Broadly speaking, a recession is a protracted economic downturn that marks the end of one business cycle and the beginning of another. There are lots of ways of calling a recession and even more methods of trying to predict one. In the end, defining and predicting recessions is more of an art than a science.
One common definition a recession used by media pundits is two consecutive quarters of negative GDP growth. Most recessions fit this definition, but not all of them. For example, there were two quarters of negative GDP growth around the time of the early 2000s recession, but they were not consecutive.
Another way to call a recession is to look at unemployment rates. For example, the Sahm rule says that a recession has started when the moving average of the national unemployment rate has risen .5 percentage points above its 12-month low.
But the most authoritative arbiter of recession is the National Bureau of Economic Research (NBER). The NBER definition of a recession is “a significant decline in economic activity that is spread across the economy and lasts more than a few months.” The NBER takes into account many indicators including unemployment, GDP, personal income, industrial production and retail sales.
The above chart displays a series of rectangles that represent NBER-dated recessions. Since 1980 there have been six NBER-dated recessions:
- 1980 recession (Jan 1980 - July 1980)
- 1981-1982 recession (July 1981 - Nov 1982)
- Early 1990s recession (July 1990 - Mar 1991)
- Early 2000s recession (Mar 2001 - Nov 2001)
- Great Recession (Dec 2007 - June 2009)
- COVID-19 Recession (Feb 2020 - April 2020)
One could easily go further back in time. According to Wikipedia there have been approximately 48 recessions since the U.S. was founded. The NBER was founded in 1920 and released its first business cycle dates in 1929. FRED displays NBER-based recession indicators going back to December of 1854.
Causes of Recession
One way to think about recessions is that they are part of a natural cycle of economic expansion and contraction. In this sense, we can view the broader “cause” of a recession as an overheating of the economy when it reaches full capacity towards the end of an expansionary period.
This overheating of the economy typically leads inflationary pressures that in turn induce the Federal Reserve to tighten monetary policy in order to cool the economy off. And while the Fed would of course prefer to cool things off without tipping the economy into a recession, all good things must come to an end. We refer to periodic expansions and contractions of the economy as “business cycles” for a reason.
The next thing to consider are “black swan” events, i.e. the bad news that either signals or triggers the start of a downturn. In the early 2000s we associate the start of the recession with the bursting of the dot-com bubble and the September 11 attacks. The Great Recession is similarly associated with the bursting of the housing bubble and the subprime mortgage crisis. The proximate cause of the COVID-19 recession was obviously COVID 19.
While bad news always has economic implications, it does not always trigger a recession. Black swan events are much more likely to have a causal impact during the apex of business cycle due to the fact that the economy is already at or beyond full capacity.
What is a Depression?
Put simply, a depression is a really severe recession. Investopedia defines a recession as “an extreme recession that lasts three or more years or which leads to a decline in real gross domestic product of at least 10%.” Depressions are relatively rare. The last one to occur in the United States was the Great Depression, which lasted from 1929 1933 and resulted in a drop of GDP of 26.7%.
Making definitive statements about when a recession has started or ended is challenging, especially in real time, but there generally three approaches that pundits rely on when trying to predict shifts in the business cycle.
The first (and most common) approach relies on looking at individual indicators and seeing how many of them are signalling a downturn or recession. The first are leading indicators, like treasury yields, business confidence, consumer confidence or a composite leading indicator. Next are what I would refer to as “contemporaneous indicators” that give a snapshot of key sectors of the economy like housing or capital goods orders. Finally, there are lagging indicators like the unemployment rate or stock prices that a recession is either underway or about to end.
A second approach relies on quantitative models that combine multiple indicators to predict the likelihood of a recession onset or the start of a new expansionary period. One example are the smoothed recession probabilities developed by Chauvet and Piger. These are calculated in real time and updated with new data in an effort to tell us whether we are entering or exiting a recession in real time.
A third approach represents the cutting edge of data science and economics on this topic. It involves using machine learning and deep learning algorithms to precise probabilities of a recession six, 12 or even 24 months in advance.