Age-specific cyclical effects in job reallocation and labor mobility

Abstract

We present an empirical analysis of job reallocation and labor mobility using matched worker-firm data for the Netherlands. Our data cover the period 1993–2002. We find that cyclical adjustments of the workforce occur mainly through fluctuations in worker entry for young and prime-age workers while for old workers they occur mainly through fluctuations in separations. Moreover, we find that employment dynamics of young workers are affected especially by national and sectoral employment fluctuations whereas employment of old workers varies especially with firm-specific employment changes.

Introduction

In the literature on dynamic labor markets two issues are discussed more or less separately. The first issue is whether workforce adjustments over the business cycle occur mainly through job creation or through job destruction; the second issue is whether employment dynamics are age-specific. Our paper studies both issues and additionally investigates diferences in cyclical effects depending on whether the cycle occurs at the firm level, the sectoral level or the national level.

The first issue concerns the question whether as a response to business cycle fluctuations firms adjust their workforce through workers entering or workers leaving. Different studies have found different results.¹ In U.S. manufacturing workforce adjustments over the cycle occur mainly via job destruction rather than through job creation (Davis and Haltiwanger, 1999, Blanchard and Diamond, 1990). This may be due to job creation being more time-consuming than job destruction (Davis and Haltiwanger, 1999). Recessions may be a time of “cleaning-up”, where outdated or unprofitable techniques and products are pruned out of the productive system and new technology is adopted. The adjustments take place in recessions, when the opportunity costs of forgone production are lowest (Caballero and Hammour, 1991). Similarly, Davis et al. (2005) find that contracting establishments in the U.S. rely on adjustments in the separation rate, not the accession rate, to bring about changes in employment. However, recently available new datasets indicate that, opposed to conventional wisdom, during the past two decades cyclical workforce adjustments in the U.S. occurred via the accession rate rather than the separation rate (Hall, 2005, Shimer, 2005). This is in line with findings for Europe, where workforce adjustments appear to occur mainly by changes in worker entry (Abowd et al., 1999) and job creation (Boeri, 1996).² For French firms, Abowd and Kramarz (2003) show this is optimal behavior given the cost structure of hirings and separations.

The second issue concerns the question whether there are age-specific differences in job and worker flows. Abowd et al. (2005) for example find that age-specific job flows may arise because older workers have different skill sets than younger workers. Studies based on both worker level data (Clark and Summers, 1981, Fallick, 1996) and firm level data (Abowd et al., 2005, Davis et al., 2005) find that young workers experience more employment dynamics than older workers. Firms may exploit business cycle fluctuations to change the composition of their workforce. Indeed, Blanchard and Diamond (1990) find clear cyclical differences in labor market flows for different age groups in the U.S.. In a recession, the highest increase in the flow from employment to unemployment is among young workers and the largest decrease in the flow from unemployment to employment is among young and old workers. Some other studies also find that both young (Clark and Summers, 1981) and old workers (OECD, 1995) bear a disproportionate share of the burden of a recession.

Age-specific effects are the result of both worker and firm decisions, which are based on human capital investments, adjustment costs and wage costs. Investments in firm-specific human capital make the employment relationship more productive. As a result, old workers who invested more in specific knowledge are less likely to quit than young workers. In a recession, firms may prefer to dismiss young workers first, because as yet they did not invest very much in firm-specific human capital and therefore have a lower productivity than older workers. It may also be easier to get rid of young workers because firing costs increase with age, tenure and wage. On the other hand, older workers may be laid off first, because they are overpaid due to the upward-sloping age-earnings profile (Lazear, 1979). Lazear (1995) combines both views in the “efficient layoff” rule, which states that both young and old workers will be laid off before prime-age workers.

This paper adds to the existing literature in two ways. First, we combine workforce adjustment and age-specificity which as far as we know have only been studied separately. We use information on job flows and worker flows to investigate whether age-specific effects in job reallocation and labor mobility exist and if so, whether these effects change over the business cycle. Second, in the analysis we distinguish between the effects of firm-specific, sectoral, and national economic shocks. This distinction allows a richer interpretation of cyclical phenomena and enables more precise policy recommendations.

We use a unique dataset consisting of matched worker-firm data with information about worker mobility. The data cover the period 1993–2002, which allows us to study job and worker flows over a full cycle.³ We find that age-specific cyclical effects in job and worker flows exist and that turnover varies most over the cycle for young and old workers. However, the effects of job loss are less harmful for young workers than for old workers. Moreover, we find that employment dynamics of young workers are affected especially by national and sectoral employment fluctuations whereas old workers' employment varies especially with firm-specific employment change.

The policy implications of our findings are twofold. First, there is not much that can be done about the cyclical employment fluctuations for young workers but policy makers should not worry about it. Second, since for many old workers separations are a one-way street, policy intervention is clearly relevant. Since firm-specific shocks are the main cause active labor market policies aimed at preventing job loss for old workers may prove to be useful.

The next section presents our data and some stylized facts. In Section 3 we report the results of our empirical analysis, where we estimate cyclical behavior of age-specific turnover. This is done both on the firm and on the worker level. Section 4 concludes.