On-the-Job Search and Business Cycle Dynamics

Recent research shows that observed labor market flows can be explained in search and matching models only by assuming either implausibly large productivity shocks (Hall 2003) or an excessively high degree of real wage rigidity even for new hires (Shimer 2003). If this is not the case, the incentive to create new jobs in a boom is weakened since workers share the returns with employers. Fewer vacancies are opened, and unemployment falls by less than is evident from the data. However, this argument relies on treating the vacancy-unemployment ratio as the relevant measure of labor market tightness, which, in turn, is a central determinant of wages in the Nash bargaining approach. We argue in this paper that on-the-job search by workers expands the pool of searchers that firms can recruit from. Since job-to-job movements are highly procyclical, so is the number of on-the-job searchers. The bargaining power of incumbent workers and new hires therefore rises by much less in a boom than would be suggested by the standard vacancy-unemployment ratio alone. Instead, the ratio of vacancies to unemployed and employed searchers is the correct measure of labor market tightness. We quantitatively assess the role of on-the-job search in a fully specified, real DSGE model with endogenous labor market flows. There are two types of jobs, good and bad, which differ by creation cost and productivity. Vacant good jobs can be filled with employed and unemployed job seekers, while bad jobs only recruit from the unemployed. We consider alternative assumptions for the bargaining process, which is complicated by the fact that employed searchers have a fallback option that differs from the value of unemployment. The critical issue here is how firms bargain with employed workers, and what fraction of the rents from a new job workers can extract. We show that on-the-job search greatly reduces the volatility of true labor market tightness, compared to a benchmark model without job search. Therefore, wages are also much less volatile. At the same time, the predicted variation of the vacancy-unemployment ratio is much higher, and thus closer to the data. We are thus able to replicate the observed labor market dynamics in a full general equilibrium context. We also extend our real framework by including money and a nominal rigidity. In monetary business cycle models with sticky prices, real labor costs enter the price setting of firms. Hence, the cyclicality of real wages as determined by bargaining between workers and firms has implications for inflation dynamics. The less cyclical real wages are, the lower we can expect inflation volatility to be. We show that on-the-job search plays a crucial role in dampening fluctuations in real wages, thereby helping to resolve the challenges to the search and matching framework of labor market dynamics posed by Shimer (2003).Search, Nash Bargaining, Real Wage Rigidity, Labor Market Dynamics, New Keynesian