Re: Kuznets Curve

["Forstater, Mathew" <ForstaterM@xxxxxxxx>]

Here is a short piece I put together for an encyclopedia entry on the topic. The actual entry is only 150 words, but this is the longer essay I wrote as background for the entry. The entry appears in the Routledge Encyclopedia of International Political Economy edited by Barry Jones.

--------------------------------------------------------------

Kuznets U-hypothesis
Mathew Forstater, UMKC

The results of the Luxembourg Income Study are in: there has been a significant rise in inequality in OECD countries since 1980 (Atkinson, Rainwater, and Smeeding, 1995).  Among the questions this raises are whether there are any implications for developing nations, the policies they pursue, and the futures that they should expect.  Conventional wisdom, in the form of the Kuznets U-hypothesis, states that in the course of economic development, as measured by growth of real per capita GNP, the size distribution of personal income will become at first more unequal, then stabilize, then become more equal, as a secular trend.  Of course, as is often the case, there are important differences between what became conventional wisdom and the initial propositions that inspired that wisdom.  This was especially so in the case of the Kuznets U-hypothesis, since Kuznets explicitly warned against applying to developing countries what he called his "collection of hunches" about the early experience of a few industrialized nations, both because it was based on "admittedly scanty evidence" and because of important historical and structural differences between the two groups of countries.  There are additional problems with the empirical work done on the Kuznets U-hypothesis, as we will see.

According to Kuznets (1965 [1955]), examination of the historical experience of a few presently developed nations shows that, at very low levels of per capita GNP, great inequality was limited by the minimal subsistence requirements of individuals with the lowest incomes.  As growth takes place, he argues, two forces will initially cause an increase in relative inequality.
The first is due to the concentration of savings among individuals in upper income groups.  This inequality in savings results in "concentration of an increasing proportion of income-earning assets in the hands of the upper [income] groups-a basis for larger income shares of these groups and their descendents" (1965 [1955], p. 264, original emphasis).  The second factor is the tendency for a shift in the industrial structure of the labor force away from agriculture toward manufacturing (also called the traditional-modern or rural-urban transition).  This increases inequality because the modern sector tends to be more unequal than the traditional, and per capita income is higher in the modern sector.  A relative increase in population employed in the modern sector is thus synonymous with the increase in the share of the population in the more unequal sector.  Since relative difference in per capita income between sectors doesn't change, inequality rises.

Kuznets posits that as growth continues, however, counter-forces emerge which result in decreased inequality.  Countering the concentration of savings in the upper income brackets are politico-legislative factors such as inheritance taxes, and demographic factors resulting in relative population decrease of upper income groups.  He also notes factors relating to social and technological change as additional forces countering the inequality-producing effects of savings concentration.

Offsetting increased inequality stemming from a shift in the industrial structure of the labor force away from the traditional towards the modern sector is "a rise in the income share of the lower groups within the non-agricultural sector of the population" (1965 [1955], p. 274).  In addition, Kuznets mentions that politico-legislative, demographic, and social considerations also come into play in reversing the trend toward greater inequality brought about in the early stages of development by this shift in industrial structure.  Thus, the forces that result in increased inequality in the early stages of growth are countered and an opposite trend toward increased equality occurs.

A great deal of empirical work has been attempted to test the hypothesis, especially since the early 1970s.  Felix Paukert (1973) conducted a cross-national study of 56 countries, which he divided into seven income groups.  He calculated the mean Gini coefficient and income per capita figures for each groups, which, when plotted on a graph and the points connected, result in an inverted U-shaped curve.  Moving from countries with a mean GDP/capita < $100 (1965 base year) to those of $101-200, there is a sharp rise in inequality.  Inequality continues to increase, although the curve flattens slightly, moving to the $201-300 grouping.  Inequality remains high but decreases slightly moving to the next group of $301-500.  It is here that Paukert identifies as the "turning point," as movement to the next grouping of $501-1000 is marked by a sharp fall in inequality (1973, p. 116).  Inequality continues to decline in the remaining groups ($1001-2000, and > $2001).

Ahluwalia (1976b) also does a cross-national study, but using regression analysis.  His sample covers 60 countries, including 41 developing, 13 developed capitalist, and 6 socialist countries.  He tests for the quadratic relationship to the log of GNP/capita using five different percentile groups as dependent variables.  These are the top 20%, the next 40%, the bottom 60%, the bottom 40%, and the bottom 20%.  The relative income of all percentile groups except the top 20% at first declines, and then rises, as GNP/capita increases.  The top 20% shows an opposite trend.  He also shows that the point at which the income share of different groups begins to change occurs at different levels of GNP/capita, first for the middle income groups and much later for the lower income groups, possibly part of a "'trickle down' process" (1976b, p. 310).
Both Paukert and Ahluwalia conclude that their findings are consistent with the Kuznets U-hypothesis, but there are several problems with this conclusion.  A problem with both analyses is their assumption that cross-national data are applicable to Kuznets' proposition, which concerned a secular trend.  Paukert justifies their use by stating that it is "a generally accepted procedure... forced on us by the scarcity and quality of historical data" (1973, p. 110).  Ahluwalia, while expressing caution, nevertheless makes a similar argument and tags the results "stylized facts."
Unavailability of one type of data, however, is not a valid justification for the use of another type.  The use of cross-national rather than time-series data for the testing of the Kuznets U-hypothesis assumes that current developing nations are reflections of an earlier stage of developed capitalist nations, and that the latter are a valid representation of the former's future.  This was explicitly denied by Kuznets himself, who emphasized the crucial importance of recognizing "the major differences between underdeveloped countries today and the presently developed countries in their pre-industrial phase, and the much greater obstacles to economic growth in underdeveloped areas that these differences imply" (1965 [1960], pp. 190-91).  There is no reason to assume that a country on a given point on the U-curve at a specific time either moved to that point from some previous point on the curve or will mover forward along the curve with growth in GNP/capita in the future.  Both Paukert's and Ahluwalia's claims to have derived secular "turning points" from cross-national data are unfounded.

Even if we accept the findings using cross-national data as "stylized facts," there are still problems.  The empirical work has largely eliminated testing the factors Kuznets postulated as causing changes in relative inequality (concentration in savings, shift in the industrial structure of the labor force, and their counter-forces), and instead imply a direct correspondence between the level of development and level of inequality.  Therefore, the causal mediators identified by Kuznets have been for the most part dropped or de-emphasized; income distribution is the dependent variable and GNP/capita the independent variable.
Paukert's data also reveals quite different results when individual countries are examined rather than their "group" mean.  Like Ahluwalia's "Cross-classification of Countries by Income Level and Inequality" (1974, pp. 8-9), Paukert's data taken individually show great variation in inequality at every level of income.  Here it should be noted that Paukert's income groups, and other divisions into "developing" and "developed" groups based solely on GNP levels, are arbitrary divisions.

Serious problems have been noted in Ahluwalia's regression equations.  He himself has stated (1976a, pp. 132-33) that "the essential complexity of a dynamic process cannot be adequately captured in a single equation."  Both Wright (1978, p. 60) and Saith (1983, p. 376) have shown that if the socialist "dummy" variable is left out, the R2 drops to zero, and if the six socialist countries are dropped from the sample, the R2 drops significantly.  Also, it is shown that the addition or deletion of two or three observations eliminates the significance of the quadratic coefficient (Saith, 1983, p. 378; Wright, 1978, p. 61).  Both authors additionally criticize Ahluwalia's equation as mis-specified due to the omission of crucial structural variables.
Other problems with data concerning income distribution, upon which these tests are based, are of mention.  Amin (1984, p. 4) points out that the tendency to falsify income reports for tax purposes, the disguising of personal consumption in the costs of firms, and the practical accounting difficulties of determining net from gross income, all cast doubt on the accuracy of measurements involved in such exercises.  Husbands and Money (1970, pp. 319-20) also warn that "any study of income inequality has to face the fact that many of the primary data on the subject are of dubious quality," and take the position that "problems of definition and the deliberate contrivances that distort income figures," demonstrated for the United Kingdom by Titmuss (1962), exist in other countries as well.

Kuznets' U-hypothesis was a self-described "collection of hunches" which proposed specific causal relationships between factors relevant to development and inequality.  He clearly and unambiguously warned against applying his "admittedly scanty evidence" of the experience of developed nations to the currently developing countries.  The empirical work that followed, of which the cases described above are representative, did not adequately heed this warning, however, and the problems with the results and conclusions of these studies have been identified.  Of even greater concern, however, are the dangerous policy implications of such scholarship (e.g., "don't worry about inequality, just keep pursuing growth").  The ever-widening gap between the rich and the poor demand such issues be handled with greater theoretical and empirical care.


Bibliography

Ahluwalia, Montek S., 1974, "Income Inequality: Some Dimensions of the Problem," in
H. Chenery, et al.: Redistribution with Growth, London: Oxford University Press.

Ahluwalia, Montek S., 1976a, "Income Distribution and Development: Some Stylized
Facts," American Economic Review, Vol. 66, pp. 128-35.

Ahluwalia, Montek S., 1976b, "Inequality, Poverty, and Development," Journal of
Development Economics, Vol. 3, pp. 307-42.

Amin, Samir, 1984, "Income Distribution in the Capitalist System," Review, Vol. 8, No.
1, pp. 3-28.

Atkinson, Anthony B., Lee Rainwater, and Timothy Smeeding, 1995, Income
Distribution in OECD Countries: Evidence from the Luxembourg Income Study, Paris, France: Organisation for Economic Cooperation and Development.

Husbands, C. T. and Roy Money, 1970, "The cross-national study of inequality: a
research note," American Sociological Review, Vol. 35, pp. 319-25.

Kuznets, Simon, 1965 [1955], "Economic Growth and Income Inequality," reprinted in
Economic Growth and Structure: Selected Essays, New York: W. W. Norton, pp. 257-87.

Kuznets, Simon, 1965 [1960], "Underdeveloped Countries and Past Growth," reprinted in
Economic Growth and Structure: Selected Essays, New York: W. W. Norton, pp. 257-87.

Paukert, Felix, 1973, "Income Distribution at Different Levels of Development: A Survey
of the Evidence," International Labour Review, Vol. 108, pp. 97-125.

Saith, Ashwani, 1983, "Development and Distribution: A Critique of the Cross-Country
U-Hypothesis," Journal of Development Economics, Vol. 13, pp. 367-82.

Titmuss, Richard M., 1962, Income Distribution and Social Change: A Study in
Criticism, London: Allen and Unwin.

Wright, Charles L., 1978, "Income Inequality and Economic Growth," Journal of
Developing Areas, Vol. 13, pp. 49-66.