Why do investors hold relatively low-performing bonds even though stocks earn much higher returns over the long-term? The question is at the heart of the equity premium puzzle, or why stocks payoff at such a higher rate of return than bonds despite bonds’ popularity with investors, a paradox that economists have long sought to explain.
Professor Emi Nakamura believes that one such explanation may lie in consumption disasters, rare events in which a country experiences a large drop in consumer expenditures as a result of severe economic conditions.
Consumption disasters directly bear on asset prices because investors will pay a lot for assets that pay off in a down economy. Such assets thus effectively act as insurance. Since investors can’t be sure when a downturn will hit, how severe it will be or how long it will last, they play it safe by investing in bonds.
Investors’ assumptions about the frequency and duration of disasters bears heavily on what investors believe the return on stocks should be relative to less risky assets like bonds; riskier investments are expected to yield higher returns, at least during normal times. For the 20th century, the equity premium — the average difference between the return on stocks and bonds — is a substantial 7 percent.
To learn more about whether rare disasters explain the size of equity premium, Nakamura worked with Robert Barro and José Ursúa of Harvard and Jón Steinsson of Columbia University’s Department of Economics to identify some of the largest consumption disasters of the last 150 years.
Nakamura and colleagues used a newly available panel data set of consumer expenditures from dozens of countries, culled from a variety of sources across a number of academic disciplines, including history and sociology, and which reached as far back as 1850 (far earlier than previously available data). The researchers then created a model that effectively functioned as a disaster filter, plugging the new data into their model to identify which economic downturns in which countries qualify as large consumption disasters.
The filter sheds light on the severity and permanence of disasters. Consumption dropped on average by 35 percent in the disasters the researchers identified. The model’s disaster cutoff is so extreme that the Great Depression doesn’t qualify as a severe disaster. (Models frequently make use of GDP as a key measure, but this reflects production rather than consumption; consumer expenditures provide a more accurate sense of how much an economic downturn hurts investors, which in turn determines how much they will discount the prices of assets that payoff badly in a recession.)
The researchers found that consumption disasters last an average of about five years; in most cases they are followed by periods of unusually rapid growth. For most countries, disasters are only partially permanent. Typically, the long term effect of a disaster is estimated to be less than half of the disasters’ short term effect. In some instances, however, disasters have very large long term effects on consumption.
While earlier work has investigated whether rare disasters can explain the size of the equity premium, previous models didn’t reflect the reality that disasters are usually partially temporary. Previous models also assumed that disasters unfolded over a single year, whereas actual disasters take several years to unfold. Here, the researchers designed their model to allow for the possibility that countries could experience both temporary and permanent disasters, and that disasters can develop over years.
“A world in which disasters are permanent is a much riskier world than one in which most countries recover from disasters,” Nakamura explains. “If investors thought that all disasters were permanent, the expected equity premium would likely double.” According to the researchers’ estimates, in a world without economic disasters, the return that equity would get over bonds is a tiny fraction above zero; when allowing for the 1.4 percent probability of a disaster happening — the actual probability suggested by the disaster filter — the equity premium shoots up, to close to 6 percent, far closer to the equity premium reflected in markets.
Economists have long understood the effect that severe market conditions have on asset prices, but until very recently there wasn’t enough historical data to backup the explanation. “If an investor believes there is a nontrivial possibility of a disaster happening — like what looks to be happening in Iceland, for instance, which the IMF forecasts will experience a 30 percent drop in consumption over the next few years, and would qualify as a disaster in our dataset,” Nakamura says, “this may have a dramatic effect on the willingness of investors to buy stocks, particularly because of uncertainty about how long and how bad the disaster could turn out to be.”
Emi Nakamura is assistant professor of finance and economics at Columbia Business School
