ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

An Analysis of Rural Shocks in India during the Pandemic

Utilising the third round of World Bank Survey data collected during September 2020, the employment shocks in rural India are analysed based on the gender, age, caste, and income quartile of individuals. Using graphical representations and simple mean tests, it is found that, on average, females in rural areas did not have an occupation relative to males in most of the states that feature in the survey. Older females and Scheduled Castes, Scheduled Tribes, and Other Backward Classes show a lower probability of retaining regular salaried jobs during the lockdown and were adversely affected owing to disruptions in daily wage work.

The industrial employment in ­India as a per cent of total employment had stagnated around 24% for the entire decade of 2010. Industrial employment in China had reached 30% in 2012 and 2013, but fell to 27.42% in 2019. In terms of the absolute number of job creation and availability, even the slightly higher percentage implies millions of additional workers in the industrial sector. Furthermore, the service sector empl­oy­ment in China has steadily risen to 47.25% in 2019, while that in India has remained around 32.28% in 2019. App­arently, this comparison is not necessary while trying to understand the impact of COVID-19 on the conditions of rural workers either in India or China, which is the main purpose of this article. Yet, a brief background should be useful to ­reflect on where the employment levels in these two growth-wise comparable countries stood just before COVID-19 affe­c­ted economic activities substantially.

A further justification to this genera­lised comparison comes from the fact that none of the developing countries have well-designed unemployment ins­urance schemes, which economic shocks such as those that are brought over by the pandemic exposes as major institutional weaknesses. Agricultural, farm and non-farm employment often provides rural insurance to large masses in developing countries. For India, that too has dro­pped 7 percentage points between 2011 and 2020, while it also fell to 25% for China. But this has been probably more than compensated for by the rise in service sector jobs. Under these circum­stan­ces, countries with higher and div­erse employment-generating capacities shall turn out to be more resilient to economic shocks. Additionally, the pandemic and the lockdown made the informal ­labour force of India, suffering over decades of job insecurity, low earnings, lack of social protection, and dependence on daily wages, even more vulnerable (ILO 2020; Chen 2012; State of Working India 2020). How did India do during the first two phases of the pandemic in terms of jobs and rehabilitation for displaced workers?

The combination of the pandemic lea­ding to various degrees of lockdowns around the world (Hale et al 2021) had disastrous effects on all economies, with no exception to India. Apart from facing the severity of steep mortality rates and breakdown of the health infrastructure att­ending to increasing caseloads, India suffered through catastrophic effects on the rural sector, which, even by the latest Census of Population (2011), hosts 69% of the population. Further, given that 50.4% of the country still lives on less than $3.20 per day based on the 2015 World Bank figures (2019), it is not hard to imagine that individuals who live on meagre monthly wages or belong to the vast majority of daily wage workers, would suffer the hardest owing to the sudden stoppage in economic acti­vities. Consequently, as mentioned by the International Labour Organization (ILO), 400 million people in India face the risk of falling back into poverty (Brookings Report 2020). Among such individuals, migrant workers, who make up 36% of the Indian population (based on the 2011 Census), experienced not only falling wages and job losses but also mass exodus and reverse migration triggered by the sudden announcement of the lockdown (Dandekar and Ghai 2020). However, this had received substantial attention in concerned forums in the subsequent months. We take this evidence forward by analysing age cohort-, gender- and occupation-wise outcomes of lockdown on the large contingent of rural workers of which ­return migrants form an important subsection.

For example, according to the initial ILO estimates, there has been a rise in unemployment and underemployment between 5.3 million (low estimate) and 24.7 million (high estimate) relative to the 2019 numbers. Subsequent analysis by the ILO has revealed that the figures were considerably under­estimated (Walter 2020). As recent papers have suggested, vulnerable groups might have been hit even harder in terms of employment and other factors by the pandemic (Deshpande 2020). Deshpande shows that although, in absolute terms, the number of male job losses is higher than female job losses, due to pre-existing gender employment gaps, the shocks have different implications for the two groups.

In this article, we present a preliminary analysis, in terms of descriptive statistics and data visualisations, of economic shocks experienced due to COVID-19 in six states covering the rural districts. Further, we describe the source and the coverage of the data in detail and offer a descriptive analysis of the rural labour market outcomes.

Data and Descriptive Statistics

 

Data source: We consider survey data collected by World Bank, IDinsight, and the Development Data Lab over May 2020 to September 2020 in India. The objective of the survey is to provide an assessment of changes in agricultural production and consumption induced by the COVID-19 pandemic. It also provides information on employment during lock­down and changes in income (wages) brought on by the pandemic. Additionally, it provides information on migrant workers and access to relief funds. Six states of India, namely Jharkhand, Raja­sthan, Uttar Pradesh (UP), Andhra Pra­desh (mP), Bihar, and Madhya Pradesh (MP), are included in the survey. World Bank (2020) mentions that four previous frames have been considered from IDinsight projects, alongside frames from the National Rural Livelihoods project. Appropriate weighting schemes have been applied in order to be able to create comparable state-level estimates. The surveys were carried out for rural areas in the six states using computer-assisted telephone interviewing (CATI) techniques. To avoid non-responses as much as possible, surveyors attempted to call back respondents up to seven times. A software called SurveyCTO has also been used to check for validation and consistency.

Sample characteristics: We have 5,200 observations for districts across six states. The average age of the sample is about 37.5 years. In terms of household size, five individuals is the average size. We have information for 2,731 observations in terms of the gender of the head of household. About 83% among 2,731 observations have a male as the head of the household. In terms of belonging to specific income quartiles, again, we have information for 2,731 observations. About 20.8% of individuals belong to the lowest quartile.

Employment: Effects by Gender

We start our analysis with the third round of data. In the first round, questions about occupation during the lockdown were not asked. In terms of overlaps, about 38% of the individuals res­ponded in the second and third rounds in which questions related to employment and occupation during the lockdown were posed. In Figure 2, we show the percentage of individuals categorised as females and non-females (males) by state who had no occupation during the lockdown period. The question asked in the survey is if an individual had any occupation during the lockdown or not. We cons­tru­cted a binary variable, which is assig­ned one for responses stating no occupation, 0 otherwise. Zero may imply one of the following: being self-employed in non-cultivation, salaried jobs in government, salaried jobs in private sector, daily wage labour in agriculture or daily wage labour in non-agriculture. Except AP, we find that, for each state, the percentage of females with no occupation is higher than males. The biggest gap is for Raja­sthan where ­almost 70% females did not have any occupation during the lockdown compared to about 52% non-females.

In Figure 2, we consider occupation to be distributed as a binary between salaried jobs in government or private or non-salaried jobs. It shows the percentages of females in each state holding salaried jobs during the lockdown. Not surprisingly, the percentage of employment for any state in salaried jobs is less than 10%. However, we do not find any clear gendered pattern for employment. In some states like MP, Bihar and uP, more females are employed in salaried jobs relative to males. The opposite is true in Jharkhand and Rajasthan. For AP, the percentages are almost similar. Finally, Figure 4 shows the percentage of females by state who are employed in daily wage jobs, distributed between agricultural or non-agricultural activities. Again, based on the figure, we do not find any obvious pattern on employment gap between females and males across states.

Next, we conduct t-tests to investigate if the group means between the groups are significantly different or similar. We consider no occupation, occupation in salaried jobs and occupation as daily wage labourers by gender as the groups. In Table 1, we report the t-statistics, the degrees of freedom and the probability for the alternative hypothesis stating that the means of the groups (females and males) are significantly different from each other. In column (1), we ­report the t-statistics along with the probability for no occupation during lockdown. In column (2), we report the same for salaried jobs, and in column (3), we present the t-statistics along with the probability for holding daily wage occupation. Interestingly, the t-test is negative and significant for column (1). This means that during the lockdown, on average, more fem­ales had no occupation compared to males. However, the t-statistic is not significant for column (2) suggesting that there is no significant difference between the emp­loyment of females and males in salaried jobs. In column (3), the test is positive and significant suggesting that more males were employed in agricultural or non-agricultural daily wage jobs relative to females.

Employment effects by age and gender: Next, we consider if the type of emp­loyment, including no occupation, has a pattern by age. The median age of our sample is 35 years. We construct a dummy assigned 1 if individuals are above the median age. Once again, we conduct t-tests by different employment types (including no occupation) based on the two age groups. In Table 2, we report the ­results. Interestingly, we find that older adults (above the median age) are more likely to not have any occupation relative to individuals below the median age. At the same time, we find that the t-statistics for salaried jobs and daily wage are both positive and significant. This implies that younger adults (individuals below median age) are more likely to have daily wage as well as salaried jobs relative to older adults.

In Figure 4, we present the percentages of individuals having no occupation based on the constructed age dummy, as well as gender. Based on the figure, we cannot identify any meaningful pattern by gender and age. For some states like AP and UP, we find that more females above the median age (35) did not have an occupation during lockdown relative to males. For all the other states, Raja­sthan, MP, Bihar and Jharkhand, we find the opposite to be true—more males above the median age (35) were not employed relative to females. To explore further, we perform t-test of mean difference for the two groups—no occupation during lockdown for females above 35 years and no occupation during lockdown for females below and equal to 35 yearsOur t-test results show that the null hypothesis can be rejected. The t-statistic for the alternative hypothesis stating that the group means are different from each other is significant at the 5% level. The t-value is negative suggesting that females above median age are more likely to not have any occupation relative not only to males, but also fem­ales below the median age. As far as government jobs are concerned, we do not find any significant difference between the group means. In the case of daily wage jobs, the t-statistic is positive and significant implying that females above 35 years of age are more likely to be in daily wage jobs. We do not present a ­table for these sets of results keeping space constraint in mind but they are available on request.

Employment effects by caste and i­ncome: As the final set of our results, we consider employment patterns of indivi­duals by caste and income groups. We construct a dummy assigned 1, if an individual belongs to Scheduled Caste (SC), Scheduled Tribe (ST) and Other Backward Caste (OBC); 0 otherwise. Zero can indicate general caste or other caste. It is well known that the existing literature has documented earning and education differences between general caste and SCs, STs or OBCs (Hoff and Pandey 2006; Borooah and Iyer 2005; Das 2003; Unni 2001; Lakshmanasamy and Madheswaran 1995). We conduct t-tests for all employment groups, including no occ­upation. In the case of no occupation during the lockdown, we do not find a significant difference between the two groups—belonging to general or other caste versus belonging to SC, ST or OBC. But for salaried jobs, we actually find the t-statistics to be positive and significant. The difference is the mean of group 1 (bel­onging to general or other group) and the mean of group 2 (belonging to SC, ST or OBC). The positive sign would imply that individuals belonging to general or other caste had more salaried jobs on average, relative to individuals belonging to SC, ST or OBC. In the case of daily wage workers, we find the opposite to be true.1

Finally, we check if not having an occ­upation during the lockdown, having a salaried job or being a daily wage worker is contingent on belonging to lower- or upper-income quartiles. We construct a dummy assigned 1 if an individual bel­ongs to the upper quartiles—third and fourth; 0 would indicate belonging to the first or the second quartile. We run the t-tests for no occupation, salaried jobs, and daily wage occupation. The t-statistic is not significant in any of the cases.

Concluding Remarks

Employment levels in the agricultural sector in India were falling for some years preceding the onset of the economic crisis owing to the COVID-19 pandemic since a large number of workers in India, whether regular salaried, employed in some public sector jobs, or daily wage earners belong to the rural and semi-­urban landscape in the country. It is also widely acknowledged that a substantial number of semi-skilled and unskilled workers migrate to agriculturally and industrially more active states. During the economic crisis, such as the spread of the pandemic, these workers returned to their places of origin for economic support from their families.

The rural areas were also less affected by the spread of the virus. However, farm and non-farm activities in rural areas are strongly related to economic activities in urban areas. Disruption in the economic activities in cities spill over to rural areas and affects livelihoods. The rural jobs were affected and the aspect that recent studies have not quantified yet accounts for gender gaps in employment owing to the crisis. We have demonstrated that during the lockdown, the probability for having no occupation was significantly higher for females in almost all states, except AP, and this is vindicated by the simple group mean test we conducted. We also subsequently showed that older workers are less likely to retain jobs during the pandemic, which is not surprising, but obviously no less distressful given that social security, except for pockets of central- and state-level supports, is rather non-uniform across the length and breadth of the older population in the country. Furthermore, since females over 35 years of age, which we find as the median age in our sample, are more likely to be in daily wage jobs, disruptions in economic acti­vities through full or partial lockdown are borne disproportionately by older women. Workers belonging to SC, ST and OBC classifications were also adversely affected as compared to general caste wor­kers with salaried jobs, while the reverse was true for daily wage work.

Overall, the lockdown and economic disruptions have been responsible for considerable hardships among the rural population in India. It has also led to non-uniform effects across gender, caste and age groups, which we used to break down the general observations. This short and descriptive article has been able to further identify the groups that have been the most adversely affected due to the pandemic and qualify for econo­mic and social support.

Note

1 We do not present the t-values but they are available on request.

References

Borooah, V and S Iyer (2005): “Vidya, Veda and Varna: The Influence of Religion and Caste on Education in Rural India,” Journal of Development Studies, Vol 41, No 8, pp 1369–1404.

Dandekar, A and R Ghai (2020): “Migration and Reverse Migration in the Age of COVID-19,” Economic & Political Weekly, Vol 55, No 19, pp 28–31.

Das, M B (2003): “Ethnicity and Social Exclusion in Job Outcomes in India: Summary of Research Findings,” Unpublished paper.

Deshpande, A (2020): “The Covid-19 Lockdown in India: Gender and Caste Dimensions of the First Job Losses,” Working Papers 30, Ashoka University, Department of Economics, New Delhi, India.

Hale T, N Angrist, R Goldszmidt et al (2021): “A Global Panel Database of Pandemic Policies (Oxford COVID-19 Government Response Tracker),” Nature and Human Behaviour, Published online 8 March, pmid: 33686204.

Hoff, K and P Pande (2006): “Discrimination, Social Identity, and Durable Inequalities,” American Economic Review, Vol 96, No 2, pp 206–11.

ILO (2020): “ILO Monitor: COVID-19 and the World of Work,” International Labour Organization, (Fourth edition) Updated Estimates and Analysis, Geneva: International Labour Organization.

Kesar, S, R Abraham, R Lahoti, P Nath and A Basole (2020): Pandemic, Informality, and Vulnerabi­lity, Impact of COVID-19 on Livelihoods in ­India,” CSE Working Paper, 2020–01https://doi.org/10.3929/ethz-b-000428008.

Lakshmanasamy, T and S Madeshwaran (1995): “Discrimination by Community: Evidence from Indian Scientific and Technical Labor Market,” Indian Journal of Social Sciences, Vol 8, pp 59–77.

Unni, J (2001): Earnings and Education among Ethnic Groups in Rural India, NCAER Working Paper Series 79, National Council of Applied Economic Research.

Walter, D (2020): Implications of Covid-19 for Labour and Employment in IndiaIndian Journal of Labor Economics, Doi:10.1007/s41027-020-00255-0.

World Bank (2019): “Poverty and Equity Brief, South Asia, India,” viewed on 25 February 2022, povertydata.worldbank.org, www.worldbank.org/poverty.

— (2020): “COVID-19-Related Shocks in Rural India 2020,” Rounds 1-3; Version date: 2021-01-12; viewed on August 2021.

 

Dear Reader,

To continue reading, become a subscriber.

Explore our attractive subscription offers.

Click here

Updated On : 18th Jul, 2022
Back to Top