One of the themes of the GM debate goes like this. On the one hand, the UAW is the problem, because it’s the high cost of union labor (and in particular, union retiree health benefits) that is crippling U.S. automakers. On the other hand, the UAW negotiated for those benefits fair and square, giving up higher current wages as part of the bargain, so it’s the fault of management for making promises they couldn’t keep. On the third hand, the UAW should have realized that when you negotiate for retirement benefits from a private corporation, one of the risks you take is that that corporation might go bankrupt. (For one example of these arguments, see Room for Debate at the NYT.)
Instead of touching that question any more than I already have, I wanted to raise the larger issue of whether unions are bad for business – which is what you would assume, given the lengths many companies go to in order to prevent unions from gaining collective bargaining rights. In general, this is a hard question to answer empirically. While you can observe differences between companies with unions and companies without unions, there is a huge problem of selection bias: since companies with unions are unlike companies without unions in many ways, you can’t say whether any differences in outcomes are due to the effect of the unions themselves, or due to the effect of other factors that would be there regardless of the unions.
John DiNardo and David Lee have an elegant way of getting around this problem in a 2004 paper, “Economic Impacts of New Unionization on Private Sector Employers: 1984-2001.” (The real economists out there probably know this paper already.) Instead of comparing all companies with unions to all companies without unions, they focus on companies where the union certification vote either barely won or barely lost, since these two companies are very similar to each other except for the treatment effect (having collective bargaining rights). This isolates the effect of unionization from other characteristics of the companies in question. They find that unions that barely win an election are successful in obtaining a collective bargaining agreement. Otherwise, however, the effect of successful unionization is insignificant on the company: differences in wages, employment, productivity, and output are all insignificant.
The UAW, historically, is a special case which people can debate for as long as they want. But the evidence is that in recent decades unions are not dangerous to firm survival.
Update: I forgot to add a link to a shorter summary of the work.
By James Kwak