Can Clean Cooking Transitions Reshape Time Allocation and Economic Activity? Experimental Evidence from Rural India

The study evaluates alternative policy interventions designed to promote sustained use of liquefied petroleum gas (LPG) among rural households in India that already possess LPG connections but continue to rely partly on traditional biomass fuels such as firewood and dung. The interventions are implemented through a cluster-randomised controlled trial across rural districts in Odisha, Jharkhand, and Telangana.

Villages were randomly assigned into one of four intervention groups or a control group. The first intervention provided a price incentive in the form of a ₹150 subsidy for each 14.2 kg LPG refill cylinder during the intervention period. The subsidy was delivered at the time of refill purchase in order to reduce affordability and liquidity constraints associated with sustained LPG usage.

The second intervention focused on behavioural barriers to regular LPG use. Households received behavioural messaging through SMS reminders and periodic field visits. The messages highlighted the health benefits of LPG use, cooking time savings, refill planning, social norms, and the convenience of sustained LPG usage. Messages were delivered in local languages and were intended to reinforce regular refill behaviour and reduce dependence on biomass fuels.

The third intervention aimed to reduce transaction costs and supply-side barriers related to LPG refills. Households received assistance with refill booking, refill scheduling, and communication with LPG distributors through field facilitators and local coordination support.

The fourth intervention combined all three components simultaneously, including refill subsidies, behavioural messaging, and refill convenience support.

Households assigned to the control group continued their normal LPG purchasing and cooking practices without additional intervention during the study period.

The intervention period lasted approximately six months, from May to October 2025. The study evaluates the effects of these interventions on LPG refill behaviour, fuel stacking, cooking-related time use, and household economic activity.

The study is a multi-arm cluster-randomised controlled trial (RCT) designed to evaluate whether alternative policy interventions can promote sustained use of liquefied petroleum gas (LPG) among rural households in India. The trial focuses on households that already possess LPG connections but continue to rely partly on traditional biomass fuels such as firewood, dung, and crop residue. The primary objective is to examine whether affordability support, behavioural interventions, and convenience-enhancing measures can increase the frequency and consistency of LPG usage and reduce fuel stacking behaviour.

The study was conducted across six rural districts in three Indian states: Kandhamal and Koraput in Odisha, Simdega and West Singhbhum in Jharkhand, and Bhadradri Kothagudem and Mulugu in Telangana. These districts were selected because they exhibit high dependence on biomass fuels while also experiencing expanding LPG access under existing clean cooking programmes. Villages were selected through stratified random sampling based on LPG penetration, infrastructure conditions, accessibility, and feasibility of field implementation. The final sample consisted of 90 villages and approximately 1,800 households, with about 20 participating households per village. Eligible households were those with LPG access but irregular or incomplete LPG usage patterns.

Randomisation was conducted at the village level to minimise spillover effects and information contamination across households. Villages were stratified according to state, district, village size, and baseline LPG usage levels before assignment into experimental groups. The sample was evenly divided across five groups consisting of a control group and four intervention arms, with 18 villages allocated to each arm.

The first intervention arm provided a direct price incentive designed to reduce affordability and liquidity constraints associated with LPG refills. Participating households received a subsidy of ₹150 for each 14.2 kg LPG refill cylinder during the intervention period from May to October 2025. The subsidy was delivered immediately at the time of refill purchase as a point-of-sale cash transfer. Each household was eligible for a maximum of six subsidised refills during the intervention period.

The second intervention arm focused on behavioural barriers to sustained LPG usage. Households received two SMS reminders per week along with one field visit per month. Messages were delivered in local languages including Odia, Hindi, and Telugu. The behavioural content highlighted the health benefits of LPG use, time savings from cleaner cooking, refill planning, social norms, habit formation, women’s time savings, and refill convenience. The intervention was informed by behavioural economics concepts such as salience, reminders, framing, peer effects, and habit reinforcement. Monthly field visits were used to reinforce message delivery using standardised communication scripts.

The third intervention arm aimed to reduce transaction costs and supply-side frictions related to LPG refill access. Households received assistance with refill booking, refill scheduling, communication with LPG distributors, and coordination of refill delivery logistics. Booking assistance was provided through phone calls, WhatsApp communication, and direct support from field facilitators assigned to nearby villages.

The fourth intervention arm combined all intervention components simultaneously. Households in this group received refill subsidies, behavioural messaging, field reinforcement visits, and convenience-enhancing refill support throughout the intervention period. Separate monitoring records were maintained to track implementation and compliance for each component of the intervention package.

Households assigned to the control group continued their existing cooking and LPG purchasing practices without any additional support during the study period. This group provided the counterfactual against which the effects of the intervention arms were evaluated.

Baseline data collection was conducted between March and April 2025 and included information on household demographics, cooking fuel usage, refill behaviour, time allocation, economic activity, and perceived barriers to LPG use. Village-level randomisation was completed at the end of April 2025, after which interventions were implemented between May and October 2025. A midline survey was conducted in November 2025 to capture short-term behavioural responses, while an endline survey was completed in March 2026 to measure sustained impacts on LPG usage, time allocation, and economic outcomes.

The primary outcomes of the study include LPG refill frequency, the share of meals cooked using LPG, and the persistence of fuel stacking behaviour. Additional outcomes include cooking-related time use, time spent collecting biomass fuels, labour-force participation, household enterprise activity, and perceptions related to affordability, convenience, and reliability of LPG usage. Administrative refill records, where available, were also used to validate self-reported measures of LPG consumption.

2025-05-31

2025-10-31

The primary outcome variables of the study relate to sustained clean cooking fuel usage and household time allocation. The main endpoints include LPG refill frequency, measured as the number of LPG refill cylinders purchased during the study period; the share of meals cooked using LPG as the primary cooking fuel; and the persistence of fuel stacking behaviour, defined as continued simultaneous use of LPG and traditional biomass fuels such as firewood or dung.

The study also examines cooking-related time allocation outcomes, including daily time spent on cooking activities, time spent collecting traditional fuels, and total household domestic work time. Additional primary outcomes include labour-force participation, weekly hours worked, and household enterprise activity in order to assess whether reductions in cooking-related time burdens are associated with greater economic engagement.

Sustained LPG usage will be measured using multiple indicators constructed from household survey responses and, where available, administrative refill records from LPG distributors. The primary LPG usage indicators include the number of LPG refills during the intervention period, the percentage share of meals cooked using LPG, and a binary indicator of fuel stacking that captures whether households continue to use biomass fuels alongside LPG.

Cooking-related time allocation outcomes will be constructed using structured household time-use modules. Daily cooking time will be measured as the total number of minutes spent on cooking activities, meal preparation, and fuel management. Time spent collecting traditional fuels will be separately recorded in minutes per day. Total domestic work burden will be calculated as the sum of time spent on cooking, meal preparation, fuel collection, and related household activities.

Economic activity outcomes will be constructed using household-level labour and enterprise information. Labour-force participation will be measured as a binary indicator capturing whether any adult household member engaged in paid work during the reference period. Weekly hours worked will measure the total number of hours spent in wage employment, self-employment, or household enterprise activities. Household enterprise activity will be coded as a binary indicator capturing whether the household operated any income-generating enterprise or self-employment activity.

The study will also construct a women’s autonomy index for heterogeneity analysis. The index will combine indicators related to household decision-making authority, women’s mobility, control over time allocation, and ability to participate independently in paid work. Variables used in constructing the index include decision-making over cooking fuel choice, household expenditure decisions, permission to work outside the home, and control over personal time use. The index will be standardised to range between 0 and 1, with higher values indicating greater autonomy.

Perceived barriers to sustained LPG use will also be measured using Likert-scale responses related to affordability, refill access, supply reliability, and preference for traditional fuels. These variables will be used to examine behavioural and operational mechanisms underlying treatment effects.

Secondary outcome variables include intermediate behavioural and operational indicators related to sustained LPG usage and household perceptions. These include perceived affordability of LPG refills, refill delays, refill booking difficulties, awareness of the health and time-saving benefits of LPG use, and satisfaction with LPG access and refill services.

Additional secondary outcomes include behavioural indicators such as refill planning behaviour, cooking frequency, number of daily cooking events, and meal complexity in order to examine potential rebound effects associated with increased LPG usage. The study also examines treatment compliance indicators, including receipt of subsidy transfers, exposure to behavioural messaging, participation in field visits, and use of assisted refill booking support.

Further secondary outcomes include household perceptions regarding barriers to LPG usage, including affordability constraints, supply reliability, convenience of refill access, and continued preference for traditional biomass fuels. Enumerator observations regarding LPG stove presence, cylinder condition, and kitchen type will also be used as supplementary validation indicators.

Perceived affordability of LPG will be constructed using household responses to Likert-scale questions assessing whether households consider LPG refills financially manageable. Refill delays will be measured as the number of days between refill booking and cylinder delivery, while refill booking difficulty will be measured using ordinal responses capturing the perceived ease or difficulty of arranging LPG refills.

Awareness-related outcomes will be constructed from survey responses measuring household recall of behavioural intervention messages related to health benefits, time savings, cost savings, and social norms associated with LPG use. Satisfaction with LPG access will be measured using Likert-scale responses regarding refill availability, delivery reliability, and overall convenience of LPG services.

Behavioural adjustment and rebound-effect outcomes will be constructed using information on the number of meals prepared daily, the number of cooking events per day, and self-reported changes in meal complexity or cooking intensity following increased LPG usage. These indicators are intended to capture whether reductions in cooking time are partially offset by increased cooking frequency or more elaborate meal preparation.

Treatment compliance indicators will be constructed using household-level records on whether the household received subsidy transfers, behavioural SMS reminders, field visits, or assisted refill support during the intervention period. A composite compliance index will also be created by aggregating exposure to the assigned intervention components.

Perceived barriers to LPG use will be measured using Likert-scale responses related to affordability constraints, refill access difficulties, supply unreliability, and continued attachment to traditional cooking fuels. Enumerator observations regarding LPG stove presence, cylinder condition, and kitchen characteristics will be used as supplementary validation measures to assess consistency between reported and observed LPG usage patterns.

The study uses a cluster-randomised controlled trial (RCT) to examine whether different policy interventions can increase sustained use of LPG for cooking among rural households in India. The trial is conducted across rural villages in Odisha, Jharkhand, and Telangana among households that already have LPG access but continue to rely partly on traditional biomass fuels such as firewood and dung.

Villages are randomly assigned to one of five groups: a control group and four treatment groups. The treatment groups receive different interventions designed to address barriers to regular LPG usage. One group receives a small refill subsidy to reduce affordability constraints. A second group receives behavioural messaging through SMS reminders and field visits highlighting the health, time-saving, and convenience benefits of LPG use. A third group receives assistance with refill booking and delivery coordination to reduce access and transaction barriers. A fourth group receives a combination of all interventions.

Baseline, midline, and endline household surveys are used to collect information on cooking fuel use, LPG refill behaviour, time allocation, household economic activity, and perceptions related to LPG use. The study evaluates whether the interventions increase sustained LPG usage, reduce reliance on traditional fuels, and affect household time use and economic outcomes.

The study employs a cluster-randomised controlled trial (RCT) to evaluate the effectiveness of alternative policy interventions aimed at promoting sustained LPG usage among rural households in India. The trial is implemented across selected rural districts in Odisha, Jharkhand, and Telangana, regions characterised by continued biomass dependence and widespread fuel stacking despite expansion in LPG access. The study focuses specifically on households that possess LPG connections but use LPG irregularly or alongside traditional biomass fuels such as firewood, dung, or crop residue.

The sample consists of 1,800 households distributed across 90 villages, with 20 households surveyed in each village. Villages were selected using a stratified multi-stage sampling strategy based on LPG penetration, rural accessibility, and feasibility of field implementation. Randomisation was conducted at the village level to minimise spillovers arising from information diffusion, shared distribution networks, and social interaction among neighbouring households. Villages were stratified by state, district, village size, and baseline LPG usage before assignment to treatment arms.

The experiment contains five groups: one control arm and four treatment arms. The control group continues under existing LPG access conditions without additional intervention. The first treatment arm provides a refill subsidy of ₹150 per LPG refill cylinder to reduce affordability and liquidity constraints associated with sustained LPG usage. Subsidies are delivered as immediate point-of-purchase cash transfers, with a maximum of six subsidised refills during the intervention period.

The second treatment arm consists of behavioural interventions designed to address informational and behavioural barriers. Households in this group receive two SMS reminders per week and one field visit per month. Messages are delivered in local languages (Odia, Hindi, and Telugu) and focus on health benefits of reduced smoke exposure, time-saving advantages of LPG, cost framing, refill reminders, social norm messaging, habit formation, women’s empowerment, and convenience of refill booking. Monthly field visits reinforce these messages using standardised communication scripts.

The third treatment arm focuses on reducing transaction costs and logistical barriers related to LPG refills. Households receive assistance with refill booking, refill scheduling, and coordination with LPG distributors. Dedicated field facilitators support households in arranging refills through phone, WhatsApp, or direct coordination with distributors, particularly in villages with weaker last-mile delivery systems.

The fourth treatment arm combines all intervention components, including refill subsidies, behavioural messaging, field visits, and assisted refill coordination. This combined intervention is intended to evaluate whether affordability constraints, behavioural frictions, and supply-side barriers jointly limit sustained LPG usage and whether integrated interventions generate larger behavioural responses.

Baseline household surveys were conducted between March and April 2025 and collected detailed information on demographics, socioeconomic status, cooking practices, LPG usage patterns, fuel stacking behaviour, time allocation, labour-market participation, women’s autonomy, and perceived barriers to LPG use. The intervention period runs from May to October 2025. Midline surveys conducted in November 2025 measure short-term behavioural responses and changes in LPG usage. Endline surveys conducted in March 2026 assess sustained impacts on fuel usage, time allocation, and economic activity outcomes.

Primary outcomes include LPG refill frequency, the share of meals cooked using LPG, persistence of fuel stacking, cooking-related time use, fuel-collection time, labour-force participation, weekly hours worked, and household enterprise activity. Secondary outcomes include refill delays, refill booking difficulties, awareness of LPG-related health and time-saving benefits, treatment compliance, cooking frequency, meal complexity, and perceptions regarding affordability and supply reliability.

The empirical analysis follows an intention-to-treat framework comparing outcomes across treatment and control groups while preserving the benefits of randomisation. Treatment effects are estimated using regression models that control for baseline household characteristics, district fixed effects, and state fixed effects, with standard errors clustered at the village level. The study also estimates treatment-on-the-treated effects using treatment assignment as an instrument for actual intervention uptake. Heterogeneity analysis examines whether treatment effects vary by gender norms, women’s autonomy, baseline LPG usage, income levels, and local labour-market conditions.

Randomisation was conducted in office using a computer-generated stratified random assignment procedure. Villages were first stratified by state, district, village size, and baseline LPG usage levels to ensure balance across treatment arms. The 90 sampled villages were then randomly assigned at the cluster level into one control group and four treatment groups using computer-generated random numbers. Randomisation at the village level was chosen to minimise spillover effects arising from information sharing, social interaction, and common LPG distribution networks among households within the same village.

The unit of randomisation is the village (cluster-level randomisation). All eligible households within a selected village were assigned to the same experimental arm. A total of 90 villages were randomly allocated across one control group and four treatment groups, with approximately 20 sampled households surveyed within each village. Cluster-level randomisation was used to reduce spillover effects related to information diffusion, behavioural interactions, and shared LPG distribution systems within villages.

Yes

90 villages (clusters)

1,800 households across 90 villages (clusters), with approximately 20 households surveyed per village.

18 villages (360 households) in the control group; 18 villages (360 households) in the price incentive treatment group; 18 villages (360 households) in the behavioural intervention treatment group; 18 villages (360 households) in the convenience intervention treatment group; and 18 villages (360 households) in the combined intervention treatment group.

For cooking-related time allocation, the study is powered to detect reductions of approximately 12–15 minutes per day, assuming a baseline average cooking time of about 183 minutes per day and a standard deviation of approximately 55–60 minutes. For labour-force participation, measured as a binary household-level indicator, the study is powered to detect changes of approximately 4–5 percentage points relative to a baseline participation rate of roughly 40 per cent. The calculations account for intra-cluster correlation arising from village-level randomisation, with assumed intra-cluster correlation coefficients ranging between 0.05 and 0.10 for the main household outcomes.