A Parametric Test for the Distinction between Unemployed and Out of the Labor Force Statuses

Whether or not to empirically consider two (employed versus not employed) or three (employed, unemployed, and out-of-laborforce) classifications in labor supply studies is a controversial issue. We develop a generalized censored probit likelihood function that nests both possibilities. A novelty of this likelihood function is that it allows researchers to test which representation of the labor market is appropriate as well as to estimate the degree to which classification errors may cloud inferences. Our empirical results demonstrate that classifying the three groups is useful to identify individuals' labor force and employment decisions separately. However, failure to incorporate classification ambiguities may result in unemployed rates that are understated and out-of-labor-force rates that are overstated.