A problem Immediately arises since the t* are not observed. For all previoui jobs (1*1,…,n-l), a first-order approximation Is the actual completed job duration. For the current job, tft Is unobserved and no reasonable proxy exists.

lower probability of quitting (and of layoff) than other individuals. Thus:
Converting (A) into continuous terms, using equations (5)-(8), and integrating yields: Electronic Payday Loans Online

Equation (9) says that the I6** segment (1-2,…,n) will have three variables associated with it in the earnings function:1* linear and quadratic experlenc< and an Interaction between experience in the 1^ job and previous experience. Each Interaction term Is negative because the higher the starting age of the j< the lower the volume of Investment in that job.

An Important Implication of (9) is that the linear current job coefficient Is likely to have a relatively stronger effect (when compared to the linear coefficients of the previous jobs). The reason is that the total duration of the current Job ie unobserved, thus making It seem as If observed current experience matters more than previous experience. Note also that the linear coefficient of previous experience segments Is given by rA a . It can be see from (6) that measures the Investment ratio that would occur early In the life cycle at a very short Job. Since little on-the-job training Is likely to take place In short Jobs, a will lie close to *ero. Moreover, It can be shown that the negative depredation rate enters the linear coefficients of Job experience [see Mincer and Polachek (1974)]. Once depreciation Is Included In the analysis, (9) provides a partial explanation of why the coefficient of previous Jobs has been found to be Insignificantly different from *ero (and sometimes even negative) In other studies.