COVID-19 and Differential Income Recovery: Survey Evidence from Indian Punjab

Abstract

The study examines changes in earnings of the workers over 1 year. The results show differential impact of coronavirus disease 2019 (COVID-19) on income of the workers. The findings suggest that workers, whose occupations were restricted in the initial phase of lockdown, continue to do worse even after lifting the restrictions. Females, casual workers and the less educated are among the worst affected by the economic shock. They continue to lag behind with no sign of catching up with other groups. The study asks for short- and long-term policy measures to address the issue.

Keywords

COVID-19 income recovery vulnerability caste gender education differential impact

Introduction

The coronavirus disease 2019 (COVID-19), in addition to costing millions of lives, proved to be the largest economic shock of modern times. Countries, in an attempt to control the spread of the virus, imposed restrictions on economic activities and movements of people (Hale et al., 2020). These restrictions along with the fear of infection led to a massive decline in economic activity causing a significant fall in income and employment (Gopinath, 2020; International Labour Organization (ILO), 2021). The economic contraction and the lockdown policies have impacted the vulnerable groups the most (Ambagudia, 2022; Azeez et al., 2021; Breman, 2020; Buffel et al., 2021; Chee, 2020; Kumar et al., 2021; Maiti et al., 2022). There are concerns that it may significantly dent the fight against poverty, food insecurity and illiteracy (Laborde et al., 2020; Lakner et al., 2020, 2021; United Nations Educational, Scientific and Cultural Organization (UNESCO), 2020; Valensisi, 2020).

Indian economy suffered considerably due to COVID-19 and the resultant nationwide lockdown. The revised estimates show that India recorded a 24.4% fall in its gross domestic product (GDP) in the first quarter of its financial year 2020–2021 (GoI, 2021). The fall in growth rate was accompanied by a sharp rise in unemployment level. Initial phases of lockdown observed unemployment level crossing 25% with more than 80% of households reporting income loss (Afridi et al., 2020a, 2020b; Bertrand et al., 2020; Deshpande, 2020; Lee et al., 2020; Mitra and Singh, 2021; Singh et al., 2020). The second quarter again witnessed a negative growth rate (−7.4%). The annual growth rate estimates show a 7.3% contraction of GDP in 2020–2021.

While the fall in economic activity negatively affected almost everyone, the vulnerable sections suffered the most. Migrant labour was among the hardest hit by the lockdown (Agoramoorthy and Hsu, 2021; Maiti et al., 2022; Ray and Subramanian, 2020). The sudden announcement left them stranded and without work. In the absence of earning, they found it difficult to survive in the city. Since the movement of transport vehicles was restricted, a large number of them started walking to their native places while avoiding detection by the police (Nayar, 2020). Mitra and Singh (2021) found that states with higher inward or outward migration witnessed a higher unemployment rate during the lockdown. Gupta et al. (2021a) found a 63% fall in income from remittances in a month proceeding lockdown announcement in West Bengal. Migrant women were faced with unprecedented hardships. Azeez et al. (2021) studied impact of the lockdown on migrant women. Migrant women suffered loss of livelihood. The loss of income compelled them to take a loan and cut spending on essential items.

Casual workers, Dalits and women suffered a higher loss of livelihood. Many struggled to survive the lockdown period. Centre for Equity Studies et al. (2020) interviewed 1405 individuals across eight states. The results show that the situation was not normalized even after 45 days of the lockdown. More than 60% of the respondents experienced a shortage of food. Nearly 11% had gone hungry for more than one day. Single mothers, Dalits and minorities suffered significantly higher during the lockdown. Ambagudia (2022) argued that the pandemic increased vulnerability among tribals in India. The historical deprivations and inadequate state support meant considerably higher economic hardships for the tribals. Chen et al. (2022) found that informal workers in Ahmedabad were earning merely 28% of their pre-lockdown (PL) income in June 2020 despite easing the lockdown.

Singh et al. (2020) examined changes in income and employment during the initial phases of lockdown in Punjab. The study found a higher impact on casual workers. They suffered more than 90% fall in income and employment in the first 7 weeks of the lockdown. Similarly, households belonging to scheduled caste (SC) and other backward castes (OBC) witnessed higher loss of income and employment than others. Gupta et al. (2021b), using CMIE data, found a 75% decline in income for daily wage workers compared with 40% for regular salaried workers between August 2019 and April 2020. Wage rate of daily wage workers fell by 40% in rural areas and 80% in urban areas. Despite the recovery of income post-lockdown, agricultural wages were 20%–30% less in October 2020 compared with the 2019 level. Non-agricultural wages were 10% less for the same period.

Mansoor and Khan (2021) studied the impact of the lockdown on Muslim casual workers in the Aligarh lock industries. The study found a 44% fall in the employment rate. Nearly 60% of the workers reported a fall in daily wages. The negative impact was especially higher on women workers. Mitra and Singh (2021) found that Indian states with higher dependency on casual wage employment experienced a higher loss of livelihood and incidence of hunger during the initial period of the pandemic. States, which were more urbanized and had a higher share of non-agricultural employment, witnessed a higher unemployment rate.

Empirical evidence not only shows differential impact of the pandemic on vulnerable sections but also suggests slower recovery of their income. Singh et al. (2021) showed that households, whose main occupation had fallen in the restricted category in the initial phase of lockdown (with no possibility of work from home), continued to lag in terms of income recovery even after lifting the restrictions. Since households in the restricted category were earning 44% less even in the PL period, the differential recovery rates meant excessive burden on vulnerable sections.

Studies showed a higher impact of COVID-19 on women workers too. Deshpande (2020), using Centre for Monitoring Indian Economy (CMIE) data, examined employment changes in India during the initial phase of the lockdown. The paper found a higher absolute fall of employment for men compared with women. However, the employed women were 20% less likely to return to work after the first phase of lockdown. Employment started recovering after easing of the lockdown restrictions in May and June 2020. However, the employment recovery showed signs of losing momentum in urban areas in July 2020.¹

The lockdown period also witnessed an increase in burden of household work for men and women. Chauhan (2021) found an increase in burden of unpaid work during the lockdown. Married unemployed women had the highest increase in unpaid work. Nearly 22.5% of married women spent more than 70 hours per week on unpaid work. The share of unemployed women, spending more than 70 hours on unpaid work, increased by 30.5 percentage points during the lockdown. Nonetheless, the period witnessed increased participation of men in unpaid domestic work and narrowing of the gender gap (see Deshpande, 2020).

While the income and employment declined across occupations and regions, Mahatma Gandhi National Rural Employment Guarantee Scheme (MGNREGS) seems to have helped rural areas to withstand the negative effects of COVID-19. Afridi et al. (2022) show that the state with better performance in terms of providing work under MGNREGS could significantly lower the employment loss. The study found that one additional workday under MGNREGS per rural resident in a district prevented the fall of employment by 7% during April–August 2020 over employment rate in January–March 2020 (the baseline employment). The same figure for rural women was 74%. Increase in employment was higher for married women and women with a school-going child. Less-educated women and those belonging to poor households benefitted more from the increase in employment. The results suggest that MGNREGS could significantly lower job loss among vulnerable sections.

The above-discussed studies prove that the vulnerable sections were excessively affected by COVID-19 and the lockdown. However, there are still many unanswered questions. The existing literature has mostly focused on income and employment impact of the pandemic in its initial phase. Studies that have analysed income changes over time are limited in their scope. Data availability is another challenge. Consumer Pyramids Household Survey (CPHS) conducted by the Centre for Monitoring Indian Economy (CMIE) is a good source of longitudinal data. In addition to its bias towards well-off households (Somanchi, 2021), the data are not suitable for comparison when the income changes are happening too fast. The survey collects data from households over 4 months. Thus, income data collected at the beginning of the survey are only comparable with data collected at the end of the survey if there is no change in income during this period. It may not be an issue in normal times when the change in income is likely to be negligible. However, larger changes within a short period make it difficult to compare groups based on CPHS data. In addition, few studies have examined long-run implications of lockdown restrictions.

This study attempts to fill these gaps by examining the long-term effect of the pandemic across sex, caste and educational categories in Mansa, Punjab. The impact of lockdown is traced by dividing workers into two categories based on income loss in the initial period and observing income recovery of the two groups. The study uses data from a repeated survey of randomly selected 55 households from the city of Mansa.

The paper is divided into five sections. Data collection and methodology is discussed in the second section. The third section analyses income changes across various socioeconomic groups among the sampled households. The fourth section uses the results of the econometric model to trace differential impact of the pandemic across groups and of lockdown restrictions. The findings of the regression model are further used to examine long-term trends in income recovery. The last section concludes the study.

Data and methodology

The study uses primary data collected from 55 randomly selected households from Mansa, Punjab, India. Households were selected through multistage random sampling. In the first stage, Mansa was divided into two parts using the railway line which passes through middle of the city. Of these two parts, the area to north of the railway line was selected using random sampling. In the second stage, the selected area was subdivided into 44 parts. Out of 44 parts, 11 were randomly selected for data collection. In each of 11 selected areas, the houses were listed and five households were selected using random sampling for the data collection. There were 79 workers in these households in the PL period. The paper is based on information on these workers. The survey was conducted in four rounds during June 2020, October 2020, January–February 2021 and April–May 2021. All selected households were revisited in each of the rounds. The first round collected information on the PL period and four phases of the lockdown determined in terms of severity of the restrictions. Specific dates of the four lockdown phases were 22 March to 19 April 2020 (P-1),² 20 April to 3 May 2020 (P-2), 4 May to 17 May 2020 (P-3), and 18 May to 17 June 2020 (P-4). The lockdown restrictions were eased with each phase and almost all occupations were allowed by P-4. Thus, the first round of data collection was conducted just when the lockdown had eased and almost all economic activities were allowed. The data collected in this round show early effects of the pandemic and lockdown.

The next three rounds collected monthly information on these households from July 2020 to March 2021. Each of these rounds was 3 months apart from each other. The gap of 3 months was chosen keeping in mind the accuracy of data and practical difficulties in the implementation of the survey. A longer recall period would have affected accuracy of the information collected. On the contrary, a more frequent survey would not only have been costly but also would have increased the chances of respondents dropping from the survey. Thus, the survey covers 1 year from April 2020 to March 2021, including information on PL income. Data of the first and second rounds were used to analyse the initial impact of lockdown and the effect of lockdown restrictions on household income. The results were published in Singh et al. (2020) and Singh et al. (2021). The present paper differs from the earlier two in terms of a detailed examination of the long-run implication of the pandemic. It tests the differential effect of socio-economic variables on workers’ earnings. It also studies impact of PL income on income recovery and whether there is a tendency for the difference in the recovery to disappear over time.

The paper estimates following regression model using the ordinary least squares method

\begin{matrix} I R_{i t} = β_{0} + β_{1} \ln P L I_{i} + β_{2} f e m a l e_{i} + β_{3} f e m a l e_{i} * t i m e_{t} + β_{4} H S_{i} + β_{5} H S_{i} * t i m e_{t} + \\ β_{6} G R_{i} + β_{7} G R_{i} * t i m e_{t} + β_{8} l o s s_{i} + β_{8} l o s s_{i} * t i m e_{t} + β_{8} S C_{i} + β_{8} S C_{i} * t i m e_{t} + β_{8} O B C_{i} + \\ β_{8} O B C_{i} * t i m e_{t} + β_{8} r e g u l a r_{i} + β_{8} r e g u l a r_{i} * t i m e_{t} + β_{8} c a s u a l_{i} + β_{8} c a s u a l_{i} * t i m e_{t} + \\ \sum_{t = 2}^{13} γ_{t} P_{t} + ϵ_{i t} \end{matrix}

(1)

Here, $i$ and $t$ refer to worker and time, respectively. $I R_{i t}$ denotes income recovery. It is income earned in a phase/month as a percentage share of PL income. $\ln P L I_{i}$ is the log of PL income per day (in Rupees). $f e m a l e_{i}$ is a dummy variable for female workers. $H S_{i}$ and $G R_{i}$ are dummy variables for higher secondary and graduate & above education with below higher secondary education as the base category. $S C_{i}$ and $O B C_{i}$ are caste dummy denoting workers belonging to an SC and OBC with ‘Others’ (workers who are neither SC nor OBC) as the base category. $r e g u l a r_{i}$ and $c a s u a l_{i}$ are dummy variables representing regular salaried employee and casual worker and ‘Self-employed’ is the base category. $l o s s_{i}$ is also a dummy variable and takes value 1 if the worker suffered an income loss of more than or equal to 50% in the first three phases (P-1, P-2 and P-3). The variable is a proxy for the lockdown restrictions (it is discussed in detail at the end of the section). $t i m e_{t}$ denotes phase or month of the survey and takes value from 1 to 13 starting with P-1 taking value 1. $P_{t}$ represents a phase dummy variable for the $t$ th period (phase/month). For example, $P_{2}$ takes value 1 if the period is P-2 and 0 otherwise. P-1 is the base category. $ϵ_{i t}$ is the error term.

Here, the log of PL income tests the impact of previous income. A positive coefficient means faster income recovery of workers who were earning higher income PL. A negative sign suggests the opposite. Female, higher secondary, graduate & above, loss, SC, OBC, regular salaried workers and casual worker are intercept dummy variables. Their coefficients measure average difference in income recovery compared with the base category of the respective variable. A negative coefficient means average income recovery to be lower (that is, higher loss of income) for the workers in the category compared with its base category. A positive indicates higher income recovery.

Each of the dummy variables is multiplied with variable time to create a slope dummy variable. It captures convergence or divergence in income recovery of dummy variable with its base category over time. Its interpretation depends on sign of the coefficient of the respective dummy variable. For example, if the coefficient of the female dummy is negative, a positive coefficient of Female × Time would suggest a convergence in male and female income recovery and a negative coefficient would mean a divergence. Phase dummy measures the income recovery over phases (and months) compared with P-1.

The model uses an explanatory variable, $l o s s$ . The variable attempts to capture the differential effect of lockdown restrictions. Since the link between income loss in initial phases and lockdown restrictions is not apparent, it needs a discussion. Lockdown restrictions had a differential bearing on income (see Singh et al., 2021). However, we did not collect data on lockdown restrictions for individual workers. Therefore, the paper captures it indirectly by dividing occupations into two categories. The first category includes workers who suffered an income loss of more than or equal to 50% during the initial three phases (till P-3). The second category of workers lost less than 50% of their income during this period. A comparison between these two categories is clear from the data (Table 1). While workers in the former category were earning more than 50% of PL income in P-1, the latter’s income had dropped to just 2.6%. Between P-1 and P-3, the income of the first category increased from 2.6% to just 6.7% compared with increase from 50.8% to 91.9%. Given the negligible income recovery in the first category, it is safe to assume that there were stricter lockdown restrictions on occupations of these workers in the initial phase and this categorisation can be used to capture the long-term impact of the lockdown.

Table 1.

Average income of workers in a phase/month as the percentage share of their pre-lockdown income.

	P-1	P-2	P-3	P-4	Jul 20	Aug 20	Sep 20	Oct 20	Nov 20	Dec 20	Jan 21	Feb 21	Mar 21
Sex
Male	19.7	32.0	37.5	48.8	86.5	89.1	95.0	96.8	103.4	101.2	93.4	105.5	93.5
Female	18.2	28.9	30.5	30.2	96.1	91.6	85.8	76.2	91.2	87.9	87.0	89.4	76.2
Caste
SC	14.2	29.6	32.8	50.5	80.6	80.2	85.3	89.0	97.1	94.2	93.5	106.6	92.7
OBC	8.9	9.1	18.7	32.1	92.9	97.6	102.1	102.7	108.4	106.6	94.5	104.3	92.5
Others	32.1	51.1	53.8	54.5	89.4	90.0	93.6	90.9	100.0	97.6	90.1	99.9	88.7
Education
Below higher secondary	3.7	6.5	18.7	45.0	76.3	74.0	79.2	83.1	88.4	84.1	87.8	99.1	85.8
Higher secondary	15.7	30.3	39.3	47.5	87.0	91.8	95.6	96.8	106.0	105.9	101.8	114.5	100.4
Graduate & above	33.1	47.7	43.8	45.6	95.6	95.7	100.1	97.1	104.6	100.7	84.4	92.6	83.4
Occupation
Self-employed	18.2	37.1	40.3	50.5	86.7	91.3	94.1	94.3	103.4	100.5	92.1	105.0	94.9
Regular salaried	41.4	49.0	46.8	45.5	102.5	99.6	106.1	105.5	110.0	110.4	94.9	104.2	89.2
Casual worker	1.4	3.9	18.2	35.8	78.6	77.6	83.4	84.5	92.3	88.9	91.3	98.9	84.7
Loss of income till P-3
Less than 50%	50.8	85.7	91.9	81.9	89.0	91.9	98.4	96.9	103.1	100.0	97.8	109.5	97.0
More than or equal to 50%	2.6	2.4	6.7	27.1	87.2	88.1	91.2	92.4	101.0	99.0	89.7	99.9	87.9
ALL	19.5	31.6	36.5	46.3	87.8	89.4	93.7	94.0	101.7	99.4	92.5	103.3	91.1

Source: Primary survey.

SC: scheduled caste; OBC: other backward castes.

In addition to $(1)$ , we have tested four more specifications of the model to ascertain validity of the results. Also, the results of equation $(1)$ are compared with results of the model without slope dummies.

Income changes since lockdown

Most of the economic activities came to a sudden halt with enforcement of curfew by the Government of Punjab on 22 March 2020 and subsequent implementation of national lockdown on 24 March 2020. The first phase of the lockdown (P-1) was the strictest and witnessed a sharp fall in income across occupations (Table 1). However, there were significant differences in the extent of loss across socio-economic categories. SC, OBC, women, workers with below higher secondary education, casual workers and self-employed suffered much higher income loss than their counterparts. Among these, workers with below higher secondary education, casual workers, women and SC continue to remain in the worse state even after lifting of restrictions. These are the categories that had lower average income even in the PL period (see Table 4 in Appendix 1).

The average income of surveyed workers surpassed the PL level in November 2020. Nonetheless, there were significant differences in income recovery across categories. While the average income for other categories hit the PL level between July and November 2020, workers with below higher secondary education, casual workers and women did not reach the mark during the period of our survey (till March 2021) and SC workers crossed the mark only once in February 2021. The respondents consider cutting down on expenditure on non-essentials as the main reason for slower income recovery. Since less educated, casual workers and women were more likely to be engaged in such works, they were more impacted.

Differential recovery does not mean that other categories did not experience income loss. All categories of workers suffered income loss. Average income reached the PL level only in November 2020 and started falling again. The temporary rise in income in November 2020 was due to rise in household expenditure during Diwali, the most important Indian festival. The festival of Diwali, known to increase the household purchases, had given push to the Indian economy in November 2020 (Nag, 2020). The increased expenditure had increased the average income of workers higher than the PL level. However, the income started declining thereafter.

Barring February 2021, the average income had a declining trend after November 2020. Workers earned just 91% of their PL income in March 2021 (after a year of the COVID-19 outbreak).³ Nevertheless, the vulnerable sections were affected much more and recovery was slower for them (the recovery is defined as income earned in a phase/month as a percentage share of PL income).

The higher average income in February 2021 was largely the result of relaxation in COVID-19 restriction and the positive sentiments owing to falling COVID-19-positive cases. Due to declining COVID-19 cases, the Government of Punjab started relaxing restrictions on gatherings and economic activities in January 2021. The maximum number of people allowed in gatherings were increased from 100 to 200 for indoor events and 250 to 500 for outdoor events and the night restrictions were lifted from 1 January 2021. The Government of Punjab further relaxed the restrictions in their order dated 28 January 2021. It led to positive sentiments regarding recovery which, in turn, resulted in higher spending. Also, January and February are months in which maximum marriages happen. The slow down on COVID-19 numbers and relaxation in restrictions had led to higher expenditure and improved income in February. The restrictions, however, were reimposed in March 2021.

Table 2 shows that workers were earning significantly less compared with their PL income till October 2020 (t test is used to test the null hypothesis that difference in income in a phase/month and PL income is equal to zero). The difference turned insignificant in November 2020. With the reversal of trend, workers again earned significantly less than their PL income in January and March 2021.

Table 2.

Difference in income in a phase/month and pre-lockdown income (in Rs.) (null hypothesis: difference = 0).

	P-1	P-2	P-3	P-4	Jul	Aug	Sep	Oct	Nov	Dec	Jan	Feb	Mar
Sex
Male	−502*** (44.4)	−376*** (54.0)	−350*** (45.3)	−299*** (32.9)	−114*** (22.9)	−81*** (23.65)	−43* (25.5)	−33 (23.1)	4 (25.0)	−17 (25.2)	−52** (24.1)	26 (27.5)	−52* (26.4)
Female	−309*** (65.6)	−280*** (69.7)	−277*** (70.6)	−277*** (69.1)	−15 (14.2)	−29 (26.7)	−71 (56.1)	−99 (64.8)	−24 (46.7)	−42 (44.2)	−56 (36.5)	−34 (47.1)	−95* (44.4)
Caste
SC	−421*** (51.2)	−325*** (45.2)	−318*** (44.0)	−237*** (36.3)	−109*** (30.8)	−99*** (31.0)	−72** (30.6)	−46 (28.3)	−7 (30.8)	−23 (30.1)	−33 (25.0)	38 (31.6)	−25 (28.8)
OBC	−445*** (48.8)	−444*** (48.9)	−401*** (52.4)	−316*** (36.7)	−53 (33.1)	−21 (33.5)	3 (32.3)	8 (30.6)	34 (33.0)	19 (36.8)	−17 (41.5)	34 (46.7)	−29 (40.5)
Others	−542*** (86.7)	−327*** (111.3)	−307*** (89.0)	−326*** (65.1)	−132*** (37.6)	−98** (39.1)	−68 (48.1)	−80* (45.1)	−21 (46.0)	−52 (44.1)	−96** (38.7)	−12 (45.4)	−108** (45.6)
Education
Below higher secondary	−482*** (65.3)	−471*** (66.7)	−418*** (65.6)	−265*** (37.1)	−126*** (31.6)	−117*** (35.1)	−102*** (32.4)	−75*** (24.3)	−61** (27.2)	−91** (32.7)	−75** (28.8)	−15 (33.7)	−80** (29.9)
Higher secondary	−503*** (50.4)	−382*** (64.7)	−322*** (52.7)	−298*** (31.7)	−120*** (34.5)	−85** (33.7)	−65* (35.7)	−55 (34.5)	−7 (37.7)	−8 (37.3)	−29 (35.5)	42 (38.2)	−34 (37.7)
Graduate & above	−439*** (86.1)	−277*** (99.2)	−314*** (84.5)	−313*** (70.8)	−62* (33.5)	−36 (35.8)	6 (43.5)	−7 (42.3)	45 (39.5)	6 (38.9)	−66* (38.6)	8 (47.6)	−74* (44.7)
Occupation
Self-employed	−609*** (61.7)	−408*** (83.4)	−389*** (68.0)	−344*** (46.6)	−140*** (32.2)	−92*** (31.7)	−66* (38.2)	−60 (36.5)	−6 (37.4)	−37 (37.1)	−77** (32.2)	18 (36.5)	−64* (36.8)
Regular salaried	−239*** (59.6)	−211*** (62.5)	−220*** (60.4)	−224*** (60.0)	−8 (40.1)	2 (46.6)	38 (41.8)	31 (40.9)	56 (45.2)	56 (45.5)	8 (58.1)	71 (67.4)	−13 (56.9)
Casual worker	−411*** (41.3)	−406*** (42.9)	−348*** (47.4)	−270*** (41.5)	−95*** (21.7)	−99*** (29.2)	−75** (27.0)	−63*** (21.7)	−29 (23.2)	−46* (22.4)	−50** (21.6)	−25 (29.0)	−82*** (28.7)
Loss of income till P-3
Less than 50%	−409*** (90.2)	−96 (100.4)	−59 (63.5)	−151*** (49.4)	−125*** (34.9)	−83** (36.1)	−27 (39.6)	−41 (36.0)	7 (39.0)	−27 (38.0)	−35 (41.5)	62 (45.9)	−42 (46.3)
More than or equal to 50%	−511*** (37.6)	−506*** (37.0)	−491*** (37.7)	−374*** (32.8)	−87*** (24.8)	−69*** (25.6)	−58** (28.7)	−42 (27.6)	−3 (27.7)	−17 (28.2)	−61** (24.3)	−6 (28.5)	−67** (26.5)
ALL	−475*** (40.0)	−363*** (47.6)	−340*** (40.2)	−296*** (29.8)	−100*** (20.2)	−74*** (20.8)	−47** (23.2)	−42* (22.5)	0 (22.5)	−21 (22.5)	−52** (21.3)	17 (24.6)	−58** (23.5)

Source: Primary survey.

Figures in parentheses are standard errors.

SC: scheduled caste; OBC: other backward castes.

, **, and *** represent significance level at 10%, 5% and 1%, respectively.

The vulnerable sections have undergone adverse consequence for a longer period. Workers with below higher secondary education earned significantly less in each period except for February 2021. For casual workers, the difference was insignificant just for 2 months, November 2020 and February 2021. SC seems to have done better in terms of the difference being insignificant after October 2020 (even though it is negative). The difference, despite being negative, is insignificant from July 2020 onwards for women and becomes significant only in March 2021. However, in the case of women, the difference is insignificant largely because of high standard error, indicating large differences in recovery within women workers. One surprising finding is that workers with graduate or above education had suffered higher losses during the second decline. They earned significantly lower income in January and March 2021. In comparison, workers with higher secondary education did much better. Although they also experienced a fall in income after November 2020, they were the least affected.

The trends and patterns of income recovery suggest that there are multiple factors associated with vulnerability. Each of these factors may add to the vulnerability of a worker. For example, a casual worker is likely to experience a high fall in income and slower recovery. This likelihood increases if he or she is an SC and has less than higher secondary education.

This conclusion is supported by difference in income recovery within a category. Box plots of income recovery show large variations in recovery within and across categories (Figure 1). A few workers were earning much higher than their PL income (nearly 300%). However, a majority was struggling to get back to their PL income level and a few were stuck at the bottom with no or negligible recovery. More than 50% of workers in almost all categories could not return to PL level even in relatively good months like November 2020 and February 2021. Median income recovery was low for vulnerable categories (as found earlier). This uneven recovery is the result of differential impact of the pandemic and lockdown restrictions. The sign of its continuity is worrying.

Figure 1.

Box plot of workers’ income as a share of pre-lockdown income (in percentage).

There are multiple reasons for the differential recovery. The pandemic and lockdown hurt income of the households and increased uncertainty. Households responded to the decrease in income and increase of uncertainty by cutting down or delaying expenditure on the non-essential goods and services. It often meant postponing expenditure on activities such as construction. These activities largely engage casual labour. Since SC and OBC were more likely to work in these occupations, the income decline was higher for them. Historical disadvantage of underprivileged sections in terms of social and economic capital is another reason for the differential recovery (see Ambagudia, 2022). The lack of social and economic capital had limited the ability of deprived sections to easily move to other occupations. Many in Others category experienced significant income decline too. However, their relatively better financial position eased their mobility. For example, son of one of the respondents had a mobile phones shop. Though he earned well during the lockdown, his income dropped after July. When his father saw no sign of income recovery, he expanded his own business and asked his son to join him. Similarly, a contractor and his son merged their businesses and let some of their workers go. Since there is no employment scheme similar to MGNREGS in urban areas, the workers from deprived section had few alternatives to their existing work.

The restriction on marriages and other gatherings had not only hurt casual workers like waiters, but also boutiques and other female dominant occupations. Though the income of these female workers witnessed sharp increase after lifting of the lockdown in July 2020, the demand did not sustain and their income started declining. The decline in expenditure on non-essential items and restrictions on large gatherings kept hurting their income. The female workers did experience improvement in their income during the month of Diwali, but it did not sustain beyond November 2020.

Determinants of income recovery

Analysis in the previous section shows differential impact of the pandemic and lockdown. There are indications that convergence of average income to their PL level is slow for vulnerable sections. Also, income recovery seems to be associated with multiple factors. The present section uses the econometric model (discussed in ‘Data and methodology’ section) to examine the statistical significance of various socio-economic factors and confirm trends and patterns found in the previous section. Regression results are presented in Table 3. Model 1 is based on equation (1) discussed in ‘Data and methodology’ section. Other models are given for comparison and to check validity of the results of model 1. Our discussion is based on model 1. Results of other models do not differ considerably from it barring dummy variable for education after dropping slope dummy variables. We have used model 4 and model 5 to compare these results.

Table 3.

Regression results.

Explanatory variables	Dependent variable: Income as percentage share of PL income
Explanatory variables	Model 1	Model 2	Model 3	Model 4	Model 5
Log of PL income per day (in Rs.)	−7.454*** (2.655)	−7.728*** (2.56)	0.141 (2.811)	−7.454*** (2.642)	−7.454*** (2.71)
Sex (Base category: “Male”)
Female	−14.273** (6.442)	−12.363* (6.388)	−17.817** (8.123)	−11.47* (6.529)	−13.631*** (4.274)
Female × Time	0.092 (0.981)	−0.309 (0.995)	0.494 (1.097)	−0.309 (1.001)
Education (Base category: “Below Higher Secondary”)
Higher secondary	−4.584 (4.655)			7.347*** (2.285)	7.347*** (2.306)
Graduate & above	12.431* (6.421)			5.351 (3.554)	5.351 (3.623)
Higher secondary × Time	1.705*** (0.615)
Graduate & above × Time	−1.011 (0.847)
Loss of income till P-3 (Base Category: “Less than 50%”)
More than or equal to 50%	−62.32*** (5.456)	−61.587*** (5.411)		−60.9*** (5.431)	−24.233*** (2.541)
More than or equal to 50% × Time	5.441*** (0.692)	5.238*** (0.679)		5.238*** (0.681)
Caste (Base category: “Others”)
SC	−14.575*** (5.444)	−18.645*** (4.999)	−21.419*** (6.19)	−16.237*** (5.162)	−4.389 (2.952)
OBC	−11.777** (5.085)	−14.734*** (5.054)	−33.204*** (6.33)	−13.945*** (5.038)	0.759 (2.812)
SC × Time	1.455** (0.661)	1.693*** (0.627)	2.224*** (0.719)	1.693*** (0.632)
OBC × Time	1.791*** (0.649)	2.101*** (0.646)	3.925*** (0.754)	2.101*** (0.642)
Occupation (Base category: “Self-employed”)
Regular salaried	2.992 (7.587)	10.305 (6.659)	26.515*** (8.038)	9.934 (7.127)	3.788 (4.883)
Casual worker	−7.014* (4.029)	−8.504** (4.087)	−12.9*** (4.753)	−7.862* (4.081)	−10.878*** (2.627)
Casual worker × Time	−0.552 (0.552)	−0.431 (0.573)	0.252 (0.616)	−0.431 (0.57)
Regular salaried × Time	0.114 (1.09)	−0.878 (0.984)	−2.173** (1.054)	−0.878 (0.983)
Time period (Base category: “P-1”)
P-2	7.443 (4.589)	7.974* (4.594)	10.659* (5.974)	7.974* (4.629)	12.244** (5.623)
P-3	7.702* (4.527)	8.765** (4.353)	14.136** (5.799)	8.765** (4.4)	17.306*** (5.104)
P-4	12.743** (5.086)	14.338*** (4.749)	22.393*** (5.38)	14.338** (4.814)	27.148*** (4.608)
July 2020	50.086*** (5.99)	52.213*** (5.591)	62.953*** (5.324)	52.213*** (5.613)	69.293*** (4.753)
August 2020	46.868*** (6.637)	49.526*** (6.136)	62.951*** (5.841)	49.526*** (6.142)	70.876*** (5.063)
September 2020	46.411*** (7.132)	49.6*** (6.407)	65.71*** (6.107)	49.6*** (6.421)	75.22*** (5.082)
October 2020	41.903*** (7.709)	45.623*** (6.811)	64.419*** (6.43)	45.623*** (6.83)	75.514*** (5.193)
November 2020	44.917*** (8.421)	49.169*** (7.334)	70.65*** (6.93)	49.169*** (7.346)	83.33*** (5.453)
December 2020	37.747*** (9.013)	42.531*** (7.697)	66.697*** (7.29)	42.531*** (7.707)	80.962*** (5.502)
January 2021	26.002*** (9.192)	31.317*** (7.595)	58.168*** (7.018)	31.317*** (7.609)	74.018*** (4.947)
February 2021	31.974*** (9.981)	37.821*** (8.021)	67.357*** (7.47)	37.821*** (8.045)	84.792*** (5.389)
March 2021	14.987 (10.573)	21.366*** (8.452)	53.587*** (7.586)	21.366** (8.476)	72.607*** (5.065)
Constant	124.831*** (25.112)	130.562*** (23.716)	32.975 (25.074)	121.642*** (24.358)	96.022*** (23.997)
Number of observations	1027	1027	1027	1027	1027
F-statistics	85.5***	102.52***	61.23***	93.24***	79.24***
R ²	0.5218	0.5114	0.4336	0.5142	0.4627

Source: Primary survey.

Figures in parentheses are robust standard errors (HC3) due to heteroskedasticity problem.

PL: pre-lockdown; SC: scheduled caste; OBC: other backward castes.

, **, and *** represent significance level at 10%, 5% and 1%, respectively.

Regression results show a significant negative effect of log of PL income on recovery. A unit increase in the log of PL income is associated with a 7.454 percentage point decline in recovery. It is understandable, as fall in income in the initial phases was lower for higher PL income. Hence, they needed a much slower pace of recovery to reach PL income. One may call it a base effect. Since the model controls for high loss in the initial phase (more than 50% till P-3), the negative coefficient of log of PL income is a positive sign.

The coefficients for SC and OBC dummy variables are negative (−14.575 and −11.777, respectively), whereas the coefficients of their slope dummy variables are positive. All coefficients are statistically significant. The results suggest a higher impact of the pandemic on SC and OBC groups, in the initial phases compared with the ‘Others’ category, but convergence in income recovery over the year. Lockdown restriction seems to have negatively affected occupations too (as shown by proxy variable, loss of more than or equal to 50%). Nonetheless, the slope dummy indicates a tendency towards narrowing the gap.

The situation is a little different when it comes to sex and occupation, as the results do not show any convergence. Female workers had a 14.273 percentage point lower recovery of income than male workers. However, the slope dummy for the females is insignificant, indicating that there is no convergence between female and male recovery over time. Results are similar for occupations. Casual workers had nearly 7% lower income recovery than self-employed with little indication of closing the gap. The slope coefficient is not only insignificant but negative. Hence, if we are committing a type II error, the matter is even worse. Regular salaried workers had done better compared with self-employed. An insignificant slope coefficient means that regular salaried workers continue to do better than self-employed workers.

Coefficients for education in model 1 present a different picture than findings of the previous section. Regression results show that higher secondary educated have suffered a higher setback as compared with below higher secondary (intercept dummy is negative and significant), whereas workers with graduate and above education did significantly better than base category. Slope dummies show the filling of the gap over the year. However, the dropping of slope dummies changes the results (model 4 and model 5). Workers with higher secondary education are significantly better than below higher secondary educated. On the contrary, there is no significant difference between the graduate and above educated and those with below higher secondary education. Findings of the previous section suggest that it is likely due to swift income recovery of higher secondary educated (Table 1). Their recovery percentage surpassed even graduate and above, after November 2020. While the graduate and above experienced a smaller income decline than the other two categories, they experienced faster deterioration of their income compared with the other two groups since November 2020. In comparison, workers, with below higher secondary education, were not only most affected but continued to struggle during the period covered by the study.

These results, along with some positive trends, suggest worrying aspects. The findings confirm differential impact of the pandemic and lockdown. Positive trends include declining gaps in income recovery among caste groups and the waning impact of lockdown restrictions. However, the differential impact shows no signs of fading for females and casual workers. They continue to show lag in recovery. Less-educated workers keep doing worse too. If the situation persists, then it is likely to widen inequalities. The situation is especially troubling because these three groups were already earning less than their counterparts in other groups. In addition, many of them fall in multiple vulnerable categories.

Conclusion

The paper studies differential impact of the pandemic on earning of the workers. It examines income changes (with respect to PL income) of workers from March 2020 to April 2021. The workers witnessed a sharp drop in their income during the lockdown. The average income did improve after the lockdown; however, it remained below the PL level till November 2020. The improvement in income in the month of November was owing to an increase in household expenditure during the festival of Diwali. The average income again started to decline in December 2020. Income recovered to the PL level in February 2021, only to start declining in March 2021 due to surge in the COVID-19 cases. The income trends and regression results show that vulnerable sections, women, SC, casual workers, less educated and workers whose occupation was under the restricted category were disproportionately affected due to the pandemic. Workers in the SC category and those whose occupation was under the restricted category had shown a tendency towards narrowing of the gap in income recovery. However, the initial differential impact seems to be enduring for women, less educated and casual workers as they continue to lag in income recovery. This differential recovery was largely a consequence of occupational segregation and difference in social and economic capital among different socio-economic categories. Since these categories were already earning less, the differential effect on income recovery may add to their vulnerability. Also, the possibility that a worker may belong to multiple vulnerable categories makes differential impact with lack of convergence in recovery highly worrying.

The study brings forth vulnerabilities among the disadvantaged sections during the COVID-19 pandemic. The differential impact requires the urgent attention of policymakers. There is a need for short- and long-term policy measures to tackle the vulnerabilities and differential outcome. The short-term policy must focus on categories that are unable to cope with the economic shock of the pandemic. The policy should be such that it can be designed and implemented in a shorter period. Since vulnerable groups may be constrained by different factors, policymakers may need to use multiple strategies to combat the problem. Extending Mahatma Gandhi Employment Guarantee Scheme to urban areas is one option. As shown by Afridi et al. (2022), an employment guarantee scheme could help vulnerable sections by preventing employment loss. More affected occupations may be helped through investment schemes. Focusing on efficient delivery of the schemes and fixing the leakages and delays could further lower job losses. However, the analysis suggests higher vulnerability among the disadvantaged sections. As Ambagudia (2022) argued, the vulnerabilities among the disadvantaged sections are a consequence of historical deprivations and inadequate government support. This raises the need for due long-term policy intervention to generally uplift the well-being of these sections through investment in education and health and creating better employment opportunities.

This study shows slower income recovery and the differential impact of the pandemic on the disadvantageous sections. The aspect about what helped in terms of fewer job losses and faster recovery still remains unanswered. While Afridi et al. (2022) showed that MGNREGA reduced job losses, the effect on the recovery of factors such as development level, education, nature of the economic activity, and landlessness remains unexplored. Little is known whether the government policies to combat the economic impact of the pandemic had any effect. Future research may attempt to answer these questions.

Footnotes

Appendix 1

Table 4.

Average income per day of the workers (in Rs.).

	PL	P-1	P-2	P-3	P-4	Jul 20	Aug 20	Sep 20	Oct 20	Nov 20	Dec 20	Jan 21	Feb 21	Mar 21
Sex
Male	692	191	316	343	393	578	611	649	659	697	675	641	718	640
Female	476	167	196	199	198	460	447	405	377	452	434	419	442	381
Caste
SC	493	72	168	175	256	385	394	422	448	486	470	460	531	468
OBC	527	82	83	126	211	474	506	529	535	561	546	510	561	498
Others	903	362	577	596	577	771	805	835	824	882	852	807	891	795
Education
Below higher secondary	494	13	23	76	229	368	377	392	419	433	403	419	479	415
Higher secondary	630	128	249	308	333	510	546	565	575	623	622	601	672	596
Graduate & above	799	360	522	485	486	736	763	805	792	844	805	733	807	725
Occupation
Self-employed	802	193	394	413	458	661	710	735	742	795	764	724	819	737
Regular salaried	629	391	418	409	405	621	631	667	660	685	685	637	700	616
Casual worker	417	7	12	69	148	323	318	342	354	389	371	368	393	335
Loss of income till P-3
Less than 50%	917	507	821	858	766	791	834	890	876	923	890	881	978	874
More than or equal to 50%	526	15	19	35	151	439	456	468	483	523	508	464	519	459
ALL	662	187	300	323	366	562	589	615	621	663	642	610	680	604
Coefficient of variation	0.70	2.02	1.89	1.60	1.23	0.74	0.78	0.80	0.78	0.73	0.71	0.75	0.75	0.77

Source: Primary survey.

PL: pre-lockdown; SC: scheduled caste; OBC: other backward castes.

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

ORCID iD

Indervir Singh

Notes

Author biographies

Indervir Singh is an Assistant Professor at the Department of Economics, Central University of Himachal Pradesh, Dharamshala. He completed his PhD at the Centre for Development Studies, Thiruvananthapuram (affiliated with Jawaharlal Nehru University, New Delhi). His doctoral thesis examines the role of law and social norms in contract enforcement in informal credit markets. His areas of research interest are law & economics and development economics. His recent work seeks to understand the labour market implications of coronavirus disease 2019 (COVID-19).

Jagdeep Singh is an Assistant Professor at the Guru Nanak College, Budladha. He has 10 years of teaching experience. He holds a PhD in Economics from Barkatullah University, Bhopal. His PhD thesis examines the impact of land size on agricultural productivity and profitability in Punjab. His research interests include agricultural economics, livelihood and labour economics. He has more than 20 research papers and 3 books to his credit.

Ashapurna Baruah is an Assistant Professor in the area of Economics & Public Policy at the Indian Institute of Management Raipur. She holds a PhD in Economics from the Jawaharlal Nehru University (JNU), New Delhi. Her doctoral thesis is an enquiry into how the changes in the rural economy affect labour market outcomes among marginalised women. Her research interests are gender & labour, employment, and rural development in particular and development economics in general. Her work has been published in reputed journals like Economic & Political Weekly and International Labour Review.

References

Afridi

Dhillon

Roy

(2020a) How has Covid-19 crisis affected the urban poor? Findings from a phone survey – I. Ideas for India, 24 April. Available at: https://www.ideasforindia.in/topics/poverty-inequality/how-has-covid-19-crisis-affected-the-urban-poor-findings-from-a-phone-survey.html

Afridi

Dhillon

Roy

(2020b) How has Covid-19 crisis affected urban poor? Findings from a phone survey – II. Ideas for India, 11 May. Available at: https://www.ideasforindia.in/topics/poverty-inequality/how-has-covid-19-crisis-affected-urban-poor-findings-from-a-phone-survey-ii.html

Afridi

Mahajan

Sangwan

(2022) Employment guaranteed? Social protection during a pandemic. Oxford Open Economics 1: odab003.

Agoramoorthy

Hsu

(2021) How the coronavirus lockdown impacts the impoverished in India. Journal of Racial and Ethnic Health Disparities 8(1): 1–6.

Ambagudia

(2022) Tribes, Covid-19 and the State in India. Journal of Asian and African Studies. Epub ahead of print 4 August. DOI: 10.1177/00219096221117073.

Azeez

AEP

Negi

Rani

AAPSK

(2021) The impact of COVID-19 on migrant women workers in India. Eurasian Geography and Economics 62(1): 93–112.

Bertrand

Krishnan

Schofield

(2020) How Are Indian Households Coping Under the COVID-19 Lockdown? 8 Key Findings 198. Becker Friedman Institute for Economics, University of Chicago, 11 May. Available at: https://www.chicagobooth.edu/research/rustandy/blog/2020/how-are-indian-households-coping-under-the-covid19-lockdown

Breman

(2020) The pandemic in India and its impact on footloose labour. Indian Journal of Labour Economics 63(4): 901–919.

Buffel

Yarker

Phillipson

, et al. (2021) Locked down by inequality: older people and the COVID-19 pandemic. Urban Studies. Epub ahead of print 6 September. DOI: 10.1177/00420980211041018.

10.

Centre for Equity Studies, Delhi Research Group and Karwan-e-Mohabbat (2020) Labouring Lives: Hunger, Precarity and Despair Amid Lockdown. Centre for Equity Studies. Available at: http://centreforequitystudies.org/wp-content/uploads/2020/06/Labouring-Lives-_Final-Report.pdf

11.

Chauhan

(2021) Gendering COVID-19: impact of the pandemic on women’s burden of unpaid work in India. Gender Issues 38(4): 395–419.

12.

Chee

(2020) COVID-19 pandemic: the lived experiences of older adults in aged care homes. Millennial Asia 11(3): 299–317.

13.

Chen

Grapsa

Ismail

, et al. (2022) COVID-19 and informal work: evidence from 11 cities. International Labour Review 161: 29–58.

14.

Deshpande

(2020) Early effects of lockdown in India: gender gaps in job losses and domestic work. Indian Journal of Labour Economics 63: 87–90.

15.

GoI (2021) Provisional Estimates of Annual National Income 2020-21 and Quarterly Estimates of Gross Domestic Product for the Fourth Quarter of 2020-21. Ministry of Statistics & Programme Implementation, The Government of India. Available at: https://static.pib.gov.in/WriteReadData/specificdocs/documents/2021/may/doc202153111.pdf

16.

Gopinath

(2020) The great lockdown: worst economic downturn since the great depression. IMF Blog, 15 April. Available at: https://blogs.imf.org/2020/04/14/the-great-lockdown-worst-economic-downturn-since-the-great-depression/

17.

Gupta

Zhu

Doan

, et al. (2021a) Economic impacts of the COVID−19 lockdown in a remittance-dependent region. American Journal of Agricultural Economics 103(2): 466–485.

18.

Gupta

Malani

Woda

(2021b) Explaining the Income and Consumption Effects of COVID in India (No. 28935; NBER Working Paper). Cambridge, MA: National Bureau of Economic Research.

19.

Hale

Webster

Petherick

, et al. (2020) Oxford COVID-19 Government Response Tracker. Blavatnik School of Government. Available at: https://www.bsg.ox.ac.uk/research/research-projects/coronavirus-government-response-tracker

20.

International Labour Organization (ILO) (2021) ILO Monitor: COVID-19 And the World of Work. 7th ed. ILO. Available at: http://www.ilo.org/wcmsp5/groups/public/—dgreports/—dcomm/documents/briefingnote/wcms_767028.pdf

21.

Kumar

Udah

Francis

, et al. (2021) Indian migrant workers’ experience during the COVID-19 pandemic nationwide lockdown. Journal of Asian and African Studies 57(5): 911–931. https://doi.org/10.1177/00219096211046278

22.

Laborde

Martin

Swinnen

, et al. (2020) COVID-19 risks to global food security. Science 369(6503): 500–502.

23.

Lakner

Gerszon Mahler

Negre

, et al. (2020) How Much Does Reducing Inequality Matter for Global Poverty? World Bank Group. Available at: http://iresearch.worldbank.org/PovcalNet/

24.

Lakner

Yonzan

Mahler

, et al. (2021) Updated estimates of the impact of COVID-19 on global poverty: looking back at 2020 and the outlook for 2021. World Bank Blog, 11 January. Available at: https://blogs.worldbank.org/opendata/updated-estimates-impact-covid-19-global-poverty-looking-back-2020-and-outlook-2021

25.

Lee

Sahai

Baylis

, et al. (2020) Job loss and behavioral change: the unprecedented effects of the India lockdown in Delhi. Working Paper No. 2020-65. Becker Friedman Institute for Economics, University of Chicago. Available at: https://bfi.uchicago.edu/working-paper/job-loss-and-behavioral-change-the-unprecedented-effects-of-the-india-lockdown-in-delhi/

26.

Maiti

Sharma

Gupta

(2022) Unfolding the impact of COVID-19 on reverse migrants in Uttar Pradesh. Journal of Asian and African Studies. Epub ahead of print 22 June. DOI: 10.1177/00219096221106080.

27.

Mansoor

Khan

(2021) On employment, wage and consumption: impact of nationwide lockdown on Muslim casual workers. The Indian Economic Journal 69(3): 501–519.

28.

Mitra

Singh

(2021) The COVID-19 pandemic and livelihood loss: lessons for the future. Economic and Political Weekly 56(24): 51–60. Available at: https://www.epw.in/journal/2021/24/special-articles/covid-19-pandemic-and-livelihood-loss.html

29.

Nag

(2020) Robust Diwali shopping is another sign that Indian economy is reviving. The Print, 13 November. Available at: https://theprint.in/economy/robust-diwali-shopping-is-another-sign-that-indian-economy-is-reviving/543226/

30.

Nayar

(2020) The long walk: migrant workers and extreme mobility in the age of corona. Journal of Extreme Anthropology 4(1): E1–E6.

31.

Ray

Subramanian

(2020) India’s lockdown: an interim report. No. 27282; NBER Working Paper. Cambridge, MA: National Bureau of Economic Research.

32.

Singh

Baruah

(2020) Income and employment changes under COVID-19 lockdown: a study of urban Punjab. Millennial Asia 11(3): 391–412.

33.

Singh

Baruah

(2021) Differential impact of COVID-19 by occupation: survey findings from Indian Punjab. Asian Economics Letters 2(4): 1–5.

34.

Somanchi

(2021) Missing the Poor, Big Time: A Critical Assessment of the Consumer Pyramids Household Survey. SocArXiv. https://doi.org/10.31235/OSF.IO/QMCE9

35.

Tleubayev

Kozhakhmet

(2022) Short and medium-term impact of the COVID-19 pandemic on the work outcomes of men and women: empirical evidence from Central Asia. Journal of Asian and African Studies 57(5): 932–944.

36.

United Nations Educational, Scientific and Cultural Organization (UNESCO) (2020) Literacy Teaching and Learning in the COVID-19 Crisis and Beyond at Heart of the International Literacy Day. UNESCO, September 14. Available at: https://en.unesco.org/news/literacy-teaching-and-learning-covid-19-crisis-and-beyond-heart-international-literacy-day

37.

Valensisi

(2020) COVID-19 and global poverty: are LDCs being left behind? European Journal of Development Research 32(5): 1535–1557.