Assessing New Keynesian Theory Part II

Print/Save PDF

As promised last week, this post elaborates on NK ambitions to make friction-augmented general-market-equilibrium (GME) macro theory settled doctrine. No nonmarket modeling, including analysis of rational exchange that occurs inside highly specialized firms, is needed or, increasingly, tolerated.

In the illustrative article featured last week (“The State of Macro”, Annual Review of Economics (2009)) Olivier Blanchard concluded that the state of New Keynesian (NK) macroeconomics is good. Given that NK theory has no place for involuntary job loss – a phenomenon thought outside of the NK bubble to be an indispensable characteristic of macro instability – his judgement does not rise above bravado.

It is helpful that Blanchard, in the same article, goes on to describe the NK agenda to introduce unemployment into mainstream modeling: “How does one think about and introduce unemployment in a macro model? Here, fortunately, we can build—and are building—on a parallel effort, developed over the past 20 years by, in particular, Peter Diamond, Chris Pissarides, and Dale Mortensen (thus the name, DMP model; for a presentation, see, for example, Pissarides (2000)). In this approach, unemployment arises from the fact that the labor market is decentralized, where, at any time, some workers are looking for jobs, while some jobs are in need of workers. This has two implications: First, by necessity, there is always some unemployment—and, symmetrically, some vacancies. Second, as it takes time for a worker to find another job, and for a firm to find another worker, both the worker and the firm have some bargaining power. This implies that the wage—and by implication, the cost of labor, employment, and unemployment—depends on the nature of bargaining.” Comment: The unemployment produced in DMP modeling is voluntary; the unemployment produced in the instability that interests policymakers is involuntary. The two variants differ greatly; it is foolish to substitute one for the other. NK theorists also never adequately model rational, evidence-consistent “bargaining”. GEM theory demonstrates that modeling rational-behavior workplace equilibrium, not bargaining, provides by far the most evidence-consistent explanation of wage rigidities and involuntary job loss.

“This approach has proven extremely productive on its own. In contrast to the representative agent approach, it forces one to take into account the fact that the labor market is a market characterized by large flows, e.g., flows of job destruction and creation and flows of workers between employment, unemployment, and nonparticipation. It allows one to think about the effects of labor market institutions on the natural rate of unemployment. It also allows one to think about whether and how fluctuations affect reallocation, and whether some of the fluctuations may be due to variations in reallocation intensity. The model is sufficiently realistic in its description of the labor market that it can be confronted to the data, be it micro data on workers, micro data on firms or, even better and increasingly available, matched panel data on workers and firms.” Comment: Is this hand-waving to distract us from NK theorists being unable to accommodate evidence-consistent involuntary job loss? The unemployment described here is still voluntary; it is a fact that almost all of the job loss produced in recession is involuntary. The capacity of the DMP model to adequately explain mass layoffs in a recession is a pipe dream.

“The central question, however, whenever we explore the implications of a specific imperfection for macro fluctuations is twofold: First, how does such an imperfection affect the dynamic effects of shocks on activity? Second, does it lead to the presence of other shocks, which may be an important source of fluctuations in activity? In the context of labor markets, we have only begun to explore the answers. Crucial to the answer is the response of real wages to labor market conditions (see, for example, Shimer 2005, Hall 2005). Decentralized wage setting implies the existence of a wage band within which both the firm and the worker are willing to continue their relationship. The existence of such a band implies that, so long as it stays within the band, the real wage may move less than the boundaries of the band. In less formal terms, the presence of a band allows for more real wage rigidity than would be implied by a competitive labor market. This real wage rigidity does not by itself have implications for existing matches, which remain profitable so long as the wage remains within the band. If (a big if and clearly an additional assumption), however, the same real wage is also paid to new hires, then real wage rigidity has important implications for fluctuations: Combined with nominal rigidities, more real wage rigidity implies less pressure of activity on inflation; this in turn implies stronger and more persistent effects of shifts in aggregate demand, and stronger and more persistent effects of supply shocks such as increases in the price of oil, on activity. The presence of a wage band implies that real wages can be more rigid than their competitive counterparts.” Comment: The wage rigidities produced by the hypothesized bands are insufficiently robust to generate involuntary job loss; they are quickly liquidated by rational wage recontracting. This direction of NK research demonstrates the pressing mainstream need for chronic wage rent paid in information-challenged workplaces, microfounded in the generalized-exchange theory. Moreover, adverse demand disturbances produce, in combination with workplace-center nominal wage rigidities, evidence-sized involuntary job loss; real shocks produce relatively tiny numbers.

“The questions, however, are whether real wages, particularly the real wages of new hires, which are the wages relevant for the hiring decisions of firms, are indeed more rigid, and if so, why. This is also a hot topic of research. Theoretical work based on the exploration of constraints across workers’ wages within a firm and empirical work based on micro evidence on the wages of existing workers and new hires are proceeding apace. The next stage appears to be an integration of the market frictions that characterize the DMP model with those of efficiency wage models, which can explain wage setting within firms and, in particular, the relation between wages paid to existing workers and to new hires.” Comment: The relative rigidity of new-hire wages is not particularly important in evidence-consistent modeling. GEM theory rationally motivates LEV downward labor-price rigidity for all the firm’s workers, powerfully rooting involuntary job loss in optimization and decision-rule equilibrium. Original efficiency-wage theory modeled by Solow and Annable has turned out to be crucially important to stabilization-relevant macroeconomics.

In closing, there is much wrong with NK labor pricing that has been produced in pursuit of Blanchard’s research agenda, especially in the context of the alternative provided by the intuitive generalization of exchange. Consider four summary critiques. First is the casual rejection of Occam’s Razor. GEM modeling produces evidence-sized involuntary unemployment from the interaction of microfounded wage rigidities and adverse nominal demand disturbances; the workplace-equilibrium story is much more straight-forward than whatever Blanchard was peddling. Practitioners, who are puzzled (and often amused) by the Blanchard-outlined research, easily recognize the intra-firm narrative. Second, market-friction general equilibrium (FGME) modeling simply assumes, not derives,  a super friction capable of evidence-consistent suppression of wage recontracting. Barro’s critique rooted draws on the immediacy and power of forced job loss to persuasively reject the existence of the hoped-for market friction.

Third, any significant joblessness produced in the NK actual market-friction model is voluntary. An extraordinary volume of NK research has been directed at generating a tiny amount of counter-cyclical movement in frictional (voluntary) unemployment. The macro academy has looked the other way as market-centric NK theorists conveniently ignore involuntary job loss that occurs in the millions in periodic recessions. Finally, the NK designation of market frictions as the cause of counter-cyclical unemployment is badly inconsistent with the plentiful evidence of how wages are set in highly specialized economies. When are mainstream theorists going to pay attention to the full range of macro facts?

Blog Type: New Keynesians   

Write a Comment

Your email address will not be published.