Recession

A recession is a contraction in overall economic activity, dated in the United States by the Business Cycle Dating Committee of the National Bureau of Economic Research. NBER states its traditional definition as "a significant decline in economic activity that is spread across the economy and that lasts more than a few months," and describes the committee's view that while each of the three criteria, "depth, diffusion, and duration," needs to be met individually to some degree, "extreme conditions revealed by one criterion may partially offset weaker indications from another."

Two features of that arrangement surprise people. The authority is real but not official: NBER is a private nonprofit, and it notes that when the Commerce Department began publishing its chronology of peaks and troughs in 1961 it did so "without endorsing them," and that "there is no alternative business cycle chronology compiled or published by the US government." And the dating is historical rather than current. The committee waits until a turning point is beyond doubt, which means a recession is identified well after it has begun and a recovery well after it has started.

The rule of thumb most people quote deserves a careful correction rather than a debunking. NBER's own answer is that "most of the recessions identified by our procedures do consist of two or more consecutive quarters of declining real GDP, but not all of them," and it names the exception: "in 2001, for example, the recession did not include two consecutive quarters of decline in real GDP." So the two-quarter guide is usually right about NBER's recessions and is not the test anyone official applies. NBER gives four reasons for not adopting it. It does not identify economic activity with real GDP alone; it considers the depth of the decline; its main chronology is monthly rather than quarterly; and in looking at quarterly production it gives equal weight to real gross domestic income, whose difference from GDP, "called the 'statistical discrepancy'," was "particularly important in the recessions of 2001 and 2007-2009." That last point is absent from almost all consumer coverage, which treats GDP as the measure.

The indicators the committee actually watches are monthly and several. NBER lists real personal income less transfers, nonfarm payroll employment, real personal consumption expenditures, manufacturing and trade sales adjusted for price changes, employment as measured by the household survey, and industrial production, and then says plainly that "there is no fixed rule about what measures contribute information to the process or how they are weighted in our decisions." There is no formula to anticipate, which is part of why forecasting the announcement is not a useful exercise.

The single most under-published fact about the term is that these labels describe direction and not level. NBER: "recessions and expansions refer to the direction of change in economic activity, not its level," and because activity in the trough month can be well below the level at the previous peak, "the first few months or even years of an expansion can witness a level of activity that is below the previous peak." Its own example is the expansion that began in June 2009, after which real personal income less transfers did not exceed its prior peak until July 2011 and nonfarm payroll employment did not exceed its prior peak until May 2014, nearly five years into an official expansion. A household that felt the recovery was a fiction was not wrong about its own circumstances. The label was measuring something else.

Unemployment, which is the indicator most people actually watch, is the one least able to say where the economy is in the cycle. NBER describes it as "a trendless indicator that moves in the opposite direction from most other cyclical indicators," notes that it reached a low of 4.4 percent in May 2007 and had risen only to 5.0 percent by the December 2007 peak, and that after the March 1991 trough it "continued to rise for 15 months." It confirms turning points late rather than signaling them early.

Two categories in common use are not NBER categories at all. It "does not define a special category called a double-dip recession," treating a second contraction as either part of the same episode or a new one depending on the strength and duration of the upturn in between, and it "does not separately identify depressions in its business cycle chronology." A depression is a colloquial description rather than an entry in the record.

One reassurance about the delay: the dates are slow but they are stable. NBER states that since the Dating Committee was created in 1978 "there have not been any changes to previously-announced business cycle turning points." The caution buys accuracy rather than covering for revisions.

The committee reviews monthly and quarterly indicators, waits until a peak or trough is unambiguous, and then announces the month. NBER says there is no fixed timing rule "because the committee waits long enough to avoid any doubt about the existence of a peak or trough," and publishes the record of how long that has taken. The shortest elapsed time was four months, for the February 2020 peak, and the longest was 21 months, for the March 1991 trough announced in December 1992.

Reading the whole table is what makes the lag concrete, because peaks and troughs are not announced with the same delay. Peaks have been announced between 4 and 12 months afterward, the December 2007 peak taking 12 months. Troughs have taken between 8 and 21, with the June 2009 trough taking 15 months and the November 2001 trough 20. So the news that a downturn has begun is late, and the news that it has ended is later still. A household living through 2008 and 2009 learned that the recession had started in December 2007 during December 2008, and learned that it had ended in June 2009 during 2010. Both prices and the recovery had moved first. That is the reason a recession is not a signal a portfolio can be traded on: by construction, the confirmation arrives after the events it confirms.

One convention is worth knowing for anyone computing a length from published dates. NBER counts "the first month of the recession is the month following the peak and the last month is the month of the trough." So a peak in March and a trough in September is a six-month recession covering April through September, not seven months. Applied to 2020, a February peak and an April trough gives a two-month recession, which NBER classified as one anyway on the ground that "the subsequent drop in activity had been so great and so widely diffused throughout the economy that, even if it proved to be quite brief, the downturn should be classified as a recession."

What the NBER approach gets right

• Uses several monthly indicators rather than one quarterly number, which is why it caught 2001, the one recession NBER names as lacking two consecutive quarters of declining real GDP.
• Gives real gross domestic income equal weight with real GDP, and NBER records that the gap between them mattered particularly in the 2001 and 2007-2009 recessions.
• Waits for certainty, and the record shows the payoff. No previously announced turning point has been changed since 1978.
• Publishes its criteria, its indicator list, and its own announcement lags, so the limits of the exercise are documented rather than implied.

What it cannot do for a household

• Cannot tell you where you are now. Peaks are announced months later and troughs later still, so the current state is never officially known.
• Describes direction, not level, so an expansion can be under way while employment and income remain below their previous peaks for years.
• Has no formula and no fixed weights, so the announcement cannot be anticipated from a checklist.
• Says nothing about any particular household, industry, or region, all of which can contract while the national chronology shows an expansion.
• Offers no help with the indicator most people watch, since unemployment can rise before a peak and keep rising long after a trough.