Use the data in MINWAGE.DTA for sector 232 to answer the following questions.
(i) Confirm that lwage232t and lemp232t are best characterized as 1(1) processes. Use the augmented DF test with one lag of gwage232 and gemp232, respectively, and a linear time trend. Is there any doubt that these series should be assumed to have unit roots?
(ii) Regress lemp232t on lwage232t and test for co integration, both with and without a time trend, allowing for two lags in the augmented Engle-Granger test. What do you conclude?
(iii) Now regress lemp232t on log of the real wage rate, lrwage232t = lwage232t -lepit, and a time trend. Do you find co integration? Are they "closer" to being co integrated when you use real wages than nominal wages?
(iv) What are some factors that might be missing from the co integrating regression in part (iii)?