of Guinea highlights that there are cases in which the opposite can happen. Table 5, a simplified version of the opportunity profile presented in Table 9 in appendix II, clarifies this point. The effect of parental occupation on child outcomes depends on area of birth: on average, in Guinea, having a father employed in agriculture is associated with low consumption. By contrast, being born in the region of Labe to parents working in the agricultural sector implies that one belongs to the type with the 19 Figure 5: Number of types and number of regressors 1.2 Guinea 1 non-parametric/parametric Nigeria 2010-11 Tanzania 2009-10 Nigeria 2012-13 Tanzania 2010-11 Uganda 2009-10 Niger .8 Malawi Comoros Uganda 2010-11 Ghana Rwanda .6 Madagascar .4 0 .2 .4 .6 .8 types/regressors Source : Calculations based on the surveys listed in Table 1. Table 5: Non-linear impact of circumstances: the case of Guinea Birthplace Parental occupation Per capita consumption Rest of Guinea agriculture 843.10 Rest of Guinea other 1,117.60 Labe other 1,272.88 Labe agriculture 1,805.57 Note : This example is obtained by aggregating data in Table 9. best outcome prospects. The effect of birthplace and parental occupation on consumption are clearly not linear. This is not merely a statistical feature, but it has a clear economic meaning: Labe is one of the main centeres of national and international agricultural trade flows (FEWS, 2013). Therefore, an individual who was born into a farming family in Labe has the best possible condition in terms of economic opportunities. It is clear that, in the specific case of Guinea, the parametric procedure neglects the interaction between parental occupation and area of birth. In sum, among the main reasons for the possible inconsistency between the parametric and nonparametric approaches, we find that the small number of observable characteristics and the possibly high number of values they may take assume do play an important role. In fact, our results demonstrate that a high number of regressors tends to make parametric estimates higher than nonparametric estimates. However, the assumption of a linear effect of circumstances on outcome, implicit in the parametric approach, can provoke a downward bias in IOp. 20 The adoption of a parametric approach does not necessarily to impose linearity or log linearity. On the contrary, we should try to estimate the best possible approximation of the process that transforms effort and circumstances into outcome. The only limitation to the introduction of a non-linear transformation of regressors and interaction terms is data availability. Unfortunately so far the empirical literature has paid little attention to this aspect. 5.1 Detecting the contribution of the specific circumstances In this section, we analyze the degree of association between each circumstance and the level of individual consumption to shed light on the relative importance of the different cir- cumstances in determining IOp. We are aware that this analysis does not identify the causal effect of each circumstance on IOp; unobservable determinants of the individual outcome are likely to be correlated with the observable circumstances preventing a causal identi- fication (see Ferreira and Gignoux 2011 for a discussion). Nevertheless, we believe that the description of the different degrees of association may help provide an interpretative framework for our estimates of IOp across the Sub-Saharan African countries considered. In doing so, we are going beyond the distinction between IOp and other inequalities; we are decomposing IOp by source. The same level of IOp can have a rather distinct meaning depending on the relative importance of different circumstances. A country in which ethnic inequalities have a prominent role may appear different from a country in which the main channel of transmission of wealth is parental education or occupation. In this analysis, an important aspect to consider is how circumstances are coded. As we have already discussed, ordinal and categorical variables are recoded by a set of dummy variables (as many as the number of values the circumstance assumes). Now, ceteris paribus, the share of total IOp explained by a circumstance is an increasing function of the number of dummies used to measure total IOp. If, as in Madagascar, birth location assumes 397 possible values, this circumstance is likely to explain more variability than the circumstance ethnicity, which is captured in the same country by 24 dummies. Therefore, in looking at the partial effect of a circumstance, one should consider both the share of total IOp explained by that circumstance and the number of variables used to describe total IOp. Table 6 shows the share of IOp explained by each circumstance (left) and the number of variables used to explain each circumstance (right). The partial IOp of each circum- stance is obtained by applying the Shapley value decomposition of the between-type Gini coefficient.12 Birth location has a small role in Malawi, where 42 of the 71 regressors explain only 29 percent of total IOp. Meanwhile, birth location appears to be the most important factor explaining IOp in Niger and in Comoros. At the time of the survey, 2004, the Comoros were in a period of relative stability after Colonel Azali Assoumani was elected president 12 For this analysis, we rely on the algorithm proposed by Araar and Duclos (2009). 21 in 2002 and agreement about the federal institutional framework was reached. However, one must bear in mind that these islands have been historically characterized by political instability because of conflicts between islands. On the one hand, the political power is traditionally concentrated in Grand Comore (Ngazidja); on the other, Anjouan - the second largest island - has the archipelago’s largest port, which has brought economic development and power to the island (IMF, 2006). Birthplace is a relatively easy piece of information to collect and often refers to a detailed partition of national territory. This variable correlates with a number of other variables such as ethnicity, mother tongue, and family wealth. Therefore, in the interpretation of this decomposition, one should bear in mind that any inequality because of omitted circumstances that correlates with observable circumstances is captured by the latter. This could be the case of Rwanda, where ethnicity is not observed and inequality across 18 regions of birth explains 29 percent of total IOp. Ethnicity is observed in the Democratic Republic of Congo, Ghana, Madagascar, Malawi, and Niger. The Democratic republic of Congo is one of the countries in the world with the highest level of ethnic diversity (Goren, 2014). Its 374 recorded ethnic groups explain a large share of total IOp (68 percent). In Uganda, ethnicity consistently explains nearly 50 percent of total IOp. Although inequality across ethnic groups is much smaller in Mada- gascar, only 17 percent of total IOp, it is explained by inequality between only 24 ethnic groups. Parental occupation and education explain half of total IOp in Nigeria. The role of parental education also appears important in Guinea (47 percent) and Rwanda (36 per- cent). This is consistent with what is observed in the opportunity profiles of the two coun- tries (see appendix II): in both cases, individuals with parents employed in the agricultural sector crowd the lowest position of the ranking. Parental occupation has an important role also in Ghana (35 percent) and the Democratic republic of Congo (20 percent). In the case of the latter, 20 percent of total IOp is explained by only nine dummies describing parental occupation. Parental education is the most frequently observed circumstance (it is missing only in Uganda and Niger). Its role is heterogeneous with a share ranging from 57 percent in Tanzania to 5 percent in the Comoros. Finally, in Malawi parental education explains a high share of IOp (52 percent); a share that is explained by only 14 of the 71 dummies. 22 Table 6: Shapley value decomposition of IOp Share of IOp (%) regressors Country Birth location Ethnicity Parental occ. Parental edu. Birth location Ethnicity Parental occ. Parental edu. Total Comoros 2004 84.19 10.76 5.06 55 0 23 12 90 Congo 2012 67.88 20.07 12.05 0 374 9 17 400 Ghana 2013 40.75 34.89 24.36 0 63 18 42 123 Guinea 2003 38.22 46.68 15.10 44 0 18 35 97 Madagascar 2005 59.16 17.28 23.56 397 24 0 24 445 Malawi 2010 29.34 18.19 52.47 42 15 0 14 71 Niger 2011-2012 87.36 12.64 40 10 0 0 50 Nigeria 2011-2012 53.86 46.14 0 0 22 18 40 Nigeria 2012-2013 52.64 47.36 0 0 22 18 40 Rwanda 2000 29.03 36.14 34.83 18 0 20 38 76 Tanzania 2009-2010 49.39 50.61 0 0 27 13 40 Tanzania 2010-2011 43.46 56.54 0 0 27 13 40 Uganda 2009-2010 52.44 47.56 57 43 0 0 100 Uganda 2010-2011 50.57 49.43 57 43 0 0 100 Source : Calculations based on the surveys listed in Table 1. Note : The share of regressors is the number of regressors that describe the circumstances divided by the total number of regressors. 23 6 Conclusion Inequality in Sub-Saharan African countries is generated by many factors. The area of birth, ethnicity or the education level of the parents are, for instance, among the most important factors. IOp, that is, the extent to which these kinds of factors determine the outcomes of individuals in adulthood, contribute to increase overall inequality, and violate principles of fairness. Although the empirical literature on IOp measurement has proliferated in recent decades, there are few contributions that focus on IOp in Sub-Saharan African countries. The lack of estimates for this part of the world is mainly associated to the lack of reliable data on individual outcomes and circumstances. This paper utilized 13 reliable household consumption surveys to assess IOp in 11 Sub- Saharan African countries. All information about exogenous factors provided by these surveys are used. These encompass information on region of birth, parental education and occupation, and ethnicity. We complement the analysis by estimating the partial effect of each circumstance in determining IOp and the adjusted measure of IOp proposed by Elbers et al. (2008). Overall, IOp is high in every country in this analysis relative to countries with similar characteristics. The portion of total inequality that can be explained by the effect of factors outside the individual’s control is between 30 percent and 40 percent in the Sub-Saharan African countries considered. It is around 25 percent in South Africa (Piraino 2015) and about 20 percent in China (Zhang and Eriksson 2010); it ranges between 24 percent and 50 percent in Latin American countries (Ferreira and Gignoux 2011). However, IOp is variable across the Sub-Saharan African countries considered, and countries with higher total inequality do not always show higher IOp. With respect to the ranking of countries, while our results are robust to the choice of the inequality measure, they appear to be less robust to the choice of the estimation method. From a methodological point of view, our analysis shows that some of the tools proposed in the literature for the measurement of IOp in Western countries need to be handled with caution in the analysis of IOp in Sub-Saharan Africa. In all countries analyzed, circumstances beyond the control of individuals such as ethnicity, birthplace, and parental background interact in determining individual opportunity in a much more complex way than what we typically observe in Western societies. Thus, as a focus of future research, this complexity should be examined with country-specific and more data-intensive studies to elucidate the best possible methods for determining IOp in non-Western countries. 24 Appendix I. The treatment of circumstances in nonparamet- ric estimates For the Comoros, the circumstances considered are birthplace, parental education, and father and mother’s occupation. Birthplace (originally categorised into 37 villages) is recoded into three categories: born in Grande Comore, born in Anjouan (Grande Comore and Anjouan are the largest and most important islands), others (born in other smaller islands or outside the Comoros). Parental education is coded into two categories: both parents have no education, at least one of the two has an elementary degree or higher. Father’s occupation is coded into three categories: employed in the agricultural sector, housekeeper, or other. The same coding is applied to mother’s occupation (see Table 7 in appendix II for details). In the Democratic Republic of Congo, the observable individual characteristics include father’s education, father’s occupation, and ethnicity. Father’s education is coded into number of years of education completed. Father’s occupation is recorded in 10 categories ranging from family helper to senior executive. The survey records 399 possible ethnic groups. These groups cannot be decoded using the available documentation; so, although we can identify individuals sharing the same ethnicity, we cannot group them into a ho- mogeneous macro-group as we do in other countries. This lack of information renders it impossible to construct types of sufficient size to allow inferences about average per capita consumption (32 ethnic groups contain only one respondent, and 131 contain less than 10). However, ethnicity can still be used as a dummy to explain inequality following the para- metric approach. We therefore present only parametric IOp estimates for the Democratic Republic of Congo. In the case of Ghana, we use information on three circumstances: birthplace, parental education, and ethnicity. To have a proper partition of the distribution, all three cir- cumstances are recoded. In particular, birthplace (originally represented by 17 regions of birth) distinguishes among individuals born in the northern, central, and southern parts of the country. Parental education is coded as for the Comoros. Finally, we divide the 64 ethnic groups present in the original data into four categories, according to their linguistic similarities. With this aim, we refer to the three main linguistic groups of the country: Gur, Kwa, and Mande. The fourth category encompasses the remaining ethnic groups (see Table 8 in appendix II for details). Birthplace, parental education, and parental occupation are the circumstances used for Guinea. Birthplace (originally represented by 34 villages of birth) is partitioned into seven categories depending on the region of birth: Boke, Faranah, Kankan, Kindia, Labe, Mamou, and Nzerekore. The last category refers to individuals born outside Guinea. Parental education is coded as for the Comoros, whereas parental occupation is partitioned into two categories: the first encompassing those individuals whose parents are employed in the agricultural sector, and the second encompassing all individuals who have at least 25 one parent who is employed in a sector other than agriculture one (see Table 9 in appendix II for details). The data on Madagascar provides birthplace, parental education, and ethnicity as en- dogenous characteristics. Birthplace (originally represented by about 400 villages) is based on the six administrative provinces of birth: Antananarivo, Antsiranana, Fianarantsoa, Mahajanga, Toamasina, and Toliara. However, we aggregate Mahajanga and Toamasina (on the basis of geographic distance), ending up with five categories. Parental education is coded as for the Comoros. The 25 ethnic groups present in the original dataset are grouped into three main categories on the basis of their main geographical location: costal, highlander, and others (see Table 10 in appendix II for details). Observations on Malawi, in contrast, are only available for two circumstances: birth- place and parental education. Birthplace encompasses 31 categories, one for each district, plus one category grouping individuals born outside Malawi. Parental education is coded as for the Comoros (see Table 11 in appendix II for details). Although information on mother tongue (used as a proxy for ethnicity) is available on Malawi, this information is only used for the parametric estimates because of the problem generated by the small size of the types if the partition also accounts for ethnicity. For Niger, the set of circumstances is represented by birthplace and ethnicity. Birth- places (originally indicated as one of the 40 departments of birth) is coded into nine cate- gories: the seven regions of the country (Agadez, Diffa, Dosso, Maradi, Tahoua, Tillaberi, and Zinder), the capital (Niamey), and others (individuals born outside Niger). The eth- nic groups represented in the survey are Arab, Djema, Haoussa, Kanouri-Manga, Peul, Touareg, Toukou, foreigners, and a residual made of other ethnic groups. To have types with a sufficiently large population to allow for inference, we group together Arab, Toukou, foreigners, and others (see Table 12 in appendix II for details). In the case of Nigeria, the circumstances considered are parental education, father’s occupation, and mother’s occupation. Concerning the first circumstance of education, we define the following categories: individuals with parents who have no education, individuals with at least one parent who has some primary education (not completed), individuals with at least one parent who has completed primary education, individuals with at least one parent who has some secondary education (not completed), and individuals with at least one parent who has completed secondary education or who has a higher degree. Concerning the second circumstance, father’s occupation, we define the following categories: individuals whose father is employed in agriculture or not working and individuals whose father is employed in a different sector. The same coding is used for mother’s occupation. The partitions are made of 20 types with a sample size ranging between 99 and 4,941 (see Table 13 in appendix II for details). The circumstances available from Rwanda’s data are birthplace, parental education, and parental occupation. Birthplace is characterized by six categories, each representing one of the five administrative regions of the country, and the last encompassing people born outside Rwanda. For parental education, we follow the coding used for the Comoros. 26 The third circumstance, parental occupation, is coded into two categories: the first groups individuals with parents who both work in the agricultural or fishery sector, and the second groups individuals with at least one parent who does not work in the agricultural or fishery sector (see Table 14 in appendix II for details). As for Tanzania, we observe two circumstances: birthplace and parental education. Birthplace is categorised into 26 administrative regions: Arusha, Dar es Salaam, Dodoma, Iringa, Kagera, Kaskazini Pemba, Kaskazini Unguja, Kigoma, Kilimanjaro, Kusini Pemba Kusini Unguja, Lindi, Magharibi, Manyara, Mara, Mbeya, Mjini, Morogoro, Mtwara, Mwanza, Pwani, Rukwa, Ruvuma, Shinyanga, Singida, Tabora, and Tanga. Parental edu- cation is classified into one of two categories: both parents have educational attainement below elementary level or at least one parent has completed an elementary school or higher (see Table 15 in appendix II for details). Birthplace and ethnicity are the information available for Uganda. Although the survey contains a large set of circumstances, such as parental education, parental occupation, area of birth, and ethnicity, we are forced to choose only two of them because of the large number of missing information for the other variables. In the original dataset, birthplace is distinguished into 56 districts, plus the capital city. For practical reasons, this circumstance is recoded into four groups according to the level of development of each district as measured by the Human Development Index (UNDP, 2014), that is, low development (HDI of 0.231- 0.433), lower intermediate (0.434-0.470), upper intermediate (0.472-0.498), and high (above 0.500). We also recode the 68 ethnic groups present in the original data on the basis on their linguistic origin: Eastern Lacustrine Bantu, Western Lacustrine Bantu, Eastern Nilotic, Western Nilotic, and ethnic minorities (see Table 16 in appendix II for details).13 13 This subdivision is based on information reported by the UNDP (2014) and Wairama (2001). 27 Appendix II. Opportunity Profiles Table 7: Opportunity profile: the Comoros, 2004 Rank Birth location Parental education Father occupation Mother occupation n 2004 p.c. consumption 2004 1 others none other other 29 1,844.41 2 others none housekeeper agriculture 208 1,986.20 3 others none agriculture other 71 2,088.10 4 Anjouan elementary or above other agriculture 8 2,092.21 5 Anjouan elementary or above other other 21 2,178.00 6 Grande Comore none agriculture other 94 2,240.15 7 others elementary or above agriculture agriculture 14 2,286.77 8 Anjouan elementary or above agriculture other 26 2,294.45 9 Grande Comore none housekeeper other 381 2,321.49 10 Anjouan none other other 89 2,350.16 11 Grande Comore none agriculture agriculture 643 2,351.03 12 Grande Comore elementary or above other other 53 2,380.17 13 Grande Comore elementary or above housekeeper agriculture 59 2,467.56 14 Grande Comore none housekeeper agriculture 957 2,524.50 15 others elementary or above other agriculture 2 2,546.14 16 Anjouan none housekeeper other 268 2,562.99 17 Grande Comore elementary or above agriculture other 22 2,577.82 18 others none housekeeper other 125 2,589.02 19 Grande Comore none other agriculture 173 2,608.87 20 others elementary or above agriculture other 12 2,630.26 21 others none agriculture agriculture 244 2,634.96 22 Anjouan elementary or above agriculture agriculture 28 2,735.11 23 Anjouan elementary or above housekeeper agriculture 10 2,857.03 24 Anjouan none agriculture other 259 2,988.62 25 Grande Comore elementary or above other agriculture 13 3,016.21 26 Anjouan none other agriculture 75 3,022.97 27 Grande Comore none other other 142 3,029.03 28 others none other agriculture 16 3,059.25 29 Grande Comore elementary or above housekeeper other 92 3,145.42 30 others elementary or above housekeeper agriculture 16 3,219.43 31 Anjouan none housekeeper agriculture 439 3,243.63 32 others elementary or above housekeeper other 99 3,547.67 33 Anjouan none agriculture agriculture 1,082 3,837.00 34 others elementary or above other other 27 3,940.04 35 Anjouan elementary or above housekeeper other 106 4,468.70 36 Grande Comore elementary or above agriculture agriculture 33 36,616.50 Source : Calculations based on the surveys listed in Table 1. Note : n 2004 is the sample size of each type in 2004; p.c. consumption is per capita consumption and expressed in 2011 PPP U.S. dollars. 28 Table 8: Opportunity profile: Ghana, 2013 Rank Ethnicity Birth location Parental education n 2013 p.c. consumption 2013 1 Kwa north none 793 917.40 2 others north none 282 923.73 3 Gur north none 11,519 1,103.89 4 Mande north none 244 1,285.88 5 Gur center none 1,079 1,328.15 6 Gur south none 844 1,536.28 7 Gur north elementary or above 1,722 1,550.71 8 Mande center none 78 1,561.71 9 Kwa north elementary or above 188 1,567.31 10 others center none 147 1,570.04 11 Mande south none 67 1,683.66 12 others center elementary or above 79 1,688.43 13 Mande center elementary or above 36 1,731.18 14 Gur center elementary or above 354 1,753.10 15 Kwa south none 7,852 1,792.07 16 Kwa center none 2,962 1,907.49 17 Mande north elementary or above 23 1,980.65 18 Gur south elementary or above 363 2,154.66 19 Kwa center elementary or above 3,715 2,181.44 20 others north elementary or above 34 2,257.19 21 others south none 185 2,330.63 22 Kwa south elementary or above 9,799 2,370.11 23 Mande south elementary or above 39 2,554.33 24 others south elementary or above 115 2,565.80 Source : Calculations based on the surveys listed in Table 1. Note : n 2013 is the sample size of each type in 2013; p.c. consumption is per capita consumption expressed in 2011 PPP U.S. dollars. 29 Table 9: Opportunity profile: Guinea, 2003 Rank Birth location Parental education Parental occupation n 2003 p.c. consumption 2003 1 Kankan none agriculture 1,608 634.83 2 outside Guinea elementary or above agriculture 6 648.25 3 Nzerekore elementary or above agriculture 186 724.56 4 Faranah none agriculture 1,375 737.11 5 Nzerekore none agriculture 2,288 755.25 6 Boke elementary or above agriculture 65 772.03 7 outside Guinea none agriculture 155 843.57 8 Kankan elementary or above agriculture 83 846.85 9 Kankan none other 1,532 864.71 10 Mamou elementary or above agriculture 21 892.79 11 Mamou none agriculture 1,424 935.20 12 Faranah elementary or above agriculture 80 1,002.34 13 Boke none agriculture 1,321 1,020.08 14 Kindia none agriculture 1,740 1,024.43 15 Boke none other 973 1,029.60 16 Nzerekore none other 1,204 1,032.19 17 Faranah none other 930 1,081.10 18 Kindia none other 2,413 1,082.47 19 Kindia elementary or above agriculture 98 1,134.86 20 Kankan elementary or above other 278 1,211.21 21 Nzerekore elementary or above other 542 1,213.00 22 Labe none other 1,270 1,224.88 23 Mamou none other 1,131 1,235.99 24 Boke elementary or above other 329 1,289.69 25 outside Guinea none other 361 1,345.27 26 Labe none agriculture 1,456 1,356.63 27 Kindia elementary or above other 1,049 1,394.03 28 Mamou elementary or above other 221 1,422.85 29 Labe elementary or above other 258 1,590.13 30 outside Guinea elementary or above other 137 1,660.16 31 Faranah elementary or above other 277 1,699.66 32 Labe elementary or above agriculture 55 8,370.83 Source : Calculations based on the surveys listed in Table 1. Note : n 2003 is the sample size of each type in 2003; p.c. consumption is per capita consumption and expressed in 2011 PPP U.S. dollars. 30 Table 10: Opportunity profile: Madagascar, 2005 Rank Ethnicity Birth location Parental education n 2005 p.c. consumption 2005 1 Coastal Fianarantsoa none 1,970 288.3178 2 Others Antananarivo none 17 307.1195 3 Coastal Mahajanga-Toamasina none 4,094 324.658 4 Highlanders Fianarantsoa none 1,307 331.2441 5 Coastal Toliara none 2,137 344.4594 6 others Toliara none 35 391.0437 7 Coastal Mahajanga-Toamasina elementary or above 2,977 406.876 8 Coastal Fianarantsoa elementary or above 1,143 406.9271 9 Highlanders Toliara none 296 408.4453 10 Coastal Antsiranana none 1,121 414.0577 11 Highlanders Fianarantsoa elementary or above 2,232 421.0522 12 Highlanders Antananarivo none 1,796 429.2367 13 Highlanders Antsiranana none 66 431.6902 14 Coastal Antananarivo none 65 440.788 15 Highlanders Mahajanga-Toamasina none 1,124 457.916 16 others Toliara elementary or above 51 463.7084 17 Others Fianarantsoa elementary or above 198 468.6341 18 Others Antananarivo elementary or above 17 471.0351 19 Coastal Toliara elementary or above 1,118 471.8447 20 Highlanders Toliara elementary or above 662 484.0782 21 Highlanders Mahajanga-Toamasina elementary or above 1,766 494.0706 22 Coastal Antsiranana elementary or above 1,041 506.3073 23 Highlanders Antananarivo elementary or above 3,079 509.4836 24 Others Mahajanga-Toamasina none 90 526.955 25 Others Fianarantsoa none 199 534.6658 26 Highlanders Antsiranana elementary or above 103 564.9221 27 Coastal Antananarivo elementary or above 84 658.7123 28 Others Mahajanga-Toamasina elementary or above 57 771.5343 29 Others Antsiranana none 57 782.28 30 Others Antsiranana elementary or above 49 1755.31 Source : Calculations based on the surveys listed in Table 1. Note : n 2005 is the sample size of each type in 2005; p.c. consumption is per capita consumption and expressed in 2011 PPP U.S. dollars. 31 Table 11: Opportunity profile: Malawi, 2010 Rank Birth location Parental education n 2010 p.c. consumption 2010 1 310. Chikwawa none 754 436.94 2 311. Nsanje none 795 452.60 3 313. Neno elementary or above 207 517.73 4 302. Machinga none 807 522.29 5 301. Mangochi none 861 524.74 6 313. Neno none 483 535.28 7 310. Chikwawa elementary or above 167 575.37 8 207. Mchinji none 622 608.09 9 206. Lilongwe none 1,332 649.88 10 306. Mwanza none 501 649.95 11 312. Balaka none 505 654.28 12 105. Mzimba none 851 658.78 13 101. Chitipa none 701 671.01 14 102. Karonga none 652 674.53 15 309. Phalombe none 727 676.20 16 208. Dedza none 908 683.15 17 other none 550 688.53 18 303. Zomba none 677 707.21 19 205. Salima none 671 710.82 20 204. Dowa none 780 731.48 21 103. Nkhatabay none 473 736.68 22 308. Mulanje none 944 737.53 23 312. Balaka elementary or above 235 751.51 24 306. Mwanza elementary or above 215 780.78 25 305. Blanytyre none 583 794.66 26 209. Ntcheu none 725 796.74 27 301. Mangochi elementary or above 167 798.94 28 201. Kasungu none 553 811.98 29 203. Ntchisi none 702 816.00 30 104. Rumphi none 447 819.09 31 311. Nsanje elementary or above 141 821.42 32 307. Thyolo none 950 840.31 33 204. Dowa elementary or above 311 855.84 34 202. Nkhota kota none 534 867.61 35 303. Zomba elementary or above 324 868.73 36 103. Nkhatabay elementary or above 401 880.57 37 206. Lilongwe elementary or above 406 883.34 38 207. Mchinji elementary or above 243 895.77 39 302. Machinga elementary or above 149 897.28 40 304. Chiradzulu none 690 919.62 41 104. Rumphi elementary or above 427 940.01 42 201. Kasungu elementary or above 311 942.09 43 107. Mzuzu City none 119 959.04 44 309. Phalombe elementary or above 172 961.65 45 101. Chitipa elementary or above 328 963.58 46 105. Mzimba elementary or above 423 993.17 47 308. Mulanje elementary or above 310 1,018.33 48 210. Lilongwe City none 296 1,040.87 49 208. Dedza elementary or above 244 1,079.29 50 304. Chiradzulu elementary or above 298 1,187.05 51 209. Ntcheu elementary or above 351 1,197.40 52 305. Blanytyre elementary or above 317 1,209.50 53 102. Karonga elementary or above 355 1,213.68 54 202. Nkhota kota elementary or above 222 1,281.76 55 307. Thyolo elementary or above 223 1,289.17 56 203. Ntchisi elementary or above 245 1,305.45 57 205. Salima 32 elementary or above 261 1,376.61 58 314. Zomba City none 313 1,382.68 59 315. Blantyre City none 262 1,545.74 60 107. Mzuzu City elementary or above 271 1,700.73 61 210. Lilongwe City elementary or above 516 1,754.84 62 314. Zomba City elementary or above 414 2,192.29 63 315. Blantyre City elementary or above 493 2,724.07 64 other elementary or above 222 3,021.64 Source : Calculations based on the surveys listed in Table 1. Note : n 2010 is the sample size of each type in 2000; p.c. consumption is per capita consumption and expressed in 2011 PPP U.S. dollars. Table 12: Opportunity profile: Niger, 2011-2012 Rank Birth location Ethnicity n 11-12 p.c. consumption 11-12 1 Maradi Peul 51 609.50 2 Maradi Kanouri-Manga 16 747.85 3 Tillaberi Touareg 251 753.47 4 Maradi Touareg 113 826.32 5 Tahoua Peul 29 893.59 6 Zinder Touareg 176 927.62 7 Tillaberi Djema 1,262 930.65 8 Maradi Haoussa 1,215 945.87 9 Tahoua Touareg 284 952.41 10 Dosso Djema 996 969.80 11 Dosso Kanouri-Manga 6 972.14 12 Zinder Peul 52 975.32 13 Zinder Kanouri-Manga 325 975.60 14 Zinder other 90 977.14 15 Dosso Peul 74 1,001.90 16 Dosso Haoussa 565 1,026.85 17 Maradi other 7 1,033.58 18 Zinder Haoussa 870 1,048.96 19 Tahoua Haoussa 1,108 1,066.47 20 Diffa Peul 198 1,121.57 21 Diffa Kanouri-Manga 516 1,127.07 22 Tillaberi Haoussa 198 1,219.05 23 Diffa other 170 1,224.04 24 Diffa Touareg 10 1,325.61 25 Tahoua Kanouri-Manga 7 1,358.41 26 Agadez Peul 55 1,371.44 27 Tillaberi other 14 1,432.79 28 Diffa Djema 7 1,467.50 29 Tillaberi Kanouri-Manga 6 1,510.25 30 Tillaberi Peul 89 1,549.28 31 Agadez Touareg 914 1,552.45 32 Agadez Kanouri-Manga 5 1,637.37 33 Dosso Touareg 10 1,686.33 34 Diffa Haoussa 18 1,697.75 35 Niamey Touareg 60 1,707.21 36 Dosso other 6 1,717.18 37 Agadez other 22 1,769.21 38 Niamey Haoussa 489 1,914.64 39 Agadez Haoussa 83 1,946.15 40 Niamey Djema 954 2,015.78 41 Agadez Djema 58 2,046.35 42 Niamey Peul 165 2,179.48 43 Tahoua Djema 26 2,329.03 44 Niamey other 112 2,514.84 45 Zinder Djema 33 2,531.27 46 Maradi Djema 17 2,602.06 47 Tahoua other 7 3,156.40 48 Niamey Kanouri-Manga 35 3,160.12 33 Source : Calculations based on the surveys listed in Table 1. Note : n 2011-2012 is the sample size of each type in 2011-2012; p.c. consumption is per capita con- sumption and expressed in 2011 PPP U.S. dollars. Table 13: Opportunity profile: Nigeria, 2010-2011 and 2012-2013 Rank Parental education Father sector Mother sector n 2010-11 eq. consumption 2010-11 n 2012-13 eq. consumption 2012-2013 1 none agriculture/not working agriculture/not working 4,941 995.59 4,923 1,304.95 2 primary incomplete agriculture/not working agriculture/not working 1,478 1,022.60 1,460 1,155.32 3 secondary incomplete agriculture/not working agriculture/not working 357 1,030.36 416 1,389.08 4 primary incomplete agriculture/not working other sectors 1,048 1,033.73 1,076 1,120.54 5 primary incomplete other sectors other sectors 357 1,184.65 301 1,473.13 6 primary complete agriculture/not working agriculture/not working 818 1,233.36 837 1,599.52 7 none agriculture/not working other sectors 1,471 1,235.93 1,314 1,653.99 8 none other sectors agriculture/not working 241 1,244.19 240 1,645.72 9 primary incomplete other sectors agriculture/not working 150 1,270.40 141 1,538.64 10 primary complete other sectors agriculture/not working 370 1,457.19 364 1,849.59 34 11 none other sectors other sectors 463 1,477.95 411 1,858.29 12 secondary incomplete other sectors agriculture/not working 248 1,485.66 261 1,772.21 13 primary complete agriculture/not working other sectors 459 1,526.24 442 1,779.26 14 secondary incomplete agriculture/not working other sectors 342 1,545.63 303 1,841.72 15 secondary complete or above agriculture/not working agriculture/not working 99 1,620.88 123 1,779.18 16 secondary complete or above other sectors agriculture/not working 189 1,628.39 254 2,339.49 17 primary complete other sectors other sectors 682 1,784.08 642 2,061.40 18 secondary incomplete other sectors other sectors 694 1,822.75 676 2,084.43 19 secondary complete or above agriculture/not working other sectors 216 2,006.31 175 2,602.63 20 secondary complete or above other sectors other sectors 590 2,350.06 591 2,782.42 Source : Calculations based on the surveys listed in Table 1. Note : n 2010-2011 (n 2012-2013) is the sample size of each type in 2010-2011 (2012-2013); p.c. consumption is per capita consumption and expressed in 2011 U.S. dollars. Table 14: Opportunity profile: Rwanda, 2000 Rank Birth location Parental education Parental occupation n 2000 p.c. consumption 2000 1 Kigali none agriculture 856 499.89 2 North none agriculture 1,969 503.94 3 North elementary or above agriculture 684 516.34 4 Kigali elementary or above agriculture 345 530.25 5 South none agriculture 2,744 542.20 6 West none agriculture 2,626 555.12 7 West elementary or above agriculture 996 577.33 8 South elementary or above agriculture 1,320 593.17 9 East elementary or above agriculture 531 615.80 10 East none agriculture 1,004 635.71 11 South none other 28 831.01 12 outside Rwanda none agriculture 325 907.77 13 East none other 3 938.79 14 outside Rwanda elementary or above agriculture 311 1,028.17 15 Kigali none other 8 1,295.43 16 West elementary or above other 65 1,521.86 17 outside Rwanda none other 17 1,639.40 18 South elementary or above other 78 1,678.08 19 East elementary or above other 12 1,712.88 20 North none other 3 1,777.56 21 North elementary or above other 33 1,796.77 22 West none other 16 1,840.53 23 Kigali elementary or above other 51 1,993.87 24 outside Rwanda elementary or above other 87 4,163.61 Source : Calculations based on the surveys listed in Table 1. Note : n 2000 is the sample size of each type in 2000; p.c. consumption is per capita consumption and expressed in 2011 PPP U.S. dollars. 35 Table 15: Opportunity profile: Tanzania, 2009-2010 and 2010-2011 Rank Birth location Parental education n 09-10 p.c. consumption 09-10 n 10-11 p.c. consumption 10-11 1 Kaskazini Pemba below elementary 198 629.43 218 762.69 2 Rukwa below elementary 103 693.35 104 642.80 3 Kigoma below elementary 188 706.21 214 663.65 4 Dodoma below elementary 189 784.72 211 731.62 5 Kusini Pemba below elementary 204 791.48 239 905.25 6 Kigoma elementary or above 135 792.55 220 925.11 7 Mwanza below elementary 230 809.92 332 827.18 8 Tabora below elementary 234 824.96 278 884.48 9 Shinyanga below elementary 365 827.26 462 784.89 10 Singida below elementary 122 837.61 142 817.32 11 Kaskazini Unguja below elementary 181 856.39 198 940.17 12 Ruvuma below elementary 199 859.33 222 828.95 13 Manyara below elementary 114 866.17 132 696.51 14 Mbeya below elementary 239 903.06 249 977.03 15 Rukwa elementary or above 100 917.05 133 894.88 16 Lindi below elementary 289 918.80 327 945.96 17 Mtwara below elementary 312 921.24 358 936.81 18 Tanga below elementary 204 929.60 234 790.71 19 Ruvuma elementary or above 178 930.56 231 795.08 20 Mara below elementary 97 930.74 100 1,011.30 21 Kusini Pemba elementary or above 133 939.32 155 1,080.71 22 Kaskazini Unguja elementary or above 78 955.11 104 1,155.49 23 Kusini Unguja elementary or above 54 958.46 82 1,169.14 24 Kaskazini Pemba elementary or above 105 983.33 129 1,005.68 25 Morogoro below elementary 187 992.61 207 995.97 26 Kusini Unguja below elementary 46 996.15 65 1,213.29 36 27 Tabora elementary or above 92 1,022.51 174 1,142.56 28 Kagera below elementary 157 1,047.17 195 1,050.00 29 Shinyanga elementary or above 186 1,111.98 312 1,068.13 30 Dodoma elementary or above 96 1,131.33 112 1,144.34 31 Manyara elementary or above 62 1,142.32 65 846.97 32 Lindi elementary or above 135 1,152.99 202 1,268.26 33 Iringa below elementary 214 1,158.59 218 1,039.38 34 Arusha below elementary 130 1,215.97 168 1,069.28 35 Mbeya elementary or above 158 1,218.00 250 1,274.27 36 Mjini/Magharibi Unguja below elementary 159 1,271.92 178 1,236.54 37 Mtwara elementary or above 213 1,272.47 304 1,316.40 38 Mara elementary or above 107 1,293.36 155 1,136.23 39 Mwanza elementary or above 186 1,294.52 328 1,209.00 40 Kagera elementary or above 194 1,307.17 280 1,347.72 41 Mjini/Magharibi Unguja elementary or above 289 1,322.01 341 1,505.85 42 Pwani below elementary 191 1,332.65 227 1,250.90 43 Singida elementary or above 102 1,340.07 137 1,324.37 44 Morogoro elementary or above 211 1,409.79 280 1,408.47 45 Iringa elementary or above 190 1,502.40 244 1,436.83 46 Tanga elementary or above 227 1,511.35 300 1,242.77 47 Pwani elementary or above 127 1,524.16 143 1,639.94 48 Kilimanjaro below elementary 210 1,546.54 219 1,259.20 49 Dar es salaam below elementary 146 1,609.63 167 1,606.85 50 Arusha elementary or above 108 1,616.80 140 1,433.71 51 Kilimanjaro elementary or above 276 1,938.00 336 1,950.82 52 Dar es salaam elementary or above 469 2,455.81 570 2,248.75 Source : Calculations based on the surveys listed in Table 1. Note : n 2009-2010 (2010-2011) is the sample size of each type in 2009-2010 (2010-2011); p.c. consumption is per capita consumption and expressed in 2011 PPP U.S. dollars. Table 16: Opportunity profile: Uganda, 2009-2010 and 2010-2011 rank HDI class ethnic group n 09-10 eq. consumption 09-10 n 10-11 eq. consumption 10-11 1 low Ethnic minorities 46 565.01 25 643.57 2 intermediate Western Nilotic 739 599.26 758 687.87 3 low Western Nilotic 671 655.57 630 737.31 4 low Central Sudanic 138 744.23 146 717.18 5 low Eastern Nilotic 499 765.74 502 668.88 6 high Ethnic minorities 535 796.01 432 634.67 7 intermediate Western lacustrine Bantu 646 957.96 552 1,052.61 8 intermediate Ethnic minorities 237 993.25 173 772.92 9 high Eastern Nilotic 89 1,055.22 75 874.06 37 10 low Eastern lacustrine Bantu 198 1,064.23 191 1,563.04 11 low Western lacustrine Bantu 146 1,070.37 120 1,063.68 12 intermediate Eastern Nilotic 284 1,112.25 266 902.45 13 intermediate Eastern lacustrine Bantu 880 1,166.45 901 1,009.49 14 high Western lacustrine Bantu 799 1,257.25 692 1,105.65 15 high Western Nilotic 56 1,296.31 54 837.56 16 intermediate Central Sudanic 296 1,301.76 287 945.67 17 high Eastern lacustrine Bantu 1,920 1,646.84 1,633 1,639.90 18 high Central Sudanic 15 2,499.20 17 1,935.08 Source : Calculations based on the surveys listed in Table 1. Note : n 2009-2010 (2010-2011) is the sample size of each type in 2009-2010 (2010-2011); p.c. consumption is per capita consumption and it is expressed in 2011 PPP U.S. dollars. Appendix III. MLD estimates Table 17: Nonparametric estimates (MLD) Country Sample Consumption Types Inequality IOp IOp % Max MLD Adj. IOp % Comoros 5,936 2,975 36 0.5358 0.0669 12.49 0.5225 12.81 Ghana 42,519 1,838 24 0.2949 0.0392 13.29 0.2741 14.30 Guinea 24,866 1,000 32 0.3121 0.0510 16.36 0.3071 16.62 Madagascar 28,951 415 30 0.2294 0.0179 7.82 0.2253 7.96 Malawi 30,137 855 64 0.3806 0.0744 19.54 0.3791 19.61 Niger 11,774 1,071 48 0.1562 0.0245 15.67 0.1549 15.80 Nigeria 2010-11 14,916 1,298 20 0.2623 0.0347 13.25 0.2376 14.62 Nigeria 2012-13 14,560 1,601 20 0.2603 0.0321 12.34 0.2367 13.57 Rwanda 14,112 641 24 0.3357 0.0425 12.66 0.3215 13.22 Tanzania 2009-10 9,119 1,133 52 0.2547 0.0448 17.59 0.2537 17.66 Tanzania 2010-11 11,391 1,112 52 0.2598 0.0410 15.79 0.2588 15.85 Uganda 2009-10 8,194 1,157 24 0.3450 0.0529 15.34 0.3310 15.99 Uganda 2010-11 7,454 1,039 24 0.3836 0.0558 14.55 0.3672 15.20 Source : Calculations based on the surveys listed in Table 1. Note : Per capita consumption is expressed in 2011 PPP U.S. dollars. Table 18: Parametric estimates (MLD) Country Sample Consumption Types Inequality IOp IOp % Comoros 5,936 2,975 91 0.5358 0.0875 16.34 Congo DR 39,578 1,535 402 0.2236 0.0494 22.10 Ghana 42,519 1,838 125 0.2949 0.0858 29.10 Guinea 24,866 1,000 96 0.3121 0.0352 11.27 Madagascar 28,951 415 445 0.2294 0.0719 31.35 Malawi 30,137 855 71 0.3806 0.1268 33.32 Niger 11,774 1,071 50 0.1562 0.0259 16.58 Nigeria 2010-11 14,916 1,298 40 0.2623 0.0425 16.22 Nigeria 2012-13 14,560 1,601 40 0.2603 0.0457 17.57 Rwanda 14,112 641 76 0.3357 0.0696 20.73 Tanzania 2009-10 9,119 1,133 41 0.2547 0.0590 23.16 Tanzania 2010-11 11,391 1,112 40 0.2598 0.0538 20.71 Uganda 2009-10 8,194 1,157 100 0.3450 0.0771 22.35 Uganda 2010-11 7,454 1,039 102 0.3836 0.1062 27.69 Source : Calculations based on the surveys listed in Table 1. Note : Per capita consumption is expressed in 2011 PPP U.S. dollars. 38 References Aaberge, R., Mogstad, M., Peragine, V. (2011). Measuring long-term inequality of oppor- tunity, Journal of Public Economics, 95(3), 193-204. Araar A. and Duclos J. Y. (2009), An algorithm for computing the Shapley Value, Mimeo, PEP and CIRPEE, Universite Laval. http://dad.ecn.ulaval.ca/pdf_files/shap_dec_ aj.pdf. Bourguignon, F., Ferreira, F.H.G., Men´endez, M. (2007). Inequality of opportunity in Brazil, Review of Income and Wealth, 53(4), 585-18. Brautigam, D.A, Knack, S. (2014). Foreign aid, institutions, and governance in Sub- Saharan Africa, Economic Development and Cultural Change, 52(2), 255-285. Checchi, D., Peragine, V. (2010). Inequality of opportunity in Italy, Journal of Economic Inequality, 8(4), 429-50. Cogneau, D., Mespl´ e-Somps, S. (2008). Inequality of opportunity for income in five coun- tries of Africa, in J. Bishop and B. Zheng (eds.) Research on Economic Inequality, Volume 16, Emerald Group Publishing Limited, 99-128. Corak, M. (2013). Income inequality, equality of opportunity, and intergenerational mo- bility, The Journal of Economic Perspectives, 79-102. De Ree, J., Nillesen, E. (2009). Aiding violence or peace? The impact of foreign aid on the risk of civil conflict in sub-Saharan Africa, Journal of Development Economics, 88(2), 301-313. Easterly, W., Levine, R. (1997). Africa’s growth tragedy: policies and ethnic divisions, The Quarterly Journal of Economics, 1203-1250. Elbers, C., Lanjouw, P., Mistiaen, J., Ozler, B. (2008). Re-interpreting sub-group inequal- ity decompositions, The Journal of Economic Inequality, 6(3), 1569-1721. Ellis, F. (2012). ‘We are all poor here’: economic difference, social divisiveness and target- ing cash transfers in sub-saharan Africa, Journal of Development Studies, 48(2), 201-214. FEWS (2013). GUINEA production and market flow map report, FEWS NET April 2013, http://www.fews.net/sites/default/files/documents/reports/Guinea-marketflow-final-en.pdf. 39 Ferreira, F.H.G., Gignoux, J. (2011). The measurement of inequality of opportunity: the- ory and an application to Latin America, Review of Income and Wealth, 57(4), 622-657. Ferreira, F.H.G., Gignoux, J., Aran, M. (2011). Measuring inequality of opportunity with imperfect data: the case of Turkey, Journal of Economic Inequality, 9(4), 651-680. Ferreira, F.H.G., Lakner, C., Lugo, M., Ozler, B. (2014). Inequality of opportunity and economic growth : a cross-country analysis, Policy Research Working Paper 6915, World Bank. Ferreira, F.H.G., Peragine, V. (2015). Equality of opportunity: theory and evidence, World Bank Policy Research Working Paper, 7217. Fleurbaey, M. (2008). Fairness, responsibility and welfare, Oxford University Press. Glick, P., Sahn, D.E. (2000). Schooling of girls and boys in a West African country: the effects of parental education, income, and household structure, Economics of education review, 19(1), 63-87. Hassine, N.B. (2011). Inequality of opportunity in Egypt, World Bank Economic Review, 26(2), 265-295. Goren E. (2014). How ethnic diversity affects economic growth, World Development, 59, 275-297 Lacina, B., Gleditsch, N.P. (2005). Monitoring trends in global combat: a new dataset of battle deaths, European Journal of Population, 21 (2-3), 145-166. Lambert, P.J., Aronson, J.R. (1993). Inequality decomposition analysis and the Gini co- efficient revisited, Economic Journal, 103(420), 1221-1227. Lassibille, G., Tan, J. P. (2005). The returns to education in Rwanda, Journal of African Economies, 14(1), 92-116. Lefranc, A., Pistolesi, N., Trannoy, A. (2009). Equality of opportunity and luck: defini- tions and testable conditions, with an application to income in France, Journal of Public Economics, 93(11), 1189-1207. Lloyd, C. B., Blanc, A. K. (1996). Children’s schooling in Sub-Saharan Africa: the role of fathers, mothers, and others, Population and development review, 265-298. 40 Luongo, P. (2011). The implication of partial observability of circumstances on the mea- surement of IOp, Research on Economic Inequality 19(2), 23-49. Madise, N. J., Matthews, Z., Margetts, B. (1999). Heterogeneity of child nutritional sta- tus between households: A comparison of six Sub-Saharan African countries, Population studies, 53(3), 331-343. ıguez, J. G. (2013). Inequality of opportunity and growth, Journal Marrero, G. A., Rodr´ of Development Economics, 104, 107-122. Milanovic, B. (2003). Is inequality in Africa really different?, World Bank Policy Research Working Paper, (3169). Moradi, A., Baten, J. (2005). Inequality in Sub-Saharan Africa: new data and new insights from anthropometric estimates, World Development, 33(8), 1233-1265. Peragine, V. (2002). Opportunity egalitarianism and income inequality: the rank-dependent approach, Mathematical Social Sciences, 44, 45-64. Peragine, V., Serlenga, L. (2008). Higher education and equality of opportunity in Italy, In J. Bishop, and B. Zheng, eds., Research in Economic Inequality, 16. Piraino, P. (2015). Intergenerational earnings mobility and equality of opportunity in South Africa, World Development, 67, 396-405. Ramos, X., Van de gaer, D. (2012). Empirical approaches to inequality of opportunity: principles, measures and evidences, Working Paper 259. ECINEQ. Roemer, J.E. (1998). Equality of opportunity. Harvard University Press, Cambridge, MA. Roemer J.E., Trannoy, A. (2015). ‘Equality of opportunity’, in Handbook of Income Dis- tribution, Anthony B. Atkinson and Francois Bourguignon (eds), vol. 2, Elsevier. Schultz, P. (2004). Evidence of returns to schooling in Africa from household surveys: monitoring and restructuring the market for education, Journal of African Economies, 13(2), 95-148. Thorbecke, E. (2013). The interrelationship linking growth, inequality and poverty in Sub- Saharan Africa, Journal of African Economies, 22(suppl 1), i15-i48. Tsounta, E., Osueke, A.I. (2014). ?,What is Behind Latin America’s Declining Income 41 Inequality?, IMF Working Paper N. 14/124. UNDP (2014). Human Development Report 2014 sustaining human progress: reducing vul- nerabilities and building resilience, United Nations Development Programme, New York. Van de gaer, D. (1993). Equality of opportunity and investment in human capital, Ph.D. Dissertation, Katholieke Universiteit Leuven, Belgium. Wairama B. (2001). Uganda: The marginalization of minorities. minority rights group international report. Wendelspiess, W., Soloaga, I. (2014). Iop: estimating ex-ante inequality of opportunity, Stata Journal, 14(4), 830-846. World Bank (2015). Purchasing power parities and the real size of world economies, 2015 World Bank. World Bank (2006). World Development Report 2006: equity and development, Washing- ton, DC: World Bank. Zhang, Y., Eriksson, T. (2010). Inequality of opportunity and income inequality in nine Chinese provinces, 1989–2006, China Economic Review, 21(4), 607-616. 42