Use the data set JTRAIN98 to answer this question.

(i) Add the unemployment indicator unem96 to the regression reported in equation (4.52). Interpret its coefficient and discuss whether its sign and magnitude seem sensible. Is the estimate statistically significant?

(ii) What happens to the estimated job training effect compared with equation (4.52)? Is it still economically and statistically significant?

(iii) Find the correlation between earn96 and unem96. Is it about what you would expect? Explain.

(iv) Do you think your finding in part (iii) means you should drop unem96 from the regression? Explain.