In his seminal 1943 paper on the political business cycle, Michal Kalecki (1971) argued that industrial leaders feared employment because the economic insecurity created by unemployment was necessary to keep wages low and maintain work intensity and discipline on the shop floor. On the basis of this reasoning, Kalecki concluded that governments would not use demand management policies to achieve permanent full employment. In terms of current macroeconomic debates, Kalecki had sketched the outlines of a theory of the neutral or natural rate of unemployment based on the importance of disciplinary unemployment as a regulator of unit labor costs. Kalecki's pessimism about the prospects for employment was premised in part on a view of firms in which the threat of dismissal was the central motivational device used by employers, and employees had only a tenuous connection to employers. This assumption may have been appropriate when analyzing labor markets in the United States during the 1930's. Since that time, however, the spread of unions, implicit employment contracts, and large, bureaucratically organized enterprises has resulted in a modern U.S. labor market in which many workers enjoy long job tenure and in which many firms do not appear to rely on dismissal threats as their primary motivational strategy (see David Gordon, Richard Edwards, and Michael Reich, 1982; Sanford Jacoby, 1983; and my forthcoming paper). From this perspective, it is reasonable to ask whether the presence of long-term employment relations alters the regulatory role played by unemployment. This paper examines the effect that unemployment and long-term employment relations exert on the determination of unit labor costs. The central empirical findings can be briefly summarized. First, as suggested by Kalecki, movements towards employment increase the rate of growth of wages and reduce the rate of growth of labor productivity. Second, where long-term employment relations are prevalent, the effect of unemployment on both wage and labor productivity growth is diminished.

Observers of Silicon Valley's computer cluster report that employees move rapidly between competing firms, but evidence supporting this claim is scarce. Job-hopping is important in computer clusters because it facilitates the reallocation of talent and resources toward firms with superior innovations. Using new data on labor mobility, we find higher rates of job-hopping for college-educated men in Silicon Valley's computer industry than in computer clusters located out of the state. Mobility rates in other California computer clusters are similar to Silicon Valley's, suggesting some role for features of California law that make noncompete agreements unenforceable. Consistent with our model of innovation, mobility rates outside computer industries are no higher in California than elsewhere.