A post-2015 development goal for inequality?
The debate on a post-2015 inequality target is missing a fundamental point: what inequality would we like to eliminate?
High level committees within the United Nations and other organizations are busy thinking about a framework of development objectives to succeed the Millennium Development Goals, upon their 2015 deadline. The World Bank has recently announced its own two central goals, which are intended to guide its work in the years to come: to effectively eradicate extreme poverty (by reducing it to no more than 3% of the world’s population by 2030) and to promote “shared prosperity” (through growth in the incomes of the poorest 40% of the people in each country).
Inequality played – and still plays – a big role in these discussions. A UN taskforce suggested that addressing inequalities should be “the heart of the post-2015 agenda”. At the World Bank, much of the motivation for focusing the indicator of “shared prosperity” on the bottom 40% of the distribution in each country came from a desire to combat inequality. But should the UN establish a development target or goal focused explicitly on inequality? The question has motivated a lively online discussion over the last few months: Rolph van der Hoeven argued for it, but Martin Ravallion and Stephan Klasen (both of whom actually care about reducing inequality) argued against it. Adam Wagstaff retorted that maybe we do actually need targets for reducing inequality in health outcomes, since in a surprising number of places the health achievements of the poor and those of the better-off are growing further apart, rather than converging.
To my mind, the main question missing from this discussion has been Amartya Sen’s old query: “Inequality of what?” One reason (though perhaps not the only one) why sensible people are uncomfortable with setting a target for, say, income inequality, is that we have no idea what the “optimal” level of income inequality is. What is it? A Gini coefficient of 0.3? Or 0.2? Economists and, I dare say, other social scientists have offered no meaningful answer to this question. While we can all agree that the ideal level of extreme poverty is zero, we probably would not want to say the same for income inequality or, indeed, for inequality in years of schooling, or even longevity.
And the reason is that these outcomes involve choice. I may well choose to earn less than others, if I have the freedom to spend more time with my children, or to work in a job that I particularly enjoy. Equally, I may choose not to do a PhD, while others choose to spend five years of their lives doing just that. I may choose to smoke and drink, even if I know it may well make my life shorter than those of others.
Because freedom and choice matter, equality of outcomes is not what we are after – as John Rawls, Amartya Sen and others figured out in the 1970s. That is why many of these thinkers – philosophers, economists and political scientists – focused on equality of opportunity as the appropriate goal: The right “currency of egalitarian justice”, in the words of Gerry Cohen. We may not insist that everyone lives exactly the same number of years, studies as long, or earns precisely as many dollars as everyone else. But social justice should require that everyone has the same chances to live long, study hard, and earn high incomes. Choices may differ, but equality of choice sets is what fairness is about.
The view that equity is actually defined by equality of opportunities – coupled with certain minimum attainments in the space of outcomes, such as the absence of extreme poverty, or a minimum amount of mandatory education – was put forward in the World Development Report 2006 on Equity and Development, which has been mentioned before in this debate in The Broker. (Full disclosure: I was one of the authors of that report.) So it may very well be that we do not need – or want – a target for inequality of incomes. But should we adopt one for inequality of opportunities?
You may think that such a target may be difficult to quantify. After all, how does one measure inequality of opportunity? Is it even a measurable concept? Well, it turns out that it is: building on the work of many people, including Dirk van de Gaer in Ghent and John Roemer at Yale, a number of recent measures have been proposed. It is true that the literature has not fully “coalesced” yet, and that there are a number of alternative approaches. Yet one simple, and conceptually appealing, approach has recently been applied to a number of countries.
This approach starts from the idea that inequality of opportunity is that part of inequality associated with circumstances that people do not control, such as their race, gender, place of birth or family background. Suppose you divided the population into various subgroups, each of which is homogeneous in terms of those pre-determined circumstances. If we lived in a hypothetical world of equal opportunities, there would be no difference between the income distributions characterizing each of these subgroups. So, one very intuitive measure of inequality of opportunity is simply the share of overall inequality that is accounted for by inequality between the mean incomes of those subgroups (called types).
This Index of Economic Opportunity (IEO) has been computed for a number of countries, as reviewed here. The figure below, showing IEO measures ranging from 2% in Norway to 34% in Guatemala, is drawn from that study.
So we can get a quantitative handle on inequality of opportunities – and many agree that it this is the right space on which pro-equality policies should focus. But what is the target? How should we phrase an “Equality of Opportunity MDG”, so to speak? Various versions of this “egalitarian objective” have been suggested and, generally speaking, they are variations on a Rawlsian theme: let’s look at the type(s) with the lowest levels of welfare (proxied by income, education, or whichever outcome one is concerned with), and seek to raise their average well-being as much possible, over the stipulated time period.
To make this idea feasible in practice, we would have to observe and agree on a common set of circumstance factors across the participating countries. Race and gender are easy to observe in surveys, but they are not enough. In my experience, family background (parents’ education or occupation are a start) and place of birth are essential. If we partitioned the population of each and every country into these roughly comparable types, we could compile the country’s opportunity profile (see six of them here). Policy should then focus on the poorest type(s), and monitor their performance over time. As soon as that type catches up with the next poorest type, start focusing on both. And so on and so forth.
As John Roemer once put it: “The rate of economic development should be taken to be the rate at which the mean advantage level of the worst-off type grows over time” (Journal of Economic Inequality, 2006, 4: 243) Note, incidentally, that if you squint hard enough, this is not that far from the World Bank’s new indicator of shared prosperity: one could think of opportunities for children, and use household income as the sole circumstance. Define the worst-off type as the poorest 40%, and there you have it. Personally, I would like to see the international community go further: A finer opportunity profile can be much more informative of the real pockets of deprivation and exclusion in society, and correspondingly more useful for policy. It would take two things: political will, and better household survey data.