Today, we present a guest post written by Jeffrey Frankel, Harpel Professor at Harvard’s Kennedy School of Government, and formerly a member of the White House Council of Economic Advisers. A shorter version appeared at Project Syndicate.
July 19, 2022 — On July 28, the US Bureau of Economic Analysis will release its advance estimate of economic growth in the just-completed second quarter of the year, as measured by GDP. The announcement is attracting more than the usual eager anticipation. The reason is that many observers predict that the Q2 GDP number will be negative and that this will officially confirm widespread beliefs that the economy went into recession in the first half of 2022, figuring that growth in national output is already determined to have been negative in the 1st quarter of the year. After all, isn’t a recession defined as two consecutive quarters of negative growth?
The stock market and bond markets could react to a negative number on July 28 by rising in the very short run, as investors reason that the recession reports will encourage the Fed to ease up on its path of aggressive interest rate hikes.
The chain of reasoning is probably wrong, however, as is likely to become clear in the future. There are three flaws. First, growth is as likely to turn out positive in the second quarter as negative. Second, even if the number indeed comes out negative July 28, a US recession is not defined as two consecutive quarters of negative growth, but rather is determined by the Business Cycle Dating Committee of the National Bureau of Economic Research. Third, contrary to the general belief, it is not quite true that growth in the first quarter is determined to have been negative. Thus, even if the pessimists turn out to be right about the BEA announcement on July 28, subsequent events will likely prove them wrong about the timing of a recession.
Some economists, including me, would guess that growth was more likely positive in the second quarter than negative. But, for the sake of argument, let us proceed on the assumption that the reported second quarter change in GDP is negative. After all, the Atlanta Fed’s GDPNow model is estimating a second quarter growth rate of minus 1.5 % per annum, based on data available through July 15. (IHS Markit, another real-time estimator, is also negative.)
A US recession is not defined as two consecutive quarters of negative growth. Rather it is determined by the Business Cycle Dating Committee of the National Bureau of Economic Research, a private non-profit research institution. The NBER looks at a variety of indicators, rather than following a mechanical two-quarter-GDP rule.
This comes as a surprise to most people. It is true that the rule-of-thumb of two consecutive quarters of falling GDP is used as the criterion for a recession in most advanced countries. But not in all countries.
In the US, the BEA officially recognizes the role of the NBER. It shouldn’t be that surprising. Private non-profit institutions also produce such important economic indicators as Consumer Confidence and the Purchasing Managers Index.
Consider an historical illustration of how it makes a difference. The NBER chronology shows a three-quarter recession in 2001, with the preceding expansion peaking in the first quarter (March) and the recession trough coming in the fourth (November). If one looks at a variety of indicators, employment in particular, it is clear that there was indeed such a recession. But the episode would fail the criterion of two-consecutive quarters, because negative GDP growth in the first and third quarters of that year was interrupted by a positive number in the second quarter.
It is natural to view growth in output as the most important indicator of whether there has been a recession. But even so, and even if one were wedded to the two-negative-quarters rule, it is important to realize that growth was not necessarily negative in the first quarter as people think it was.
There are two ways to measure output. The one that gets all the attention in the US is GDP measured “on the product side,” i.e., by adding up the sectors in which goods and services are sold, like consumption. That measure of GDP growth was indeed negative in the first quarter (-1.6 % per annum). But growth is also measured by GDI, which is calculated “on the income side,” i.e., by adding up kinds of income, like employee compensation. The two measures should be precisely equal, in theory. But in practice there is a statistical discrepancy. It was large in the first quarter. GDI showed a positive change in that quarter (+1.8 % per annum). The average of the two measures was positive. (The average of the two is sometimes called Gross Domestic Output.)
Here comes the important part. Experts, including those in the federal government, have for some time viewed GDI as just as informative in measuring economic output as the much more famous GDP. So has the NBER committee: “The dating of the quarterly turning point dates also considers the average of real Gross Domestic Product (GDP) and real Gross Domestic Income (GDI)”.
There are two implications for July 28. The first is that, even if reported GDP shows two negative quarters in the first half of 2022, the NBER committee is very unlikely to conclude that a recession started in the first quarter. Such a conclusion would be too much at odds with GDI, with the very strong growth in employment, and with most other economic indicators in January-March.
Second, the GDP numbers will soon be revised, as is routine. Revisions are substantial. The mean absolute revision for a given quarter, going from the third BEA release to final (i.e., after mid-year “benchmark revisions”) is 1.2%, for a sample ending in 2018. This is the main reason why the NBER committee waits so long – 11 months, on average – before calling cyclical turning points.
BEA undertakes a comprehensive mid-year “benchmark revision” of the National Income and Product Accounts. Apparently the results are to be released in September this year, instead of July. At that time, the revision of first-quarter GDP could well be upward, conceivably even by enough to turn it from negative to positive. The reason to think so is that revisions of GDP in the past have often moved it in the direction of GDI, more than revisions in GDI have moved it in the direction of GDP. (In that sense, GDI may be a more reliable measure of domestic output than is GDP.)
Thus, we could have a situation on July 28 where the public and the markets, responding to media headlines, think they have been told that a US recession began in the first half of the year, only to be told the opposite two months later.
 The average lag for the 10 turning points between 1980 and 2019 was 11.7 months. For the 2020 recession, it took the NBER BCDC 4 months to call the beginning and 15 months to call the end.
This post written by Jeffrey Frankel.