Redistribution with African Characteristics

ABSTRACT

Progressive income redistribution is central to the development agenda in Sub-Saharan Africa, but is receiving much less attention than it deserves. This paper tries to rectify this oversight by exploring and explaining two specific manifestations of progressive redistribution during the period 1990–2010: first, declining levels of market income inequality in some states of the region and secondly, increasing use of fiscal means to reduce net income inequality in a broader set of states. While the dynamics of the first may not be sustainable, fiscal redistribution could become a more permanent feature of the redistributive landscape of the region, provided some challenges are met. The cross-national trends identified here could herald the arrival of more institutionalised, less emergency-prone welfare regimes in the region, buttressed by popular demand.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Philip Nel http://orcid.org/0000-0002-1325-307X

Notes

1 Well aware of the problems that beset data sources on SSA (Klasen and Blades 2013), I use the best available cross-country evidence on income inequality (see note 3). I do not have observations on all measures for all the states in the sample and for all time periods, so the dataset should be regarded as unbalanced. In some of the regressions, the number of states in SSA drops to 36.

2 A summary income inequality measure such as the Gini is sensitive to a transfer of income from one income group to another, no matter where on the overall distribution the transfer takes place. If the transfer takes place from a group higher on the income ladder (= ‘the not poorer’) to a group lower (= ‘the not richer’), the Gini will be lower and we call it a progressive transfer. Obviously, this can take place amongst the rich (from the richest 5% to the next richest 5%, for instance) or among the poor, or among the middle class. Hence, Dalton’s insistence on the term ‘not richer’ as a general term for any group relatively lower on the income ladder than another group, which he calls the ‘not poorer’.

3 The SWIID is preferred for cross-sectional analysis because of its coverage and higher degree of comparability compared to other alternatives (see Acemoglu et al. 2015). There is a high (0.8 plus) correlation between the SWIID data on SSA and other available datasets, and the broad tendencies deduced from the SWIID data below are also reflected in these other datasets, such as the UNU-WIDER World Income Inequality Data (WIID), the All-the-Ginis dataset of Branko Milanovic, the inequality data reported in the World Bank Development Indicators, and the University of Texas Estimated Household Inequality (EHII) dataset. However, SWIID has the best coverage of SSA of all the available datasets, and its methodology ensures a standardization that others only approximate.

4 See Lambert, Nesbakken, and Thoresen (2011) for a discussion of the empirical challenges we face in operationalising income redistribution given the different ways in which pre-tax and -transfer market income distributions are conceived. Solutions are data intensive and the standardisation provided by SWIID is the best that we can do in the case of SSA as a whole for the time being.

5 All of the states that registered a reduction in market income inequality also saw their Palma ratios decline (except Cape Verde and Mauritius for which we do not have income share data). That is, the income share of the bottom 40% increased relative to that of the top 10% (we do not have Palma data for Cape Verde and Mauritius).

6 Calculated as the exports of goods and services multiplied by total natural resource rents, both as percentages of GDP. Own calculations. Based on World Development Indicators, 2015. According to the 2014 African Economic Outlook (AFDB 2014), fuels and mining products accounted for 69.5% of total African exports in 2012.

7 To arrive at the predicted values, I regress MII and the income share of the poorest 40% on a range of variables that are identified in the literature as significant predictors of income inequality, namely per capita income (GDP per capita) and its square, the age dependency ratio, a measure of the depth of human capital, the flow of inward foreign direct investment (Figini and Goerg 2006; Hisako and Hamori 2009), the value added to GDP by agriculture, and a measure of institutional quality (‘Common law’ legal origin). The predicted values of these two regressions are used as the y-axis in Figures 2(a) and (b), while resource intensity of exports is plotted on the x-axis. Appendix describes the variables used.

8 The ASPIRE data and the SWIID record a very significant (p < .05) correlation of 0.39 (r-square) between fiscal redistribution as measured here, and the contribution of the SSA social protection programmes towards reducing inequality.

9 See report of the meeting in Accra at http://www.sidint.net/content/report-pan-african-conference-inequalities, and the text of the Statement at http://africainequalities.org/conference/wp-content/uploads/2015/11/ACCRA-DECLARATION-2014.pdf