Detecting Recessions in the Great Moderation: A Real-Time Analysis
Davig, Troy, Economic Review (Kansas City, MO)
The nature of the business cycle, particularly in the United States, has changed dramatically over the past several decades. In the 1970s and early 1980s, the U.S. economy often whipsawed up and down. Since then, real economic activity stabilized considerably, entering a period economists call the "Great Moderation." With the ups and downs of the economy becoming less dramatic, it has become harder to determine in real time when the economy dips into recession.
Economists have a variety of methods to determine when the economy is entering a recession. These methods range from directly analyzing a broad spectrum of data to the formal use of recession prediction models. The National Bureau of Economic Research (NBER) uses the first approach, relying on several data series to make a determination of when the economy enters or exits a recession. Their decisions are intended to be accurate, not timely. More formal recession prediction models are designed to send a timely signal, but often do not take account of how the Great Moderation has altered the business cycle.
This article uses a framework that efficiently utilizes a large set of data in a "business cycle tracking" model. The model accounts for shifts in overall economic volatility--to capture the Great Moderation--and sends a signal when the economy is shifting between periods of low and high economic activity. Historically, the low-growth regimes identified by the model correspond closely to NBER-defined recessions. Prior to the Great Moderation, the low-growth regimes were short-lived but exhibited a sharp contraction in economic activity. During the Great Moderation, the low-growth regimes have lasted longer but do not exhibit as pronounced of a decline in economic activity. In this respect, the low-growth regimes over the past few decades have basically traded "depth" for "length" compared to the earlier era.
The business cycle tracking model can be used in different ways to extract a signal regarding whether the economy is likely heading for a NBER recession. One method uses a rule-of-thumb that signals whether the economy is "in" or "not in" a condition that will likely turn into a recession. The second method generates a probability that provides a signal of "how likely" it is that the economy is heading for a recession. The methods are complementary; the difference is that one is simple to use (i.e., the rule-of-thumb) and the other requires more sophistication (i.e., computing the probability using the model).
The first section of the article addresses the anemic growth in 2008 and how this period relates to the NBER's definition of a recession. The second section describes an economic activity index that is used as input into the business cycle tracking model. The third section outlines the business cycle tracking model. The fourth section uses the model and the economic activity index to provide a reading on the current state of the business cycle.
I. IS THE U.S. CURRENTLY IN AN NBER-DEFINED RECESSION?
Currently, the financial system in the United States is facing considerable strain. Financial markets are highly volatile, new jobs have posted a net loss every month this year through September, and the unemployment rate has risen from 4.4 percent in March of last year to 6.1 percent in September. Yet the National Bureau of Economic Research (NBER) has not yet declared the United States to be in a recession.
According to a common rule-of-thumb, the economy is not in a recession until national output declines for two consecutive quarters. Despite the weakening labor market, the economy actually expanded in the first two quarters of 2008. In the second quarter, the economy grew 2.8 percent due to productivity gains, strong export growth and a temporary fiscal stimulus package. For comparison, in some of the worst quarters in the recessions of the early 1980s, the economy contracted more than 6 percent at an annual rate. …