This piece is a follow-up to the report published in June 2020, "The COVID-19 crisis: How do U.S. employment and health outcomes compare to other OECD countries?"
In the past three to four months, the U.S. economy has recovered somewhat from the COVID-19-induced employment troughs observed in April, while the virus caseloads and deaths at first declined and then surged again, starting in June. However, the partial recovery of U.S. labor markets in the late spring and summer, and more recent trends in virus cases and deaths, do not change the fundamental fact that both employment and health outcomes for the U.S. during the pandemic have been worse than in almost any other high-income country in the world.
This brief presents an update of my earlier analysis showing unemployment rates as of April, and virus caseloads and deaths as of late May. This time, I focus on changes in unemployment rates from January of this year through July, the most recent month for which virtually all of these countries report monthly unemployment data (though they are now available for August in the U.S.).1 I also present virus caseloads and deaths for virtually each country, as of September 9. Like the earlier brief, I present data for total virus cases and deaths per capita, as well as new cases and deaths per capita (now from a seven-day moving average). And I report some virus data for the U.S. (relative to other countries) as of July 15, so that we can compare employment and health outcomes for similar points in time. The sources of these data are the same as for the earlier brief.2
Figure 1 presents a summary comparison of unemployment rate increases between January and July 2020 in the U.S. to the mean increase across the other OECD countries in part A, and a comparison of new virus cases and deaths per capita in part B. Table 1 presents similar unemployment rate increases between January and each month from April through July in the U.S. and the other OECD countries (in Part A), as well as both total and new virus caseloads and deaths per capita in the US and the other countries (in part B).3
The data in Figure 1 and Table 1 on unemployment rates indicate the following:
- Unemployment rates in the U.S. rose by over 11 percentage points between January and April before falling by 4.5 percentage points from April to July;
- Unemployment rates rose, on average, by much less in the other OECD countries, and then fell modestly after April;
- The January-April increase in the U.S. was 11 times larger than the average of other OECD country increases, and the January-July increase in the U.S. remained five times larger than the average of the others.
The most recent unemployment rate for the U.S., for the month of August, declined by nearly two percentage points. Assuming that such rates continued to slowly decline in the rest of the OECD, the increases in the U.S. would now likely be four times higher than in the OECD, rather than five times.
The data on new viruses and caseloads in the U.S. and the other OECD countries indicate the following:
- Overall virus cases per capita in the U.S. are now over four times higher in the U.S. than in the average high-income OECD country, while total deaths per capita are over twice as high; and
- New virus cases in September in the U.S. are 60 percent higher than in the average OECD country, and new deaths are five times higher.
Changes in Economic and Virus Outcomes:
|Unemployment Rates (%)||Others||USA|
|Jan – Mar 2020||0.31||0.8|
|Jan – Apr 2020||0.98||11.1|
|Jan – May 2020||1.41||9.7|
|Jan – Jun 2020||1.32||7.5|
|Jan – Jul 2020||1.22||6.6|
|Virus Cases and Deaths|
|Note: The sample of countries includes the 25 richest OECD countries (in per capita terms) with at least 4 million people. Countries reporting unemployment quarterly rather than monthly and Saudi Arabia (for which few economic data are available) are excluded. Unemployment rates are obtained from Trading Economics (www.tradingeconomics.com) on September 9. COVID-19 case numbers are obtained from Johns Hopkins University Coronavirus Resource Center (www.coronavirus.jhu.edu) on September 9. “New cases” and “new deaths” are defined as 7-day moving averages. The sample of countries for virus cases and deaths is the same as for unemployment, except that it excludes Hong Kong, for which these data are not available.|
The country-specific data on unemployment rates and virus cases/deaths for each individual country appear in the Appendix table (part A for unemployment rates and part B for virus data). These results show the following:
- The unemployment rate increase of 6.6 percentage points in the U.S. between January and July is the largest of any high-income country in the OECD;
- Total per capita virus cases are higher in the U.S. than in any other wealthy OECD country, and total deaths per capita here are higher than in all but four; and
- The new per capita caseload is the highest in the US except for two other OECD countries, while new deaths per capita are highest in the U.S. except for one (Israel).
In sum, the partial recovery of U.S. labor markets in the late spring and summer, and more recent trends in virus cases and deaths, do not change the fundamental fact that was apparent in my earlier brief: both U.S. employment and health outcomes during the pandemic have been worse than what we find in virtually all other high-income countries around the world.
Some questions and a few caveats are now in order. First, is it more appropriate to use employment outcomes rather than other economic data, like real GDP changes or household income? And, if so, is the unemployment rate the best labor market indicator to use?
I believe the answers to these questions are “yes” and “yes,” though measured unemployment rates are far from perfect. It is true that changes in real GDP in the U.S. have been more in line with those of other OECD countries; and household incomes had been maintained until recently because of relief through the CARES Act from last spring, especially through enhanced Unemployment Insurance benefits (from the Pandemic Unemployment Assistance and Pandemic Unemployment Compensation programs, or PUA and PUC).
But employment changes are an immediate and growing concern to many workers, who now risk permanent job separation more than before. Other countries have made much greater use than we have of payroll subsidies that keep workers connected to their jobs; in the U.S., the Payroll Protection Program (PPP) has expired, and was never as widely used as comparable programs in most other countries. Also, the $600 increases in weekly PUC payments have expired, and now are reduced to just $300 per week (on top of being only temporarily funded from disaster assistance funds). So unemployment has become a much graver concern in the U.S. than it was last spring.
And the unemployment rate is the single employment measure most widely and frequently reported by OECD countries. In the U.S., despite the well-known limitations of the unemployment rate (known as “U3” in the monthly reports of the BLS), it has trended fairly consistently with other measures like the employment-to-population ratio and changes in payroll jobs over the past several months.4
Another question about the results presented is that they reflect somewhat different time periods – with the labor market data for July and the COVID-19 cases/deaths in September. In each case, I have presented the most recently available data. But it is natural to ask whether there is at least a correlation between employment and COVID-19 cases in the same time period, if not a more causal one.
To ascertain the extent to which such a correlation exists, I have compiled similar data on COVID-19 cases and deaths across these countries, but as of July 15 rather than September 9. The earlier data actually tell a worse story. New cases at that point are .203 per thousand, substantially higher than those we observe in September, and new deaths are 2.92 per million – more than twice higher than the 1.34 per million we observe in September. Both numbers in July are higher than what we find for almost any other high-income OECD country.
So, if increases in unemployment rates and new virus caseloads/deaths are correlated across the U.S. and other countries at similar times, do we believe the virus rates are causing higher unemployment here? Here again the likely answer is probably “yes,” at least to some extent. Our virus caseloads in the U.S. began rising in mid-June, and death rates began rising again in July, after falling through much of the late spring. At about the same time, our labor market recovery began to flatten – with job growth and unemployment declines peaking in June and falling in July.5 Economists at the Federal Reserve Bank, among many others, also found a flattening recovery around that time, and linked it to a resurgence of virus cases.6
One final question remains that I addressed in my earlier brief in June: had the U.S. achieved the same lower unemployment rate and virus deaths as the other OECD countries, how many more jobs would American workers now have, and how many lives would have been saved? My calculations indicate that:
- An equivalent increase in the unemployment rates of the US and other wealthy OECD countries would have meant that at least 8.6 million more Americans would be employed today;7 and
- An equivalent increase in our death rates would translate into 112,540 American lives that would have been saved.
In a brief I released in early June, I reported that the U.S. unemployment rate had risen by vastly more than those of almost all other wealthy OECD countries, and our virus caseloads and deaths per capita were higher, as of April 2020.
Our labor market has recovered somewhat since that time, though progress in fighting the virus has been uneven. Comparing recorded unemployment rates as of July in the U.S. and other wealthy OECD countries, we find that the vastly larger increases in unemployment in the U.S. has declined somewhat relative to other countries, though ours remains much higher. And, comparing virus caseloads/deaths between the U.S. and other wealthy countries indicates much worse outcomes here as well, especially when we compare the most recent trends.
Furthermore, there is good reason to believe that our limited progress in fighting the COVID-19 virus has at least partially caused our continuing high unemployment rate. Had we been as successful in each measure as the other OECD countries, nearly nine million more Americans would be employed and over 100,000 would still be alive.
Looking at other economic measures, like real GDP or household income, would not generate as large a contrast between the U.S. and these other countries. But employment is a very meaningful measure – all the more so as relief efforts in the U.S. for unemployed workers have weakened and as more laid off workers become permanently unemployed. And the monthly unemployment rate is the most frequently reported measure of employment across OECD countries.
At least by these measures, the U.S. has experienced the worst of both worlds – very high unemployment and very high virus caseloads and deaths – during the COVID-19 pandemic.8
Appendix Table 1: USA v. Other and OECD Countries (Population > 4M): Employment and Virus Outcomes in 2020
Part A. Unemployment Rates in 2020 (%)
|Note: The sample of countries includes the 25 richest OECD countries (in per capita terms) with at least 4 million people. Countries reporting unemployment quarterly rather than monthly and Saudi Arabia (for which few economic data are available) are excluded. Unemployment rates are obtained from Trading Economics (www.tradingeconomics.com) on September 9.|
Appendix Table 1 (Cont’d)
Part B. Virus Cases in USA v. Other OECD Countries
|Country||Cases/1,000||New Cases/1,000||Deaths/1,000||New Deaths/1,000,000|
|Note: The sample here is the same as in Part A, except it excludes Hong Kong, for which virus data are not available. COVID-19 case numbers are obtained from Johns Hopkins University Coronavirus Resource Center (www.coronavirus.jhu.edu) on September 9. “New cases” and “new deaths” are defined as 7-day moving averages.|
- Four of the original countries report only quarterly unemployment data, and I have removed them from the sample.
- The unemployment data appear in the online website Trading Economics (https://tradingeconomics.com/country-list/unemployment-rate?continent=europe). The virus caseloads and death data appear in the Johns Hopkins University coronavirus website: (https://coronavirus.jhu.edu/data/new-cases). Data on unemployment rates for each country appear in the former website, except for the UK and Norway (since unemployment rates there were not yet available for July). Data on virus cases and deaths appear at the latter website except for Hong Kong, which no longer appears in these data.
- The means of unemployment rates and virus caseloads/deaths across the other OECD countries are unweighted, rather than weighted by country population. This enables each country to count equally in comparison to the U.S.
- In April, the unemployment rate peaked at 14.7 percent, the employment/population ratio bottomed out at 51.3 percent, and payroll jobs had dropped to 130.3 million. By July, all three measures had recovered by approximately 40 percent, relative to their values in January or February.
- In June, the unemployment rate fell by 2.2 percentage points, compared to May, while employment/population rose by 2.2 percentage points and payrolls grew by 4.8 million. In July, unemployment dropped by 0.9 percentage points, the employment rate rose by just 0.5 points, and payrolls by 2.3 million.
- For instance, see the comments made by Fed chairman Jay Powell in late July: https://www.marketwatch.com/story/fed-sees-some-pickup-in-economy-but-maintains-dovish-policy-stance-2020-07-29
- The 8.6 million job figure assumes a constant labor force participation rate, though we know our labor force participation rate has declined. Had there been no labor force decline, the higher number of employed workers would be 8.9 million. Since the other OECD countries likely suffered some labor force reductions as well, the true higher number of jobs that would exist in the U.S. is likely somewhere in between 8.6 and 8.9 million.