Clearly equation (9) cannot be estimated since the dependent variable Is earnings capacity which Is unobserved. Net earnings can be defined as ~ Cfc, so that In Yg ■ In Eg + In (1 – kg). Y Is closer to the empirically observed earnings since most Investment costs are likely to be forgone earnings. Assuming that kfc Is a small number, In (1 – kfc) * -kg. Equation (9) can then be written as:

At this point It Is perhaps appropriate to note an Important problem with equation (10)the parameters of Interest are not Identified. In particular, for a given rate of return only can be Identified; the parameter measuring the Importance of specific training, p^, cannot be estimated. Thus It Is Impossible to test directly whether the existence of specific training significantly affects the distribution of earnings. However, It can be shown that If t were known, equation (10) becomes: Electronic Payday Loans Online
Equation (11) adds In two variables that did not enter (10): R and an Interaction term between R and e . The reason that R enters negatively Into the equation is because of the existence of a positive correlation between investment and Job duration: the higher the remaining time in the current Job, the higher the incentive to invest more in the current time period t (since the payoff period to that part of investment which is specific is longer), therefore the lower current earnings. The interaction between R and ед is positive because the theory suggest more investment in longer Jobs: the longer e , the more Investment that occurred on the Job, therefore the higher the returns the individual is collecting at time t. Thus the specific training hypothesis can be tested by taking advantage of the nature of panel data. If Individuals can be observed over a relatively long period of time, then we can obtain (at least for a subset of the sample) completed time In the current Job.