Labor demand research: Toward a better match between better theory and better data

a r t i c l e i n f o JEL codes: J23 J3 J4 J5 D4 F66 Keywords: Labor demand Input heterogeneity Labor adjustment costs Wage and employment determination Multiple fixed effects Establishment-level functions At first blush, most advances in labor demand were achieved by the late 1980s. Since then progress might appear to have stalled. We argue to the contrary that significant progress has been made in understanding labor market frictions and imperfections, and in modeling search behavior and heterogeneous preferences. Perhaps most notable have been the improvements in data, in the form of longitudinal matched employer–employee data, and in techniques and algorithms (e.g. for solving heterogeneous parameter models). In short, the Cinderella status of the field is frankly overdrawn. Nevertheless, a chief lacuna remains the need for a better match between theory and data. This paper provides a critical albeit eclectic assessment of these developments, along the dimensions of the static and dynamic theory of labor demand, wage formation, and estimation, noting advances and limitations. As is conventional, somewhat greater emphasis is placed on the latter. For the general reader, research on labor demand may seem to have progressed little in the past 25 years especially when compared with other areas of research within labor economics. Thus, the static theory of labor demand has long been established; reliable estimates of the most relevant parameters have long been available; and the essential features of labor adjustment and the underlying micro-mechanisms were identified in the late 1980s. Yet we shall argue that significant progress has nonetheless been achieved in this research area. On the theoretical front, the most important developments have stemmed from the recognition of the labor market frictions stemming from union presence and intervention, together with other imperfections that impart positive slope to the labor supply curve. To this mix one would need to add search frictions and heterogeneous preferences. Furthermore, the importance of good data – especially longitudinal matched employer–employee data – to progress in labor demand studies cannot be overemphasized. In fact, benefiting from increasingly detailed data, most of the progress achieved in the recent years has come from empirical studies that have allowed for a finer characterization of the labor input and renewed interest in the study of interrelated factor demand, the interaction between the labor and product markets, and a richer description of the dynamics of labor demand. Significant progress on the …