By Force of Habit: Why Risk Feels Terrifying in a Recession

Mehra and Prescott had left the field with a wound that would not close. Stocks earn a large premium over bonds. Aggregate consumption is remarkably smooth. Standard theory says a smooth economy should not generate a large risk premium, and there was no accepted way out.

In 1999, John Campbell and John Cochrane offered one of the two or three most influential answers, and it hinges on a single idea that anyone who has ever taken a pay cut will recognize immediately.

The problem: the model assumes your fear depends on your wealth

The standard consumption-based model has a hidden and rather strange feature. It assumes that how much you dislike risk depends on your absolute level of consumption. A person consuming a lot is relaxed; a person consuming little is anxious. Since aggregate consumption is high and rises steadily over the decades, the model says people should be steadily less frightened over time, and never very frightened at all, because a bad year still leaves them consuming almost as much as a good one.

That is not how anyone actually experiences a recession.

The key idea, via analogy

Consider someone who has earned two hundred thousand dollars a year for a decade. They have a mortgage, private school fees, a certain life. Now their income drops to one hundred and fifty thousand.

In absolute terms, they are still rich. Objectively, they consume more than the vast majority of humanity. The standard model says they should barely notice, and should certainly not become dramatically more risk averse.

But of course they are terrified. Because their expenses are built around two hundred thousand. They are now scraping against a floor they had never had to think about before. Every further loss now threatens something that actually hurts. They will not take any gambles. They will pay almost anything for safety.

That is the whole paper. Campbell and Cochrane build a model where what matters is not your consumption level but your consumption relative to a slow-moving "habit": an accumulated benchmark set by your own recent past, the standard of living you have grown accustomed to. The habit adjusts, but it adjusts slowly. It is sticky. That stickiness is everything.

Now watch what this does to risk aversion:

• In a boom, consumption has run well above habit. You have a comfortable cushion between where you are and what you are used to. A small loss barely registers. You feel relaxed, and you are willing to hold stocks for a modest premium.
• In a recession, consumption falls toward the habit. The cushion vanishes. Now every further dollar of loss cuts into a standard of living you have built your whole life around. Your effective risk aversion skyrockets, even though your consumption has only fallen by a few percent.

So risk aversion is not a constant. It is countercyclical: it rises in bad times and falls in good times, and it does so dramatically, because a small drop in consumption can be a large drop relative to a nearby habit.

This one modification generates a remarkable amount of what we actually observe:

• A large equity premium. Stocks crash exactly when consumption falls toward habit, meaning precisely when people are most terrified. Their timing could not be worse. So they must offer a very large premium. You no longer need absurd risk aversion; you need ordinary risk aversion that spikes at the worst moments.
• A stable, low risk-free rate. This is the harder trick, and it is why the paper is celebrated. Mehra and Prescott's puzzle had an evil twin: attempts to raise the equity premium usually blow up the predicted risk-free rate. Campbell and Cochrane carefully engineer the habit process so that two offsetting effects cancel, keeping the risk-free rate low and steady while the equity premium stays large.
• Predictable returns. When the market is beaten down and people are frightened, they demand a huge premium, so future returns are high. When times are good and complacency reigns, they accept low premia, so future returns are low. That is exactly the pattern Campbell and Shiller had documented: a low price relative to fundamentals forecasts high returns.
• Volatility that clusters in bad times. Since risk aversion moves sharply when consumption is near habit, prices become more sensitive to news precisely during downturns. Stock markets really are more volatile in recessions.

Why it mattered

• It rescued the consumption-based model. After Mehra and Prescott, many had written off the whole framework. Campbell and Cochrane showed you did not need to abandon it, only to fix its assumption about what drives fear.
• It made time-varying risk aversion mainstream. The idea that the price of risk swings over the business cycle is now standard, and it reconciles rational asset pricing with the return predictability that behavioral economists had claimed as evidence of irrationality. Both camps can point at this paper.
• It reframes the equity premium as a timing problem, not a size problem. Stocks are not dangerous because they are volatile. They are dangerous because they lose money exactly when losing money hurts most. That sentence is the most portable insight in the whole of asset pricing.
• It gives a practical warning. Your own risk tolerance is not a fixed personality trait. It will collapse precisely when markets are cheapest and expected returns are highest. Knowing this about yourself is worth more than most investing advice.

The honest limitations

• The habit process is reverse-engineered. The specific mathematical form of the habit is chosen, with considerable ingenuity, to make the model produce the facts we already know. That is not the same as deriving those facts from first principles, and critics call it curve-fitting with extra steps.
• The implied risk aversion still gets very large. In deep recessions, the effective risk aversion the model needs is high enough to make some economists uncomfortable, even if it never reaches the absurdity of the original puzzle.
• The habit is external, and that is convenient. The model assumes your habit depends on aggregate past consumption rather than your own, which sidesteps a hard technical problem but is not obviously how habits work.
• The micro evidence is mixed. Direct evidence that individual households behave this way, with slow-moving reference points driving sharp swings in risk aversion, is far weaker than the model's macro success would suggest.
• It is one of several rivals. Long-run risk models, rare disaster models, and heterogeneous-agent models all explain a similar set of facts through different mechanisms. They cannot all be right, and the data has not clearly picked a winner.

The one-line takeaway

Campbell and Cochrane showed that if your fear of loss depends on how close you are to the standard of living you are used to, rather than on how rich you are, then a small recession makes people enormously more risk averse, which is enough to explain the equity premium without assuming anyone is irrational.