I’m reluctant to disagree with a newly minted Nobel laureate in economics, but I just can’t let Robert Shiller’s remark about the business cycle pass without comment. “The world economy is softening a bit,” he told Yahoo Finance yesterday. “There’s always a chance of another recession. It’s been six years since the last recession started -- they tend to come along with some regularity.” Regularity? Well, actually that’s the wrong word to use on this subject for a simple reason: there's nothing regular when it comes to the timing of recessions.
Yes, I know that it’s easy to assume otherwise, even if you just won a Nobel, albeit for work in asset pricing. The business cycle, although related, is another animal entirely. If you think financial markets are difficult to model (and they are), that’s a walk in the park compared with trying to find order in the ebb and flow of the economy’s major turning points. Sure, there’s always another recession coming, but that’s about the only thing you can say with anything close to certainty in this corner of economics.
Studying the business cycle is a world unto itself and so analytical skills and experiences forged elsewhere in the dismal science may be of limited help here. I don’t make this point lightly. I’ve spent quite a lot time over the past several years writing and researching a soon-to-be-published book about recessions while dissecting the business cycle on these pages (here, for instance). That includes reading through countless studies and books on this slippery subject. The one lesson that stands out is that common sense and intuition that may serve you well in other corners of economics have a habit of leading you down dead-ends with business cycle analysis.
But I digress. Turning back to Shiller’s claim that there’s some “regularity” to the arrival of recessions, I guess it depends on how you're defining your terms. We can start by looking to the gold standard for dating recessions: the National Bureau of Economic Research (NBER). There have been 33 recessions in the US since the mid-1800s, 11 since the end of World War II. Not a huge sample. In fact, it's rather thin. But it's all we've got and it's hard to argue that there's regularity in the record on recessions. The period of expansion (i.e., the length of time from the end of one recession to the start of the next one) has ranged from a short 12 months up to 10 years. Regularity, it’s fair to say, is nowhere in sight in this data set. Clocks and calendars come in handy for many things, but anticipating recessions isn't one of them.
Casually looking at the historical record may suggest that expansions die of old age; if so, recession risk mechanically rises as growth rolls on. But a closer review casts serious doubt on this idea, as the NBER data clearly shows. Not surprisingly, formal studies also find minimal evidence in support of the idea that economic expansions are destined to expire after a certain length of time, as explained by Francis Diebold and Glenn Rudebusch in a 2001 paper ("Five Questions about Business Cycles") via the San Francisco Fed (pdf):
The evidence indicates that expansions show no effects of aging or duration dependence, whereas contractions are increasingly likely to end with age. Thus, while prewar business cycle analysts were perhaps accurate in their assessment of expansion duration dependence, post-war commentators are not justified in suggesting that a business cycle peak is more likely to occur as an expansion ages.
That’s one reason why predicting downturns is so challenging. In fact, history also shows that it’s devilishly difficult to accurately call the start of new recessions in real time shortly after these events have started. All’s not lost, however, as my upcoming book—Nowcasting The Business Cycle—will explain. By standing on the shoulders of giants in the field of business cycle analysis you can develop quite a lot of valuable insight in the art/science of recognizing those times when the economy is slipping over to the dark side.
In order to improve our skills in estimating recession risk in real time we must abandon the notion that casual observation and rules of thumb are useful. Quite a lot of what passes as reasonable or informed thinking in the crowd on the topic of the business cycle is misleading if not patently false. That’s understandable since the only way you can cut through the noise on this subject is to spend a lot of time with the data and the associated theories. That’s no one’s idea of fun, but context and perspective don't come cheap in this branch of macro.
In any case, let’s note that regularity is reserved for the tides, the seasons, and birthdays. But presuming anything close to a reliably rhythmic pattern with matters of the business cycle is assuming facts not in evidence.