Return to our data on women’s labor force participation and consider the possibility of adding A_i, the age of the ith woman, to the equation. Be careful when you develop your expected sign and functional form because the expected impact of age on labor force participation is difficult to pin down. For instance, some women drop out of the labor force when they get married, but others continue working even while they’re raising their children. Still others work until they get married, stay at home with young children, and then return to the workforce once the children reach school age. Malcolm Cohen et al., for example, found the age of a woman to be relatively unimportant in determining labor force participation, except for women who were 65 and older and were likely to have retired.11 The net result for our model is that age appears to be a theoretically irrelevant variable. A possible exception, however, is a dummy variable equal to 1 if the ith woman is 65 or over, 0 otherwise.

a. Look over the data set in Table 13.1. What problems do you see with adding an independent variable equal to 1 if the ith woman is 65 or older and 0 otherwise?

b. To get practice in actually estimating your own linear probability and logit equations, test the possibility that age (A_i) is actually a relevant variable in our women’s labor force participation model. That is, take the data from Table 13.1 and estimate linear probability and logit versions. Then use our specification criteria to compare your equation with the parallel version in the text (without A_i). Explain why you do or do not think that age is a relevant variable.