Looking for Labour Market Rents with Subjective Data

A long-running debate in labour economics concerns the interpretation of industry and occupational wage differentials. One view is that these reflect compensating differentials for mostly unmeasured aspects of jobs: utilities are equalised between industries and occupations, but wages are not. Alternatively, there may be labour market rents: identical workers in some industries and occupations have better jobs (as in higher utility) than do others. This paper proposes a trick by which it is not necessary to measure job disamenities in order to test for the presence of rents. The correlation between the estimated 2-digit occupation coefficients from wage and job satisfaction regressions is examined. If wage differentials are rents, then “highpaying” occupations should also be “high satisfaction”. If wage differences are purely compensating differentials then no relationship should be found. An analogous approach is taken for industry dummies. Using eleven waves of British panel data, it is shown that there is a persistent positive and significant correlation between the estimated occupational dummies, but no relation between the industry dummies, both in level and panel regressions. Analysis of spell data shows that highrent occupations are reached through promotion, rather than worker-induced mobility, suggesting the existence of tournaments within firms. Last, I show that high wage rent occupations are also high social status (as defined by recent sociological research), whereas status has no relation with the non-rent part of wages. JEL Codes: C30, J28, J31, J41, M51.