Additional findings: a. the estimated differences are not greatly changed when the sample is restricted to workers not covered by union contracts b. the estimated differences are quite stable across time and across countries c. there is no evidence that working conditions are worse in high-wage industries, so that the differences do not appear to represent compensating differentials d. workers in industries with higher estimated wage premia quit much less often. Thus, the differences appear to represent genuine rents e. workers who change to a job in a different industry on average experience wage changes approximately equal to the difference in the estimated wage premia for the two industries f. when workers lose their jobs because of plant closings, the wage cuts that workers experience when they find new jobs are much higher when the jobs they lost were in high wage industries