Inequality Decomposition by Factor Component : A Rank-Correlation Approach Illustrated on the Taiwanese Case

The issue of measuring the relative importance of income factors in the distribution of household income is at the heart of studies of changes in income inequality. Usual decompositions methods mostly based on Shorrocks (1982) framework allow for a broad evaluation of the contribution of income sources. These decomposition procedures however suffer from two strong drawbacks: i) the share of inequality imputed to any income source is independent of the inequality measure used and, ii) the impact of changes in the marginal distribution of income sources cannot be disentangled from the effect of changes in correlation between sources. This paper generalizes the methodology briefly and intuitively initiated by Burtless (1999) and provides a general framework for inequality decomposition by factor components. This method, based on the concept of rank correlation, has two main advantages over usual decomposition procedures: i) it allows for a decomposition of observed changes in the whole income distribution and not just of a specific inequality index and, ii) it enables to isolate the specific impact of marginal distribution changes of income sources, as well as that of changes in correlation between sources. This paper illustrates the application of the methodology on Taiwanese data over the 1979-94 period. After a long phase of economic development with decreasing inequality, Taiwan has experienced a widening of income disparities since the end of the 70s. Many studies emphasize the role played by changes in family structure and especially to a rise in assortative mating of spouses, but fail to measure the specific impact of rising correlation between spouses' income. Using the methodology developed in the first part of the paper, we find that changes in the earning structure of household members have been strongly unequalizing and that the main force comes from the rise in the correlation between spouses' earnings