Occupational Specificity of Human Capital

In this paper we document substantial returns to occupational tenure. Everything else being constant, ten years of occupational tenure are likely to increase wages by at least $19\%$. Moreover, we show that when occupational experience is taken into account, tenure with an industry or an employer have little importance in explaining the wage one receives. This finding is consistent with occupational specificity of human capital. We utilize the newly released by the Panel Study of Income Dynamics (PSID) Retrospective Occupation-Industry Supplemental Data Files that retroactively recode the reported occupations and industries of household heads and wives for the 1968-1980 period. We argue that the methodology employed by the PSID in constructing the Retrospective Files - different from the one employed in the original coding of the occupation and industry affiliation data - minimizes the error in identifying true industry and occupation switches, a crucial step in constructing consistent occupational and industry tenure. We use the Retrospective Files to document that the originally coded occupation and industry affiliation data in the PSID is often incorrect, to evaluate various methods for identifying genuine occupation and industry switches, and to demonstrate that the use of raw or improperly corrected occupational or industry data may bias the results of an empirical investigation substantially.