The Distribution of Personal Income: Complex Yet Over-Simplified

Abstract

After years of apparent stability, the distribution of personal income in a number of countries has exhibited significant changes, as is illustrated in Figures 1 and 2 by the Gini coefficients of inequality for ten OECD countries. Since the late 1970s there has been a substantial rise in the Gini coefficient in the United States and the United Kingdom, although this may have ended in the 1990s. The pattern in other countries is mixed. Some have shown rising inequality, but others, notably those in Figure 2, demonstrate a variety of time paths which are not readily summarised.

The subject of this paper is the basic concept under consideration: the distribution among households of disposable income. I argue that this concept is, at the same time, both challengingly complex and misleadingly over-simplified. The complexity and the simplification need to be borne in mind when interpreting the evidence in Figures 1 and 2. This paper considers (1) the intricacy of the phenomenon to be explained and the limited contribution to such an explanation of what appears in most economics textbooks as the “Theory of Distribution”, and (2) how it is over-simplified as a measure of the welfare of households, or as the input into an assessment of the justice or injustice of a particular distribution. In each case, the paper puts forward four reasons, on the one hand, why the observed distribution of personal income is more complex than typically assumed, and four reasons, on the other hand, why the observed distribution is too crude to capture concerns of economic welfare. The points are far from novel, but they are often lost from sight in current discussions about rising income inequality.

It should be noted that the series for different countries are based on different definitions.The United States figures, for instance, are for gross income and are unadjusted for household size, whereas the United Kingdom figures relate to disposable income adjusted for household size using an equivalence scale. Not only the levels are not comparable, but also differences in definition may affect the measurement of trends over time.