Nowhere does the absence of microfounded meaningful wage rigidity (MWR) do more damage than in the analysis of business cycles. The evidence from modern model-building research is conclusive. Stabilization-relevant rational-behavior macroeconomics without MWR cannot be done. Yet NK theorists continue to try, for two reasons. First, the subject is obviously important. A decade ago, a recession generated multi-trillion dollars in measured costs and threatened much more. It would be nice to have a mainstream macro model that is useful, not useless, in such perilous circumstances. (Recall that “useless” was the judgment of stabilization policymakers.) Second, most NK theorists are unaware that the GEM Project has derived MWR from optimizing behavior organized by general decision-rule equilibrium. Microfounding MWR is still widely thought to be a problem that is too difficult to solve, burdening research on the inherent cyclicality of highly specialized economies with unhindered wage recontracting.
The GEM Blog has made sport out of ever-shameless NK theorists, dwindling in number, who continue the futile attempt to rigorously model recognizable cycles without MWR. I cannot resist making mock. A post two weeks ago looked at Jan Eeckhout and Ilse Lindenlaub’s “Unemployment Cycles”, recently published in the American Economic Journal: Macroeconomics. They modeled cycles, analysis which the article asserts to be evidence-consistent, by extending the voluntary joblessness produced in labor-market search theory to include employed workers looking for a better job. Their evidence consistency permits ignoring forced layoffs, which they must know – everybody else does – accounts for most of the increase in unemployment in recessions.
How does their theory of unemployment get away with omitting the most important, by far, class of unemployment? The answer is as simple as it is unfortunate. A necessary condition of rational-behavior involuntary job loss is microfounded MWR, which uniquely suppresses wage recontracting. NK theorists have long given up microfounding MWR, apparently because the problem is too difficult. But that’s OK. Since the AEJ:M editors have no interest in how to microfound MWR, it is does not harm a paper’s prospects to pretend, even if titled “Unemployment Cycles”, that forced layoffs do not exist. Hamlet can still be called Hamlet without the Prince.
More silliness. In a recent issue of the American Economic Review, Paul Beaudry, Dana Galizia, and Franck Portier (BG&P) present a paper titled “Putting the Cycle Back into Business Cycle Analysis”. From the authors: “Are business cycles mainly a response to persistent exogenous shocks, or do they instead reflect a strong endogenous mechanism which produces recurrent boom-bust phenomena? In this paper we present evidence in favor of the second interpretation and we highlight the set of key elements that influence our answer. The elements that tend to favor this type of interpretation of business cycles are (i) slightly extending the frequency window one associates with business cycle phenomena, (ii) allowing for strategic complementarities across agents that arise due to financial frictions, and (iii) allowing for a locally unstable steady state in estimation.”
This post will not conduct a detailed walk through the analysis, which is another example of the “angels-dancing-on-the-head-of-a-pin” obsession that is familiar in modern stabilization theory. Suffice it to point out that their endogenous-mechanism argument is little more than the familiar search for evidence, however weak, that supports mainstream market-centric general-equilibrium modeling. It does not surprise; the defense of that model class has become the principal goal of NK research.
Playing by modern rules, BG&P rely on “spectral density properties of many trendless macroeconomic aggregates, such as work hours, rates of unemployment, and risk premia”. This class of evidence is a favorite of EK theorists largely because it accommodates a wide range of explanatory theories. The big tent includes both those that BG&P promote and reject. The weakness of spectral density evidence has become so well known that it must be acknowledged. From the authors: “the presence of such [evidence] does not necessarily imply strong endogenous forces”. BG&P use it anyway; they have little else. They must ignore other, much more powerful evidence that inconveniently supports the main alternative approach, i.e., exogenous macro shocks, powerfully propagated by nominal demand disturbances. (See below.) An example is the on-point evidence that contracting aggregate spending is always followed by contracting employment and output. No vague spectral-density analysis is required for that insightful fact.
Early Keynesian analysis. Decades ago, EK theorists devised an analytic framework for business-cycle modeling that is much more closely aligned with the evidence than modern NK research. The impulse-propagation model posits some exogenous macro shock that is, given their assumption of downward wage rigidity capable of suppressing labor-price recontracting, powerfully propagated by weakening nominal demand and its associated multipliers. The GEM Project has substantially strengthened the EK argument, both microfounding MWR and usefully separating nominal demand disturbances (DD) into two classes. Stationary demand disturbances (SDD) are eventually self-correcting, helped along by automatic stabilizers augmented by central-bank “lean-against-the-wind” intervention. Nonstationary demand disturbances (NDD), by contrast, feature contracting total spending that overwhelms automatic stabilizers and orthodox central-bank intervention. If not contravened, extreme instability induces rapidly cumulating forced job and income loss, collapsing output and profits, impaired lending by and funding for financial institutions, gathering price deflation, widespread debt default, wealth destruction, and cataclysmic depression. The GEM Project has identified an underappreciated keystone concept here: investor/lender assessment of the credibility of stabilization authorities’ trend real-side (full-employment) objective. (See the website’s e-book, chapter 6.)
The EK impulse-propagation framework robustly explains actual instability that occurs in highly-specialized market economies. It also usefully orients effective policy responses around the discretionary management of total spending. It was, and is, the real deal.
BG&P, along with NK model-builders still working on business-cycle theory, reject the EK framework. Why? The reason is not that it is less powerful. It is much more powerful. It is not that it explains fewer of the cyclical facts. It explains many more facts. (It good place to start is its unique, in combination with microfounded MWR, explanation of the existence and dominating size of forced layoffs in recession. The market-centric internal-mechanism approach never has, or will, come close to explaining involuntary job loss.) Instead of worrying about how weak their model is, BG&P aim their guns at the EK impulse-propagation modeling for assuming MWR. A failure to microfound is an invitation to the modern-theory dustbin. How embarrassed will BG&P be when they find out that, years before their paper was published, the GEM Project had microfounded MWR? An interesting question for histories of macroeconomics is who’s at fault for all the wasted model-building time and effort.
Blog Type: New Keynesians San Miguel de Allende, Mexico