Stock, Watson, & Misinterpreting Macro Evidence

The damage done to macroeconomics by the dominating mainstream impulse to defend, no matter the cost, market-centric modeling is breathtaking. It has produced both analysis that does not come close to being stabilization-relevant and evidence interpretation that is carefully organized to ignore the most important facts. This post looks at the latter.

Convenient evidence. The implicit collusion among mainstream analysts to limit attention to evidence that supports consensus market-centric thinking has been remarkably effective. A leadin examgple is Stock and Watson (1999). Assigned the task of setting the empirical stage for the (still) most recent Handbook of Macroeconomics survey of contemporary thinking, they concentrated on covariances among production, interest rates, prices, productivity, employment, investment, income and consumption with basic measures of the business cycle. The time-series approach was pioneered by Hodrick and Prescott (1997) and has become the standard for identifying and organizing relevant macro evidence.

Here’s the problem. Their pile of covariances tells us little about the behavior of highly specialized economies. Most problematic, they provide an embarrassingly low bar for assessing the stabilization-relevance of competing macro theories. It is difficult to think of a model that is not consistent with the Stock-Watson findings. Their data certainly don’t eliminate the Early Keynesian Neoclassical Synthesis, the banishment of which from mainstream debate has been a successful goal of modern gatekeeping in the academy.

Illustrative of important evidence being ignored, a diligent student could read the three volumes of Taylor and Woodford’s Handbook, from cover to cover, and never learn that involuntary job loss (IJL) is the most powerful source, by far, of the cyclical movement of employment and unemployment. Why? It surely isn’t that employment instability has become unimportant, somehow having sunk beneath policymaker notice. It is instead because IJL cannot be coherently accommodated in the mainstream New Neoclassical Synthesis, a fatal flaw that academic gatekeepers have concluded is best left unmentioned. “Job loss” appears once in the Handbook index and turns out to refer to voluntary quits, the only kind of job separation that can be analyzed  within the ubiquitous market-centric continuous-equilibrium Search/Match/Bargain framework. Meaningful wage rigidity, the preferred explanation of involuntary unemployment by the great Early Keynesians, has become so unimportant that it no longer deserves mention in a 1,745 page survey of contemporary macroeconomics.

Important evidence. The GEM Project identifies (and explains) a much more robust array of evidence that must be accommodated by macro theorists aspiring to be both micro-coherent and stabilization-relevant, including the following:

• Quarterly variation in employment, hours, and output are best tracked by same-direction disturbances in total nominal spending. That is a set of correlations that really matter. They demand explanation in stabilization-relevant analysis, setting a high bar for theorists.
• Unemployment deviations around its natural rate are largely attributable to variations in involuntary job loss, i.e. layoffs.
• Layoffs are concentrated in large, specialized establishments.
• Chronic wage rents are paid and jobs are rationed in large firms, while jobs are much easier to get in small firms that are characterized by effective workplace monitoring and market-wage taking.
• In contrast to the substantial payoff resulting from long job tenure in the high-wage sector, jobs paying market rates provide significantly less benefit from longer-term employment. SEV (small-establishment venue) employees optimize at the work-leisure margin, producing voluntary job turnover that is much higher, and labor-force attachment much weaker, than for LEV (large-establishment venue) workers. Given wage rents, the latter are pushed off their neoclassical market-supply schedule and optimize their on-the-job behavior rather than worker-leisure choice.
• An employee permanently expelled from a high-wage jobs confronts a difficult period of labor-price discovery. Such workers, having been paid a significant rent, must sort out their true market opportunity costs, typically experiencing frequent spells of joblessness interspersed among relatively short periods of low-wage employment before settling into a job that pays significantly less than the involuntarily lost position. The extended search process is significantly reinforced by the existence of jobless benefits.
• The rate of product-demand growth in the large-establishment venue governs the absorption of workers transferring from the low-wage sector. The nonstationary queue awaiting admission, contributing to unemployment and underemployment, is increasing in nonstationary wage rents and the subjective probability of obtaining a LEV job, as documented by the substantial evidence supporting the famous Harris-Todaro thesis. The stationary component of joblessness reflects LEV downward wage rigidity over the business cycle, producing layoffs and subsequent recalls. Job quitting, concentrated in the SEV sector, moves inversely with the length of the Harris-Todaro venue-transfer unemployment queue.
• Wages of workers successfully transferring from the small- to large-establishment venues substantially exceed the rate received on their last job, immediately capturing most of the available rent.
• The stagflation decade blew apart the interindustry wage structure, a critical result that cannot be adequately understood as the outcome of generally unanchored inflation expectations.
• Lagged inflation is more important in periodic LEV wage adjustments than inflation expectations. Their relative significance is not close.
• Pure-profit expectations, rationally constructed, are the primary determinant of capital spending. Household income is the primary determinant of consumption.

Those facts, far from exhaustive of the significant evidence systematically pushed aside in modern thinking, have been been around for a long time. They are not hard to find. (Chapter 10) The GEM Project Blog is available to whomever wants to make a serious case that the list does not provide better  guidance to stabilization-relevant macro modeling than Stock and Watson. Even more to the point, who wants to argue that the mainstream strategy of evidence-exclusion does not result from an overriding desire to defend consensus market-centric macro theory? Finally, who wants to make the case that systematically turning your back on a pile of important evidence challenging micro-coherent market-centric DSGE theory has not resulted in a generation of ill-informed macroeconomists? Something like the broad review of macro-relevant evidence found in Chapter 10 should be required reading in every serious graduate school.

Blog Type: New Keynesian Saint Joseph, Michigan