San Francisco, 1935. Photo by Dorothea Lange, courtesy the Franklin D. Roosevelt Presidential Library and Museum.
Charts help us put numbers in perspective by easing historical comparisons and highlighting trends. So how should we handle numbers like the past few weeks’ unemployment claims, which are literally off the charts?
Like any other numbers – thoughtfully and carefully, only more so.
The headline figures are certainly headline worthy. Last week the U.S. Labor Department reported that seasonally adjusted, new unemployment claims for the week ending April 4 totaled 6.6 million nationwide. The bureau also revised totals for the previous week upward to nearly 6.9 million. This followed a then-unfathomable 3.3 million reported for the week ending March 21, when coronavirus-induced lockdowns first became widespread in this country. These three weeks account for some 16.8 million reported claims.
We have never seen anything remotely like this over such a short period. Before the pandemic’s arrival, weekly unemployment claims had been hovering in the 200,000s. The total number of unemployed Americans in the labor market in February was around 5.8 million, and this figure included people who had been out of work for as long as six months in many places. At the depth of the Great Depression in 1933, a total of 12.8 million U.S. workers were counted as unemployed.
If you want to assert that COVID-19 has caused more Americans to lose their jobs than the Great Depression did, these figures will let you truthfully say as much. But what numbers say and what they mean can be considerably different things. And what I think a set of numbers means may be different than what you think it means. We interpret data according to our own experiences, perspectives and biases. It is still useful, however, to understand what the data reflects before we interpret it.
Let’s start with the recent unemployment claims figures. Note that these headline figures are “seasonally adjusted.” That’s another way of saying they are not real. Labor statisticians take raw data and alter it based on prior seasonal trends. Workers at ice cream stands, for example, probably submit more unemployment applications in November than they do in June. To avoid distortions, or alleged distortions, in the data, statisticians massage the figures to account for these trends. (I say “alleged distortions” because the summertime ice cream stand worker may become the package sorter or delivery driver who helps handle the holiday season gift-giving rush. Seasonal adjustments are supposed to account for this too.)
How can anyone seasonally adjust statistics when an unprecedented pandemic forces huge swaths of the economy to shut down practically overnight? There has never been a season when New York City restaurants, hotels and airports were forced to curtail their workforce as they now have. When we look at the raw data for unemployment claims in the past three weeks, we see 6.2 million, 6.0 million and 2.9 million, for a total of 15.1 million. This is still a huge number, but a real one – sort of.
This raw figure does not account for the unknowable number of people who have not been able to submit unemployment claims at all because they could not get through to overloaded state-run websites or phone lines. These people could not apply for benefits in person, either, because of social distancing shutdowns. So while seasonal adjustments likely overstate the numbers, these obstacles to new claims leave the figures at least temporarily understated.
However we count the recently unemployed, these figures are not directly comparable to any past unemployment claims data. This is because – by extending unemployment benefits to many self-employed individuals, including millions of “gig economy” workers – the government has altered the very definition of unemployment. How do you determine when a singer-songwriter is unemployed? She may have lost paying gigs due to canceled festivals or concerts, but she can retreat to her home studio to write or record music. She might sell that music to someone who wants to use it in a television commercial next month. She might release it on her next album. Or she may never use that music at all.
In ordinary times, self-employed human capital is not counted as unemployed. But under the expansion of benefits Congress enacted last month as part of the CARES Act, someone like our singer-songwriter is. Or she might be, at least. As of this writing, we were still waiting for federal officials to release rules that will guide the states in awarding unemployment benefits to people like our hypothetical musician, and in millions of other varied circumstances.
Is an Uber driver unemployed if he stays home pursuant to a governor’s order closing nonessential business? Probably, under the eventual federal guidelines. What if that driver continues to work delivering takeout meals as part of Uber Eats, but earns only a fraction of his usual income? How will usual income be defined, since the driver may have varied his driving hours according to demands on his time from other work, school or personal commitments?
From any point of view, and regardless of what the statistics mean, the economic pain induced by national efforts to flatten the pandemic’s curve is substantial and spread widely, if unevenly. Armies of workers are idled in industries that encompass big employers including airlines, hotels, restaurants and casinos. Employees of smaller businesses like hair salons and chiropractic offices have also been hit hard. Other fields, like journalism, accounting and law, are less affected because more of their staff can work from home. A few types of workers who make it possible for the rest of us to stay home and healthy are swamped.
In such odd times, we can have trouble understanding what statistics mean for real-world conditions. Typically a worker must be involuntarily out of work for at least a week before becoming eligible for state-administered unemployment benefits. The CARES Act encouraged states to waive the waiting period (on the federal dime), which accelerated claims. Usually, benefits would then begin to flow almost immediately. But between the volume of regular claims and the expansion to formerly ineligible claimants, a far higher percentage of claimants have been waiting several weeks to receive any money. They may have to wait for several weeks more. This delay due to volume is also affecting other coronavirus relief efforts, notably the Paycheck Protection Program that is meant to issue at least $350 billion in forgivable loans to small businesses that keep workers on their payroll, or rehire those who have been laid off.
The financial hardship produced by the pandemic does not approach the misery of the Great Depression, no matter how many people end up being counted among the unemployed. The 12.8 million who were out of work in 1933 represented about a quarter of the 51 million people in the American workforce at the time. Most of them were the sole or primary breadwinners in their homes. Today’s unemployment numbers represent about 10% of the modern workforce, which consists of many more two-income households.
The Depression’s unemployed had very little in the way of government unemployment benefits on which to fall back. There was nothing at all like the federal stimulus payments of up to $1,200 that should begin hitting most Americans’ bank accounts sometime this week, via direct deposit from the Internal Revenue Service. Social Security did not begin paying benefits until 1940. And the unemployment followed (and to a considerable extent was triggered by) a wave of bank failures that wiped out the savings that might have helped carry millions through the downturn. Federal deposit insurance only came along in June 1933, after those failures had occurred.
A better measure of today’s economic pain could come from the increase in people relying on food banks and other charities simply to eat now that they have lost their daily income. We know anecdotally that these numbers have swelled. But we don’t have the sort of meticulously curated government statistics that would tell us how this suffering compares with the postwar past. We also know that, according to one industry survey, more than 30% of residential renters had not paid their rent in the first week of April. That sounds like a lot, and it is – but we should bear in mind that in a typical nonpandemic month more than half of these late payers would have been this late anyway. A better view may come later, if we can get statistics about how many people fall more than a month behind in rent payments.
Any modern comparisons with the Great Depression are inherently misleading. They are also of little help to those in economic distress right now. The Depression was notable not only for its depth but for its duration, spanning more than a decade before our entry into World War II absorbed all the economy’s idle labor. Bad as current conditions are, if misery in the population could be charted, our current levels would not approach what Americans experienced through the 1930s.
With good choices and good luck, a real recovery from this contraction could be underway in a matter of months. We won’t know until we have more statistics. Even then, how we interpret those statistics will be a function of our personal perspectives.