Working Paper No. 521 | November 2007
Extending Oaxaca’s Approach
This paper extends the famous Blinder and Oaxaca (1973) discrimination in several directions. First, the wage difference breakdown is not limited to two groups. Second, a decomposition technique is proposed that allows analysis of the determinants of the overall wage dispersion. The authors’ approach combines two techniques. The first of these is popular in the field of income inequality measurement and concerns the breakdown of inequality by population subgroup. The second technique, very common in the literature of labor economics, uses Mincerian earnings functions to derive a decomposition of wage differences into components measuring group differences in the average values of the explanatory variables, in the coefficients of these variables in the earnings functions, and in the unobservable characteristics. This methodological novelty allows one to determine the exact impact of each of these three elements on the overall wage dispersion, on the dispersion within and between groups, and on the degree of overlap between the wage distributions of the various groups.
However, this paper goes beyond a static analysis insofar as it succeeds in breaking down the change over time in the overall wage dispersion and its components (both between and within group dispersion and group overlapping) into elements related to changes in the value of the explanatory variables and the coefficients of those variables in the earnings functions, in the unobservable characteristics, and in the relative size of the various groups.Download:Associated Program(s):Author(s):Joseph Deutsch Jacques Silber
Working Paper No. 513 | September 2007
This paper begins by proposing two cardinal measures of inequality in life chances. Using as its database a matrix in which the lines correspond to the social category of parents and the columns to the income distribution of their children, it then highlights the importance of the marginal distributions when comparing social immobility within two populations. It also shows how it is possible to neutralize differences in these margins. The idea is to adapt a method used in the field of occupational segregation measurement that allows one to make a distinction between differences in gross and net social immobility, assuming that the marginal distributions of the two populations are identical. Borrowing ideas from recent literature on the equality of opportunity, the paper then defines the concept of an inequality in circumstances curve and relates it to that of a social immobility curve.
Two empirical datasets are used to determine the usefulness of the concepts presented. The first dataset comes from a survey conducted in France in 1998 and allows one to measure the degree of social immobility and of inequality in circumstances on the basis of the occupation of fathers or mothers and the income class to which sons or daughters belong. The second dataset, drawn from a social survey conducted in Israel in 2003, is the basis for a study of social immobility and inequality in circumstances, emphasizing the transition from the educational level of the fathers to the income class of the children. Both illustrations confirm the usefulness of the analytical tools described in this paper.Download:Associated Program:Author(s):Jacques Silber Amedeo Spadaro
Working Paper No. 509 | August 2007
This paper begins with an examination of various ways of measuring unemployment and, borrowing ideas from the poverty measurement literature, proposes four new general unemployment indices. The first of these is parallel to the Sen poverty index; the second, to the Sen index’s generalization by Shorrocks; the third, to the FGT poverty index; and the fourth, to the Watts poverty index. The authors then present an empirical illustration based on Swiss data compiled at the state, or canton, level, using the so-called Shapley decomposition to determine the contribution of three components—the traditional unemployment rate, the average unemployment duration, and the inequality in the unemployment durations—to the differences between the values of the four proposed indices, both within a given canton and within Switzerland as a whole. The paper concludes with a discussion of the assumptions made about the maximum unemployment duration for the purposes of the study, and their impact on the results obtained.Download:Associated Program:Author(s):Joseph Deutsch Yves Flückiger Jacques Silber