COVID-19 Economic Downturn: What do cyclical norms suggest?
“But the universal nature delights in change, and in obedience to her all things […] from eternity have been done in like form, and will be such to time without end.” Marcus Aurelius, Meditations, Book Nine
Business cycle downturns come in many forms. Some are big, others small. Some are long, some are short. Some result from policy errors or euphoric booms, while others are the consequence of external events. Fortunately, unlike Marcus Aurelius’ classically stoic view of all things, they are not eternal.
Nevertheless, downturns have some common features and regularities. Among those that have been reasonably stable over much of the past half century are the relationships among unemployment, activity and federal budget deficits. Using these, we explore the impact of the U.S. COVID-19 economic downturn that began last month.
To sum up, recent labor market developments already make clear that we are in the midst of the deepest recession since the 1930s. In fact, the coordinated shutdown of a large swath of the American economy has made this plunge more rapid than that of the Depression. Whether we are at the start of a second Depression depends greatly on how long we keep the economy in a state of suspended animation.
If the lockdown extends from weeks to months, the short-term pain will turn into long-term scarring. The longer it takes to reopen businesses safely, the more damage we will do to the many linkages and networks (including lender-borrower, supplier-user and employer-employee relationships) that make up the fabric of the economy. As the wave of bankruptcies grows, damage to the financial system will increase, as will the resulting harm to the economy’s productive capacity.
While many aspects of the current experience are without precedent, history does help us gauge some aspects of what is happening around us. We start with what is arguably the most famous feature of business cycles: the link between unemployment and aggregate economic activity. In 1962, Arthur Okun observed that there is a stable relationship between short-run deviations of economic activity from trend and the unemployment rate (see here). Recently, Ball, Leigh and Loungani argue that “Okun’s Law” is strong and stable both in the United States and in most advanced economies.
With that in mind, we plot the deviation of annual real GDP growth from its norm on the horizontal axis and the four-quarter change in the unemployment rate on the vertical axis (see chart below). Consistent with many prior estimates of Okun’s Law in the United States, the slope of the fitted line is -0.50. This means that for every one percent decline of economic growth from its norm, the unemployment rate rises by one-half of one percentage point.
United States: Growth deviation from norm versus four-quarter change in the unemployment rate (quarterly), 1985-2019
Note: The normal rate of growth is the CBO measure of the potential growth rate. Source: FRED.
At this stage in the COVID-19 downturn, we know more about the unemployment rate than about economic activity, so we can use Okun’s Law to help us estimate current-quarter real GDP growth.
To do so, however, we first need to estimate how much the unemployment is likely to rise from its March level of 4.4%. While the reported unemployment rate is not based directly on weekly unemployment insurance readings, these data still provide us a useful view of what is likely to happen. Between mid-March and mid-April, the sum of all initial unemployment claims was a staggering 22 million people: that alone is 13½% of the 163 million persons in the labor force. Adding these newly unemployed to those already counted in early March leads us to estimate an April unemployment rate of 18%―a post-Depression record. We plot this on the following chart as the red diamond. (On April 20, Coibion, Gorodnichenko and Weber reported a private survey showing that most March-April job losers left the labor force. This would limit the rise of the still-to-be reported official April unemployment rate. In that case, measuring unemployment as we have would correspond better than the official unemployment rate to changes in activity and the budget deficit.)
U.S. unemployment rate, 1929-2020
Source: FRED (NBER Macrohistory Database from 1929-1942; BLS since 1948)
Given the path the economy is on, it would not surprise us if the unemployment rate in the current cycle peaks above 20%, as it did in the Depression. Nevertheless, if we assume unemployment remains at 18% from April through June, Okun’s Law implies a record four-quarter plunge in real GDP of –28.8%. Even assuming a substantial GDP drop in the first quarter of 2020, this means that the plunge from Q1 to Q2 could exceed –70% at a seasonally adjusted annual rate (SAAR). (Since 1947, the previous record was –10% in the first quarter of 1958.)
Now, –70% is far lower than the Congressional Budget Office’s latest estimate of a quarterly SAAR drop of –28%. It also is well below the private sector forecasts we have seen: these are typically in the range of –30% to –40%. However, the Okun’s Law estimate is consistent with the recent report that business shutdowns are depressing daily U.S. activity by 29%. If those conditions persist through June, then one would expect second-quarter real GDP to decline by 75% SAAR. In other words, the Okun’s Law estimate—while subject to wide error—hardly seems outlandish, while forecasts of smaller second-quarter declines probably anticipate that activity will pick up during May and June from an April trough.
What about the link between unemployment and the federal deficit? The answer (again) is that the relationship looks reasonably stable: As a rule of thumb, a one-percent rise of the unemployment rate over four quarters is associated with an equal increase in the deficit-to-GDP ratio (see chart).
United States: Four-quarter changes of the unemployment rate and the federal budget deficit (Percent of GDP), 1960-2019
Sources: FRED, BLS and BEA.
However, a recession-induced deterioration in the government budget has two distinct parts. The first is integral to the government’s tax and expenditure systems. As production and income fall, revenues decline. And, as employment declines, unemployment insurance benefits―which are administered by states but largely paid by the federal government―rise (see our earlier post). In addition to these automatic stabilizers, unemployment increases lead policymakers to boost spending, cut taxes, or both. These discretionary fiscal actions expand the deficit as well. Using estimates from the Congressional Budget Office, we reckon that about 60 percent of the observed 1:1 historical link between the unemployment rate and the deficit arises from automatic stabilizers, while the remaining 40 percent is a result of discretionary policy.
So, even without discretionary fiscal actions like the $2.2 trillion CARES Act, one should expect that a 13.5 percentage-point year-to-year increase in the unemployment rate would correspond to an 8-percentage-point rise of the federal deficit-to-GDP ratio to about 13.5% of GDP. That level is more than four percentage points higher than the previous postwar high of 9.3% in the first quarter of 2010. The CARES Act could add a further 10 percent of GDP to the deficit ratio. And, with further fiscal easing still likely, it would not be surprising to see the federal deficit at least temporarily enter the World War II range above 20% of GDP this year (see here).
Of course, what we really want to know is how long the coronavirus war will last. For that, however, U.S. business cycle history provides little guidance. Indeed, as far as we know, this is the first documented pandemic-induced U.S. downturn. (Correia, Luck and Verner provide preliminary estimates suggesting that states with higher than average exposure to the 1918 flu experienced relatively larger drops in manufacturing output. The timing of the flu epidemic also roughly corresponds to the brief 1918-19 recession.)
Aside from the duration and depth, business cycles also vary greatly in the strength of the recovery. How fast the economy returns to its pre-COVID-induced recession level depends critically on the maintenance and restoration of the mechanisms that are key for the economy’s capacity to operate. At the top of our list is the ability to revive or establish new businesses, to restore workers to the productive firms where they were previously employed, and to reestablish the flow of credit to healthy borrowers.
Will the government’s disaster relief and central bank’s crisis mitigation policies be sufficient to keep these networks alive and sustain productive capacity? From our perspective, the aggressiveness of the Fed bodes well. Moreover, fiscal policymakers have delivered large relief measures more quickly than ever before. However, in addition to our uncertainties about the evolution of the pandemic, we do not yet know whether the federal government will act with sufficient speed and effect to counter what are sure to be powerful contractionary forces, including the enormous pressures on state and local budgets and on indebted households and businesses.