key: cord-0903363-ogb56xg9
authors: Palomino, Juan C.; Rodríguez, Juan G.; Sebastian, Raquel
title: Wage inequality and poverty effects of lockdown and social distancing in Europe
date: 2020-08-11
journal: Eur Econ Rev
DOI: 10.1016/j.euroecorev.2020.103564
sha: c8abe71f673f95865c2fbbf82fc12c5a9aac0a6d
doc_id: 903363
cord_uid: ogb56xg9

Social distancing and lockdown measures taken to contain the spread of COVID-19 may have distributional economic costs beyond the contraction of GDP. Here we evaluate the capacity of individuals to work under a lockdown based on a Lockdown Working Ability index which considers their teleworking capacity and whether their occupation is essential or closed. Our analysis reveals substantial and uneven potential wage losses across the distribution all around Europe and we consistently find that both poverty and wage inequality rise in all European countries. Under four different scenarios (2 months of lockdown and 2 months of lockdown plus 6 months of partial functioning of closed occupations at 80%, 70% and 60% of full capacity) we estimate for 29 European countries an average increase in the headcount poverty index that goes from 4.9 to 9.4 percentage points and a mean loss rate for poor workers between 10% and 16.2%. The average increase in the Gini coefficient ranges between 3.5% to 7.3% depending on the scenario considered. Decomposing overall wage inequality in Europe, we find that lockdown and social distance measures produce a double process of divergence: both inequality within and between countries increase.

The dramatic and unprecedented intensity of the shock due to the Covid-19 pandemic has highlighted the importance of measuring the economic consequences of "social distancing". The lockdown measures implemented in many countries around the world will likely have a significant negative impact on their GDP. Thus, the global economy is expected to shrink by 3% this year. Asia will not have economic growth for the first time in 60 years, the economies of the US and Europe are projected to contract between 8% and 13% (IMF, 2020) , and the global job losses are estimated to be over 200 million (ILO, 2020) . However, the effect of the pandemic will not only take place at the aggregate level and is likely to have distributional implications (Furceri et al., 2020) . The social distancing imposed by governments to limit the spread of the pandemic has caused an asymmetric effect on the labour market: discounting essential occupations like health services and food sales, only the jobs not closed by the lockdown that can be done from home ("teleworkable") will be not impeded. This asymmetry of the supply shock implies that the economic costs of social distancing could be significant, not only in terms of negative GDP growth rates but also in terms of higher wage inequality and poverty rates. In this paper, we analyse the potential effects of social distancing on wage inequality and poverty in absence of any compensating public policy across Europe.

Recent studies have provided estimates of the supply shock caused by the emergency regulation imposed to contain the spread of Covid-19 (Dingel and Neiman, 2020; Hicks, 2020) . They have evaluated the possible economic consequences of social distancing -without considering the subsequent effects that may occur on the demand side-by calculating to what extent occupations can be performed from home (teleworking). Mongey et al. (2020) find that workers with less ability to work from home have been indeed more unable to follow the 'stay at home advice' -using geolocation data-and that they have suffered higher unemployment increases.

Studies focused on the demand side have shown that, due to fear of infection, economic activity dropped and individuals changed their consumption habits even before lockdown was formally imposed (Goolsbee and Syverson, 2020) . In a more comprehensive analysis, del Rio-Chanona et al. (2020) provide quantitative predictions of supply and demand shocks associated with the Covid-19 pandemic. These studies, mainly focused on the U.S. economy, have analysed the consequences of social distancing in the job market, at the occupation and industry levels, without delving into the study of poverty and inequality. We find two exceptions. First, Irlacher and Koch (2020) obtain a substantial wage premium (higher than 10%) in a Mincer regression for German workers performing their job from home, and a lower share of teleworkable occupations in poorer German regions. Second, Brunori et al. (2020) study the short-term effects that two months of lockdown have had on the Italian income distribution. By using a static microsimulation model, they find a non-negligible increase of poverty and inequality in Italy.

To estimate the impact of social distancing on wage inequality and poverty across Europe, we concentrate on the legal restrictions (supply side) due to the closure of non-essential occupations and workers not being able to perform their activities at home during the lockdown period.

Additionally, we consider a range of scenarios with further partial closures of certain activities, caused by mandatory capacity limitations and by the change in individual consumption habits to prevent contagion. Despite its relevance, our analysis does not consider additional indirect effects in supply (as shortages propagate through supply chains) and in demand (when the loss of labour income for some workers further reduces consumption). These effects are difficult to estimate at this early stage, and the legal and voluntary restrictions imposed to prevent the spread of the pandemic already provide a clear framework to study the consequences for inequality and poverty of having a particular productive structure. Under a lockdown, the asymmetry of the restrictions may affect economies in a different way just because their productive structure is not the same: countries who are specialised in outdoor and non-essential activities like tourism will, in principle, suffer more from the lockdown and the capacity limitations. We thus restrict ourselves to measuring the potential impact on wage inequality and poverty of enforced and voluntary social distancing.

In general, if occupations with higher wages are more teleworkable, we should observe an increase in wage inequality due to lockdown and the implementation of social distance measures within each country analysed. However, whether this happens, and the intensity of this change, will depend on the structure of the economy and the extent of essential and closed occupations under the lockdown. The wide set of European countries in our sample -with a variety of productive structures-will also allow us to test if different productive structures imply different potential effects on wage inequality and poverty under lockdown and social distancing. Note that if that is the case, inequality would not only increase within countries, but also between nations, which could exacerbate the problem of cohesion in Europe.

The first step to measure the changes in wage inequality and poverty across Europe due to lockdown is to calculate the index of teleworking at the occupational level. Dingel and Neiman (2020) found that 37 percent of jobs in the United States can be done entirely from home. What is the share of occupations that allow teleworking in Europe? Following these authors, we use fifteen questions from the Occupational Information Network (O*NET) database such as, 'is the work done outdoors?' or, 'does it require significant physical activity?' to calculate the probability of teleworking for each occupation. We then use the 2018 European Union Labour Force Survey (EU-LFS) occupational structure to translate these probabilities into the European context. Finally, we match the EU-LFS occupation teleworking index and the 2018 European Union Statistics on Income and Living Conditions (EU-SILC), which provides detailed information on wage at the individual level. After this process, we have for every worker in the EU-SILC database the individual index of teleworking (according to her occupation) and the wage.

A lockdown implies that some activities -like healthcare or food chain related jobs-will become essential while others will be closed. When the occupation is essential, workers will be not affected by lockdown regardless of their capacity to work from home. When a certain economic activity is closed -like hospitality-working is not at all possible. For the remaining economic activities, only teleworking is allowed. Consequently, during the lockdown we need to adjust our index of teleworking for the workers whose occupation is essential or closed, to obtain an individual measure that summarizes the capacity of each worker to keep active under the lockdown. We will call this measure Lockdown Working Ability (LWA) index. The next step is to estimate the wage loss due to the lockdown. Because not all workers are able to perform their job at home and some activities are closed or limited (or less demanded to avoid infection), there are potential wage reductions for a significant part of the labour force. To simulate these wage losses, we consider four possible scenarios: two months of lockdown and two months of lockdown plus six months of only partial functioning of the closed activities at 80%, 70% and 60% of capacity. In addition, we consider in Appendix E another three scenarios where partial functioning of closed occupations −at 80%, 70% and 60% of full capacity− lasts for nine months. With these proposals we intend to measure not only the effect of the lockdown, but also the impact of the de-escalation period imposed to contain the spread of Covid-19.

Although each European country may have followed slightly different lockdown and deescalation strategies, the core of the social distancing enforcing policies has been similar in most of them. For that reason, and to ensure that in our analysis differences across countries are mainly due to their productive structure, we simulate the same scenarios for all European countries.

The last step is to measure the changes in wage inequality and poverty across countries, and the variation of wage inequality between and within-countries due to the lockdown. For this task, we first compute the Lockdown Incidence Curve (LIC), which represents the relative change in the wage of individuals ordered by centiles, and the related changes in the mean 'growth' rate of the poor (Ravallion and Chen, 2003) and headcount poverty index. Then, we use the Gini coefficient and the Mean Logarithmic Deviation (MLD) to calculate the changes in wage inequality. The first measure is a popular index of inequality which is widely used in the literature, while the second measure fulfils some properties which are necessary for our analysis of inequality decomposition. In particular, the MLD is the only inequality index that is additively decomposable into a between-group and a within-group component (Bourguignon, 1979; Shorrocks, 1980) and has a path-independent decomposition (Foster and Shneyerov, 2000) .

Our results show that poverty increases for the headcount index and the mean loss rate of the poor in all countries for all simulations, although these changes vary greatly with the country and simulation under consideration. For the four simulated scenarios, we estimate for 29

European countries an average increase in the headcount poverty index that goes from 4.9 to 9.4 percentage points and a mean loss rate for the poor that goes from 10% to 16.2%. Likewise, wage inequality increases for the Gini coefficient in all countries for all simulations. For example, the average increase in the Gini coefficient varies between 3.5% and 7.3% for Europe as a whole. We thus find that poverty and inequality changes are sizeable in all countries and they increase with the duration of the lockdown. When we decompose overall inequality in Europe, both within countries and between countries inequality increase, producing a double process of divergence in wage inequality in Europe. But this increase in wage dispersion is not symmetric, and the within-countries inequality component increases more than the betweencountries inequality component. Thus, the increase in the within-countries inequality component ranges from 5.0% to 12.1%, while the change in the between-countries inequality component goes from 2.5% to 4.0%, depending on the scenario under consideration.

The observed increases in poverty and inequality are in general greater in Eastern and Southern countries than Central and Northern countries and are inversely related to the LWA index. To further understand these differences across European countries we need to analyse separately the three components of the LWA index: teleworking and essentiality (which increase the working ability of occupations during the lockdown), and closure (which implies a lower working ability during the lockdown and subsequent de-escalation periods). We observe that the occupational and industry structure of Eastern European countries presents, in general, medium average levels of closure, but low average levels of essentiality and teleworking. On the contrary, Northern and Central European countries show large values of essentiality and teleworking, and low average levels of closure. Meanwhile, workers in Southern European countries have medium average levels of essentiality, rather low levels of teleworking and, especially, the highest average levels of closure, due to the large preponderance in their productive sector of activities that imply the agglomeration of large groups of people like recreation or hospitality.

The rest of the paper is structured as follows. In Section 2 we present the LWA index and the methodology applied to calculate the wage losses and changes in inequality and poverty. In Section 3 we highlight the main results obtained for Europe. Finally, Section 4 concludes.

To analyse the working ability of individuals during the lockdown, we need first to estimate occupational teleworking. For this task we use three different databases (see Appendix A). First, we make use of teleworking information acquired using key attributes and characteristics of occupations from the American O*NET database (Dingel and Neiman 2020) . Second, we use the latest 2018 wave of EU-LFS (2020 release) -with detailed employment and occupational information for European countries-to accurately obtain occupational teleworking information for the European occupational categories. Finally, this information is combined with the rich socioeconomic data -crucially, salaries-from the 2018 wave of EU-SILC (2020 release). This strategy, which is explained in detail in Appendix A, allows us to calculate the teleworking capacity of European workers.

However, teleworking capacity is not the only determinant of workers ability to effectively work and keep their wage during the lockdown period. We need to consider that some Having calculated the teleworking index and identified the essential and closed occupations, we construct the Lockdown Working Ability (LWA) index. This measure summarizes the capacity of individuals to work under a lockdown taking into account not only the value of their occupation's teleworking index, but also if such occupation is essential ( ) or closed ( ). The central idea is that workers in essential occupations can work during the lockdown regardless of whether the occupation can be teleworked or not. On the contrary, workers in closed activities cannot work at all to the extent that their overall activity has been closed. In all remaining cases, working capacity will depend on the share of that occupation that can be teleworked.

Formally, the first step in constructing the LWA index requires to split the population of workers into three groups according to the occupation of each worker ∈ {1, 2, … , }. If the individual has an occupation that is neither essential nor closed, the value of her index will be equal to the value of her index of teleworking, ∈ [0,1]. If the person has an essential job ( = ), we will compute the LWA index as

is the essentiality score given to the occupation of the individual. Thus, for partially essential occupations (0 < < 1), the non-essential share of the occupation (1 − ) can work during lockdown only to the extent that it is teleworkable. Finally, if the person has a job that is closed ( = ), we will calculate the LWA index as = (1 − ) , where ∈ (0,1] is the close score given to the closed occupation of the individual. In partially closed occupations (0 < < 1), the non-closed share of the occupation (1 − ) can work to the extend that is teleworkable. In summary, the Lockdown Working Ability index is calculated as follows:

for all ∈ {1, 2, … , }.

As shown in Figure 1 , the average LWA index varies significantly across European countries: Our LWA index varies significantly not only by countries but also by gender, type of contract −permanent or temporary−, type of work −full or part time−, and level of education (see Table   1 ). According to their LWA index, women jobs are less affected by social distancing than men jobs in all European countries. Interestingly, the biggest difference between both sexes is found in the Nordic countries: Norway (23 points), Denmark (22 points 

The next step is to calculate the potential wage loss due to the lockdown for every individual in the population. Following the facts observed in Europe during the pandemic, we adopt four possible scenarios. As a lower bound, we assume a simple scenario: two months of lockdown (case i). 4 Here we suppose that occupations can be developed at full capacity (100%) after the lockdown. Next, we assume two months of lockdown and six months of partial functioning of closed occupations at 80% (case ii), 70% (case iii) and 60% (case iv) of full capacity. The rationale for these scenarios is that governments may not allow a fully functioning of closed occupations after the lockdown to avoid a new outbreak of the virus and that individuals may voluntarily choose to stay home, which reduces their consumption, to avoid infection (Goolsbee and Syverson, 2020) . For example, some activities like arts, entertainment, recreation, restaurant, hotel and transportation are still under a large negative shock in production and consumption. 5 Consequently, to be more informative we have adopted a range of capacity constraints that goes from 100% to 60% of full capacity for a period of six months. In addition,

given the uncertainty about the duration of legal and voluntary restrictions, in Appendix D we have further replicated our results for another three scenarios where partial functioning of closed occupations −at 80%, 70% and 60% of full capacity− lasts for nine months instead of six. We simulate the same scenarios for all countries so that differences across countries are mainly due to their distinct productive structures, isolating our analysis from the influence of particular mitigation measures adopted by each European government.

Using the LWA index, we calculate the wage loss ( ) experienced by every individual during the lockdown according to the four simulated scenarios described. For the first case (i) the equation we estimate is the following:

where −1 is the annual wage of individual in period − 1 (before the lockdown) and

represents the duration of the lockdown in annual terms, i.e., = 2 12 . Because measures the capacity of individual to work under a lockdown, the term 1 − represents the incapacity of worker to perform their job under a lockdown. Consequently, the wage loss 4 The duration of the lockdown in many European countries has been approximately of two months. For example, the lockdown went from the 23rd of February to the 3th of May in Italy, from the 15th of March to the 4th of May in Spain, and from the 17th of March to the 11th of May in France. 5 For these activities Chetty et al. (2020) have found, as of July 08 th 2020, a decrease in spending compared to the pre-lockdown levels in the range of between 32.9% and 48.3% in the US. Based on weekly credit card expenditure data, BBVA Research (2020) has found, as of July 12 th 2020, a negative interannual growth rate for these spending groups ranging between 20% and 40% in Spain. experienced by workers under a lockdown is the proportion of annual wage they lose due to their inability to work during the shutdown period. For the cases (ii), (iii) and (iv) we apply the same equation (2) unless the individual has a closed occupation, in which case, we need to additionally consider the wage loss due to the partial closure of 20%, 30% or 40% of their occupation for six (or nine) additional months. For these scenarios the relevant equation is:

where 1 = { 1 = 0 ≠ is the indicator function, represents the duration of the partial closure ( 6 12 or 9 12 ), and is the proportion of closure (0.2 for case (ii), 0.3 for case (iii) and 0.4 for case (iv)). To give the intuition of expressions (2) and (3), we provide an example.

Suppose we are interested in calculating the wage loss for three workers with different occupations under a lockdown of 2 months and a posterior partial closure of some activities of 20% (keeping 80% of capacity) for 6 months. The first worker is a physician, the second works as a clerk and the third one is a waiter. Their pre-lockdown ( − 1) annual salaries are 1 −1 , 2 −1 and 3 −1 , respectively. The occupation of the first worker is essential with = 1 so we apply equation (2) to calculate her wage loss: 1 = 1 −1 · 2 12 · 0 = 0, i.e., the doctor does not lose earnings during the pandemic. The second worker has an administrative occupation that in our example is neither essential nor closed so her index is equal to the value of her index of teleworking, for example, 0.7. Then, according to equation (2) her wage loss will be 2 = 2 −1 · 2 12 · (1 − 0.7) = 2 −1 · 0.6 12 . Finally, the occupation of the third worker is closed with = 1 so the value of her index is 0. In this case, we apply equation (3) Closely related to the LIC we have the mean loss rate for the poor, a measure that reveals the average wage loss for workers below the poverty line. 8 Let ( ) denote the headcount index defined as the proportion of workers whose salary is less than , where is the poverty line (set at 60% of the median wage). Then, the mean loss rate for the poor ( ) is defined as the area under the LIC up to the headcount index divided by the headcount measure, which can be expressed as:

When < 0 for all < ( ) one can conclude that the lockdown was poverty-augmenting.

In addition to representing the LICs for the set of European countries and compute the above Unfortunately, the Gini index is not additively decomposable into a between-group and a within-group component. Its decomposition includes also a residual term which cannot be assigned to either component. For this reason, we use the MLD index in the last part of our analysis, where we decompose the overall estimated change of inequality in Europe into its between-countries and within-countries components. The MLD belongs to the Generalized Entropy class, which is the only class of inequality indices that is additively decomposable into a between-group and a within-group component (Bourguignon, 1979 and Shorrocks, 1980) . Moreover, the MLD has a path-independent decomposition, so the result of the decomposition is independent of which component (between-group or within-group) is eliminated first (Foster and Shneyerov, 2000) . The MLD ( ) is defined as: 

where 1 is n-coordinated vector of ones. Expression (7) provides a breakdown of overall wage inequality into between-group and within-group terms. The between-group component

( 1 1 1 , 2 1 2 , … , 1 ) is the level of wage inequality that would arise if each worker in a country enjoys the mean wage of the country, and the within-group component ∑ ( ) =1 is the weighted sum of wage inequalities within different countries.

How large are the wage losses experienced by workers during the lockdown? Are they evenly distributed? A visual approximation to the wage loss across the wage distribution can be achieved by looking at the LIC curves. In Figure 3 we represent the LICs for some European countries under two months of lockdown (case i). 9 We highlight the LIC for two extreme cases, Romania, and Cyprus, where the main wage losses take place at the upper and lower part of the wage distribution ( Figure 3 , panel a) and for Germany, the UK and France (Figure 3, panel b) ,

where wage losses tend to decrease with the percentile that the worker occupies.

In our two-month lockdown scenario, wage losses are already sizeable and can be superior to 10% at some parts of the distribution in some countries, varying significantly with the centile of the workers. Also, we find that the shape of the distribution of wage losses is similar for the rest of scenarios and the main difference lies only in the larger size of the wage drops. 

We formally capture the potential impact of social distancing on poverty with two measures: the mean loss rate for the poor ( ) and the changes in the headcount poverty index ( ). 10 Table 2 (columns 2 to 5) shows the values of for the four simulations under consideration, which indicate that the lockdown is poverty-augmenting in the four scenarios for all European countries. Under the most conservative scenario, a two-month lockdown, we find a mean loss rate for the poor of 10% for Europe on average. If we consider an additional 6-month period of partial functioning of closed activities at 60% of full capacity (scenario iv), the mean loss rate for poor workers in Europe increases up to 16% (19% if the period of partial functioning lasts for nine months). By countries, the highest loss rate for the poor is found in Cyprus (between 12.3% and 22.4%), while the smallest one happens in Romania (between 3.1% and 4.0%).

The share of workers under the poverty line also increases significantly for the four simulated cases in all European countries (see Table 2 , columns 6-14). 11 For Europe on average, this increase varies between 4.9 percentage points under two months of lockdown and 9.4 percentage points if we consider an additional six months during which closed activities are partially functioning at 60%. These values imply that −in absence of compensating policies− the percentage of poor people in Europe (those with earnings below 60% of the pre-lockdown median in the country) may substantially increase even if lockdown does not last long. If closed 10 The contribution of growth and redistribution to poverty reduction over the period 1981 to 2010 in 124 countries, including Europe, is estimated in Bluhm et al. (2018) . The relationship between the components of social mobility (growth, dispersion and exchange mobility) and inequality is estimated for 15 countries of the European Union in Prieto et al. (2008) . 11 The headcount poverty index gives additional relevant information about the consequences of the pandemic for poverty since this index is not necessarily consistent with the LIC (Essama-Nssah and Lambert, 2009). Eastern and Southern countries than Central and Northern countries. Thus, we find that the highest increase in poverty according to the headcount poverty index is in Croatia (between 8.5 and 15.3 points). On the contrary, the smallest percentage increase in poverty according to the headcount index happens in Switzerland for case i (2.6 points) and Denmark for cases ii (3.7 points), iii (4.1 points) and iv (4.7 points). If we look at the relationship between these changes in the headcount poverty index and the average LWA we find a negative correlation with 2 = 0.35 for the first scenario ( Figure 4 , panel a) and 2 = 0.51 for the fourth scenario ( Figure C1 , panel a, in Appendix C). When we consider the Gini of LWA within countries instead of the average LWA the correlation becomes positive with 2 = 0.34 for the first scenario (Figure 4 panel b) and 2 = 0.56 for the fourth scenario ( Figure C1 , panel b, in Appendix C). 

The uneven distribution of potential wage losses across the distribution is prone to have an impact on wage inequality as well. European countries already have important differences in pre-lockdown inequality (Table 3, pre-lockdown wage inequality ( = 25.4) . Other countries with low levels of inequality before the lockdown are Sweden, Czechia, Belgium and Norway, all of them with a Gini index slightly below 0.30. On the other extreme of the spectrum, we find the countries with the highest level of pre-lockdown wage inequality: Bulgaria, Ireland, the UK and Spain, all of them with a Gini index above 0.40.

When comparing wage inequality after the lockdown with the baseline, it is observed that absolute and relative changes in inequality are sizeable and increase in all countries with the duration of the partial closure of some activities (Table 3) . For example, the relative increase of the Gini coefficient, at scenario (i), ranges from 2.2% (the Netherlands) to 4.9% (Cyprus). At the more severe scenario (iv), relative increases in inequality range from 4.9% in France to 10.9% in Cyprus (if the period of partial functioning lasts nine months these rates become 7% in France and 15% in Cyprus).

A scrutiny of the absolute changes in the Gini coefficient −which makes the change scale independent of the previous level of inequality− reveals that Cyprus is again the European country where inequality increases the most for the four simulation cases (between 1.9 and 4.2 Gini points) followed by Ireland (between 1.7 and 3.4 points). On the other hand, the smallest absolute change in inequality is found in Norway (0.7 points) for scenario (i), in Norway and France for scenarios (ii) (1.2 points) and (iii) (1.5 points), and in France and Denmark for scenario (iv) (1.8 points). These variations in inequality also tend to be related with the LWA index. In Figure 4 Given the link between the values of the LWA index and the increases in wage inequality (and poverty), it is not surprising that the observed changes are in general greater in Eastern and Southern countries than Central and Northern countries. As mentioned before, Northern and Central Europe present large average levels of essentiality and teleworking, and low average levels of closure. Meanwhile, Eastern Europe shows the opposite pattern, i.e., low average levels of essentiality and teleworking, and medium-high average levels of closure. The partial closure of activities that imply the agglomeration of large groups of people like tourism significantly penalises Southern Europe, whose countries tend to have the highest average closure scores. While some tertiary jobs like administrative, programmers and accountants are not affected by partial closures, others are strongly impacted by the measures to fight covid-19 and the fears to Table 3 . Wage inequality changes in Europe.

Note: CL1, CL2 and CL3 refer to partial closure for 6 months at 80%, 70% and 60% of full capacity, 2m is 2 months, and ∆ A G and ∆ R G are the absolute and relative changes in the Gini index. Bootstrapped standard errors are in parenthesis. be infected: hospitality, restaurants, arts and entertainment. For this reason, the impact on inequality and poverty is also related to the share of these specific tertiary occupations. 12

Our findings show an increase in inequality for all European countries but, would inequality changes be different enough to increase inequality between countries? The answer is yes, although they are smaller than the inequality changes occurring within countries. In Table 4 we show the results of the decomposition of wage inequality for all European workers. Our simulations show an increase in overall inequality in workers' salary of 4.26% in Europe (0.423 to 0.441) according to the MLD index under a lockdown of two months. For the same scenario, the changes in the between-and within-countries inequality components are 2.46% (from 0.125 to 0.128) and 5.01% (from 0.298 to 0.313), respectively. For two months of lockdown and six months of partial functioning of closed activities at 60% of capacity (scenario iv) we find the following increases in inequality for Europe on average: 9.68% in overall inequality, 3.98% in between-countries inequality and 12.07% in within-countries inequality. When the period of partial functioning of closed activities at 60% of capacity lasts nine months the corresponding increases are: 14.00% in total inequality, 4.81% in between-countries inequality and 17.85% in within-countries inequality.

Both components of overall inequality increase, but the within-countries inequality component increases significantly more than the between-countries inequality component in all scenarios. 13

That is, cohesion between European countries decreases with the lockdown, though the main change in wage inequality happens within European countries. With the partial closure of some activities, changes get larger and the double process of wage divergence (between and within countries) deepens.

The emergency measures adopted to contain the spread of Covid-19 all around the world are largely based on social distancing and closure of high-risk productive activities. The paralysis of production imposed by the contention measures during the lockdown and the capacity limitations driven by official restrictions and by consumers precautionary behaviour will thus have an uneven impact on workers from different occupations and industries. Our analysis reveals a sizable potential increase in poverty and inequality across Europe. Table 4 . The between-and within-countries inequality components in Europe.

Note: ∆ A is the absolute change in wage inequality and ∆ R (%) is the relative change in wage inequality. Standard errors are in parenthesis.

Poverty will increase under our simulations in all countries. Under the most conservative scenario, a lockdown of two months, we estimate a mean loss rate for the poor of 10% and an increase in the headcount index of 4.9 percentage points on average in Europe, with the change ranging from 2.6 points (Switzerland) to 8.5 points (Croatia). Likewise, wage inequality increases under a lockdown of two months, being the change in the Gini coefficient equal to 3.5% for Europe on average, with changes ranging between 2.2% (Netherlands) and 4.9%

(Cyprus). Considering a more severe scenario with 6 months of partial closure at 60% of full capacity after a two-month lockdown, we estimate a mean loss rate of 16.2% for the poor workers in overall Europe, a rise of 9.4 percentage points in the headcount poverty index and a Gini increase of 7.3% on average for Europe. Our results also highlight that lockdown measures are likely to worsen cohesion in Europe both between countries and, especially, within countries. Our simulations show that between-countries inequality will increase in Europe between 2.46% and 3.98%, while within-countries inequality will increase between 5.01% and 12.07%.

In general, we find a greater increase of both poverty and inequality in Eastern and Southern

Europe than in Northern and Central Europe. Workers tend to have a lower and more unequally distributed ability to work under the shutdown and social distancing in the economies of Eastern and Southern Europe than in the Northern and Central European countries. These differences across European areas get larger with the severity of the measures to fight and prevent Covid-19

infection.

Impacto de la COVID-19 sobre el consumo en España en tiempo real y alta definición

Poverty accounting

Decomposable Income Inequality Measures

Distant and unequal. Lockdown and inequalities in Italy

How did COVID-19 stabilization policies affect spending and employment? A new realtime economic tracker based on private sector data

Supply and demand shocks in the COVID-19 pandemic: An industry and occupation perspective

How many jobs can be done at home?

Tax progression and inequality of income distribution

Measuring pro-poorness: A unifying approach with new results

Path Independent Inequality Measures

Will Covid-19 affect inequality? Evidence from past pandemics

Decreto-Legge 23 febbraio 2020, n. 6. Retrieve at: www

Fear, Lockdown, and Diversion: Comparing Drivers of Pandemic Economic Decline 2020

Occupational exposure to social distancing: A preliminary analysis using O*NET data

ILO Monitor 2nd edition: COVID-19 and the World of Work

World Economic Outlook Update

Working from home, wages, and regional inequality in the light of Covid-19

What is pro-poor growth

Poverty equivalent growth rate

Which workers bear the burden of social distancing policies?

A study on the relationship between economic inequality and mobility

Measuring pro-poor growth

Global inequality when unequal countries create unequal people

The Class of Additively Decomposable Inequality Measures

An economic definition of poverty

An axiomatic characterization of the Watts index