What Is A Recession?

This article is an excerpt from my book “Recessions: Volume I” (Section 1.2). This is a topic that has come up in online conversations. My argument is that one can have multiple definitions of recession, and the concern is: what exactly are we trying to capture with the term “recession”? From what I am seeing in online discourse, some people seem to feel that anything bad about the economy is a “recession,” which is too vague to useful.

Also please note that I am using the same charts as in the book, and hence, they are out of date. The book was published in 2020, but the book was largely written in 2019.

The first challenge in discussing recessions is defining what exactly a recession is. One rule of thumb is that a recession is anytime there are two consecutive quarters of declining real gross domestic product (GDP). Although I use the “two consecutive quarters of declining GDP” definition for some countries (as discussed in the next section), we probably want to work with a different definition.

(In case the reader is unfamiliar with the concept of GDP, it is the total value of goods and services produced in the domestic economy. It does not count every single transaction, like the resale of goods between wholesalers and retailers, nor financial transactions. Nominal GDP is the value of domestic production in current dollar terms and is equal to nominal domestic income. Real GDP – which is what is used in the recession definition – adjusts nominal GDP to remove the effect of inflation. In a high inflation environment, nominal GDP might not decline in a recession.)

NBER Recession Definition

In the United States, the National Bureau of Economic Research (NBER) has a business cycle dating committee that makes determinations of the start and end months of recessions. (The fact that they work with a monthly frequency is valuable for recession dating, as recessions tend to be short-lived.)

The NBER has a brief qualitative description of their methodology on their website: https://www.nber.org/cycles/recessions.html. They write:

The NBER's Business Cycle Dating Committee maintains a chronology of the U.S. business cycle. The chronology comprises alternating dates of peaks and troughs in economic activity. A recession is a period between a peak and a trough, and an expansion is a period between a trough and a peak. During a recession, a significant decline in economic activity spreads across the economy and can last from a few months to more than a year. Similarly, during an expansion, economic activity rises substantially, spreads across the economy, and usually lasts for several years.

The definition refers to “economic activity.” One way of expressing that concept is the level of real GDP, hence there is an overlap with the “two consecutive quarters of declining GDP.” However, the NBER Business Cycle Dating Committee uses a broader definition of activity.

The Committee does not have a fixed definition of economic activity. It examines and compares the behavior of various measures of broad activity: real GDP measured on the product and income sides, economy-wide employment, and real income. The Committee also may consider indicators that do not cover the entire economy, such as real sales and the Federal Reserve's index of industrial production (IP). The Committee's use of these indicators in conjunction with the broad measures recognizes the issue of double-counting of sectors included in both those indicators and the broad measures. Still, a well-defined peak or trough in real sales or IP might help to determine the overall peak or trough dates, particularly if the economy-wide indicators are in conflict or do not have well-defined peaks or troughs.

By using other indicators, it is possible to fix start and end months. It also avoids some concerns with real GDP. For example, GDP included imputed measures of value-added, which will not rise and fall with the business cycle.

The chart above shows the post-1949 history of real GDP growth and recessions. The annual percentage change in real GDP is depicted, with the recession periods in the United States (as defined by the NBER committee) shaded in pink. As expected, growth rates fell in recessionary periods. One thing to note is that showing the annual percentage change, we are comparing real GDP in a quarter with its level four quarters earlier. This smooths out the wiggles in quarterly data. One side effect is that the growth rate (on this measure) remained positive throughout the 2001 recession, while real GDP shrank in the first and third quarter of 2001 (using the current measure of GDP, which may be revised in the future as a result of changes to definitions). Since the shrinkage did not occur in consecutive quarters, this recession would not qualify under the “two consecutive quarters of declining real GDP” definition.

Definitions Vary Across Countries

For the United States, the National Bureau of Economic Research (NBER) has managed to become the authoritative source for recession dating, whereas for other countries, recession dating determination is less clear. In this text, I stick with the NBER definition for the United States and treat other countries on a case-by-case basis. In my figures, I indicate what method is used to generate the recession bars. Since I think we should be more concerned about the side-effects of recessions than formal declarations of recessions, my preference is for flexibility in definition. (I return to this in Section 2.5.)

Rather than attempt to list the various bodies that date recessions in other regions, I will just use Canada as an example. In Canada, the C.D. Howe Institute has created a Business Cycle Council which has a mandate to date recessions: https://www.cdhowe.org/council/business-cycle-council. The chart below shows the post-1990 history of real GDP growth and the recession dates as defined by that committee. There are two declared recessionary periods: March 1990 to April 1992, and October 2008 to May 2009.

Considerable Leeway in Definitions

From a practical perspective, one might argue that the bar used by the C.D. Howe committee for declaring a recession is too high.

The top panel of the figure above shows the Canadian unemployment rate and the C.D. Howe Business Cycle Council's recession dates. (Note that this council was formed after the 2008 Financial Crisis, and so the earlier recession dates were based on applying academics’ methodologies to past data.) The unemployment rate rose by just over 1% in 2001, in a relatively rapid fashion. Although the downturn was somewhat localised, it certainly felt like a depression for anyone associated with the technology industry.

The bottom panel shows the rise in unemployment rate versus the trailing minimum value over the past 12 months (including the current month, so the minimum of this time series is zero). We see a big jump in 1992, as well as 2001, that were not classified as recessions (based on continued growth in other economic activity series).

I would argue that the rise in the unemployment rate overshadows the other activity indicators for two sets of reasons.

• Normative policy perspective. We should react to a rise in the unemployment rate based on ethics, as well as the cost to society created by unemployment, as well as the long-lasting effects of job loss (referred to as “hysteresis”).

• Inflation-targeting perspective. In the post-1990 environment, the only plausible candidate for “capacity constraints” in the domestic developed economies revolves around the labour market – rather than the commodity markets. There is a massive excess of manufacturing and service sector productive capacity, so that growth in output alone is not enough to trigger a plausible inflation risk. (Note that although I view NAIRU to be a questionable concept, I accept that labour market “tightening” (vaguely defined) will eventually pose wage inflation risks. As we have discovered repeatedly since the 1990s, an unemployment rate that is drifting lower tells us little about inflation risks, but a rapid rise probably is telling us something.)

My comments above on the importance of the rise in the unemployment rate are not purely my opinions: the Bank of Canada reacted to the weakness in the labour market (chart above). From a bond market participant's perspective, the drop in the target rate from 5.75% in January 2001 to 2% in 2002 is what you are paid to forecast, not what the C.D. Howe Business Cycle Council thinks.

However, it is possible to get a definition that is perhaps too sensitive to economic fluctuations. The figure above shows Canadian real GDP, and the “turning points” defined by the OECD leading economic indicator. (URL: http://www.oecd.org/sdd/leading-indicators/oecdcompositeleadingindicatorsreferenceturningpointsandcomponentseries.htm.) The OECD has these indicators for most (if not all) of the countries in their data coverage universe.

The shaded regions show where the OECD leading economic indicator dropped from a local peak to a local trough. It does align very well with the acceleration and deceleration of real GDP growth, so these indicators could be useful for some purposes. (For example, financial markets might be sensitive to such acceleration.)

However, the “negative turning points” appear to be too numerous to be taken as a definition of a recession. For example, we saw very little in the way of a rise in the unemployment rate during two of the episodes after 2010, and only about 0.5% in the third (bottom panel) – although that 0.5 rise would have breached the trigger proposed by Sahm (Section 2.5).

Is There a “Best” Dating Methodology?

My interest here is in the theory of relatively deep recessions. They do not have to be as devastating as the crisis in 2008 but need to be deeper than the fluctuations that seem to be captured by the OECD indicators. The reason is straightforward: these big dips in activity typically result in large errors in economic forecasts, which is my theoretical preoccupation.

As a result, I have a primarily qualitative interest in recessions, and do not see much value in trying to find the “best” recession-dating methodology. Instead, I argue that we might be interested in different aspects of recessions and we might need a different set of criteria for each aspect. In the United States, the NBER-defined recession dates have done a very good job of coinciding with most aspects of interest, but the experience in other countries seems to be somewhat more muddled (as in the Canadian post-2010 experience), and so recession dating is going to be more controversial.