Equilibrium and Meltdown, by George A. Waters: Equilibrium in economics is such a ubiquitous concept some might be surprised to hear that it is also controversial. It is hard to imagine doing much economic analysis without examining the intersection of supply and demand, but whether we should focus exclusively on equilibrium outcomes is a critical question. Though the importance of equilibrium might seem like an esoteric debate, the attitude toward this question has deep implications for economists’ views on the correct policy response to the current economic crisis.
Milton Friedman believed in equilibrium. An example of the intensity of his belief was his argument against requiring medical licenses for doctors on the grounds that more doctors would be allowed to practice, and market forces would weed out poor doctors and improve the quality of care. This argument makes perfect sense if we focus on the equilibrium outcome, but most people are instinctively repelled by this proposal. The concern arises from the consideration of how the world gets to such an equilibrium. To find out who the bad doctors are, some patients have to try them, get bad treatment and tell others about it. Since patients have little expertise in making judgments about doctors, a poor doctor could practice for many years without detection. I know this happens even with licensing requirements, but the situation could easily be worse without them. An equilibrium outcome cannot be divorced from the path to arrive at that state.
This issue arises in the labor market, with implications for macroeconomics. If labor supply always equals labor demand then there is no such thing as involuntary unemployment. Workers don’t work because the wage offered is too low. While undoubtedly true for some, the picture of unemployment must be more complex. Real Business Cycle (RBC) models of the macro-economy focus on equilibrium in the labor market, leading to the remark that, in such a framework, the Great Depression was really the Great Vacation. There are other views of the labor market in macroeconomics. Traditional Keynesian approaches focus on sticky wages (and prices) and have a more natural explanation for unemployment. For example, in a recession firms prefer to use layoffs rather than lower wages. Labor search explicitly model the matching of unemployed workers and vacant positions.
Dynamic Stochastic General Equilibrium (DSGE) models are currently the dominant approach in macroeconomics. They are descendents of RBC models in that they are general equilibrium models with microfoundations, meaning they focus on household and firm behavior rather that make direct assumptions about the relationships between macro variables. However, the DSGE label applies to wide variety of models including monetary policy models with price and wage stickiness. However, the vast majority of DSGE models used for policy analysis still study fluctuations around a unique equilibrium.
The desirability of fiscal stimulus in response to the current crises has been a point of contention producing much debate and, unfortunately, a number of muddled arguments. One notable exception comes from James Hamiltonwho points out that the housing, financial and auto industries are going through necessary contractions, and the process of those workers finding new jobs, which could require retraining and relocating, will be a long, painful process that no fiscal or monetary stimulus can alleviate. He goes on to add that government spending that increases productivity, such as infrastructure improvements, or maintains services, such as block grants to states, could still be beneficial. I would add that spending to stimulate demand and prevent unnecessary contraction of healthy industries could help as well. Nevertheless, his point that the government should allow the labor market to adjust and not try to guarantee a certain level of employment at all times is very much on target.
Of course, fiscal policy should not be considered in isolation from other aspects of the economic crisis. The housing market continues to fall affecting households and putting the financial sector is in danger of collapse with serious effects on the economy, pushing real estate prices further still. Monetary policy has reached the zero lower bound for the Federal Funds rate, raising the possibility of a deflationary spiral. Whether to Fed can avoid such a liquidity trap with “quantitative easing,” is another important issue. Targeting longer maturity rates, now that the fed funds rate is essentially fixed, is an untried idea, so far.
And what does macroeconomic theory have to tell us about the present situation? As a macroeconomist, I’m sorry to say, not enough to give confident answers to the policy questions, though we do have some helpful tools at our disposal. Labor search models have the potential to analyze the frictional unemployment that Hamilton says should be tolerated. Natural ways to describe the deflation that can arise in a liquidity trap include have been developed using a model with multiple equilibria or at least entertaining the possibility that the economy is not in equilibrium for significant periods of time. Furthermore, one can think of recessions in general and financial crises in particular as shifts between multiple equilibria. However, standard DSGE models with monetary and fiscal policy are single equilibrium models and do not incorporate these possibilities. When Tom Sargent, one of the godfathers of the DSGE approach, says that “the calculations that I have seen supporting the stimulus package are back-of-the-envelope ones that ignore what we have learned in the last 60 years of macroeconomic research,” he may be right in a narrow sense, but his comment is more indicative of how far macroeconomic theory has to go to properly analyze complex economic environments such as the one we face.
To be more specific, we would like to know what combination of fiscal and monetary policy can avoid deflation, maintain aggregate demand so that firms do not fail unnecessarily while allowing for necessary adjustments in the labor market. To analyze such questions, we need a model incorporating fiscal and monetary policy, a financial sector, labor search while allowing for alternate equilibria or out-or-equilibrium behavior that can be interpreted as a deflationary liquidity trap and/or a recessionary state. Many macroeconomists would disagree with the details of this prescription, but most know that a model with sufficient sophistication to give a confident policy recommendation does not exist. That back-of-the-envelope calculations are all we have should not be surprising. In Keynes’ words, macroeconomists should be humble.
We can better understand economists’ responses to the stimulus question through the lens of their preferred models. Those who have spent their careers working with single equilibrium RBC models tend to view the stimulus as wasteful, contributing only to debt and inflation but not productivity. Those who view unemployment as unused resources in an out of equilibrium labor market believe government spending could put those resources to work. Paul Krugman has advocated forcefully for the stimulus, which is unsurprising given the nature of his research. One of his major contributions is to show how multiple equilibria can arise in a model of international trade where industries have increasing returns to scale. The notion of an economy shifting to a better equilibrium with a push, like a stimulus, is natural for him.
One reason models with multiple equilibria are not standard parts of policy analysis is that their implications for data analysis are difficult to interpret. Interestingly, the most common macro-econometric methods that does allow for multiple equilibria is the regime shifting model of James Hamilton, which estimates a probability of switching in or out of a recessionary state in each time period. As noted above, he has expressed some skepticism about the stimulus. I suspect that he considers the current recession as a persistent change in our natural rate of growth or as necessary “creative destruction.” Alternatively, the economy could be in the process of shifting to a new state with high saving, low consumption, investment and employment. Olivier Blanchard describes the crisis in these terms and advocates for policy to reinvigorate demand to help the economy to shift back. Distinguishing between a temporary shift to a new equilibrium and a persistent change in productivity is hard, which points up the subtle issues arising when we take the idea of multiple states seriously.
Despite the difficulties, there is good reason to extend our understanding of models with multiple equilibria. There has been some loose talk among economists that the current crisis is not a garden-variety recession. The language of multiple equilibria could help formalize this idea: Normal recessions are dips below trend for output growth that can be handled with monetary policy, but we are experiencing a potential shift to a bad state that should be avoided using all available policy tools. If the keys are not to be found under the streetlight of a unique equilibrium, we must look in the dark alley. The restriction to a unique equilibrium is not the only problematic feature of many macro models. Most micro-founded models have representative agents, such as a single household representing the consumption, saving and labor supply decisions for the whole economy. A quick glance out most windows reveals that not all households are the same. The vast majority also assume rational expectations, which assumes all economic actors use all available information to form forecasts, a rather strong assumption.
It is not necessary to resort to arguments based on multiple equilibria to argue for the stimulus. Standard Keynesian arguments support government expenditure to employ unused resources. Some argue against spending on the grounds that government debt will grow unsustainably. Rising national debt is a serious concern, but a temporary stimulus need not lead to an insoluble problem.
This essay is not an argument for throwing all of macroeconomic theory overboard. Understanding each of the models described above gives insights into particular issues. My primary point is that many of the arguments dismissing fiscal stimulus as a solution to the current crisis rely on strong underlying assumptions about equilibrium and the impact of government spending. Arguments against the stimulus due to concerns about the debt and the necessity of large labor and financial market adjustments are quite valid. The policy problem comes down to a difficult judgment weighing these costs against the ability of fiscal and/or monetary policy to halt a deepening of the crisis. Given the state of theory, it is not surprising that there are vast disagreements among our most eminent economists. Those who believe there is a clear answer to the question of whether fiscal stimulus will work (whatever that means) should recognize that macroeconomics needs to re-examine its most closely held beliefs.
1 In the discussion of macroeconomic models, “multiple steady states” is more accurate than “multiple equilibria” which could also encompass sunspot equilibria arising in indeterminate models. I refrain from using “steady state” to avoid excessive jargon.
2 The “treasury view” and complete crowding out arguments against the stimulus are prominent examples.
3 See his Econbrowser post “The Paradox of Thrift” of February 8, 2009. Unfortunately, he followed this with post titled “How Much is a Trillion” (March 3, 2009), which offers as much insight as the title suggests.
4 Bullard and Cho (Journal of Economic Dynamics and Control v. 29(11) and Evans, Guse and Honkapohja (European Economic Review v. 52(8) examine liquidity traps as distinct steady states. The latter argues for fiscal stimulus when monetary policy has reached the zero lower bound of its interest rate target.
5 See his Economic Focus column in The Economist (January 29, 2009). He does not explicitly talk about multiple states.
6 The idea of multiple equilibria or steady states is far from new. Cooper and John (Quarterly Journal of Economics 103 (August, 1988)) discuss a switch to an inferior steady state as a “coordination failure.”
7 Peter Diamond, among others, has cautioned against relying on representative agent models for policy analysis.