Abstract
The theories of distributive justice attaching importance to equality of opportunity assert that inequalities caused by circumstances, which are uncontrollable by a person, need to be fixed. As one of these theorists, Roemer has proposed the idea, namely equality of opportunity principle, that if a society is partitioned into equivalence classes by profiles of circumstances, then persons who have the same percentile among equivalence classes can compare their responsibilities with one another. Based on this idea, we propose a measurement method for the degree of inequality under the condition that inequality of opportunity is regulated. Specifically, we assume that if inequality of opportunity does not exist, then the persons who have the same percentile among equivalence classes should have the same level of social resources. Under this assumption, a vector of social resources is virtually rebuilt by the mean of each group where the percentiles correspond to one another. In order to verify how the degree of inequality decreases, we compare the two Gini coefficients before and after the regulation. In general, when we partition a group into disjoint subgroups by a certain index, the number of samples in each subgroup is not equal to one another. Therefore, it is hard to make one-to-one correspondence of the percentiles. For this, resampling procedures of a data is needed.
As one of the applications of our method, we analyze the effect of inequality of opportunity in terms of sex on the individual income based on the SSM 1985 and 1995 data. As a result, we find that the inequality of opportunity in terms of sex increases gradually in all job categories except for blue-collar workers of large corporations and farmers.
