The Art of Labormetrics
Using a wide array of examples from the literature and from original estimates, this essay examines the pitfalls that make good empirical research in labor economics as much art as science. Appropriateness and cleanliness of data are considered, as are problems of extreme observations and interactions. The validity of attempts to produce exogeneity using instrumental variables and natural experiments' is examined, as are the treatment of selectivity and unobservable individual effects. Testing empirical results to ensure that they make sense is stressed along with the importance of clear, economical and useful presentation of those results.