Animal spirits and rational expectations

Keynes' Consuming Ideas On Economic Intervention : NPR

This is the first of several posts discussing my new working paper “Old Keynes and New-Fisher: a Model of Animal Spirit Driven Recessions”.

The idea of “Animal Spirits” — that individuals are animated not by rational economic calculation, but by instincts, proclivities and emotions — is of course an old one, going back to Keynes, but over the last decade its importance has become increasingly obvious. It’s hard to open Twitter or a newspaper without examples — Tesla, Bitcoin, Gamestop come immediately to mind. On a macroeconomic level, the frenzied increase in real estate prices in the run up to the 2008 financial crisis also immediately comes to mind. But even beyond the economic realm the rise of conspiracy theories like Qanon on a large scale show how beliefs of large groups can animate social actions.

In the midst of the Covid pandemic, are businesses and individuals making decisions about hiring, investment, and consumption based upon rational calculation of expected probability of vaccinations, re-openings, and recoveries? Or are they swayed by individual proclivities, rumors, and media narratives? I can’t resist letting Keynes have his say:

Our basis of knowledge for estimating the yield 10 years hence of a railway, a copper mine, a textile factory, the goodwill of a patent medicine, an Atlantic liner, a building in the City of London amounts to little and sometimes to nothing. In fact, those who seriously attempt to make any such estimate are often so much in the minority that their behavior does not govern the market.

Keynes, General Theory

As a macroeconomist, the most important question related to animal spirits is if they can cause recessions with involuntary unemployment. One natural channel for animal spirits to have an effect would be through the financial system. Animal spirits could drive up the price of the stock market or another asset to dizzying heights. A crash naturally ensues, with bank failures / widespread freezing of credit then leading to recessions. This is a focus of one branch of the current literature on animal spirits.

But there is a more direct channel through which expectations can lead to unemployment — the effective demand channel. If consumers and investors develop a bout of pessimism, they will cut their spending. Firms, facing lower demand, will cut hiring, which in turn leads to worse pessimism and further spending cuts. Although the basic idea is simple, formalizing it in a model raises a number of important questions. The first question is whether to model expectations as “rational”.

The idea of Animal Spirits would on the surface appear to be at odds with the technical concept of “Rational Expectations”, or the idea that in an economic model an individual’s subjective expectations of an event should coincide with the mathematical expectations of said event. However, the concepts are not necessarily at odds. Take for example, a sudden outbreak of “animal spirits” which causes people to become optimistic about the price of bitcoin. They then purchase bitcoin en masse, driving up the price. If the equilibrium price happened to coincide with their initially optimistic projections, their animalistic expectations would technically be ‘rational’. Another way of thinking about rational expectations is that people’s beliefs become a self-fulfilling prophecy — the simple act of believing the economy will tank in fact causes the economy to tank, exactly in line with expectations.

Most of the macro literature that writes about ‘animal spirits’ actually works within the rational expectation framework. And a lot of it would seem at first glance to really capture the idea that fluctuations in optimism and pessimism can cause recessions. In Liquidity Traps and Jobless Recoveries by Schmitt-Grohe and Uribe (2017), negative expectations of inflation and employment cause agents to cut spending, leading to a low employment / inflation equilibrium that is self-fulfilling. Even better, the equilibrium features involuntary unemployment. Heathcote and Perri (2017) have a model with recessions caused by a similar channel. In Fiscal Policy in an Expectations-Driven Liquidity Trap, Mertens & Ravn (2014) have a self-fulfilling recession caused by low expectations. Roger Farmer has written papers in which depressions are caused by self-fulfilling expectations about the economy or the stock market.

At a first glance, then, the ideas of Keynes seem to be well represented by these modern macro models. Surprisingly, however, the policy conclusions that fall out of the three strands of this literature on ‘self-fulfilling recessions’ is decidedly anti-Keynesian. In Schmitt-Grohe and Uribe (2017) and Heathcote and Perri (2017), the best way to ensure the economy never reaches a bad equilibrium is to raise interest rates through an interest rate peg. In Mertens & Ravn (2014), the best way to raise output is to cut government spending. And in Farmer’s models, government spending has no impact on output. These surprising effects of government policy I refer to as Neo-Fisherian results.

Both classes of Neo-Fisherian effects are closely driven by the expectations that the policies induce. Interest rate pegs in Schmitt-Grohe and Uribe (2017) eliminate any potential rational expectation equilibrium with inflation below target. When a peg is introduced, agents immediately realize there is no possibility of a bad equilibrium, raising expectations of output and causing the economy to converge to full employment. In Mertens & Ravn (2014), a cut in government spending immediately raises expectations of output, increasing spending and employment.

These results are startling, and seem on the surface unrealistic. How do agents know that the ‘bad equilibrium’ goes away if the central bank pegs interest rates? What causes agents to become optimistic about the future when government increases spending? And why is the traditional Keynesian multiplier reversed in sign?

Stay tuned for next week’s post.

EG Blog Post: What are the implications of reopenings for the economic recovery?

See full post on the Equitable Growth Blog here, and see the working paper here.

When states reopened after the Covid shutdowns, what happened to consumer spending? Most previous research surprisingly finds little effects of the reopenings on spending. Our work goes against this, and in fact finds substantial effects of reopenings.


Why are our results different? We have more detailed data that can examine the spending categories that are directly affected by the shutdown policies. In addition, we do a much more careful job in distinguishing between different policies that are likely to curtail spending. For example, while we find that policies that forced retail businesses to close had substantial effects on spending,

Does the fact that restrictions cause lower GDP / employment mean we should reopen asap? No! The main point of restrictions is to restrict economic activity that can spread Covid.

As states further reopen it is likely the remaining restrictions on activity (capacity limits, large events) will become binding constraints. It’s important not to chase false idols of GDP and employment, but continue to prioritize public health.