Business cycle length and the probability of a recession has been saved
The length of a recovery has little to do with the probability that a recession might occur. Business decision makers should look elsewhere to gauge where the economy might be headed.
View the Behind the Numbers collection, a monthly series from Deloitte’s economists.
Housing bubbles bursting, Lehman Brothers failing, a $700 billion dollar rescue package…it doesn’t seem so long ago. Yet the trough—which is to say, the end—of the last business cycle downturn was June 2009. It’s been more than five years.1 Enough time to start asking when the next recession might happen. Some people like to point to the length of the recovery as a signal that we should start to get worried.
But while good business planners always worry about recessions, the length of a recovery has little to do with the probability that a recession might occur. Modern business cycle thought—and data about recessions—suggests that business decision makers should look elsewhere to gauge where the economy might be headed.
Many people learn that a simple rule of thumb—two quarters of negative growth—defines a recession. US economic researchers, however, use recession dates determined by a committee of economists who are affiliated with the National Bureau of Economic Research (NBER). By agreeing to a common dating scheme, economists can devote their arguments to how and why, rather than when, recessions occur.
The NBER committee looks at a wide variety of indicators to decide the month in which the recession starts (the previous month is the “peak” (or high point of the cycle) and the month in which it ends (the previous month is the “trough” or low point). NBER publishes cycle dates and methodology here.
The average US business cycle expansion since the end of World War II has lasted 56 months.2 Figure 1 shows the length of expansion after each postwar business cycle peak (identified by year).The standard deviation of those 11 expansions is 35 months, which implies that there is a roughly 95 percent chance that an expansion will last between zero and 126 months.3 It’s hard to draw a lot of conclusions from this—which itself is an important observation. You might think that expansions have become longer over time, but notice the very long expansion of the 1960s (peak in 1969) and the very short expansion that ended in 1981. It’s true that the last three expansions have been unusually long. And the current expansion (not on the chart, since there is no peak yet) is already longer than most of the 1950s–70s expansions. So it is possible that something—the role of services in the economy, perhaps—has changed. But the sample is very small, and the uncertainty is very high. That makes it impossible to tell when the economy might be “due” for a recession.
Truth is, most economists who study business cycles don’t actually think of them as cycles. To understand why, it helps to know something about the history of thinking about business cycles—or trade cycles, or banking panics, or recessions, or depressions, or any of the other colorful terms attached to these events.
The world’s first industrialized economy—the United Kingdom—suffered banking panics in 1825, 1847, 1866, and 1890. The fact that the financial system—and the economy—appeared to experience a crisis about every 20–25 years suggested to observers that there was some regular force at work in “modern” economies that led to a cycle of booms and busts. But what was this force? Speculation ranged from the sunspot cycles to climate and weather to the behavior of various economic actors. Karl Marx and Joseph Shumpeter (strange bedfellows indeed) both told stories in which capitalist “overproduction” or “overinvestment” led to periodic crises. While these were not necessarily bound by a time schedule, both viewed cycles—regular up and down behavior—as a natural feature of economic behavior.
As economists turned to statistical techniques, however, they discovered that the “cycles” that appeared in the data were likely illusions. It is not difficult to generate data that looks cyclical but is purely random. Figure 2 shows the graph of a random “process”4 that could easily pass for a typical business cycle measure. Statistical tests cannot distinguish standard data series such as GDP and employment from data series generated by these random “processes.”
As a result of this discovery, economists have mostly rejected the idea that business “cycles” represent some inherent, regular (in a time or structural sense) feature of the economy.5 Instead, modern business cycle economists view the economy as experiencing random “shocks.” Some shocks (technological improvements, booms in neighboring countries, better functioning financial markets) are positive, and some (financial market failure, problems abroad, wars, poor policy choices) are negative. Economic behavior translates these shocks into what appear to be regular patterns in the data. But it’s all really random. A typical popular modern macroeconomics textbook doesn’t even mention the term “business cycle,” preferring the term “economic fluctuations.”6
It follows that each recession—like an unhappy family in Leo Tolsty’s Anna Karenina—is “unhappy in its own way.” Each “cycle” is the result of a set of unique circumstances and shocks. This very powerful idea explains a lot of the mystery of business cycles—especially the fact that recessions seem to be almost completely unpredictable.
So the accepted theory tells us that a shock will create the next recession. Do we know what the shock will be and when it will occur? That’s impossible. By its nature, the shock will be something we can’t anticipate. What we do know is this: The probability of a shock likely has little or nothing to do with how long it’s been since the last recession.
The recent recession has set economists to look not at the shocks themselves, but at the economic and financial conditions when the shock hits. The same shock might devastate an economy which is financially overextended, but have a much smaller impact on an economy with more conservatively financed businesses. The nature of business also matters a lot—economies that depend on durable goods (easily forgone as income falls) will be more sensitive to shocks than economies that depend on services.
To estimate the probability of a recession, we should keep an eye on potential imbalances in the economy. Examples include investment above what common sense might indicate, overly optimistic financial arrangements, geopolitical problems, and even articles claiming that the business cycle has been “solved.” These are the factors that determine the extent to which a “shock” gets propagated across the economy. The length of the current expansion is at best a very minor factor compared to these real drivers of the business “cycle.”