What We Learn from Estimating the Genetic Contribution to Inequality in Earnings: Reply
In his criticism of several articles I have written alone or with colleagues, Arthur Goldberger concentrates on two issues. The first is the statistical methodology that yields our estimate that about 40 percent of the variance in earnings of white males at age 50 is attributable to differences in genetic endowments. He suggests the estimates rely on improper or overly strong assumptions. The second issue is whether in his words, the whole effort is misguided and that our results have no implication for policy. Clearly if the whole effort is misguided, any attempt to understand the statistical issues or to improve the methodology would be barren; hence, I concentrate initially on the question of what one can and cannot learn from (unbiased) estimates of the contribution of genetic endowments to inequality of earnings. There are some extremely important questions that we can answer and other important questions that we cannot answer with this information. Unfortunately people have tried to answer the unanswerable ones in the heated debate in the IQ literature. I believe Goldberger fears that some economists will mistakenly try to use my results to answer the last set of questions. It is helpful to conduct the analysis within the context of a human capital model in which parents and their children are assumed to invest optimally. We begin by assuming that a person's earnings depend upon his marginal productivity, which is a function of his skills. Let us assume further as numerous writers including Gary Becker and James Meade have done that a person's skills depend upon his genetic endowments (G) and his environment or investments in human capital (N). For simplicity let us also assume that a person's observed phenotypic earnings (Y) are related linearly to his genotype and environment as shown in equation (1).
- Export
- BibTeX
- Sources
- openalex