Key Takeaways
- Approximately five million people in the U.S. are unemployed or have left the workforce since the pandemic began, and it is difficult to predict whether their decision was short term or permanent.
- It has been suggested that extended federal unemployment benefits offered during the pandemic have kept workers on the sidelines. Our analysis indicates this is not the case.
- States that ended the extended benefits earlier than the September deadline also largely reopened their economies sooner, so the reopening effect helped these states boost employment even before those benefits ended.
- The reason workers aren't filling jobs today seems to stem more from the decision to drop out of the workforce entirely, indicating a structural shift rather than a temporary change.
Once again, meager job gains in September did not live up to business demands. Even as extended unemployment benefits expired nationwide, and children returned to classrooms, people did not turn up for work. Business managers complain that they can't find employees, and the significant mismatch between job openings and labor turnover proves their point: job openings are now almost twice that of hires (see chart 1).
Chart 1
The Bureau of Labor Statistics (BLS) reported 194,000 job gains for September, with jobs added averaging 561,000 per month over the last three months. If the current pace of monthly jobs growth holds steady at around 500,000, the U.S. would only get back to the pre-pandemic trend in second-quarter 2023. However, based on the recent softer-than-expected job reports, there is a concern that the pace has started to slow closer to 230,000, which would mean the job market wouldn't get back to its pre-pandemic trend until fourth-quarter 2025 (see chart 2).
Chart 2
The biggest reason why the U.S. economy can't grow faster is supply constraints, in goods and services, including workers. Enormous supply constraints are crippling business capacity. This has resulted in much lower economic activity than would be the case if business operations were running smoothly. With that in mind, we revised our 2021 GDP forecast down a full percentage point to 5.7% in September from 6.7% in June.
The outward shift of the Beveridge curve, which shows a higher level of unemployment than at previous points in time when the vacancy rate was the same, highlights the labor mismatch in the U.S. economy (see chart 3). With the lackluster September employment gains despite the boom in demand, the question remains: where are the workers?
Chart 3
Possible reasons for the mismatch in the job market include extended unemployment benefits or the pandemic-induced savings cushion, allowing potential workers to take a wait-and-see approach. Some other explanations may be a location mismatch, skills mismatch, family and childcare constraints, or fear of getting infected with COVID-19, which could encourage workers to leave the workforce entirely. Indeed, labor market conditions since the pandemic began highlight a possible structural shift in the labor force, with 60% of these missing workers people who have left the workforce entirely (see chart 4).
Chart 4
In this report, we try to gauge the factors behind a worker's decision to return to the workforce or remain on the sidelines. Approximately five million people have left the workforce, and it is difficult to predict whether their decision was short term or permanent. Given the shortages, wages are already climbing higher, with the Atlanta Fed's Wage Growth Tracker at 4.2% year over year, its highest since 1995, illustrating the significant impact of the supply constraints. Also significant--the tracker data is less susceptible to compositional or demographic changes, unlike the BLS wage data. It remains to be seen whether potential workers will be lured back into the workforce, stabilizing wages.
Are Extended Unemployment Benefits The Main Culprit?
We believe extended unemployment benefits had little impact on a worker's decision not to return to the workforce. Those arguing that benefits are the culprit say they created an incentive for people to remain unemployed rather than accept a job (often customer-facing jobs). Many have suggested that extended unemployment benefits increased people's reserve wages, limiting worker interest in accepting a job at pre-pandemic wages.
Anecdotally, businesses complained that workers refused to accept their job offers as they stayed home collecting federal unemployment benefits on top of state-specific benefits. The federal government's extended unemployment benefits during the pandemic equaled $600 per week (the amount paid as an extended unemployment booster under the CARES act). This meant those who received the extended benefits were at one point essentially earning an hourly increase of around $15, on top of the average state-wide hourly benefit of around $8 to $9, for a total of around $24 per hour. After the extended unemployment benefit booster was brought down to $300 per week, adding the $320 average state-level unemployment benefit (without boosters), the overall unemployment benefit averaged around $15.50 an hour--a much higher reservation wage than pre-pandemic, particularly for people at the lower end of the wage spectrum. The extended unemployment benefit on an hourly basis was more than double the federal minimum wage of $7.25. It was also higher than the average hourly earnings for many customer-facing jobs hit by the pandemic, such as those in leisure and hospitality and retail. Average hourly wages are about $13 for U.S. retail sales associates or hotel guest receptionists and $16 for waitstaff, according to ZipRecruiter.
If extended benefits played a significant role in a worker's decision to accept a job, the jobs data would likely show a difference in hires between the 26 states that ended the extended benefits before Sept. 6, 2021, and those that did not (see definition and diagnostics in the appendix). For the "early-end" states that ended the benefits in May, one would expect to see their unemployment rates fall significantly relative to the "no-early-end" states that kept the benefits until September. Alternatively, one may also see a pickup in labor force exit rates once the extended benefits expired, either for early retirement or other reasons. But, based on a comparison of the data, the jobs data indicates that there was not a big difference between the two groups after the early end date. The unemployment rate didn't drop significantly in early-end states relative to no-early-end states.
States that reopened sooner saw unemployment drop before benefits stopped
Instead, we found a significant drop in the average unemployment rate for the early-end group well before May. In April, the average unemployment rate for this group was just 4.5%, only 79 basis points above its February 2020 rate. By way of comparison, the rate for the no-early-end states was 6.0% in April, 2.26% higher than its pre-pandemic rate (see chart 5). This suggests that the impact of extended unemployment benefits on a worker's decision to remain unemployed and collect benefits may be overestimated.
The initial comparisons indicated that the states that exited the extended unemployment program had a greater increase in employment than those that didn't exit early. However, it's likely not because the benefits stopped. It's because they saw more economic activity. It appears that the early-end states reopened sooner, so they saw a greater need to create jobs to meet demand. The early-end states also saw a pickup in labor exit rates. This is likely because some people may have already planned to exit the workforce but stayed on as unemployed until those benefits expired.
Chart 5
Regression analysis affirmed it's not about extended benefits
In addition to the correlative evidence, we performed a multi-regression analysis, which controlled for competing factors, including the reopening effect, to provide more robust evidence that the policy to extend unemployment benefits had little impact on the decision to remain unemployed.
To answer whether the federal government's extended benefits did indeed slow down hires--and increase unemployment rates--we analyzed joblessness in states that ended the extended unemployment benefits early compared with those that kept them through the September deadline. This has been done before, but unlike other studies by the Wall Street Journal in August and Indeed Hiring Lab in June, we controlled for state reopening, COVID-19 conditions, labor force participation, home stay, and work travel conditions. The appendix section explains how we performed our regression analysis in more detail.
Our stand-alone regression results (see table 1) found negative significant association between early-end states and their jobless claims relative to no-early-end states, controlling for labor force participation, state reopening, and COVID-19 status. This confirms our earlier analysis, which found that early-end states saw lower unemployment rates than the other group well before they exited the program. We had similar findings for both state reopening proxies where regression results found that the bigger the reopening, the lower the unemployment rate. We found strong direct negative association between joblessness and labor force participation, as given a weak job market and smaller opportunity cost from staying in the job market, people are more likely to leave the market than otherwise.
However, the results in the full sample set do not explain whether the decision to end the extended unemployment benefit by itself is impactful. To answer, we split the sample into two categories: before May 2021 and after April 2021. We present the results in columns 2 and 3 of table 1. All indicators in columns 2 and 3 were significant.
Column 2 shows that early-end states had significantly lower joblessness rates than the no-early-end states, before they made the announcement in May to end the extended unemployment benefits. In column 3, which tests the period relating to the termination decision and actual implementation, we found muted results on the reopening proxy and the state_early variable. We interpret these results to mean that the early termination of extended unemployment benefits in and of itself had little impact on joblessness, after the May announcement. Furthermore, the lower coefficients for both the reopening and the state_early indicators lose their impact on the joblessness months after the early-end decision as the two groups converge on full reopening statues and the end of extended unemployment benefits.
We interpret these results to mean that the termination of extended unemployment benefits, announced in May, enacted in June, for most early-end states, in and of itself had little impact on the unemployment rate.
Table 1
| Unemployment Regression Results | ||||||||
|---|---|---|---|---|---|---|---|---|
| DEP. VAR.: Continued Claims (CUI) | Full sample | Before May 2021 | After April 2021 | |||||
| STATE_EARLY | -1.208*** | -1.427*** | -0.616*** | |||||
| (-0.0897) | (0.1200) | (0.0694) | ||||||
| LABPART | -0.199*** | -0.238*** | -0.0737*** | |||||
| (-0.0107) | (-0.0139) | (-0.006) | ||||||
| WORKPLACE | -11.33*** | -13.14*** | -3.596*** | |||||
| (-0.979) | (-1.230) | (-0.710) | ||||||
| ATHOME | 0.162*** | 0.174*** | 0.130*** | |||||
| (0.0200) | (0.0246) | (0.0137) | ||||||
| LOGCASE | 0.454*** | 0.358*** | 0.365*** | |||||
| (-0.0983) | (-0.107) | (-0.0724) | ||||||
| CONSTANT | 8.194*** | 11.85*** | -0.144 | |||||
| (0.946) | (1.055) | (0.723) | ||||||
| Observations | 3777 | 2797 | 980 | |||||
| Adj R² | 0.761 | 0.723 | 0.47 | |||||
| Year-week FE | YES | YES | YES | |||||
| Source: Bureau of Labor Statistics, Google Trends, Department of Labor, Department of Transportation, Centers for Disease Control and Prevention, S&P Global Economics calculations. | ||||||||
It's The Reopening Effect
"Alabama has an unemployment rate of 3.8%, the lowest in the Southeast, and significantly lower than the national unemployment rate. Our Department of Labor is reporting that there are more available jobs now than prior to the pandemic. Jobs are out there," continued Gov. Kay Ivey. "We have announced the end date of our state of emergency, there are no industry shutdowns, and daycares are operating with no restrictions. Vaccinations are available for all adults." (Gov. Kay Ivey, May 2021. Office of Alabama Governor.)
We believe one big reason why early-end states exited the benefits program earlier was because of the reopening effect, as was the case in Alabama. These states opened their economies sooner than those that remained in the extended unemployment program, so there was more economic activity, more jobs, and a greater need for workers to fill demand. Because there isn't a specific indicator that tracks state reopening over time, to get a sense of state reopening status, we used several indicators (see appendix for more information). The data we tracked in all cases indicated that the early-end states did reopen sooner and at a faster rate than no-early-end states.
During the height of the pandemic, in-room dining plunged by an astonishing 100% through April 2020, relative to the same period in 2019, for both groups (see chart 6). The two groups tracked each other during the recovery through September 2020. However, as the virus resurged in the fall and winter of 2020, seated diners in the no-early-end group dropped by 30%, to 68% below pre-crisis levels, while the early-end group fell by a smaller 10%, to 44% below pre-crisis levels. The no-early-end group recovered a significant amount of its winter drop, but its rate of in-room dining remains below that of early-end states.
Chart 6
A similar dynamic played out in workplace mobility and job postings data (see charts 7 and 8). Both groups saw significant declines in job openings at the height of the crisis. But after June 2020, the early-end group began to see more job postings than the no-early-end group. The job postings gap largely continued through September 2021 at which point it appeared to have closed.
Chart 7
Chart 8
Although we have proven that extended unemployment benefits had little impact on a worker's decision to remain unemployed, the U.S. still has over 10 million job openings to be filled. Thus, we have not answered our original question--where are the workers? We now point to another reason for the missing workers: the large drop in U.S. labor participation since the pandemic began.
Millions Of Worker Departures May Reflect A Structural Change
According to our calculations, over three million workers have left the workforce since February 2020. The exit of people 16 years or older from the workforce during the pandemic drove the labor participation rate down to 60.2%, a 47-year low. As of September, the rate is still at a 44-year low of 61.6%, from a 41-year low of 63.3% in February 2020. Including unemployed workers, this means the U.S. economy lost the productivity of over five million people of working age during the pandemic, with lost labor supply since February 2020 accounting for 60% of that five million (see chart 4). Job openings totaled over 10 million in July, twice the five million lost workers in September. Since July, the number of job openings only climbed higher. Job openings since February 2020 added an additional 3.4 million postings in August.
Interestingly, labor participation by group highlights the larger drop in labor participation for the no-early-end group, which remains 1.74 percentage points below its pre-pandemic level (see chart 9). The early-end group, which had a lower participation rate before the pandemic, also saw a significant drop in labor participation, though two-thirds the size of the no-early-end group. Moreover, the labor participation rate recovered more of its loss in early-end states, now down 1.1 percentage points from the pre-pandemic level.
Chart 9
Chart 10
Comparing job market conditions since the pandemic began by group, no-early-end states lost jobs sooner than the early-end states, largely because they suffered more COVID-19 cases sooner (see chart 14 in appendix). And even when cases started rising faster in early-end states, the unemployment rates did not dramatically increase, likely given their decisions to remain open (see charts 11 and 12). Early in the pandemic, no-early-end states also saw more labor force exits, relative to the early-end states. However, since July, the labor force exit rate for early-end states has doubled (though it is still below no-early-end states). This suggests that the extended benefits may have encouraged some unemployed workers to stay in the workforce until their benefits expired.
Chart 11
Chart 12
The possible explanations for people's decisions to leave the job market vary from childcare or family-care constraints, early retirement, health reasons, and pandemic fears. The stimulus checks and little spending during the pandemic helped households build a substantial nest egg once quarantine ended, giving some people the option to wait for the right opportunity before reentering the job market.
Some of those two million unemployed workers, relative to February 2020, who were collecting benefits may have decided to retire once their benefits expired. And swollen savings, now about $2.5 trillion more than 2019 average household savings, also give people more reason to be selective. But given heightened demand for labor, it would be safe to assume that most unemployed workers will be hired in the near future.
But what about those who left the workforce, which now make up about 60% of the five million missing workers since February 2020? The recent drop in the labor participation could be explained by older workers retiring earlier for various reasons, including fear of the pandemic, childcare constraints from hybrid schooling and closed daycare centers, or increased family care responsibilities. The much larger relative drop in the labor participation rate among women between ages 25 and 54 since the pandemic began seems to make that clear (see chart 13). The location mismatch was also likely a significant factor in a person's decision to stay out of the market. Many people who moved out of cities during the pandemic have yet to return. The skills mismatch, which has been a problem for years, has only become worse during the pandemic. Many of these factors apply to the unemployed workers.
Chart 13
Indeed, the reason workers aren't filling jobs seems to stem more from the decision to drop out of the workforce (and not collect unemployment benefits), rather than staying in the workforce (and collecting unemployment benefits if unemployed). Overall, the labor force lost eight million workers at the start of the pandemic. As of September, five million workers are still on the sidelines or have checked out of the workforce completely.
Appendix: Regression Analysis Results
In our regression analysis, our dependent variable is weekly insured unemployment (continuous) claims by state. With no specific indicator that tracks state reopening over time, we considered several proxies, with weekly data at the state level. We first used weekly dine-in statistics from Open Table in major cities from 40 states (including Washington, D.C.) as a proxy for state reopening, though this limited the universe. We then used mobility data from the DOT, covering all cities and states. We finally used Google trends workplace mobility data tracking distance from work, as a measure for openness, but also for location mismatch. We included monthly labor participation by state to measure the size of the workforce relative to the working population. We added monthly labor participation rates by age and gender at the national level to gauge the relative impact of decisions to leave the workforce for childcare or take an early retirement (state-level data was not available). We added an indicator on COVID-19 status by state to measure quarantine conditions and to capture a potential worker's infection risk concern. Finally, to measure whether extended unemployment benefits influenced a worker's decision to accept a job offer, we included a dummy variable identifying states that ended extended benefits earlier than the Sept. 7 cut-off. The indicators used in this analysis are explained in more detail below (see table A5).
We ran univariate tests on the indicators and found that for most indicators the difference between the two groups was significant (see table A1). The difference between the two groups was only insignificant for labor participation data by gender and age. This is not surprising given the limitations on the data because only national data was available. In line with our earlier analysis, the insured unemployment rate (continuing claims) average for the early-end group was 4.35%, over two percentage points lower than the mean insured unemployment rate of 6.58% for the no-early-end group. The summary statistics on data we tracked (OpenTable, Workplace, and Athome) in both cases also indicated that the early-end states did reopen at a faster rate than no-early-end states. The COVID-19 case load was also significantly higher for early-end states relative to no-early-end states. Summary data for labor participation indicators at the state level also indicated that no-early-end states saw more workers exit the labor market than early-end states. Results on labor participation by gender and age were insignificant, because only monthly data at the national level was available. We ran diagnostic tests on these variables, finding little presence of multicollinearity in our data structure for all indicators (see table A2). We find significant multicollinearity for the reopening indicators and gender labor force regressors. We tested these regressors separately in columns 4 and 5 in table A4.
With this level of comfort about the analysis, we ran regression models to test the reason behind the labor mismatch (the outward shift in the Beveridge curve; see chart 3) that we observe in the data over the sample period: March 7, 2020, the month when extended unemployment benefits started, through Sept. 11, 2021, the week after the expiration of the benefits (see table A1 for the model parameters). The dependent variable is continued unemployment claim i (Cui) in week t. The independent variables are STATE_EARLY, a dummy that takes the value of 1 if the state ended its federal unemployment benefits before Sept. 4, 2021, and zero otherwise; state reopening, labor participation, and COVID-19 status.
To proxy states reopening, we ran several indicators (OpenTable, Workplace, and Athome). We observed a moderately high correlation (0.616) between OpenTable and Workplace, which shows up in the variance inflation factor of 2.76 for OpenTable. This correlation impacts the significance of OpenTable in the full sample analysis. Additionally, OpenTable data is not available for 20% of states. This makes it difficult to extrapolate relevance of findings to these 10 states. As a result, we exclude OpenTable variable in the multi-regression analyses. All reopening indicators are significant, indicating that early-end states opened earlier than no-early-end states. The most powerful regression outcome came using the workplace indicator, as seen in the table below. We added the labor participation rate by state [LABPART]. We included the natural logarithms of COVID-19 cases by state (LOGCASE) to measure outbreaks by region with its knockdown impact on quarantine, economic activity, and job market conditions.
Regression results
Stand-alone regression results in table 1 above find significant results with correct signs for all regressors in the full sample, with the exception of gender and age indicators. Finally, the model has good fit of the data: running from 47% to 76%. The fit in the "after April 2021" is influenced by the small sample size and period.
We find negative significant association between early-end states and their jobless claims relative to states that held out until the benefit expiration period, controlling for labor force participation, state reopening, and COVID-19 status. This confirms our earlier analysis, which found that early-end states saw lower unemployment rates than the other group well below they exited the program. We had a similar finding for state reopening proxies (WORKPLACE and ATHOME) where regression results found that the bigger the reopening, the lower the unemployment rate. We found strong direct negative association between joblessness and labor force participation, as given a weak job market and smaller opportunity cost from staying in the job market, people are more likely to leave the market than otherwise.
However, the results in the full sample set do not explain whether the decision to end the extended unemployment benefit by itself is impactful. To answer, we split the sample into two categories: before May 2021 and after April 2021. We present the results in columns 2 and 3 of table 1. All indicators in columns 2 and 3 were significant.
Column 2 shows that early-end states had lower joblessness than the no-early-end states before they made the decision to end benefits early--in line with the qualitative analysis we presented earlier. We also found consistent results on reopening proxies, labor participation, and COVID-19 status. All readings were significant to probability greater than 1%. In fact, the absolute coefficients on all variables are larger in this column. In column 3, where we test the period relating to the termination decision and actual implementation, we had more muted results on the reopening proxy. Additionally, the coefficient of the state_early variable, at -0.616***, is less than half its size in the "before May 2021" run. We interpret these results to mean that the early termination of extended unemployment benefits in and of itself had little impact on joblessness, after the May announcement. Furthermore, the lower coefficients for both the reopening and the state_early indicators lose their impact on the joblessness months after the early-end decision because the two groups converge on full reopening statues and the end of extended unemployment benefits.
The COVID-19 caseload remained significant and positive across all model runs (see columns 1 through 3). This is not surprising given that the higher the caseload, the more likely states would see business closures and some form of self-quarantine, which means less demand for services and less need to retain staff. We find negative significant readings for LABPART, the labor participation rate by state, which indicates a higher labor participation rate is associated with a lower unemployment rate. That suggests that the lower the unemployment rate, the higher the number of people entering the workforce, perhaps seeing more opportunities in the job market. Alternatively, a higher unemployment rate would suggest lower employment opportunities, leading to people dropping out of the workforce or retiring early.
In the interaction analysis (see table A4), the stand-alone results (except for state_early and logcase) are consistent with those in table A1. Our interactions between state_early and other regressors are largely significant. This means the state_early dummy has a residual effect on unemployment claims via the remaining regressors. Given the coefficients of determination remain unchanged from those in table 1, we focus on the stand-alone models.
Chart 14
Chart 15
Table A1
| Descriptive Statistics | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Full Sample | No early end | Early end | T-Test | |||||||||||||
| Variables | Observations | Mean | Standard deviation | Mean | Mean | t-stat | p-value | |||||||||
| UNEMP_RATE (%) | 3836 | 7.028 | 3.493 | 7.96 | 6.131 | 16.8 | 0 | |||||||||
| LOGNOTWORK | 3836 | 13.958 | 1.068 | 14.085 | 13.837 | 7.25 | 0 | |||||||||
| LOGCIVPOP | 3836 | 14.942 | 1.034 | 15.072 | 14.818 | 7.65 | 0 | |||||||||
| LOGEMP | 3836 | 14.395 | 1.013 | 14.516 | 14.278 | 7.3 | 0 | |||||||||
| LABOR PART | 3836 | 62.403 | 3.928 | 62.525 | 62.287 | 1.9 | 0.06 | |||||||||
| OVERALL PRIME | 3836 | 81.224 | 0.566 | 81.222 | 81.226 | -0.2 | 0.828 | |||||||||
| WOMEN PRIME | 3836 | 74.857 | 0.612 | 74.856 | 74.859 | -0.15 | 0.873 | |||||||||
| MEN PRIME | 3836 | 87.74 | 0.564 | 87.737 | 87.743 | -0.3 | 0.774 | |||||||||
| WOMEN OLD | 3836 | 70.532 | 0.434 | 70.532 | 70.532 | -0.1 | 0.933 | |||||||||
| MEN OLD | 3836 | 58.999 | 0.594 | 58.998 | 59.001 | -0.15 | 0.888 | |||||||||
| CONTINUED CLAIM (%) | 3836 | 5.445 | 4.761 | 6.577 | 4.356 | 14.85 | 0 | |||||||||
| WORKPLACE | 3836 | -0.268 | 0.095 | -0.3 | -0.237 | -21.75 | 0 | |||||||||
| STAY AT HOME | 3836 | 23.742 | 3.84 | 25.054 | 22.479 | 22.05 | 0 | |||||||||
| CASE_RATE | 3777 | 4910.123 | 4475.884 | 4362.773 | 5439.515 | -7.45 | 0 | |||||||||
| *** p<0.01, ** p<0.05, * p<0.1. Source: Bureau of Labor Statistics, Google Trends, Department of Labor, Department of Transportation, Centers for Disease Control and Prevention, S&P Global Economics calculations. | ||||||||||||||||
Table A2
| Test For Multicollinearity | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Variables | (1) | (2) | (3) | (4) | (5) | VIF | ||||||||
| STATE_EARLY | 1 | 1.2 | ||||||||||||
| (2) workplace | 0.332* | 1 | 1.43 | |||||||||||
| (3) athome | -0.335* | -0.451* | 1 | 1.45 | ||||||||||
| (4) labpart | -0.03 | 0.011 | 0.200* | 1 | 1.06 | |||||||||
| (5) case_rate | 0.120* | 0.364* | -0.329* | 0.031 | 1 | 1.21 | ||||||||
| *** p<0.01, ** p<0.05, * p<0.1. Source: Bureau of Labor Statistics, Google Trends, Department of Labor, Department of Transportation, Centers for Disease Control and Prevention, S&P Global Economics calculations. | ||||||||||||||
Table A3
| Analyses With Gender Labor Participation | |||
|---|---|---|---|
| DEP. VAR: WORK MOBILITY | Full sample | Before May 2021 | After April 2021 |
| Early_State | 0.784*** | 1.333** | 0.278 |
| -0.288 | -0.535 | -0.286 | |
| lnotwork | -1.278*** | -1.223*** | -1.311*** |
| -0.135 | -0.23 | -0.143 | |
| logcase | 2.870*** | 3.150*** | 2.467*** |
| -0.449 | -0.656 | -0.641 | |
| Constant | -19.72*** | -25.42*** | -13.50** |
| -3.961 | -5.284 | -6.159 | |
| Observations | 1350 | 650 | 700 |
| Adj R | 0.522 | 0.509 | 0.397 |
| Week FE | YES | YES | YES |
Table A4
| Interaction Analyses | |||
|---|---|---|---|
| DEP. VAR.: Continued claims (CUI) | Full sample | Before May 2021 | After Apriil 2021 |
| WORKPLACE | -19.71*** | -21.80*** | -9.954*** |
| -0.909 | -1.136 | -0.645 | |
| STATE_EARLY | -4.840*** | -5.365*** | 20.16*** |
| -1.274 | -1.627 | -1.772 | |
| WORKPLACE × STATE_EARLY | 9.956*** | 9.833*** | 7.264*** |
| -1.013 | -1.142 | -0.971 | |
| LABPART | -0.206*** | -0.239*** | -0.0740*** |
| -0.0172 | -0.0217 | -0.0126 | |
| LABPART × STATE_EARLY | 0.0698*** | 0.0814*** | 0.0152 |
| -0.0201 | -0.0256 | -0.0139 | |
| LOGCASE | 0.0651 | -0.0118 | 0.687*** |
| -0.104 | -0.116 | -0.0854 | |
| LOGCASE × STATE_EARLY | 0.248*** | 0.218*** | -2.167*** |
| -0.0344 | -0.0411 | -0.187 | |
| Constant | 12.82*** | 15.91*** | -1.737* |
| -1.271 | -1.503 | -0.891 | |
| Observations | 3929 | 2949 | 980 |
| R-squared | 0.766 | 0.732 | 0.455 |
| Year-month FE | YES | YES | YES |
Table A5
| Variable Definition | |||||
|---|---|---|---|---|---|
| Variable | Aggregate level; frequency | Definition | Formula | Source | Source link |
| CUI | State; weekly | Traditional continuing unemployment claims | 100*continued claims/covered employees | Department of Labor | https://oui.doleta.gov/unemploy/claims.asp |
| WORKPLACE | State; weekly | Time spent at workplaces | As provided | Opportunity Insights Economic Tracker | |
| ATHOME | State; weekly | Percent of people staying at home in each state | 100*persons at home/ (persons at home + persons not at home) | Bureau of Transportation Statistics | https://data.bts.gov/Research-and-Statistics/Trips-by-Distance/w96p-f2qv |
| STATE_EARLY | State; weekly | Dummy that takes the value of 1 if a state opted out of the pandemic-related federal unemployment benefits and 0 otherwise | See definition | USA Today | https://www.usatoday.com/story/money/2021/07/01/unemployment- |
| LABPART | State; monthly | Monthly labor participation rate for each state | As provided | Bureau of Labor Statistics | https://www.bls.gov/lau/rdscnp16.htm |
| LOGNOTLABOR | State; monthly | Natural logarithm of monthly number of persons not in the labor force per state | See definition | Bureau of Labor Statistics | https://www.bls.gov/lau/rdscnp16.htm |
| LOGCASE | State; weekly | Natural logarithm of weekly case rate per state | See definition | Centers for Disease Control and Prevention; Opportunity Insights Economic Tracker | |
| PRIME_LABPART | National; monthly | National monthly average of labor participation for 25-54 age group | As provided | Federal Reserve Economic Data | |
| MPRIME_ACTRATE | National; monthly | Activity rate for men in 25-54 age group | As provided | Organization for Economic Cooperation and Development | |
| WPRIME_ACTRATE | National; monthly | Activity rate for women in 25-54 age group | As provided | Organization for Economic Cooperation and Development | |
| MOLD_ACTRATE | National; monthly | Activity rate for men in 55+ age group | As provided | Organization for Economic Cooperation and Development | |
| WOLD_ACTRATE | National; monthly | Activity rate for women in 55+ age group | As provided | Organization for Economic Cooperation and Development | |
Table A6
| State Groups | |
|---|---|
| Early end | No early end |
| Alabama | California |
| Alaska | Colorado |
| Arizona | Connecticut |
| Arkansas | Delaware |
| Florida | District of Columbia |
| Georgia | Hawaii |
| Idaho | Illinois |
| Indiana | Kansas |
| Iowa | Kentucky |
| Louisiana | Maine |
| Maryland | Massachusetts |
| Mississippi | Michigan |
| Missouri | Minnesota |
| Montana | Nevada |
| Nebraska | New Jersey |
| New Hampshire | New Mexico |
| North Dakota | New York |
| Ohio | North Carolina |
| Oklahoma | Oregon |
| South Carolina | Pennsylvania |
| South Dakota | Rhode Island |
| Tennessee | Vermont |
| Texas | Virginia |
| Utah | Washington |
| West Virginia | Wisconsin |
| Wyoming | |
This report does not constitute a rating action.
| U.S. Chief Economist: | Beth Ann Bovino, New York + 1 (212) 438 1652; bethann.bovino@spglobal.com |
| Contributor: | Joseph Arthur; joseph.arthur@spglobal.com |
| Research Contributor: | Arun Sudi, CRISIL Global Analytical Center, an S&P affiliate, Mumbai |
No content (including ratings, credit-related analyses and data, valuations, model, software or other application or output therefrom) or any part thereof (Content) may be modified, reverse engineered, reproduced or distributed in any form by any means, or stored in a database or retrieval system, without the prior written permission of Standard & Poor’s Financial Services LLC or its affiliates (collectively, S&P). The Content shall not be used for any unlawful or unauthorized purposes. S&P and any third-party providers, as well as their directors, officers, shareholders, employees or agents (collectively S&P Parties) do not guarantee the accuracy, completeness, timeliness or availability of the Content. S&P Parties are not responsible for any errors or omissions (negligent or otherwise), regardless of the cause, for the results obtained from the use of the Content, or for the security or maintenance of any data input by the user. The Content is provided on an “as is” basis. S&P PARTIES DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, ANY WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE OR USE, FREEDOM FROM BUGS, SOFTWARE ERRORS OR DEFECTS, THAT THE CONTENT’S FUNCTIONING WILL BE UNINTERRUPTED OR THAT THE CONTENT WILL OPERATE WITH ANY SOFTWARE OR HARDWARE CONFIGURATION. In no event shall S&P Parties be liable to any party for any direct, indirect, incidental, exemplary, compensatory, punitive, special or consequential damages, costs, expenses, legal fees, or losses (including, without limitation, lost income or lost profits and opportunity costs or losses caused by negligence) in connection with any use of the Content even if advised of the possibility of such damages.
Credit-related and other analyses, including ratings, and statements in the Content are statements of opinion as of the date they are expressed and not statements of fact. S&P’s opinions, analyses and rating acknowledgment decisions (described below) are not recommendations to purchase, hold, or sell any securities or to make any investment decisions, and do not address the suitability of any security. S&P assumes no obligation to update the Content following publication in any form or format. The Content should not be relied on and is not a substitute for the skill, judgment and experience of the user, its management, employees, advisors and/or clients when making investment and other business decisions. S&P does not act as a fiduciary or an investment advisor except where registered as such. While S&P has obtained information from sources it believes to be reliable, S&P does not perform an audit and undertakes no duty of due diligence or independent verification of any information it receives. Rating-related publications may be published for a variety of reasons that are not necessarily dependent on action by rating committees, including, but not limited to, the publication of a periodic update on a credit rating and related analyses.
To the extent that regulatory authorities allow a rating agency to acknowledge in one jurisdiction a rating issued in another jurisdiction for certain regulatory purposes, S&P reserves the right to assign, withdraw or suspend such acknowledgment at any time and in its sole discretion. S&P Parties disclaim any duty whatsoever arising out of the assignment, withdrawal or suspension of an acknowledgment as well as any liability for any damage alleged to have been suffered on account thereof.
S&P keeps certain activities of its business units separate from each other in order to preserve the independence and objectivity of their respective activities. As a result, certain business units of S&P may have information that is not available to other S&P business units. S&P has established policies and procedures to maintain the confidentiality of certain non-public information received in connection with each analytical process.
S&P may receive compensation for its ratings and certain analyses, normally from issuers or underwriters of securities or from obligors. S&P reserves the right to disseminate its opinions and analyses. S&P's public ratings and analyses are made available on its Web sites, www.standardandpoors.com (free of charge), and www.ratingsdirect.com and www.globalcreditportal.com (subscription), and may be distributed through other means, including via S&P publications and third-party redistributors. Additional information about our ratings fees is available at www.standardandpoors.com/usratingsfees.
Any Passwords/user IDs issued by S&P to users are single user-dedicated and may ONLY be used by the individual to whom they have been assigned. No sharing of passwords/user IDs and no simultaneous access via the same password/user ID is permitted. To reprint, translate, or use the data or information other than as provided herein, contact S&P Global Ratings, Client Services, 55 Water Street, New York, NY 10041; (1) 212-438-7280 or by e-mail to: research_request@spglobal.com.