COVID-19 pandemic, losses of livelihoods and uneven recovery in Pune, India

The study investigates the long-term socio-economic impacts of the COVID-19 pandemic on slum dwellers in Pune, India, focusing on losses in employment, income, and consumption expenditures. Prior research documented large immediate shocks in 2020 (job losses, income reductions, food insecurity) and partial macroeconomic recovery by late 2020–2021, with unequal effects exacerbating pre-existing disparities. In India, strict lockdowns in 2020 and a severe second wave in 2021 disrupted livelihoods, especially in urban informal sectors. Official recovery indicators may not capture the informal economy, where vulnerable groups—women, youth, and daily wage workers—were disproportionately affected. This study asks whether individuals living just above subsistence level in urban slums experienced sustained recovery two years after the pandemic’s onset, and how consumption patterns evolved. Using a four-round panel (2018, 2019 pre-pandemic; 2020, 2022 post-pandemic onset), the study aims to provide evidence on persistent welfare impacts and heterogeneity by socio-demographics.

Global and Indian studies report substantial immediate impacts of COVID-19: increased unemployment, reduced incomes, heightened food insecurity, and rising poverty (e.g., Major et al., 2020; Egger et al., 2021). Recovery was uneven, tracking and worsening existing inequalities (Dang and Nguyen, 2021; Reichelt et al., 2021). Indian evidence shows drastic initial employment and income losses among informal and low-income groups (Afridi et al., 2020a,b; CES, 2020; Kesar et al., 2021), with urban areas harder hit than rural (Kang et al., 2021). Some macro indicators and employment rates rebounded by late 2020–2021 (Economic Survey of India 2021–22; APU, 2021), but many remained detached from the labor market and incomes stayed depressed (Bertrand et al., 2020; Lahoti et al., 2021; Abraham et al., 2022). The second wave in 2021 brought renewed disruptions and increased food insecurity (Rains, 2022; Veluguri et al., 2022). Longer-run effects of pandemics and shocks can persist (Jordà et al., 2022; Newhouse, 2005). Few studies extend beyond 2020–2021; Jha and Lahoti (2022) document persistent urban income losses and increased poverty and inequality up to late 2021. This study contributes long-term panel evidence for urban slum residents living just above subsistence.

• Study setting and sample: Slum settlements in Pune, Maharashtra, India. Original sample from an RCT (Dec 2018–Oct 2019) on a savings intervention (Steinert et al., 2022). Participants were adults (18+) with at least some monthly or weekly income (formal/informal work, remittances, or social welfare), representing poor but not necessarily the poorest households. Sampling covered 250+ slum settlements with very high population density and limited services.
• Panel design: Four rounds: Round 1 (2018, in-person baseline, n=1525), Round 2 (2019, in-person endline, n=1421), Round 3 (Oct–Dec 2020, phone, n=851), Round 4 (Feb–Mar 2022, phone, n=513). Phone numbers from R1–R2 enabled R3–R4.
• Attrition: R1→R2 attrition 6.8%. R3 complete data for 851 (56% of R1); R4 complete data for 513 (34% of R1). Balanced panel size n=411. Post-2019 respondents tended to be more educated, own more assets, and be more likely working than R1; the panel has more females, married, educated, and working individuals relative to R1.
• Measures: Employment status (salaried/self-employed/unemployed); self-reported loss of livelihood since the pandemic (full/partial); decreases in earned income; household shortage of financial resources; coping strategies (use savings, borrow, cut overall/food expenses, sell assets). Monthly per-capita expenditures by respondent on food groups (all four rounds) and non-food (R1, R2, R4; not collected in R3). Total consumption expenditure computed where food and non-food available. Expenditures deflated to 2018 real terms using Maharashtra CPI indices (food price index for food; general index for non-food).
• Empirical approach: Descriptive comparisons across rounds; for balanced panel (n=411), fixed-effects models with individual and time fixed effects to assess changes in total, food, and non-food expenditures over time. Logit models estimate the odds of experiencing any loss in livelihood and shortages of household financial resources since the pandemic, using 2019 socio-demographic covariates (age, sex, marital status, education levels, employment type, caste, religion, household size, assets) and a treatment indicator from the original RCT as controls. Standard errors reported; significance indicated in tables.
• Limitations acknowledged include selection due to attrition, non-representativeness beyond the sampled sub-population, and measurement differences between in-person and phone modes.

• Employment and livelihood losses:
  • Not working (balanced panel): 29.5% (2018), 37.5% (2019), 53.5% (2020), 40.1% (2022).
  • Among those employed in 2019, loss of livelihood due to the pandemic: 49.0% complete and 26.4% partial in 2020; 21.8% complete and 35.0% partial in 2022.
  • Decrease in earned income among those employed in 2019: 86.8% (2020) and 60.9% (2022).
  • Household financial shortages since the pandemic: 87.5% (2020) and 75.0% (2022) in the panel (similar figures in the full samples: ~88% and ~72%).
• Vulnerability gradients (logit results): Younger individuals, those with lower education (primary/secondary vs none; higher education), and those not in salaried jobs had significantly higher odds of losing their livelihood. Lower education was also associated with higher odds of household financial shortages. No strong gender effect detected within this sample conditional on 2019 employment.
• Coping strategies (panel): Used savings 70.8% (2020) and 53.8% (2022); borrowed 16.5% (2020) and 21.2% (2022); cut expenses 26.5% (2020) and 15.3% (2022); cut food expenses 19.7% (2020) and 12.2% (2022); sold assets 5.6% (2020) and 3.4% (2022).
• Expenditure dynamics (balanced panel):
  • Total monthly per-capita expenditure was stable pre-pandemic (2018–2019) but fell sharply post-pandemic; authors note a ~50% decline by 2022 compared to pre-pandemic levels.
  • Food expenditure relative to 2018: −10% (2019), −40% (2020), −25% (2022); partial recovery by 2022 but still below pre-pandemic.
  • Non-food expenditure showed a steep decline by 2022.
• Food group expenditures (Rs., balanced panel): Cereals fell in 2020 and did not recover to pre-pandemic levels by 2022; lentils, vegetables, and meat/eggs decreased in 2020 but recovered by 2022; sugar increased post-pandemic vs 2019; milk products rose by 2022; expenditures on unhealthy items (fried snacks, cakes, beverages) declined pre-pandemic and remained below pre-pandemic in 2022.

Findings indicate that macro-level recovery in GDP and unemployment masked persistent and uneven recovery among slum dwellers living just above subsistence. A large share experienced complete or partial livelihood losses in 2020, and more than half still reported losses by early 2022. Income losses translated into reduced expenditures, with substantial and lasting declines in total and non-food spending and only partial recovery in food spending by 2022. Younger and less educated individuals and those without salaried positions were more vulnerable to livelihood losses, reinforcing pre-existing inequalities. Despite national-level evidence of greater challenges for women to regain employment, this study did not detect strong gender disparities within its sample, possibly due to composition and statistical power. Food consumption patterns reveal potential nutritional concerns: while some food groups recovered by 2022, cereals did not return to pre-pandemic levels and expenditures on unhealthy items remained low, suggesting constrained budgets and changing access. Limited asset sales suggest that, so far, households avoided deep erosion of long-term asset bases, but continued financial stress risks future impoverishment and adverse health and productivity outcomes. Overall, the results underscore the protracted and unequal nature of recovery for urban informal and low-income populations.

The study documents sustained and uneven recovery from COVID-19 among slum dwellers in Pune two years after the pandemic’s onset. Despite macroeconomic rebounds, many continued to face livelihood and income losses, with younger, less educated, and non-salaried workers being most affected. Consumption expenditures remained depressed, and food spending had not fully returned to pre-pandemic levels. Policy implications include the need for extended and more substantial social protection: improved targeting and coverage of cash transfers (beyond limited Jan Dhan transfers), leveraging and strengthening existing platforms (NREGA, PDS, ICDS) while addressing exclusion errors, and diversifying food support beyond cereals to promote dietary quality. Longer-term reforms are recommended to enhance social security for informal workers and to mitigate vulnerability to future shocks. Future research should continue tracking vulnerable populations to understand the persistence of impacts and evaluate interventions that can accelerate equitable recovery.

• High attrition over four years with selection concerns; post-2019 respondents are relatively more educated and asset-owning, limiting generalizability.
• Non-representative sample focused on individuals just above subsistence in Pune slums; findings do not generalize to the broader population.
• Observational design precludes causal inference; only two pre-pandemic rounds limit disentangling pandemic effects from secular trends.
• Mode differences between in-person (2018–2019) and phone surveys (2020–2022) may affect measurement reliability and recall accuracy.
• Only respondent-level expenditures were collected (not entire household), and non-food expenditures were not collected in 2020, complicating total expenditure comparisons across all rounds.
• Limited temporal resolution post-2020 (only 2020 and 2022) restricts analysis of the second wave’s timing and recovery trajectory.
• The study examines primarily economic distress (employment, income, expenditure) and does not capture multidimensional distress (e.g., social, psychological) due to phone survey constraints.