Using high-frequency household surveys to describe energy use in rural North India during the COVID-19 pandemic

The study investigates how the COVID-19 pandemic and associated socio-economic shocks affected household energy use in rural India, where transitions to clean fuels (electricity and LPG) remain fragile amid the prevalence of biomass fuels. Prior to the pandemic, India had made substantial progress toward universal electrification (Saubhagya, 2017) and clean-cooking access (PMUY, 2016), with 95% of households connected to grid electricity and widespread access to LPG by 2020. However, pandemic-related mortality, lockdowns, and economic disruptions threatened these gains. The government introduced a three-free-LPG-cylinders scheme for PMUY beneficiaries (April–September 2020) to mitigate hardship. The purpose of this research is to document changes in energy-use patterns and socio-economic conditions during the pandemic in Jharkhand and Bihar using repeated high-frequency phone surveys, and to evaluate the utility of such surveys for capturing dynamic household energy behaviors.

The paper situates its inquiry within emerging evidence that the pandemic increased economic hardship and disrupted markets in low- and middle-income countries, potentially altering household energy use and prompting reversion to polluting fuels. Prior studies have documented energy insecurity and changes in electricity consumption during COVID-19 in various contexts, but less is known about impacts in settings where clean fuel use is tenuous and biomass is dominant, available, and perceived as free. The authors build on this literature by focusing on rural India, providing fine-grained, repeated measures of energy-use changes and linking them to socio-economic hardships and policy responses (such as the three-free-cylinders LPG scheme).

The study comprises two phone-based, repeated survey designs conducted by local enumerators.

Jharkhand panel survey:

• Sampling and timing: Six survey rounds between July 2020 and July 2021 administered to a panel drawn from a representative 2019 baseline of 1,440 rural households; 882 total households participated across rounds, with 600 surveyed each round. 261 households completed all six rounds; 310 completed 4–5 rounds; 107 were surveyed once.
• Participant characteristics (baseline 2019): Household heads predominantly men (86%); 58% literate; 39% no formal education; >50% non-general caste; 45% Below Poverty Line ration cards. Energy status at baseline: 85% grid-electrified (additional 2% micro-grid/solar); 24% used kerosene lamps as primary lighting; two-thirds stacked a polluting cooking fuel with LPG; one-third used only polluting fuels; exclusive LPG use <1%. Between 2019 baseline and COVID-19 surveys, one-third of households without LPG acquired it.
• Survey content and duration: 10–15 minutes focusing on lighting and cooking fuels, costs and accessibility of modern fuels, refill and subsidy timelines, and socio-economic hardship.
• LPG context: Cylinder-recirculation model; households order refills via distributor; pay full market price upfront (approx. INR 960 for 14.2 kg in Jan 2022) with subsidy later deposited.
• Hardship measurement: Specific domains—(1) increased prices/difficulty accessing food; (2) reduced income/hours/other financial help; (3) loss of employment/inability to find work—collapsed into overall hardship indicator for regressions.
• Analysis: Quasi-binomial logistic regressions with household and round fixed effects and household-clustered standard errors relating hardship (binary) to: kerosene use for lighting; any polluting cooking-fuel use; and primary reliance on a polluting cooking fuel. Alternative models assessed domain-specific hardships and included LPG refill costs.
• Implementation: Surveys translated, piloted, and administered via SurveyCTO by trained enumerators calling from home; data processed in R 4.1.2.

Bihar high-frequency survey:

• Sampling and timing: From a database of 38,000 phone numbers, 450 households across eight districts were randomly recruited. Eligibility: primary cooks ≥18 years who used both biomass and clean fuels in the week before the January 2021 baseline. Of these, 203 participants were randomly selected for weekly calls over eight consecutive weeks (January–April 2021).
• Participant characteristics: Nearly all primary cooks female (≈99%); two-thirds of household heads had secondary education or higher; one-quarter general caste; ~60% Other Backward Class; main income sources included day labour and own-farm agriculture; median monthly expenditure INR 8,000.
• Survey content and duration: Baseline ~20 minutes; weekly follow-ups 5–10 minutes. Weekly modules recorded hours cooked in prior four days, fuels used, household energy tasks (cooking staples, non-cooking tasks like water heating, space heating, fodder preparation), and fuels per task. Reasons for fuel choices were elicited specific to the observed stack (pro-LPG vs pro-biomass/anti-LPG sub-themes).
• Implementation and data: High follow-up completion (all eight weekly surveys for nearly all participants); phone-credit compensation (INR 50 baseline; INR 20 each follow-up); data processed in R 4.1.2.

Ethics: Approvals from Johns Hopkins University IRB and Morsel Research and Development IRB (and exemption at Emory for Jharkhand). Oral informed consent obtained.

Outcomes and measures across both surveys captured dynamic fuel stacking/switching, LPG refill access and pricing, subsidy timing, and awareness/use of the three-free-cylinders scheme, contextualized against COVID-19 waves and lockdowns.

• Socio-economic hardship prevalence (Jharkhand): Approximately two-thirds to over three-quarters of households reported pandemic-related hardships, with peaks aligning with COVID-19 waves (rounds 1; 4–5).
• LPG access and costs (Jharkhand): Nearly all households perceived higher LPG refill prices; around half reported longer wait times for subsidy deposits; about one-third reported increased difficulty acquiring refills during round 4 compared to pre-pandemic. Delivery vs pickup varied by round (26–41% delivered; others traveled median 5 km).
• Biomass collection (Jharkhand): 75% collected firewood (not purchased) at baseline; 50–80% reported no change in collection amount, frequency, or difficulty across rounds.
• Lighting (Jharkhand): ~90% primarily used grid electricity across rounds; kerosene use remained common (~50% at various points) but declined later. About 81% reported using multiple lighting fuels. Kerosene usage compared to pre-pandemic: 75% same, 11% more, 16% less in some rounds.
• Cooking fuel changes (Jharkhand): Marked fluctuations; in round 1, three-quarters used polluting fuels as much or less than pre-pandemic; in round 4, about half reported using polluting fuels more, and over half reported using LPG less than pre-pandemic.
• Regression associations (Jharkhand): Reporting any socio-economic hardship (vs none) was associated with: • 1.5× odds of using kerosene for lighting (95% CI: 0.8–2.6), • 2.5× odds of primarily using a polluting cooking fuel (95% CI: 1.4–4.3), • 2.9× odds of any polluting cooking-fuel use (95% CI: 1.5–5.5), controlling for household and round fixed effects. Results robust to including LPG refill costs and similar across hardship domains.
• Awareness and uptake of LPG relief policy (Jharkhand): 74% aware of the three-free-cylinders scheme in the first round, mostly via gas agencies (50%), neighbours (30%), TV (15%); about 90% of those aware had availed at least one free refill early; by study end, >90% were aware and had obtained at least one free refill. Convenience similar to typical refills.
• Bihar fuel stacking and switching: Despite baseline stacking eligibility, exclusive LPG use in prior four days occurred in 35% of observations; exclusive biomass 6%. When stacking, LPG was primary in 83% of instances. Households switched stack category week-to-week about 50% of the time; 99% switched at least once during the eight-week period.
• Cooking tasks and duration (Bihar): Average cooking time ~3.5 hours/day; similar tasks regardless of stack. LPG used more frequently than biomass on average, but hours per day for LPG vs biomass were similar. Staples (rice, lentils, vegetables, roti, tea/snacks) commonly prepared; tea/snacks often LPG-only; meat/fish (less common) often biomass-only.
• Non-cooking tasks (Bihar): Infrequent—bath water heating (~5%), fodder cooking (13%), space heating (7%); typically done with biomass, with occasional LPG use for bath water heating.
• Motivations for fuel choice: • Jharkhand: Liked LPG more during the pandemic due to ease, convenience, speed, cleanliness; some cited improved LPG accessibility. Liking LPG less was overwhelmingly due to high refill costs, prompting biomass reliance; non-use of LPG in prior four days was almost always due to insufficient gas/conserving it (96%). • Bihar: Exclusive LPG users cited preference for LPG (~90%) and sufficient LPG (~30%). Stackers cited the same pro-LPG reasons but used biomass because LPG is too costly to use exclusively; about half noted some tasks are better with biomass (too much LPG required or insufficient LPG power). About 28% preferred biomass regardless, and 25% preferred biomass for outdoor cooking. Exclusive biomass users commonly lacked LPG (≈90%); about one-third cited LPG expense.
• Policy impact context: In Bihar and Jharkhand, average PMUY beneficiary refill consumption (April 2020–March 2021) was 4.7 and 3.5 refills, respectively; the three-free-cylinder scheme could substantially increase effective clean cooking days and reduce expenditures. Nationally, 141 million free refills were delivered to 75 million PMUY beneficiaries, ~INR 82 billion in benefits.

The findings show that pandemic-related socio-economic hardships were associated with increased use of polluting fuels and reduced use of clean fuels for both lighting and cooking in rural Jharkhand, addressing the core question of how shocks affect clean energy transitions. Week-to-week data from Bihar reveal highly dynamic fuel stacking and frequent switching, with exclusive LPG use common when LPG was available and affordable. Households prefer LPG for its convenience, cleanliness, and speed, but the high cost and occasional access constraints led to biomass supplementation, especially for specific tasks. The widespread awareness and uptake of the three-free-LPG-cylinders scheme highlight that targeted subsidies can buffer against regressions to polluting fuels during shocks. Overall, the study underscores the fragility of recent clean energy gains and the importance of affordability and accessibility in sustaining clean fuel use. Methodologically, it demonstrates that high-frequency phone surveys can capture temporal variability and responses to shocks that would be missed by infrequent surveys, informing responsive policy and program design.

This study contributes timely, granular evidence that COVID-19-related hardships in rural North India increased reliance on polluting fuels and reduced clean fuel use, while a targeted LPG relief policy achieved high awareness and uptake. It demonstrates the value of repeated, high-frequency phone surveys for tracking dynamic energy-use behaviors during shocks and more stable periods. Policy implications include the need to maintain and target subsidies for vulnerable households to ensure LPG affordability and accessibility, thereby strengthening resilience of clean energy transitions. Future research should expand repeated surveys to additional states and populations, evaluate optimal survey frequencies balancing resource demands and information gains, and integrate surveys with in-person validation or sensor-based measurements to improve accuracy.

• Geographic and sampling scope: Limited to selected districts in Jharkhand and Bihar; findings may not generalize beyond these areas. Differences in survey frequency and duration between states were convenience-based, though leveraged to assess feasibility.
• Sample representation: Reliance on mobile phones may underrepresent women and marginalized groups; the sample may be relatively higher income compared to typical biomass-using households.
• Survey instrument: Potential imperfections and translation issues despite pre-testing; some questions identified for revision.
• Data collection constraints: No in-person verification due to COVID-19; self-reported measures may be subject to recall or social desirability bias.
• Pandemic context: Pervasive disruptions to mobility, work, and health may not be fully captured by the survey and may limit comparability to non-pandemic conditions.
• Panel composition: Irregular participation across Jharkhand rounds (600 per round from a pool of 882 households).