Micro-Macro Coherence Is Not Enough

Print/Save PDF


Jean-Claude Trichet, the head of the European Central Bank during the global instability crisis that began in 2007, sharply criticized the macroeconomics community three years later: “Macro models failed to predict the crisis and seemed incapable of explaining what was happening to the economy in a convincing manner…. [W]e felt abandoned by conventional tools.” Trichet’s tough talk, along with similar criticism from stabilization authorities around the world, was supposed to push the profession to do better. It hasn’t worked. Bad advice from mainstream macroeconomists just keeps coming, piling up at an apparently accelerating pace. The most critical issue today for macroeconomists is how to get the profession to take its relevancy problems seriously.

The GEM Project proposes two simple gatekeeping protocols for the mainstream dissemination of research that asserts stabilization-policy relevance:

  • The policy objective must be clearly specified and supported by the pertinent evidence;
  • Macro disturbances used to motivate model behavior must also be clearly specified and consistent with available evidence.

Compliance with those rules would follow from requiring authors to explicitly consider the nature and practical relevance of their policy objectives and macro shocks (and their propagation), enabling better assessment of the seriousness of claims to policymaker attention. Whenever macroeconomists claim policy significance for their work, the new protocols would join micro-macro coherence, today’s predominant requirement for research to be eligible for mainstream debate and dissemination. We have learned, over and over again, that the absence of such empirical relevancy creates analytic thickets in which policy advice that is groundless, misleading, and damaging often thrives.

The need for improved protocols is illustrated by a recent example from the raging epidemic of search/match/bargaining model misuse: Philip Jung and Keith Kuester’s “Optimal Labor-Market Policy in Recessions”, appearing last year in the American Economic Journal: Macroeconomics. The paper’s claim on policymaker attention is made immediately clear in its abstract: “Within a search and matching model with risk-averse workers, endogenous hiring and separation, and unobservable search effort, we show how to decentralize the constrained-efficient allocation by a combination of a production tax and three labor-market policy instruments: vacancy subsidies, layoff taxes, and unemployment benefits. We derive analytical expressions for the optimal mix of these over the business cycle. Calibrating the model to the US economy under the assumption that wages are rigid, we find that hiring subsidies and layoff taxes should rise considerably and persistently in recessions. The optimal variation in unemployment benefits, in contrast, is quantitatively small and short-lived.”

Policy objective. From J&K (p.125): “Optimal policy first and foremost reduces the fluctuations in unemployment. This is achieved by the vacancy subsidies and layoff taxes.” How should unemployment fluctuations be adequately specified? It is hard to go wrong by starting with the most fundamental characteristic about cyclical unemployment, i.e., it is mostly generated by involuntary job loss. In coherent macro modeling, forced job separation requires meaningful wage rigidity (MWR), which is defined by its capacity to suppress labor-price recontracting. (Chapter 2) It follows that, for policy-relevant work, properly specifying the J&K macro objective requires careful attention to labor pricing actually observed in the U.S. economy. If you do not understand authentic MWR, you cannot come close to adequately assessing the benefits and costs of vacancy subsidies and layoff taxes.

Most S/M/B modelers painstakingly ignore both forced layoffs and downward nominal wage inflexibility, greatly damaging their claim to stabilization-policy relevance. But J&K are an exception, conceding the neoclassical logic that links wage rigidity to unemployment periodically rising above its socially efficient level. However, in their attempt to attach meaning to their wage rigidity, the J&K analysis falls apart. J&K resort to the analytically convenient assumption of counter-cyclical bargaining power in order to suppress profit-seeking firms’ labor-price recontracting. In their world, workers respond to nominal wage cuts in recession by separating from their jobs, shutting down the stylized firm.

The belief that workers’ bargaining power increases in cyclical downturns is of course silly, not coming close to passing practitioners’ laugh test. It is also silly that the curiously enhanced power would be used to voluntarily eliminate, in recession (!), one’s own job. There is almost nothing in this story that is consistent with well-known facts. J&K hope to get away with indefensible assumptions because counter-cyclical bargaining power has been relied upon in published work by other theorists, providing citations that provide cover for the absence of plausible explanation. Their root problem is depressingly familiar: the necessity of some implausible wage-setting scheme resulting from not knowing how to derive MWR from axiomatic model primitives in the context of continuous general decision-rule equilibrium. J&K are stuck, just like everybody who tries to make coherent S/M/B modeling account for recognizable cyclical unemployment gets stuck. Microfounding MWR is apparently difficult, so getting stuck is understandable. What is inexcusable is jumping from stuck model-building to confidently pushing groundless stabilization-policy advice.

Macro disturbance. J&K animate recessions, with unemployment rising “to socially inefficient levels”, by positing productivity regress, an analytically convenient shock famously used in real-business-cycle analysis. J&K simply ignore the critical problem of coming up with a convincing story or evidence that supports of the general use of that class of productivity shock. The most damaging consequence of using RBC regress is crowding out the use of adverse nominal demand disturbances (NDD), either as an initiating shock or in the powerful propagation of a real shock. Active demand management, if it had a role in the J&K model, would alter the benefits and costs of vacancy subsidies and layoff taxation.

NDD is fundamental to the proper understanding of the cyclical behavior of temporary and permanent layoffs, involuntary unemployment, employment, weekly hours, output, wage income, profits, asset prices, investor confidence, investor rational inaction, and the rest of the array of pertinent evidence. Any model excluding NDD cannot adequately capture involuntary job loss and cannot be stabilization-policy relevant. J&K ignore NDD because microfounded MWR is beyond their grasp.

Blog Type: Policy/Topical Saint Joseph, Michigan

Write a Comment

Your email address will not be published.