On the Implications of Unobserved Technology and Preference Shifts for Aggregate Labor Demand and Supply

Abstract

The macroeconomic approach to the labor market aims at explaining aggregate labor market phenomena that have been present in many industrialized countries such as the United States or the western European countries in the postwar period. These phenomena include the permanent and simultaneous presence of unemployment and unfilled job vacancies as well as the fact that aggregate hours worked, employment and unemployment strongly co-move with the business cycle. At the same time, the amount and composition of labor supplied to the market and employed in production have substantially changed over the last decades. Understanding the driving forces and economic mechanisms that lead to the outcome in the labor market is of high importance to grasp the business-cycle fluctuations as well as the evolution of long-run growth of the economy as a whole.<br /> A large strand in the macroeconomic labor literature builds on the seminal work of Mortensen and Pissarides 1994 who employ search frictions in the labor market in order to explain the parallel existence of unemployment and unfilled vacancies in equilibrium. In their model, posting a vacancy is costly for firms, and matching in the labor market takes time depending on the tightness of the labor market, i.e., the ratio of unemployed workers seeking employment and open vacancies required to be filled. Firms and workers who meet in the labor market bargain over the wage given their economic conditions such as labor productivity or unemployment benefits.<br /> The dynamic version of the Mortensen-Pissarides model aims at replicating the procyclical fluctuations of employment and countercyclical fluctuations of unemployment respectively. Here, shocks to labor productivity increase the incentive for firms to post vacancies and therefore decrease unemployment. This means that the fluctuations in the labor market are prominently driven by labor demand. Building on this baseline model, a variety of issues have been raised in recent years in order to improve the empirical performance of the model, i.e., its ability to replicate the employment and unemployment fluctuations it aims to explain. For example, the introduction of rigid wages is conjectured to improve the model with respect to the volatility of the key variables (see Shimer 2005 or Hall 2005). Another issue of importance is the question whether the data support the assumption of a constant job separation rate, i.e., the probability with which an employed worker is separated from an existing employment relationship (see Fujita and Ramey 2006 or Ramey 2008).<br /> A related part of the literature incorporates search-and-matching in the labor market as in Mortensen and Pissarides into a general macroeconomic real-business-cycle (RBC) setup (see Merz 1995 and Andolfatto 1996 and also den Haan et al. 2000). This research was initially motivated to improve upon the empirical performance of the RBC model. In contrast to the standard model, it provides a more sophisticated framework which links the developments in the labor market to the key macroeconomic aggregates such as output, capital investment and consumption. Fluctuations in the basic RBC model are (mainly) driven by shocks to technological progress or more generally aggregate supply. As labor productivity shocks in the more general setup of the Mortensen-Pissarides model are widely interpreted as technology shocks (Shimer 2005), the RBC framework indeed provides the natural framework to link the cyclical movements of the labor market variables to the induced driving force of the business-cycle. However, the RBC paradigm about the cyclical movements of the variables and the sources of the business cycle is heavily disputed in the literature (see for example Gali 1999). In the tradition of this dispute, Chapter 1 challenges the conventional view that cyclical labor market dynamics are mainly driven by technology shocks by highlighting the role of shocks to the aggregate demand (here preference shocks) as a complementary source of fluctuations in employment and unemployment. Based on this analysis, Chapter 1 also contributes to many of the issues raised with respect to the Mortensen-Pissarides model such as its empirical performance, or the endogeneity of job separations.<br /> Another part of the macroeconomic labor literature has focused on the long run developments of labor demand. Also here, technological progress constitutes the main determinant of the labor demand of firms in the medium to long run. A wide literature has documented that this demand has shifted in favor of higher rather than lower educated workers in recent decades, so-called skill-biased technological progress (for example Katz et al. 1992 or Autor et al. 1998). Analyzing these shifts in the composition of inputs to aggregate production via a growth decomposition framework reveals not only the skill-bias of technological progress but also the degree of substitutability or complementarity between high and low skilled labor as well as capital in aggregate production (Krusell et al. 2000). Building on aspects of the above mentioned RBC literature as well as results and methods of Chapter 1, Chapter 2 investigates the implications of skill-biased technological change for the business cycle. Over and above linking the conventional driving forces of business-cycles and long-run growth to the developments in the labor market, Chapter 2 therefore attempts to identify sources of cycles and growth that originate in the labor market itself.<br /> Apart from labor demand, aggregate labor market supply evolves as a complex result of individual economic choices which are potentially affected by wages, labor market policies such as unemployment benefits or labor taxes, but also preferences for education and skills or social norms and culture. Shifts in the magnitude or composition of labor supply and consequently labor input are an important factor of actual and potential economic growth. In fact, this highlights the importance of the underlying heterogeneity between different workers in the labor force for the aggregate developments. Like labor demand, labor supply has shifted towards a higher ratio of high skilled workers in the recent decades (see again Katz et al. 1992 or Autor et al. 1998. This may be due to the aforementioned increase in skill demand, but is also attributable to an increased preference for education. Together with compositional changes with respect to skill, the age composition of aggregate labor supply has also changed. As more and more young persons tend to stay in education longer, they enter the labor force later in life, but are also better educated. Further, changes in the age composition of the population towards a higher share of older persons also affect the age composition of the labor and work force.<br /> In addition to growth, skill supply shifts have also importantly affected the business-cycle fluctuations of employment as shown in Chapter 2. Apart from this, the labor force in the aforementioned literature (and the first two chapters of this dissertation) was assumed to be constant. However, fluctuations in labor market participation behavior are present in the data, affect the aggregate labor supply and play an important role for the labor market business-cycle fluctuations (for a recent study on this, see Veracierto 2008. In the presence of labor market frictions as assumed in Chapter 1, a rise in labor market participation initially increases the pool of unemployed and therefore decreases the probability to find a job. Participation may increase over the cycle in response to rising wages, i.e., higher opportunity costs of staying out of the labor force. This effect strongly depends on the underlying wage elasticity of the labor supply and is often associated with the so-called "added worker" effect of persons, often females, in a liaison with a working partner.<br /> In addition to cyclical fluctuations, labor market participation has substantially increased over time, especially for females. This rise in participation has often been associated with cohort effects. Cohort effects generally encompass any factor associated with a particular birth year and constitute unobserved sources of labor supply shifts between different generations of workers. These shifts are often reflected in choices made early on in life such as marriage, maternity leave and/or education and are related to changes in the underlying preferences or cultural factors. Chapter 3 attempts to disentangle shifts in labor market participation that are due to movements of the business cycle or a change in labor market policies from those that are due to unobserved cohort effects.