امور مالی پر هزینه خارجی و پویایی بازار کار

We study the role of agency frictions and costly external finance in cyclical labor market dynamics, with a focus on how credit-market frictions may amplify aggregate TFP shocks. The main result is that aggregate TFP shocks lead to large fluctuations of labor market quantities if the model is calibrated to the empirically observed countercyclicality of the finance premium. A financial accelerator mechanism thus amplifies labor market fluctuations by rendering rigidity in real wage dynamics. In contrast, if the finance premium is procyclical, which the model can be parameterized to accommodate, amplification is absent, and labor-market fluctuations display the Shimer (2005) puzzle.

This paper studies the role of costly external finance in the dynamics of labor markets. The starting point of the model is that firms require working capital to finance their operating costs, and the focus of the analysis is on how credit-market frictions may amplify neutral technology shocks. The environment in which this question is studied brings together a benchmark business-cycle model of financial frictions and a benchmark business-cycle model of labor search-and-matching frictions. The main result is that aggregate technology shocks can lead to large cyclical fluctuations of labor market quantities—in particular, unemployment, vacancies, and labor-market tightness, the quantities identified by Shimer (2005) as failing to be explained by standard search models. The framework quantitatively accounts for the empirically observed large fluctuations of labor markets very well, even though it is calibrated to the cyclical nature of financial conditions rather than to the cyclicality of labor markets. The model thus provides a joint explanation of some salient financial-market and labor-market dynamics. The property of the model economy that is crucial for amplification is a countercyclical external finance premium. In a version of the model featuring instead a procyclical external finance premium—which the model can be parameterized to accommodate—no amplification occurs, and the model displays the Shimer (2005) volatility puzzle. A broad message of the paper is thus that costly external finance can play an important role in amplifying shocks into the labor market, but it is not financing frictions per se that are important. Rather, the cyclical behavior of financing costs is crucial for the amplification mechanism; in particular, the mechanism imparts rigidity to the real wage. Real wage rigidity has been the main theme in the recent DSGE literature, as summarized by Rogerson and Shimer (2011). The cyclicality of the finance premium is governed by a single parameter in the model economy, the elasticity of firms' idiosyncratic productivity with respect to aggregate total factor productivity (TFP). Once this parameter is selected via simulated method of moments to match U.S. empirical evidence on the dynamics of the finance premium—in particular, a contemporaneous cyclical correlation with GDP of about −0.50—all other parameters regarding credit markets and labor markets hardly matter quantitatively for the response of the labor market to shocks to aggregate TFP. Furthermore, the model's predictions of the cyclical fluctuations of key labor-market quantities matches well cyclical fluctuations observed in the U.S., even though the model is calibrated to match the cyclical properties of the finance premium, not to match the cyclical properties of labor markets. The amplification the model displays is thus not merely qualitative in nature, but also a good quantitative fit. The mechanism of the model turns on how fluctuations in aggregate TFP shift the distribution of firms' idiosyncratic productivity, an effect referred to as a “technology spillover” or a “productivity correlation.” If there is no technology spillover, then the finance premium is (mildly) procyclical and labor-market dynamics in the face of TFP shocks are similar to those predicted by baseline DSGE search models such as Andolfatto (1996) and Merz (1995) despite the presence of credit market frictions. On the other hand, if technology spillovers are sufficiently positive—specifically, if an improvement in aggregate TFP raises sufficiently the mean of the distribution from which firms draw idiosyncratic productivity—the finance premium is countercyclical. Because firms borrow to finance their inputs, a countercyclical finance premium leads to sharper expansions of firm activity, including hiring activity, during aggregate upturns and sharper pullbacks of firm activity during aggregate downturns. A financial accelerator mechanism thus amplifies labor market fluctuations. The financial accelerator effect accounts for 60 percent of the model's ability to improve on standard search models in explaining labor market fluctuations. This channel operates by sharply reducing a firm's idiosyncratic risk of bankruptcy for a given size positive aggregate TFP shock, which lowers the bankruptcy premium charged by the firm's lenders. A lower finance premium in turn allows net-worth-constrained firms to expand activity, including new job-vacancy creation, more than otherwise. The other 40 percent of the model's mechanism operates through a direct productivity channel. At the firm level, productivity is the sum of an aggregate component and an idiosyncratic component. If aggregate TFP shocks shift positively the distribution of a firm's idiosyncratic productivity, a firm's effective productivity moves more than without the positive spillover. This direct productivity correlation induces sharper adjustments, including hiring adjustments, in response to shocks than if there were no productivity correlation between the macro level and the micro level, even if the cyclical behavior of the finance premium remained unchanged. Nascent evidence from firm-level studies is suggestive of the type of positive technology correlation present in our model. By constructing new measures of firm-level productivity, Petrin et al. (2011) document, among many other micro-macro supply-side empirical relationships, this type of productivity correlation. Of particular relevance for the calibration of our model, Petrin et al. (2011) compute an annual correlation between aggregate productivity growth, as measured by an aggregate Solow residual, and growth in firms' technical efficiency, which is a measure of firm-specific technology, in the range of 0.79–0.89. Our model, which is driven by only an aggregate TFP shock, portrays this high correlation in the extreme, assuming a correlation of unity. Nonetheless, we are still left with the task of selecting the appropriate elasticity of idiosyncratic productivity with respect to aggregate TFP, which we do via simulated method of moments to match the cyclical properties of the finance premium. The positive technology spillover in the model in this paper is also virtually identical to a key mechanism underlying Faia and Monacelli's (2007) study of optimal monetary policy in a New Keynesian model featuring financial frictions and perfect labor markets. At a theoretical level, we think it is important to know that a modeling strategy that has proven useful in a very different branch of the business-cycle literature turns out to also be important for the question under study in this paper. At an empirical level, the work cited above by Petrin et al. (2011), in addition to recent work by Foster et al. (2008), Acemoglu et al. (2012) and Oberfield (2012) (the latter two document and model this positive relationship), adds some realistic foundation to the Faia and Monacelli (2007) assumption.1 Given this positive relationship, the main question of this paper is how much amplification is induced in labor markets due to productivity shocks. Regarding the question addressed in this paper, the study most closely related is Petroksy-Nadeau (2009). Our work shares with his the basic ideas that financing frictions may induce an amplified response of the labor market to aggregate TFP shocks and that the cyclicality of the finance premium is important for the transmission mechanism. In these respects, the two studies are highly complementary. Several modeling choices, however, most importantly the ones that govern the precise amplification mechanism, distinguish our work from Petroksy-Nadeau (2009). First, as already noted, the way in which we construct our model allows for both a countercyclical as well as a procyclical external finance premium. The calibration of this central part of the transmission mechanism is guided by suggestive evidence on firm-level productivity and aggregate evidence on the dynamics of the finance premium; these features are intrinsically linked in our model because the cyclicality is governed by the degree of technology spillover. Because it can admit a procyclical finance premium, our setup thus especially highlights the centrality of a countercyclical finance premium, rather than simply the existence of financing frictions per se, in the amplification of TFP shocks. A second distinction between our work and Petroksy-Nadeau (2009) is that his model does not feature physical capital accumulation, whereas our model does. Third, in Petroksy-Nadeau (2009), financing frictions are assumed to affect only recruitment costs, which, at roughly 2 percent, are a small share of firms' total input costs. A more reasonable empirical view is that a (much) larger share of firms' input costs are subject to working capital requirements, and that the costs subject to working capital requirements are not merely recruitment costs. I take the broader view that all of firms' ongoing operating costs—wage payments and capital rental payments—require short-term working capital, whereas recruitment costs may or may not require financing. Empirically, this view is more in line with Buera and Shin's (2008) finding that a majority of firms' costs require working capital. Theoretically, this broader view motivates our adoption of Carlstrom and Fuerst's (1998) “output model” of all-encompassing financial constraints, rather than the more commonly employed “investment model” specification of Carlstrom and Fuerst (1997), Bernanke et al. (1999), and much of the DSGE literature on financial frictions, in which it is only investment goods that are subject to financing frictions. Another theoretical reason that leads us to adopt the output specification is that it has the most potential to interact with labor market frictions. Search and matching frictions directly impinge on the neoclassical “labor wedge” studied by Chari et al. (2007), Shimer (2009), and others. As Carlstrom and Fuerst (1998) show, if financial constraints affect firms' entire input bill, financing frictions appear in equilibrium as wedges in all factor markets, including labor markets. In contrast, with the investment specification, financing frictions dichotomize away from the labor wedge and instead appear only as an investment wedge. Of course, general equilibrium effects would likely cause financing frictions affecting investment wedges to interact with labor wedges arising from labor market frictions. But, in principle, a more promising and direct route to studying the interactions is if they both affect the same wedge. Moreover, this output-model formulation also allows us to connect to the results of Arseneau and Chugh (2012), who have developed a new notion of labor market wedges for general equilibrium labor search environments. The rest of the paper is structured as follows. Section 2 develops the baseline model. Section 3 presents the concept of the external finance premium, which is standard in business-cycle agency cost models, and presents the intuition for why positive technology spillovers have the potential to lead to a countercyclical finance premium and in turn to amplify labor-market fluctuations. In Section 4, we draw on the concept of efficiency for labor search environments developed by Arseneau and Chugh (2012) to show that, in contrast to Carlstrom and Fuerst (1998), financing frictions do not create an independent wedge in the economy's consumption-leisure margin; rather, the surplus-sharing rule by which wages are determined completely absorbs the effects of financing costs. Section 5 presents our main quantitative results, including a detailed parsing of the results in Section 5.4. As an application, Section 6 briefly assesses how well our model performs in explaining the sharp decline in employment and sharp rise in unemployment experienced during the current credit-market-induced recession. Section 7 provides a number of quantitative robustness exercises with respect to the labor-market and credit-market parameters of the model. The results are hardly changed compared to the baseline model, which strengthens the case that the productivity correlation is the most important feature of the model. Section 8 concludes.

We have developed a model in which shocks to aggregate TFP lead to large fluctuations in labor markets, and the amplification is mediated through financial conditions. The model performs well in quantitatively accounting for the volatility of vacancies, unemployment, and labor-market tightness, the quantities identified by Shimer (2005) as failing to be explained by standard search models. The key to the amplification is a countercyclical external finance premium driven by a positive spillover from aggregate TFP to firms' idiosyncratic productivity. Because of firms' financing needs, a countercyclical finance premium makes input costs cheaper during booms and more costly during recessions. This leads to much sharper swings in firm recruiting efforts over the business cycle, and thus much sharper swings in labor markets, than in standard search models of the labor market. Cyclical movements in financial conditions are responsible for 60 percent of the model's amplification, with the direct productivity effect of the spillover responsible for the other 40 percent. The key parameter of the model is the elasticity of the mean of firms' idiosyncratic productivity with respect to aggregate TFP. Firm-level empirical evidence suggests that this elasticity is positive, which is the direction of this relationship needed for the core mechanism of our model to operate. Premised on a positive elasticity, the numerical value for the elasticity is then selected to match the observed countercyclicality of the finance premium with respect to GDP. Thus, the model quantitatively matches labor-market volatility well even though the model is calibrated to the dynamics of financial conditions, not to the dynamics of labor markets. A number of extensions suggest themselves. One may be to allow for endogenous job separation, which has been known since den Haan et al. (2000) to potentially matter for many business-cycle issues. Our model performs well on the labor-market amplification dimension despite only exogenous separations, but a mechanism such as den Haan et al. (2000) or Fujita and Ramey (2007) may improve the model along the propagation dimension. Another extension would be to model credit-market frictions even more deeply, which is sure to be an active area of research in the coming years. Introducing other sources of variation besides just TFP shocks may allow for deeper connection with the rich firm-level evidence offered by studies such as Foster et al. (2008) and Petrin et al. (2011). Issues of fiscal and monetary policy are also likely to be interesting ones to explore in the type of framework we have developed. Perhaps most importantly, further empirical resolution on firm-level productivity and its correlation structure with aggregate shocks is needed. In our model, the elasticity of idiosyncratic productivity with respect to aggregate TFP is simply a parameter. Phenomena at the micro level for which it may stand in are, among others, selection effects in firm entry and exit and non-constant returns in production technologies. In a deeper micro-founded model, one would not want to label these phenomena “productivity,” even though in our representative-agent model they appear as such. Nonetheless, we view our model and results as suggesting that credit-market frictions may be an important driver of labor market fluctuations.