World income distribution: which way?

I. INTRODUCTION

Several inter-related issues concerning income distribution in the world have come to the forefront in recent years. A major stimulus has been the many efforts to renew growth theory since the seminal works of Lucas [1988] and Romer [1990], which pose new questions for distribution. Studies of income distribution have also been helped by improvements to the database, despite considerable remaining shortcomings. The new empirical evidence on distribution is not always congruous, however.

One of the contentious issues is whether there is a general tendency for income distribution within countries to become more unequal over time in the wake of economic growth and globalisation. Extended and improved data now allow this question to be addressed for some 50 countries. The estimates suggest that in a few countries, income distribution has become significantly more uneven, and in another small number of countries, significantly more even. In the great majority of the 50 countries, however, income distribution, measured in different ways, has remained more or less unaltered between the individual years compared [Li et al., 1998; Sala-i-Martin, 2002]. Some investigations, based on cross-country observations, find economic growth to have a neutral effect on income distribution in general (for example, Dollar and Kraay [2002]). Ravallion [2001] shows that 'this aggregate picture hides more than it reveals', that is, the individual country experiences are much too split for an average to make much sense. (1)

Also the reverse link has received renewed interest. In contrast to earlier conventional wisdom, a string of studies in the mid 1990s found an equal distribution of incomes within countries to be beneficial for economic growth in subsequent periods [Alesina and Rodrik, 1994; Persson and Tabellini, 1994; Clarke, 1995; Perotti, 1996]. More recent findings suggest that this result must be qualified [Deininger and Squire, 1998; Forbes, 2000; Barro, 2000]

Another intensively studied question is whether there has been a change in absolute poverty over time. The World Bank [2000/01] and associated researchers [Chen and Ravallion, 2001] find the share of absolute poor in the developing world to have marginally declined (from 28 to 23 per cent) between 1987 and 1998, and the number of poor people to have remained more or less unaltered (at about 1.2 billion). These results are now contested. Reddy and Pogge [2002] and Deaton [2001] question the method used to construct the P$I poverty line. Sala-i-Martin [2002] finds a considerably lower incidence of poverty in the world and, most notably, that the P$1-a-day poverty rate has declined dramatically, from 20 per cent in the early 1970s to 5 per cent in 1998.

In the past few years, an extensive literature has studied a related controversial question: How has the relative distribution of incomes developed over time, across countries and globally? This is the main question to be addressed in the present article. The author will attempt to explain why the various studies have reached completely different results. In order to save space, only studies published in the most recent years are covered.

The most alarming estimates emanate from the United Nations Development Program [UNDP, 1999] and the World Bank [2000/01]. Both find that the ratio of per capita incomes in rich and poor countries (defined differently; see below) has doubled since 1960, while according to other studies, the distribution has remained relatively stable (see Table 1). The two studies reporting the most encouraging picture of world income distribution are Melchior et al. [2000] and Sala-i-Martin [2002]. Applying different distribution measures, they conclude that income inequality across countries (and globally) was substantially reduced between 1970 and 1998.