The question of inequality effects of the Covid-19 pandemic has not been widely addressed due to a general lack of information about recent individual or household incomes. Some early studies have examined simulated datasets or collected new household surveys (O’Donoghue et al. 2020, Almeida et al. 2021, Clark et al. 2021). Results from these studies suggest that inequality decreased after the pandemic’s outbreak, primarily due to massive government transfers without which inequality would probably have increased following job losses among low-paid workers. However, studies of history find that inequality tends to increase during pandemics (Furceri et al. 2020, Galletta and Giommoni 2020)
In a new study (Angelov and Waldenström 2021), we estimate the inequality impact of the Covid-19 pandemic using population-wide administrative earnings registers in Sweden collected at a monthly level both before and during the pandemic for all working individuals. By comparing the dispersion of income during the same months in 2019, 2020, and early 2021, we attain measures of the impact of the pandemic on the average and inequality of earnings. We also access register data on individual-level take-up of government Covid-19 support, which allows us to analyse their distributional impact. Finally, we analyse annual pre- and post-tax market incomes for 2018, 2019 and 2020. The annual incomes are less precise with respect to the impact of the pandemic, which broke out a few months into 2020, but offer a broader view of income inequality by including incomes from all sources, some transfers, and all taxes paid.
The distribution of monthly earnings increased during the pandemic
Figure 1 shows the evolution of average pre-tax earnings in 2019-2021 among working individuals 18-64 years old. The wage earners are divided into six quantile groups based on the size of earnings, separated by percentile thresholds (P): the three lowest income quartiles P0-25 (consisting mainly of younger, part-time workers), P25-50 and P50-75, and the top earnings quartile divided into three subgroups, P75-90, the lowest nine tenths P90-99 and the top hundredth P99-100.
Several interesting patterns emerge. First, comparing the earnings levels suggest that the pandemic depressed average earnings by around 5%. The fall was the largest in April-May and in the lowest earnings quartile. Second, the within-year variation of earnings differs quite much across the distribution. Low-income earners experienced higher overall variance and their earnings peak in the summer’s holiday season. Middle-income earners enjoyed more stable incomes, and received pay rises in June (holiday pay) and December (extra-wage payments). High-income earners experienced spikes in earnings in December (mainly small business owners’ wages, which affect the amount of next year’s low-tax dividends) and March (chief executives’ variable remuneration).
Figure 1 Wage income in different parts of the income distribution, 2019-2021
Figure 2 presents the earnings inequality development during the period, measured as the commonly used Gini coefficient (which spans between 0, no inequality, and 1, maximum inequality). The Gini increased roughly one point, about 2.5%, after the outbreak of the pandemic. This increase is not particularly large; during the 1990s recession in Sweden, the labour earnings Gini increased by 15% between 1990 and 1994.
Figure 2 Earnings inequality before taxes, individuals 18-64 years, 2019-2021
A more precise picture of how the pandemic affected earnings across the distribution is offered in Figure 3. This figure shows the marginal effects of the pandemic on earnings conditional on being employed during the whole observation window, estimated separately for different earnings quantile thresholds using so-called RIF regressions. To identify the effect, we exploit the variation from both yearly trends and within-year changes. The results suggest that the earnings impact was most negative for low-income earners. At the absolute bottom of the distribution, average earnings decreased by 6%. In the middle, they fell by 2.4% and in the top hundredth they decreased by 1% (with a confidence interval between 0% and 2%). These estimates confirm the regressive nature of the pandemic’s economic impact on the Swedish labour market.
Figure 3 Earnings impact of the Covid-19 pandemic across the earnings distribution (RIF regressions)
Besides affecting the earnings among the employed, the pandemic also had an impact on unemployment. This is illustrated in Figure 4 showing the share of employees who had a salary during January-February but no salary, that is, unemployment, during March-December in 2019 and 2020. The figure shows that in the bottom quartile, the share of workers that became unemployed increased threefold, from 6% in 2019, which is assumed to be a normal within-year transition rate, to 18% in 2020. Employees in other quantile groups did not experience such a drastic rise in unemployment spells, indicating that labour supply effects could account for a large part of the observed overall inequality increase during the Covid-19 pandemic.
Figure 4 Unemployment share in March-December among workers with a salary in January-February, 2019-2020
The distributional impact of the government’s Covid-19 support
We have access new registers that record all individual payments of the government’s Covid-19 support directed to firms and individuals. By linking these payments to the earnings registers, we can calculate counterfactual earnings distributions by subtracting the support money from the individual earnings.
Two main counterfactual scenarios are implemented. In the first policy simulation (‘PS1’), we reduce the earnings of employees by the exact amount of the support received. This corresponds to a situation in which employees work fewer hours, and receive less income, but retain their jobs in line with an anti-unemployment model like Germany’s Hartz-IV programme. In ‘PS2’, we instead let employees keep their full-time employment, and full-time salary, but shrink the number of employees until the total wage bill matches the employers’ supportless payments. This procedure is close to the traditional Swedish labour market model, which nurses full-time employment and the sacking of young, low-wage workers according to a ‘last in, first out’ scheme.
Figure 5 presents the results of these simulations. Both scenarios generate a markedly higher inequality level than what was actually observed. The Gini coefficient would have increased two to three times more than it did during the pandemic’s initial months and slightly less thereafter had the government not extended any income support.
Figure 5 Covid-19 support and earnings inequality: Simulated counterfactual outcomes
Note: Gini-coefficient of monthly earnings among working adults aged 18-64. ”Actual” signifies observed outcomes. The simulations reflect outcomes without government Covid-19 support money in two scenarios: sustained employment but fewer hours worked (PS1) or sustained full-time jobs but higher unemployment (PS2). PS3 denotes the PS1 simulation plus a redistribution of all support to individuals in the bottom earnings quartile.
The finding that inequality increased during the pandemic in Sweden goes counter to what has been found in other countries (Stantcheva 2021). One possible explanation is that the direct government Covid-19 support has been relatively small in Sweden (IMF 2021). To examine this, we make a third simulation, ‘PS3’, in which we investigate whether the actual amounts of support in Sweden could have, hypothetically, decreased inequality, should the money have been distributed among those with the lowest earnings. We first remove all the support (as in PS1) and then add this support by distributing it to individuals in the lowest earnings quartile. The results show that inequality decreases when doing this, thus emphasising the role of government support for inequality during the pandemic.
Annual total income inequality
In a supplementary analysis, we examine yearly income tax returns during 2018-2020 of all adults (including old-age pensioners). Annual incomes cannot be used to draw firm conclusions about the pandemic, but these data contain incomes from all sources, including capital, self-employment and taxable transfers as well as all taxes paid.
Figure 6 shows that average incomes fell in the bottom of the distribution, partly due to lower earnings (as in the monthly analysis) and to losses among the self-employed (the grey negative fields). Incomes the upper half of the distribution increased somewhat.
Figure 6 Average incomes and their composition, adult population, 2018-2020
Figure 7, finally, presents Gini coefficients for annual total incomes of adult individuals. Inequality increased significantly in 2020 by about 5%, both before and after taxes. This is a larger increase than was observed for monthly earnings, and the main explanation is the addition of self-employment deficits and lower earnings among pensioners.
Figure 7 Inequality in the total annual income among adult individuals, 2018-2020
Our analysis of administrative tax registers show that the Covid-19 pandemic has increased income inequality in Sweden. Government support policies mitigated the increase, and simulations show that without them, inequality would have increased two-three times more. We also find suggestive evidence that if the Swedish government would have distributed the amount of support differently, inequality might even have decreased during the pandemic, as seen in some other Western economies.
Almeida, V, S Barrios, M Christl, S De Poli, A Tumino, and W van der Wielen (2021), “The impact of COVID-19 on households’ income in the EU”, Journal of Economic Inequality, forthcoming.
Angelov, N and D Waldenström (2021), “COVID-19 and income inequality: Evidence from monthly population registers”, CEPR Discussion Paper 16333.
Clark, A E, C D’Ambrosio, and A Lepinteur (2021), “The fall in income inequality during COVID-19 in four European countries”, Journal of Economic Inequality, forthcoming.
Furceri, D, P Loungani, J D Ostry, and P Pizzuto (2020), “COVID-19 will raise inequality if past pandemics are a guide”, VoxEU.org, 8 March.
Galletta, S and T Giommoni (2020), “Pandemics and inequality”, VoxEU.org, 3 October.
IMF (2021), Fiscal Monitor April 2021, Washington D.C.
O’Donoghue, C, D M Sologon, I Kyzyma, and J McHale (2020), “Modelling the distributional impact of the COVID-19 crisis”, Fiscal Studies 41: 321–336.
Stantcheva, S (2021), “Inequalities in the times of the pandemic”, Economic Policy, forthcoming.