The Effect of the Covid-19 Pandemic on Inequality

3.1 Incomes, consumer spending, and poverty

Different data sources provide information on different aspects of the distribution of household income, spending, and consumption throughout the pandemic. BSZ produce comprehensive estimates of the distribution of U.S. incomes over these years. Leaving out transfers, the incomes of the highest and lowest income groups declined most sharply in 2020. These same groups recovered most strongly in 2021. The top 10 percent recovered their pre-pandemic income within 12 months, and lower-income groups recovered within 21 months. BSZ find that labor earnings for the bottom half of the distribution increased more rapidly than for the upper quartiles between January 2019 and September 2022.

Government transfers had a substantial equalizing effect during the pandemic. BSZ report that “after accounting for taxes and … transfers, average disposable income for the bottom 50% was nearly 20% higher in 2021 than 2019.” Kollar (2023) examines pre-tax and post-tax U.S. household income inequality using data from the Current Population Survey. In this data, the Gini index for pre-tax income increased steadily since 2007, and the 2020 and 2021 increases were similar to this trend. Post-tax income in the pandemic period includes transfer payments made through the tax system, such as the Economic Impact Payments and the Expanded Child Tax Credit. These transfer programs generally expired in 2022, so that, although pre-tax inequality declined modestly in 2022, post-tax inequality increased considerably in that year.

Government responses similarly reduced inequality effects in Europe. Clark et al. (2021) showed that inequality initially increased in France, Germany, Italy, and Spain, then fell due to government intervention. Menta (2022) examined related data on poverty. These studies find that, in these European countries, inequality and poverty peaked around May 2020.

Consumer spending fell more for higher income groups at the start of the pandemic. However, the savings balances of households lower in the income distribution did not decrease on average. Using credit- and debit-card data, Chetty et al. (2024) find that households in top-quartile income areas spent 34 percent less in mid-April 2020 than in March, whereas households in bottom-quartile areas spent only 23 percent less. Using Chase checking account and credit- and debit-card data, Cox et al. (2020) find that in March 2020 the pandemic brought about a large spending decline throughout the income distribution. Later in the pandemic, spending by low-income households recovered more quickly than high-income expenditures.

Using U.S. Consumer Expenditures Survey data, Meyer et al. (2022) find similar results for the entirety of 2020; consumption declined only for higher income households, partly because of reductions in travel and entertainment spending. These authors attribute income gains for lower-percentile households to government policy choices which averted a decline in their consumption expenditures. These payments were predominantly saved, or used to pay off debt, rather than spent (Armantier et al., 2021). Lower-income households spent somewhat more and paid down debt more in the short run, relative to higher-income households, according to a variety of data sources and researchers (Coibion et al., 2020; Parker et al., 2021). Garner et al. (2024) show that high-income consumers spent more on expensive travel and restaurants, so the decline in these industries reduced consumption inequality in 2020, then it returned to its previous level. Tauber and Van Zandweghe (2021) show that the 2020–2021 increase in durable goods expenditures reflected both a shift in consumer demand away from services and increased disposable income due to short term government transfers.

The same basic pattern holds true for countries outside the United States. Hacıoğlu-Hoke et al. (2021), using transaction data from the UK, show that in the early stages of the pandemic higher-income people reduced their consumption expenditures the most, and labor earnings fell most for low-income people. Government transfers played an important role in moderating the decline in earnings. Similarly, Bounie et al. (2023) show that in France, in the first wave of the pandemic, spending fell more for high-expenditure households. Low-expenditure households showed an increase in debt.

Because the data for many studies of consumer spending come from banks or credit-card providers, they exclude the poorest of the poor and therefore miss some material hardship such as food insecurity. Meyer et al. (2022) review the literature and find conflicting results. Bitler et al. (2020) compared estimates from the pre-Covid National Health Interview Survey with the Covid Impact Survey and found food insecurity was almost three times greater than pre-COVID rates in June 2020. Karpman and Zuckerman (2021) found from an Urban Institute survey that between December 2019 and December 2020, the share of U.S. adults reporting household food insecurity in the past 12 months declined from 23.9 percent to 20.5 percent. The two data sources are not directly comparable in content or timing, but the combination suggests that the government response to COVID reduced hunger by December 2020.

3.2 Effects of the pandemic period on specific groups

3.3 Difference between wealthier and developing countries

Wealthy and poor countries experienced the crisis quite differently (Miguel & Mobarak, 2022). Deaton (2021) examined aggregate data for many countries through December 2020. Richer countries suffered more pandemic deaths and lost more GDP per capita. Inequality between countries had narrowed since 2000, especially when countries are weighted by population. During the pandemic, population-weighted international income inequality increased largely because incomes in India declined.

Yonzan et al. (2021) estimate the amount by which 2020 and 2021 incomes were lower than pre-pandemic projections for each percentile of the global population. Their estimates assume constant within-country income distributions, but allow countries' incomes to grow at different rates, with consequences for the global income distribution. In both years, losses were largest for people between the 10th and 30th percentiles of income. Income recovery has so far been greater in the better-off countries, with the top decile enjoying the strongest recovery.

Labor force participation rates dropped more sharply in developing countries than in advanced countries (IMF, 2021, Fig. 1.7 and IMF, 2023b, Figure 1.1). Bluedorn et al. (2021) found that unemployment increased more in developing countries. In both advanced and developing countries, increases in unemployment were generally similar for men and women.

Workers in developing countries often did not have access to systems of social support that would enable them to stay home and avoid the pandemic. Goolsbee and Syverson (2021) show that in the United States, many people changed their behavior to avoid COVID, apart from any government mandates. Miguel and Mobarak (2022) point out that many people in developing countries, with less of a social safety net, did not have this option. The International Labor Organization (ILO) (2020) estimated that 2 billion people worked in the informal economy, largely in developing countries, and that 1.6 billion of these were significantly affected by COVID, leading to an estimated 60 percent decline in their income.

Developing countries have a smaller proportion of elderly residents, who would be more susceptible to dying from COVID. Developing countries also have a greater number of workers in the informal sector, less effective medical systems, reduced access to vaccines, and lesser access to overall resources to meet the many challenges of the pandemic. On balance, these challenges tended to lower incomes in poor countries in the COVID period, increasing inequality between countries.

4.1 Telework

The Covid pandemic dramatically increased the amount of telework, in many instances permanently. An estimated 34.5 percent of U.S. establishments increased telework for their employees in response to the pandemic (BLS, 2021). Among establishments that increased telework during the pandemic, 60.2 percent expect to continue when the pandemic is over. Worker expectations of post-pandemic telework increased between mid-2021 and early 2022 in 10 of 12 countries surveyed (Aksoy et al., 2022). Telework may directly benefit affected employers and employees, for example by increasing productivity and/or reducing costs of office space and commuting. In addition, telework can change where workers live, what they consume, and where they buy, and such changes may have distributional consequences.

A survey conducted by Barrero et al. (2021) reports that U.S. employers plan to conduct 22 percent of future work at home, about half of what the surveyed workers prefer. The authors examine five factors that will help sustain telework: (1) the stigma of working at home has been reduced; (2) inertia against telework was overcome; (3) workers and firms both invested heavily in fixed costs that facilitate telework; (4) some residual fear of closeness to others may remain; and (5) the pandemic spurred supportive technological advances.8 These authors believe that increased telework carries benefits that will accrue disproportionately to better-off workers. The authors additionally foresee large time savings from reduced commuting and substantial increases in productivity.

Davis et al. (2021) similarly conclude that sustained increases in telework will be large and most beneficial to higher-paid U.S. workers. They argue that the pandemic accelerated adoption of technologies that permanently raise many high-skill workers' productivity when working from home. They develop a model in which high-skill workers allocate their time between working at home or in the office. The model predicts that in the long run, high-skill workers will work from home much more than prior to the pandemic, and that this will widen inequality. The findings are consistent with other literature that finds that technical advances in production disproportionately benefit high-skilled workers. The pandemic data bear this out. In the pandemic period, work-from-home rates were highly correlated to education level. In the January 2021 Census Household Pulse Survey, graduate degree holders were 71 percent more likely to report that someone in their household was working from home than those with high school education or less were (Dey et al., 2021; Rothwell & Smith, 2021).

Telework increased more for higher-paid workers. Among workers in the highest quartile of weekly pay, the percentage of workdays worked at home increased from 34.4 percent in 2019 to 59.7 percent in 2021, while the corresponding increase was only 9.5 to 15 percent for workers in the lowest pay quartile (Bureau of Labor Statistics 2020, 2022).

Bonacini et al. (2021) reach similar conclusions for Italy: a long-term increase in telework favors male, older, more highly educated, and better paid Italian workers. Overall average pay increases, and inequality also widens. Bachelet et al. (2021) examine how increased telework affects German workers, concentrating on time and financial savings from reduced commuting, rather than on pay. Time savings are greater than financial savings. Some benefits accrue to workers at all income levels, but higher income workers gain more, so inequality increases.

Telework can be a useful accommodation for workers with disabilities. According to Schur et al. (2020), from 2009 to 2018, 5.7 percent of disabled workers in the U.S. worked primarily from home, compared with 4.6 percent of those without disabilities. The share was greater among disabled workers with mobility impairment (6.7 percent), those with “difficulty outside home” (7.2 percent), and those with more education. The authors found, however, that disabled workers were likely to work in occupations where telework was not common. Ne'eman and Maestas (2023) report that during and after the pandemic, as telework became more common, employment of people with disabilities increased notably.

The geographic distribution of employment changed during the pandemic because of telework. The effect of these changes on income distribution is unclear. Li and Su (2023) show that net migration shifts towards suburbs and smaller cities, with a disproportionate contribution from high-income households. Housing prices increased in areas with a large proportion of people working from home (Kmetz et al., 2022).

Increased telework among higher-paid workers shifts consumption of housing and other goods away from city centers and toward outer suburbs and smaller cities and towns. In the short run, this shift reduced incomes of workers who provided services in city centers. The long-run effects on income inequality are not clear.9 Telework is less common for the occupations and the telecommunications systems available in developing countries.

4.2 School interruption

The COVID-19 pandemic shut down schools across the globe, throwing educators and students into an unfamiliar world of virtual learning. It disrupted education for an estimated 1.6 billion students at its peak (World Bank, 2021). Such a widespread disruption led many economists to predict significant and long-lasting scarring effects, including learning loss (Blaskó et al., 2021), lower lifetime earnings (Azevedo et al., 2020), and increased inequality (Agostinelli et al., 2020).

A meta-analysis of 42 studies from 15 countries examined 291 estimates of learning loss (Betthäuser et al., 2023). They estimate that students lost 35 percent of a year's learning, with greater losses in math than in reading. Losses were more severe for students from families of lower socioeconomic status.10 This study also suggests that educational losses spiked early in the pandemic and then stabilized. Learning losses increased as more school was missed (Kane and Reardon, 2023; Schady et al., 2023). Jakubowski et al. (2024) showed that the duration of school closures had an important effect on mathematics scores, especially for lower-achieving students.

Studies agree that students suffered substantial learning losses during the pandemic. The Northwest Evaluation Association Center (NWEA) examined test scores for millions of students in the United States. Using these data, Kuhfeld et al. (2022) reported severe learning losses; that study also emphasized that high quality tutoring and summer instruction can make up for learning losses. A subsequent analysis (Lewis & Kuhfeld, 2022) mentioned that many students had begun to rebound and catch up on their lost learning. Unfortunately, for a while further work found little evidence that students were actually recovering their lost learning (Lewis & Kuhfeld, 2023). NWEA assessments are based on elementary school students. The influential 2023 U.S. National Assessment of Educational Progress is based on 7th and 8th graders. The important National Assessment data also did not find any sign of the rebound that initially appeared in the NWEA data.

However, recent evidence from state test data is beginning to establish solid evidence that U.S. students are recovering important portions of their learning losses. Halloran et al. (2023) report that the proportion of students who were regarded as proficient declined during the pandemic. They show that, by 2022, students had regained 37 percent of their losses of proficiency in math and 20 percent of their losses in English. Fahle et al. (2024) looked at test scores and showed that students recovered 32 percent of their losses in math and 26 percent in reading from spring 2022 to spring 2023. These studies provide important evidence of test score recovery.11

Across studies, learning losses were worst for low performing and minority students. The NWEA studies found that the largest learning gaps were found among minority students (Hispanic, Black, American Indians, and Alaskan Natives). Minority students seem unlikely to recover fully. Fahle et al. (2024) report, however, that some states or local areas were able to reduce learning losses substantially.

The pandemic, and the associated stress and family pressures, similarly hampered the development of pre-school children, who suffered significant declines in verbal, non-verbal, and overall cognitive performance compared to their pre-pandemic counterparts (Deoni et al., 2022).

Catch-up programs to help students recover from learning losses, such as tutoring or summer instruction, typically require substantial expenditures. Lower-income communities tended to have greater learning losses in the short run, and less resources for recovery. If many students do not make up their learning losses, then the pandemic will substantially reduce their lifetime incomes, increasing inequality, and a less-skilled workforce will lower future GDP (Hanushek, 2023; Jakubowski et al., 2024).12

4.3 Scarring of individual careers

In recessions, output growth usually slows down for many years, largely because many workers drop out of the labor force for long periods, and job displacement reduces both current and future earnings (Arulampalam et al., 2001; Martin et al., 2015). The effect on future earnings has been called scarring. After the covid recession, however, employment and labor force participation recovered relatively quickly in the industrialized countries.13 Incomes of most groups have recovered and the lowest 10 percent of workers obtained considerable real income increases.

It appears that relatively little career scarring occurred from the COVID recession, partly because it was brief. Schmieder et al. (2023) provide additional insight. Most such scarring occurs when workers are forced out of high-income jobs and must accept lower-paid jobs in other establishments. Many workers displaced during COVID were employed in low-wage jobs in hospitality and leisure or in retail trade. These jobs do not have a large positive wage premium that can be lost; workers displaced from these jobs are generally able to find similar employment elsewhere. Such wage losses increase with the duration of unemployment, and the COVID recession was relatively brief, so the process of scarring did not have much time to develop.14

The fact that labor force participation rates have recovered suggests that not much scarring occurred in the United States. This may be especially true for those, mostly women, who were forced to stay home to take care of children or others who could not go to school or to a day care center as usual. Many women could go back to work as soon as childcare centers or schools reopened. More generally, many workers retained a long-term connection with their employers.

Growth in many advanced countries has quickly recovered from the COVID crisis, suggesting not many workers were forced to drop out of the labor force permanently. These countries have on average returned to their earlier growth rates. The recovery has generally not gone as well for developing and lower income countries, where growth is still substantially below the rates the IMF projected prior to the crisis, raising the prospects for scarring in those countries (IMF, 2023b, Figure 1.1).

4.4 Government debt

Due to recession and increased expenditures, debt increased substantially in the pandemic in many countries.15 Developing countries were not generally as able to borrow through normal channels as rich countries were. Multilateral agencies, such as the IMF, the World Bank, and regional development banks, had to step in to help developing countries during the COVID pandemic. These loans helped to offset a collapse of government revenues and the withdrawal of private capital (Bulow et al., 2020, Chart 1); however, as a result, more countries are now near debt distress. The indebtedness of developing countries had grown prior to the pandemic (Estavao & Essl, 2020), but the increase in food and fuel prices associated with the Russian invasion of Ukraine and the worldwide increase in interest rates intended to slow inflation added to the burden of government debt in developing countries. These challenges will tend to increase income inequality across countries.