Strikes as a Result of Imperfect Information

This paper proposes a model of the bargaining process that recognizes that negotiators have only imperfect information at their command. The author argues that the probability of a strike occurring will increase if one party bases its perception of the opponent's position on the same variables employed to form its own position, when the opponent's position is, in fact, based on different variables. Strike probability is therefore assumed to be related to movements in variables of interest to only one side. The model is tested on a sample of fourteen bargaining relationships drawn from the series Wage Chronology , published by the Bureau of Labor Statistics, which details settlements between particular firms and unions over a period of about thirty years. Estimates of strike probability are derived from the maximum-likelihood logit procedure. The findings suggest that although both relative and absolute wage increases in a firm have a negative effect on strike frequency, relative changes are more important. In addition, factors that tend to reduce the potential that the parties will misperceive the positions of one another also reduce the probability of a strike occurring. The analysis shows, for example, that strikes are less likely to occur in bargaining relationships whose agreements follow patterns. Finally, after controlling for fixed effects, the author finds that a strike is less likely to occur if one had taken place at the time of the previous negotiation, indicating that the strike serves a learning process for the parties.