Animal spirits and business cycles

In a 1992 American Economic Review article, Peter Howitt and Preston McAfee (H&M) introduce a "stationary rational-expectations equilibrium in which an extraneous random variable, called animal spirits, causes fluctuations in unemployment."

https://ir.lib.uwo.ca/cgi/viewcontent.cgi?article=1485&context=economicsresrpt

They explain that "research in business cycle theory has been guided almost entirely by the view that fluctuations in the overall level of economic activity are the result of exogenous shocks to the fundamental conditions of a dynamically stable economic system. By this view, booms and recessions are attributable to random changes in the availability of profitable investment opportunities, the propensity to save, the stance of macroeconomic policy, population, international terms of trade, the distribution of demand, [sunspots,] etc., either contemporaneous or lagged, or to the arrival of information signalling such changes."

A competing view states that fluctuations would occur even if fundamental conditions remain unchanged. The view has two variants: (1) Fluctuations are endogenous and the economy is a nonlinear system that exhibits periodic equilibria or chaos; (2) Fluctuations occur due to random waves of optimism and pessimism -- this is Keynes' "animal spirits", can also be found in the work of Mill and Hayek.

H&M construct a model of a rational-expectations animal-spirits business cycle. They model animal spirits as an exogenous random variable that follows a two-state Markov process (high, low). When spirits are high, firms expect a high level of employment, and hence a high level of aggregate demand. High demand reduces the expected cost of contacting a potential customer, leading firms to hire more workers, and expectations are fulfilled. When spirits are low expectations are also fulfilled.

The paper has two aims. First, show that business cycles do not depend on the assumption that fluctuations in aggregate employment are driven by fluctuations in the expected rate of inflation, which induce workers to vary the amount of labor offered for sale. This is not how people think. Second, show how people could acquire the beliefs underlying an animal spirits cycle, even if they don't have rational expectations but rather form expectations through a plausible adaptive learning scheme. H&M show that Bayesian updating induces convergence to the equilibrium even if people start with no definite belief that animal spirits affect the profitability of hiring.

Their model has (strong) assumptions such as costly matching in the labor market (a lifetime contract) and a "thin-market externality that makes production more profitable when others are producing a lot". My initial thought on the paper is that it is a useful improvement on the sunspot models that rely on things such as the effect of expected inflation on labor supply. Since animal spirits are not tied to something more concrete than a Markov process, their model in this form doesn't seem that different from a sunspot model. But I think it's useful for thinking about the mechanics of how expectations can drive business investment and how to model it.