COVID-19 will raise inequality if past pandemics are a guide
Davide Furceri, Prakash Loungani, Jonathan D. Ostry, Pietro Pizzuto 08 May 2020
The tragic death toll from COVID-19 has been accompanied by the upending of millions of livelihoods as governments take necessary steps to limit the spread of the virus. More jobs were lost in the US in March this year than over the entire Great Recession of 2008-09 (Coibion et al. 2020), with workers with less than college education taking the largest hit according to early evidence. Globally, job loss is estimated to be over 200 million, with 40% of the global workforce employed in sectors that face high risk of displacement and with limited access to health services and social protection (ILO 2020). Such workers will face challenges in regaining their livelihoods even after economies start to recover.
Recognising the disproportionate burden of the pandemic on low-skilled workers, a recent poll of economists found that the vast majority are concerned that COVID-19 will raise inequality (IGM 2020). Our results strongly support this concern (Furceri et al. 2020). We find that major epidemics in this century have raised income inequality, lowered the share of incomes going to the bottom deciles, and lowered the employment-to-population ratio for those with basic education but not for those with advanced degrees.
Pandemics raise inequality
While there is much to be learnt from the experience of pandemics that occurred prior to this century – see Barro et al. (2020) on the Spanish flu of 1918-19 and Jordà et al. (2020) for even earlier events – our present analysis focuses on five major epidemics of this century – SARS (2003), H1N1 (2009), MERS (2012), Ebola (2014) and Zika (2016) – which each affected several countries. We construct a dummy variable which takes the value 1 in the year in which the WHO declared a major epidemic event for a country and 0 otherwise. To estimate the distributional impact of these pandemic events, we follow Jordà (2005) and estimate impulse response functions directly from local projections.
Figure 1 shows the estimated impulse response of the net Gini to a pandemic event over the five-year period following the event. Pandemics lead to a persistent and significant increase in the net Gini measure of inequality. After five years, the Gini is above its pre-shock level by about 1.25%. Given that changes in the Gini coefficient tend to be very gradual, these are quantitatively important effects – the effect corresponds to about ½ standard deviation of the change in the Gini in our sample.
Figure 1 Impact of pandemics on inequality
Notes: The figure shows the impulse response (and 90% confidence bands) of the net Gini to a pandemic for five years after the event for 175 countries over the period 1961-2017. The baseline specification includes two lags of the dependent variable and current and two lags of the pandemic dummy variable. Gini coefficients are from the Standardized World Income Inequality Database. See Furceri et al. (2020) for details.
We find that the impact on the net Gini is larger than that on the market Gini, which suggests that policies undertaken to address previous pandemics may have ended up having regressive effects. While further investigation of this result is needed, it is worth noting that early assessments of some government programmes enacted to combat COVID-19 suggest that the rich are the major beneficiaries (e.g. JCT 2020).
Our paper describes several robustness checks of the baseline finding that pandemics raise inequality. Specifically, (1) we use the autoregressive distributed lag model (ADL) as in Romer and Romer (2010) as an alternative regression strategy; (2) we include several control variables to explain the behaviour of inequality (such as proxies for the level of economic development and measures of trade and financial globalization); and (3) we check that our results hold for the post-2000 sub-sample.
In addition, we verify that that the increase in the Gini coefficient is also reflected in changes in the relative shares of income going to various deciles. Using data from the World Bank’s World Development Indicators – available for 64 countries for our sample – we find that the shares of incomes going to the top deciles increases and that to the bottom deciles falls after a pandemic event. For instance, in our sample, the share of income going to the top two deciles is 46% on average, while the share going to the bottom two deciles is only 6% – a gap of 40 percentage points. Five years after the pandemic, this gap increases by 2½ percentage points. Recent work (Avdiu and Nair 2020) has shown that workers in low-income deciles have more limited ability to work from home than those in higher deciles. The extensive lockdowns associated with efforts to curtail the spread of COVID-19 can therefore exert a particularly adverse impact on such workers in the absence of policies to alleviate such outcomes.
Channels of transmission
As shown by Ma et al. (2020), the impact of the five pandemic events on aggregate economic activity varies across episodes and countries. We therefore augment our baseline specification to allow for the distributional effect of pandemic events to vary with their impact on economic activity. We find that for pandemic episodes associated with significant economic contractions, the effect is statistically significant and larger than the baseline effect (the medium-term effect on the Gini increases from 1.25% to about 2%). In contrast, it is not statistically significantly different from zero for episodes associated with high growth.
Figure 2 Impact of pandemics on inequality: Role of economic conditions
Notes: The figure shows the impulse response (and 90% confidence bands) of the net Gini coefficient to pandemic events for a sample 175 countries over the period 1961-2017. The red line shows the response associated with very low growth (left panel) and very high growth (right panel). The dotted green line denotes the baseline average (unconditional) effect reported in Figure 1.
In addition to output loss, a related channel through which pandemics can affect inequality is through adverse impacts on employment prospects for some groups of workers, particularly low-skilled workers. Since data on employment by skill levels are difficult to obtain for a large group of countries, we use data on employment-to-population ratios for different education levels; ILO (2020) notes that “statistics on levels of educational attainment remain the best available indicators of labour force skill levels.” Figure 3 shows the vastly disparate impacts: those with advanced or intermediate levels of education are scarcely affected, whereas the employment to population ratio of those with basic levels of education falls significantly, by more than 5% in the medium term.
Figure 3 Impact of pandemics on employment outcomes
Notes: The figure shows the impulse response (and 90% confidence bands) of the employment-to-population ratio to pandemic events for a sample of 76 countries over the period 1990-2017.
There is a growing literature on the likely aggregate consequences of COVID-19 drawing on the experience following past major epidemics. Our work examines the likely distributional effects of such episodes, finding that the effects are similar to those of other types of crises in exacerbating inequality, including by depressing employment prospects for those most vulnerable, such as low-skilled workers (de Haan and Sturm 2017). While governments have acted fast to provide stimulus, as described in Baldwin and di Mauro (2020), our results suggest that in the absence of deliberate and strenuous attempts to protect the most vulnerable segments of society, this pandemic could end up exerting a significant adverse impact on inequality. In fact, our finding that the inequality effect increases with the negative effect of pandemic events on economic activity suggests that, all else equal, the distributional consequences of COVID-19 may be larger than those in previous pandemic episodes.
Note: The views expressed in this column should not be ascribed to the institutions with which the authors are affiliated.
Avdiu, B and G Nayyar (2020), “When face-to-face interactions become an occupational hazard: Jobs in the time of COVID-19,” Brookings Review, 30 March.
Baldwin R and B Weder di Mauro (2020), Mitigating the COVID economic crisis: act fast and do whatever it takes, CEPR Press VoxEU.org eBook.
Barro, R, J Ursua and J Weng (2020), “Coronavirus meets the Great Influenza Pandemic,” VoxEU.org, 20 March.
Coibion, O, Y Gorodnichenko and M Weber (2020), “Labour markets during the Covid-19 crisis: A preliminary view, NBER Working Paper No. 27017.
De Haan, J, and J-E Sturm (2017), “Finance and Income Inequality: A Review and New Evidence,” European Journal of Political Economy 50: 171-195.
Furceri, D, P Loungani, J D Ostry, and P Pizzuto, 2020, “Will Covid-19 affect inequality? Evidence from past pandemics,” Covid Economics 12: 138-57.
Furceri, D, P Loungani and J D Ostry (2019), “The Aggregate and Distributional Effects of Financial Globalization: Evidence from Macro and Sectoral Data,” Journal of Money, Credit and Banking 51(S1): 163-198.
IGM – Initiative on Global Markets (2020), “Inequality and the Covid-19 Crisis,” 13 April.
ILO – International Labor Organization (2020), ILO Monitor 2nd edition: COVID-19 and the World of Work.
JCT – Joint Committee on Taxation (2020), Estimated Revenue Effects Of The Revenue Provisions Contained In An Amendment In The Nature Of A Substitute To H.R. 748, The “Coronavirus Aid, Relief, And Economic Security (‘CARES’), April.
Jordà, O (2005), “Estimation and inference of impulse responses by local projections,” American Economic Review 95: 161–182.
Jordà, O, S Singh and A Taylor (2020), “The longer-run economic consequences of pandemics,” VoxEU.org, 8 April.
Ma, C, J Rogers and S Zhou (2020), “Global economic and financial effects of 21st century pandemics and epidemics”, Covid Economics 5: 56-78.
Romer, C D and D H Romer (2010), “The Macroeconomic Effects of Tax Changes: Estimates Based on a New Measure of Fiscal Shocks,” American Economic Review 100(3): 763-801.