Voucher-based liquefied petroleum gas subsidies achieve nearly exclusive clean fuel use when affordable delivery is available

Abstract

Introduction:

Fuel subsidization can promote clean fuel adoption and reduce household air pollution. We investigated whether a voucher-based system could be as effective as a home-based fuel delivery program in achieving exclusive use of liquefied petroleum gas (LPG) within a randomized controlled trial in Puno, Peru.

Methods:

In the first year of the trial, we delivered unlimited LPG to intervention homes, while comparison group participants continued baseline cooking practices. In the second year, comparison group participants received vouchers covering the cost of two 10 kg LPG tanks per month but did not receive tank delivery. We assessed voucher use and compared LPG use between the voucher and home delivery schemes. We conducted qualitative interviews with 15 women to understand perceptions and use of the vouchers and LPG. Data were collected in 2020 but remain relevant given the unchanged LPG system in Peru.

Results:

Voucher use was high (97.6% of vouchers were used). Participants receiving vouchers used LPG for 94.3% of cooking minutes, which was only slightly lower than the percent of cooking done with LPG by participants who received free home delivery (98.2%) (p<0.001). Voucher participants cooked with biomass for 84.7 minutes per week, within the range estimated as acceptable for meeting the World Health Organization interim air quality targets (<1–3 hours per week). Facilitators of voucher and LPG use included the ease and speed of cooking with gas, availability of LPG delivery in the region, access to several nearby LPG distributors, owning a second back-up LPG tank, and reinforcement of exclusive LPG use by study fieldworkers.

Conclusion:

A voucher-based system for providing subsidized LPG, paired with a reliable LPG infrastructure and behavioral motivation, effectively achieved near-exclusive clean fuel adoption. Findings can guide the design of LPG subsidization programs, which are increasingly important amid rising global LPG prices.

1. Introduction

Globally, efforts to reduce the burden of household air pollution have focused on promoting clean fuel adoption through a range of strategies.¹ These include financial policies such as taxes on polluting fuels, tax credits for clean fuels, subsidies that reduce the market cost of clean fuels, and vouchers provided to households that cover some or all of the cost of clean fuel.² Other efforts include regulation that restricts or bans certain polluting fuels or technologies, behavior change campaigns that raise awareness of the benefits of clean fuels, and standards for stove efficiency or emission requirements.²

Most existing liquefied petroleum gas (LPG) subsidization programs, including the Fondo de Inclusión Social Energético (FISE) program in Peru, rely on a voucher-based system of delivery in which households must bring their voucher and LPG tank to a retail location to receive a refill or request and pay for delivery.³ However, these programs typically do not subsidize enough fuel to enable exclusive LPG use.⁴ To date, most programs promoting clean fuels have been unable to achieve exclusive clean fuel use and near-abandonment of traditional biomass stoves,⁵ which is necessary to achieve health benefits.⁶

Recent research studies have found that when LPG was provided for free and delivered directly to households, participants used the LPG stove for most of their cooking tasks.^7-10 This suggests that when economic and transportation barriers to LPG adoption are removed, near exclusive adoption of LPG is possible. However, these studies provided unlimited free LPG, free household delivery, and installation and verification of the LPG tanks in households upon delivery. Thus, it remains unclear whether household delivery of LPG played an important role in the exclusive adoption achieved, or if removing fuel costs alone is sufficient regardless of the delivery mechanism.

This paper seeks to explore whether provision of free LPG (up to 20 kg per month, which was previously estimated as necessary to meet household energy needs⁴) through study-provided vouchers, without free delivery or installation, could achieve similar levels of near-exclusive LPG use as a program with free, unlimited fuel delivery. We aimed to explore the impact of delivery mechanism (i.e., vouchers vs. home-based delivery) on exclusivity of LPG use to understand whether national voucher-based programs could achieve greater LPG adoption by integrating direct household delivery into their programs.

2. Methods

2.1. Location

The study was based in rural areas of Puno, a southern Andean region of Peru located at 3,825 meters altitude. People in the region traditionally cook with wood or animal dung in a clay stove called a fogón. The Cocinas Perú program previously provided free LPG stoves to poor families, and, at the time of our study, the FISE program provided a subsidy to poor households covering approximately 5 kg of LPG per month.³ As a result of these efforts, access to and use of LPG has been increasing in rural Puno; however, most people continue to cook primarily with biomass and use LPG as a supplemental fuel when in a rush or during rainy weather.¹¹

2.2. Data Collection

2.2.1. Cardiopulmonary outcomes and Household Air Pollution (CHAP) trial

This research was carried out as part of the second year of the Cardiopulmonary outcomes and Household Air Pollution (CHAP) trial.^12,13 Details of the trial have been published elsewhere.^12,13 Briefly, we recruited participants on a rolling basis over one year by screening a simple random sample of women from a household census of the region to identify those who used biomass fuels daily. We obtained informed consent and randomized 14-16 participants per month to intervention or comparison groups between March 2017 and February 2018, for a total of 180 women (aged 25 to 64 years). For the first year of the trial, intervention participants (n = 90) received a free LPG stove and tank with 10 kg of LPG, education and reinforcement for LPG use, and free LPG tank replacements delivered to their homes. Comparison group participants (n=90) were asked to continue their baseline cooking practices.

At the beginning of the second year (13 months after randomization), comparison group participants received the same LPG stove, tank with 10 kg of LPG, and education that intervention participants had received in the first year. For one year following their receipt of the LPG stove, comparison group participants were given two vouchers per month, which each covered the entire cost of a 10 kg LPG tank replacement. This amount was determined to correspond with the 19.2 kg average LPG use per month observed among intervention households in the first year of the CHAP trial.⁴ Comparison group participants were instructed on how to use their voucher to obtain LPG tank replacements from any of several designated LPG retailers. They could bring their empty LPG tank and voucher to the distributor to receive the replacement tank or pay the LPG distributor for delivery (when available). We attempted to limit voucher leakage (i.e., selling/giving vouchers or LPG tanks to others) by requiring redemption through designated LPG providers with a study ID and reinforcing non-sharing during monthly fieldworker visits. Vouchers were valid for one month. Providers were reimbursed after submitting the used vouchers to study staff. This contrasts with intervention participants whose LPG tank replacements were delivered for free directly to their household by study fieldworkers during the first year of the trial.

2.2.2. Quantitative data collection

A survey was administered on socio-demographic characteristics such as education, household assets, animal ownership, employment, number of household members, participation in the FISE program, and stoves present in the household. Fieldworkers also administered a quarterly behavioral survey, which assessed cooking practices including number of cooking sessions per day, amount of time spent cooking, stoves used, and cooking tasks completed. Voucher use was tracked using the vouchers that were returned to the study team by LPG providers, which was mandatory for them to receive their reimbursements.

LabJack Digit-TL stove use monitors (SUMs)¹⁴ were installed on both traditional and LPG stoves in all intervention and comparison group households to continuously monitor stove use. Fieldworkers visited households monthly to download SUMs data and conduct a fuel use survey. Some households voluntarily destroyed their traditional stoves after realizing the LPG provided was sufficient to meet their needs. At monthly visits, SUMs were removed from any traditional stoves that were destroyed voluntarily during the trial and reinstalled in any traditional stoves that were reconstructed. A fuel use survey included questions on LPG tank purchases, LPG voucher use (including both trial-provided vouchers and governmental-provided FISE vouchers), and spending on LPG.

2.2.3. Qualitative Methods

During the second year of the CHAP trial, we conducted in-depth qualitative interviews on perceptions of the voucher-based delivery mechanism. To capture the diverse range of characteristics observed across trial participants, we purposively selected 15 participants from the comparison group of the trial with varying levels of LPG use, voucher redemption rates, FISE beneficiary status, wealth quintiles, ages, and household sizes. We determined our sample size based on thematic saturation, stopping recruitment once interviews consistently revealed recurring themes with minimal new insights emerging.¹⁴ A fieldworker who was bilingual in Spanish and Aymara and trained in qualitative interviewing methods conducted all interviews in the participants’ homes. All interviews were recorded using digital handheld audio recorders. A fieldworker took handwritten notes during the interviews on key themes and contextual information.

2.3. Analysis

2.3.1. Quantitative data

We analyzed voucher tracking and survey data to determine the percentage of vouchers used by comparison group households (i.e. those that received vouchers), which providers they exchanged the vouchers with, and reasons why any vouchers were not used. SUMs data were processed in R using a previously published algorithm to identify cooking periods on both fogones and LPG stoves based on temperature spikes (R Core Team, Vienna, Austria).^8,16,17 We used independent two-sample t-tests to compare the percentage of cooking done with LPG, based on both self-report and SUMs data, between participants receiving home delivery vs. vouchers, using StataSE version 15 (StataCorp LLC, College Station, TX, USA).¹⁸

2.3.2. Qualitative data

Qualitative interview recordings were transcribed into Spanish and translated into English. The first author reviewed all transcripts to inductively identify key themes related to factors influencing LPG and voucher use. The technique of memoing was used to capture key information related to each theme within each interview.¹⁹ The first author coded all transcripts in Taguette to identify quotes relevant to each theme.²⁰ This information was then compiled across all transcripts and synthesized into a summary of key points within each theme to understand usage and perceptions of the vouchers.

3. Results

3.1. Participant characteristics

Characteristics were similar between intervention (home delivery) and comparison (voucher) participants (Table 1). Participants in both groups had on average six years of education and most were in the poor or poorest wealth quintiles. Households had on average four total members (including the participant) and about half had ever participated in the governmental LPG subsidization program (FISE). Most reported prior ownership of an LPG stove or tank, indicating access to an extra LPG tank apart from the one delivered by the trial. Slightly more than half owned pigs and two-thirds owned dogs (animals that are commonly fed cooked food). Table 2 shows characteristics of participants who completed qualitative interviews.

Table 1.

Characteristics of participants in the CHAP trial at baseline.

	Voucher participants		Home delivery participants
	n	% or mean (SD)	n	% or mean (SD)
Age (years)	90	47.9 (11.1)	90	48.7 (9.1)
Years of education	90	6.4 (3.3)	90	6.1 (3.4)
Wealth quintile
1 (Poorest)	50	55.6%	51	56.7%
2 (Poor)	37	41.4%	32	35.6%
3 (Middle)	3	3.3%	7	7.8%
Total # household members	90	3.7 (1.4)	90	4.0 (1.6)
Number of pigs owned
0	36	40.0%	38	42.2%
1-2	43	47.8%	41	45.6%
3+	11	12.2%	11	12.2%
Number of dogs owned
0	33	36.7%	24	26.7%
1	38	42.2%	54	60.0%
2+	19	21.1%	12	13.3%
Household was ever enrolled in FISE	46	51.1%	42	46.7%
Owns two LPG tanks	78	86.7%	63	70.0%
Round trip time to nearest LPG outlet (minutes)*	86	72.9 (46.0)	86	60.0 (36.8)

^*Mode of transportation varies by participant, including walking, public bus, bicycle, motorcycle, car or a combination. For example, 40 minutes walking to the main road, 20 minutes waiting for a bus, and 30 minutes riding the bus equals 90 minutes in both directions, for a total of 180 minutes round trip time.

Table 2.

Characteristics of comparison group participants receiving LPG vouchers who participated in qualitative interviews.

Participant	% of cooking done with LPG	% vouchers used	FISE beneficiary	Wealth quintile	Age	Total # household members	Round trip time to nearest LPG outlet (min)
A	99.8%	91.7%	Yes	Poor	62	2	120
B	93.7%	100%	Yes	Poor	45	6	25
C	88.6%	100%	No	Poor	64	2	60
D	99.9%	91.7%	No	Poor	54	2	45
E	92.0%	100%	No	Poorest	64	2	180
F	100%	100%	No	Poorest	45	3	180
G	100%	100%	Yes	Poor	53	3	60
H	94.8%	100%	No	Poor	47	4	30
I	95.2%	100%	Yes	Poorest	47	5	60
J	99.8%	91.7%	Yes	Poorest	65	1	30
K	100%	100%	No	Poor	52	3	36
L	100%	100%	Yes	Poor	45	4	30
M	100%	100%	Yes	Poor	46	5	30
N	99.5%	100%	Yes	Poor	35	7	180
O	84.6%	95.8%	No	Poor	55	3	120

3.2. Voucher usage

Usage of the two LPG vouchers provided per month was extremely high. On average, participants used 97.6% of the vouchers (range 87.5%-100%). Most participants (94.4%; n=85) used at least one voucher per month and 62.2% (n=56) used both vouchers every month (100% of vouchers). Participants who did not use both vouchers every month had 1 month (27.8%; n=25), 2 months (7.8%; n=7), and 3 months (2.2%; n=2) in which they used fewer than two vouchers. Reasons for not using the vouchers included still having gas (n=7), providers not having LPG available (n=2), delays in receiving the vouchers (n=1), poor service from LPG providers (n=2), and expired vouchers (n=2); others reported no reason (n=20). Qualitative data suggests that LPG providers may have occasionally lost vouchers that they received as payment from participants. Thus, participants may have used more vouchers than were returned to and recorded by project staff.

3.3. Exclusivity of LPG use

The voucher scheme achieved near-exclusive LPG use. Voucher participants used LPG for 94.3% of all cooking minutes and 97.5% of all cooking events, according to stove use monitors (Table 3). In surveys, voucher participants reported using LPG for 99.5% of cooking minutes and 99.3% of cooking events. The rate of LPG use seen among voucher participants who received a limited quantity of LPG (20 kg per month) was slightly lower than the rate achieved among intervention participants who received unlimited LPG delivered directly to their homes (Figure 1). However, the average number of minutes per week in which voucher participants used biomass (84.7 minutes) falls within the range considered consistent with meeting the World Health Organization interim air quality targets (<1–3 hours per week).^6,21

Table 3.

Percent of cooking done with LPG according to stove use monitors and quarterly survey data.

	n	Intervention (unlimited free LPG and home-based delivery)	n	Comparison (20 kg free LPG per month and voucher system)	p-value
Percent cooking minutes done with LPG- SUMs, % (SD)	90	98.2% (2.9)	90	94.3% (8.8)	<0.001
Percent cooking minutes done with LPG- Survey, % (SD)	90	99.3% (0.03)	90	99.5% (0.03)	0.59
Number cooking minutes per day with fogón, mean (SD)*	90	3.5 (6.0)	90	12.1 (25.7)	<0.01
Percent cooking events done with LPG- SUMs, % (SD)	90	99.2% (1.3)	90	97.5% (3.8)	<0.001
Percent cooking events done with LPG- Survey, % (SD)	90	99.4% (0.03)	90	99.3% (0.03)	0.83

^*NOTE: Average daily fogón use is calculated across the full year. Occasional fogón cooking events translate into a small average number of minutes per day when spread over 365 days.

Figure 1.

Percent of cooking done with LPG by intervention group participants in year 1 (with unlimited, free LPG delivered to their homes) and by comparison group participants in year 2 (with 20 kg free LPG per month that was not delivered)

In Figure 2, we plot the percent of days with biomass fuel use by study month; free LPG stove and vouchers enabled comparison group participants to significantly reduce their biomass use compared to year 1 before they received an LPG stove or vouchers from the trial. Overall, comparison group households cooked with biomass on 72.2% (SD 22.8) of monitored days during year 1 of the trial and on 5.7% of monitored days (SD 9.1) during year 2 (p<0.001).

Figure 2.

Percent of days in which comparison group participants used biomass in year 1 (prior to receipt of an LPG stove from the trial) and year 2 (after installation of an LPG stove and receipt of vouchers for two tanks of LPG per month)

3.4. Factors influencing LPG and voucher use

3.4.1. Preference for gas

All women interviewed described preferring LPG to the traditional stove, which aligns with the quantitative data showing near-exclusive LPG use under the voucher scheme. Women appreciated that they could cook more quickly with LPG. They also appreciated how LPG is cleaner, meaning their pots, kitchens, and clothes no longer got stained with soot or smelled of smoke. They described having become accustomed to cooking with LPG and not wanting to return to their traditional stove.

“We prefer the gas because we cook quickly. It helps me have time to take out my animals. Then, I’m knitting. And after that, I light the stove and put the vegetables on to cook. And while they cook, I have time to knit and feed the animals. It doesn’t matter if we are tired and wake up late because we can cook quickly… Also, with gas, we don’t get dirty. With the fogón, we always got smudged with soot. The pots and clothes got smudged too. But now [with gas], it’s cleaner… [My fogón] is no longer working, it’s broken into pieces.”

– Participant K

3.4.2. Two tanks per month generally sufficient

Most women said two tanks of LPG per month (~20 kg) were sufficient to meet all their cooking needs, including cooking meals for their family, cooking for their dogs and pigs, and heating water for tea and bathing. Women described how they sometimes needed more LPG in months when their family members were visiting. Some women reported purchasing additional LPG, either at full price or using a FISE voucher.

3.4.3. Exclusive LPG use reinforced by study staff

All women recognized that the study fieldworkers who delivered their LPG vouchers every month first downloaded data from the stove use monitors on their LPG and fogón stoves. The project staff also encouraged the women to cook only with their LPG stove and not to use the fogón. Women commented that this motivated them to abandon the fogón to please the staff and ensure continued receipt of their LPG vouchers.

“I no longer cook [with the fogón], because it has this keychain [referring to the SUM] that keeps track. Those who bring the vouchers always check it and they would know if I’m cooking. When the young man comes to bring the voucher, he tells me not to cook with the fogón. He says that’s why the keychain is there. It can catch me. There’s also one hanging in the gas stove. It keeps track and they review it, so I never cook with the fogón. Otherwise, I would cook there sometimes.”

– Participant A

3.4.4. Value for delivery

Most women said they preferred to pay to have the LPG tanks delivered to their house vs. going to the store themselves. In survey data, the 78 people who reported paying for delivery indicated that it cost an average of 2.6 soles [US $0.79] (SD 0.8 soles [US $0.24], range 1–5 soles [US $0.30–$1.52]), compared to a reported LPG tank cost of 36 soles [US $10.91] (SD 0.4 soles [US $0.12], range 35–37 soles [US $10.61–$11.21]). Many participants were older and said they would be unable to carry the heavy LPG tank even a short distance to the store. The small community LPG providers typically did not offer delivery; thus, women often used distributors in the nearby town instead. Most women said deliveries were made within 30 minutes, which is less than the time it would take them to go to the store themselves.

Additionally, women described appreciating that the delivery men would install the tank for them and check to make sure that the stove was functioning properly, making them feel safer.

“The delivery men from Acora [nearby town] connected the tank and checked if it was okay… Once he fixed my stove. The knobs on my stove wouldn’t move and I was only cooking with the two burners. When he came, he fixed it for me… Once it was smoking a lot, so he told me to wash the knobs and burners.”

– Participant C

A few participants who lived very close to an LPG distributor or who had husbands who were willing and able to help brought their LPG tank to the store to avoid the delivery fees. Some women also reported exchanging the tank themselves due to difficulties in calling for deliveries, for example, because they did not have a phone, did not have cell signal, or did not have enough credit to call. Other women relied on family members or neighbors to help them call for delivery.

Quantitative data showed that people living in communities with an LPG provider were more likely to purchase LPG from that provider than people living in other neighboring communities (Table 4; Figure 3). However, even participants in communities with LPG providers were more likely to purchase LPG from Acora, from where they could get deliveries (Table 4).

Table 4.

Probability of purchasing LPG from community providers (Thunco and Culta) and providers in a nearby town (Acora).

LPG Provider Location	Probability of people from the community using the provider compared to people from other communities		# people purchasing from the provider who lived in a different community	Probability of people from the community using an Acora provider
	Coefficient	p		Coefficient	p
Thunco	8.8	p<0.01	2	11.6	<0.001
Culta	5.5	p<0.01	3	3.1	0.21

Figure 3.

Map of CHAP participant homes, by community, and LPG distributor locations.

“There are providers [in my community], but I would have to carry the tank and it’s heavy… For that reason, I call Acora [LPG provider in a nearby town] and they bring it straight to my house. I would go to the store, but I’m sick and it’s heavy, so they bring it here from Acora, and they install it for me. It would be good if I could go to the store because it would be free, but I don’t go because it’s too heavy…”

– Participant A

3.4.5. Multiple providers enhance accessibility and reliability

Almost all women interviewed were aware that there were multiple LPG providers in the area. Several women said they switched from one provider to another when they were dissatisfied with one’s service. Knowledge that they had a choice of providers also made women less afraid to stand up to the providers when there were problems. Women described preferring providers who have long hours, answer the phone right away when people call, deliver quickly and in the timeframe promised, always have LPG tanks in stock, supply high quality tanks, and are responsive to problems and willing to replace faulty tanks. If one provider was out of stock, women said they called a different provider and thus avoided interruptions to their LPG stove use. Quantitative data supported these findings, indicating that most participants (74.4%; n=67) used different distributors during their 12 months of receiving vouchers. Only 25.6% (n=23) exchanged all of their vouchers with the same distributor.

“Over there is another store where some people exchange their gas tanks. But I don’t trust them. Sometimes the store is closed. Where would I buy gas on those days? [The store I use] is always open… Once I got gas from Acora but they didn’t arrive quickly… If I call them today, they won’t arrive today. They won’t come to deliver until tomorrow… And sometimes they don’t have the screw tanks… On the other hand, my store always has the screw tanks.”

– Participant D

Despite having a choice of providers, most women said they generally used the same one once they identified a good one. Providers came to know them and where they lived so they could make deliveries more quickly. Also, some women said if their LPG tank ran out before the next voucher arrived, their trusted provider would give them a tank in advance, trusting that the participant would bring them the voucher as soon as it arrived. This practice was more often described as occurring with women who used community providers, as the larger providers (i.e., those in the Acora town center) had more clients and fewer personal relationships with their clients. Women who ran out of LPG and could not get another tank immediately said they sometimes used a family member or neighbor’s stove, or cooked with their fogón.

3.4.6. Two LPG tanks facilitate uninterrupted use

Women who had a second LPG tank in their house described how the second tank facilitated their uninterrupted use of the LPG stove. If their LPG tank ran out in the middle of cooking, they were able to quickly switch to their second tank, and then replace the empty tank later at their convenience. Those who did not have an extra tank had to wait for a delivery or use their fogón to finish cooking.

“When the tank is empty, I call to order a new tank… But I keep cooking until it completely runs out… By the time the gas delivery man arrives, the tank is empty… Once I was cooking fiambre, chuño, potatoes on one burner and fava beans were boiling. They weren’t cooked. I left them there and called the provider. He delivered the gas in 20 minutes and only then I could finish cooking… I was thinking about buying another tank. If I had another one, I wouldn’t have to wait for the tank to run out before replacing it.”

– Participant H

3.4.7. Preference for paper vouchers

Participants described liking the paper-based vouchers. Many noted that they are old and do not know how to use or cannot see a cellphone and thus would not be able to use an electronic voucher. However, some women mentioned being nervous about losing the paper voucher or about it getting destroyed by children or mice. Nonetheless, they did not find it difficult to carefully store the paper voucher. Of note, the project voucher was delivered directly to participant homes; in comparison, participants in the FISE program are required to go somewhere to pick up their FISE vouchers, which some participants said they did not have time to do. Participants also noted that the project vouchers were always delivered on time, whereas their FISE vouchers were often late.

“I don’t know how to handle a cell phone. [Electronic vouchers] could be a problem for me. My daughters know but I don’t know how to use the cell phone. What would I do? How would I see it?… The paper vouchers are good for me… It’s like the incentives I receive on my mobile bank account. I have banking on my phone, but I can’t see it. When I get a deposit, I don’t know. [Electronic vouchers] would be the same for me… I would rather use the paper vouchers. It would be more practical with the cell phone, but I can’t use it.”

– Participant B

4. Discussion

A voucher scheme in which participants received paper vouchers to cover the cost of two LPG tanks (~20 kg) per month was successful in achieving near-exclusive use of LPG in Puno, Peru. LPG use was similarly near-exclusive when providing vouchers or free home-based delivery, representing a substantial increase over the ~20% of cooking reportedly done with LPG prior to receiving the vouchers.¹⁶ For voucher participants, availability of delivery in the region facilitated voucher use, and many participants chose to pay for the convenience of delivery. Having access to a variety of nearby LPG distributors, owning a second back-up LPG tank, and receiving behavioral reinforcement to use LPG and abandon traditional stoves also facilitated use of the vouchers and LPG.

Previous research has demonstrated that participants use LPG almost exclusively when it is provided for free, in unlimited quantity, delivered directly to participant homes, and promoted through behavioral reinforcement.^7-10 Our research demonstrates that free home delivery is not a necessary requirement for achieving high adoption of clean fuel use. Participants were willing to invest either effort (by going to a distributor to exchange their LPG tanks) or money (by paying for delivery) to obtain LPG. The fact that participants reported valuing LPG and preferring to cook with it over the traditional stove likely motivated their willingness to invest resources and/or energy into obtaining the cleaner fuel, as did reinforcement of exclusive LPG use by study staff. Nonetheless, having vouchers that covered the full cost of the LPG was necessary for participants to afford near-exclusive LPG use. While some participants were willing to pay for a small amount of LPG per month to supplement the vouchers, others filled the gap by using their traditional stove. Similarly, other studies have found that although people see benefits of LPG, financial constraints often prevent adoption.^22,23

These findings suggest that a voucher scheme for providing LPG subsidies could be successful in enabling exclusive clean fuel use if the vouchers sufficiently cover the cost of the amount of fuel needed to meet daily energy demands. Our research corroborates the findings of Pollard et al. (2018)³ that the FISE subsidy amount (which covered about 50% of the cost of one LPG tank per month at the time of our study) was insufficient for sustaining near-exclusive LPG use (which previous research showed required about two LPG tanks per month).⁴ It is also important to note that paper vouchers in our project were hand delivered to participant homes by study fieldworkers, while FISE vouchers must be picked up at a central location or obtained through electronic platforms. Although the older women participating in our research preferred the paper vouchers and noted that fieldworkers delivering the vouchers motivated their exclusive use of LPG, this system would be harder to implement in practice. LPG providers who make deliveries could possibly play a role in facilitating older participants to access electronic vouchers, as well as in encouraging exclusive LPG use to enable participants to fully realize the potential health and other benefits of clean fuel use.^24-26

Our study found that when vouchers covered the cost of two LPG tanks per month, participants achieved near-exclusive LPG use. This finding is particularly relevant given the substantial increase in LPG prices over recent years. In 2020, at the time of our study, the cost of a 10 kg LPG cylinder in Puno was approximately 36 soles (around $10 USD). By 2025, this price rose to approximately 55 soles (around $14.80 USD). Despite this increase, the FISE voucher amount has remained unchanged at 20 soles (approximately $5.40 USD). Consequently, the affordability of two LPG tanks per month has decreased over time, underscoring the necessity for an effective subsidy system. Although some LPG vendors have changed over time—some going out of business and others emerging—and fuel prices have increased, the overall system for obtaining LPG in Puno remains the same. Households still exchange empty tanks for filled ones, have the option to pay for delivery (which is more often available from urban vendors), and face a concentration of vendors in Acora city, with fewer available in surrounding rural communities. Thus, our study's identification of structural facilitators essential for a successful subsidy scheme, as well as the effectiveness of subsidy delivery mechanisms, remain pertinent.

While further research is needed to determine households' willingness to pay for LPG and the factors influencing this within a changing economic context, our findings provide valuable evidence to inform the development of a successful subsidization scheme aimed at achieving exclusive LPG use. More research is needed to identify sustainable financing mechanisms that could be implemented by the government to expand the current LPG subsidy to cover the cost of two LPG tanks per month—the amount necessary for achieving exclusive LPG use.⁴ Future work is needed to assess the feasibility of such a subsidy increase given rising fuel prices and evolving economic conditions. Consideration of the health, environmental, and time-saving benefits of clean cooking could strengthen the case for such investment.^24-26 Integrating energy conservation strategies with LPG subsidies could help lower overall fuel demand, making it more feasible to finance a smaller amount of LPG while still meeting household energy needs.²⁷ Without an increased subsidy, households would likely revert to stove stacking, although with more LPG use than pre-trial, as was seen in an analysis of sustained LPG use by intervention participants following one year of free LPG fuel and delivery.¹⁶

4.1. Strengths and limitations

Strengths of our study include the use of mixed methods data, demonstrating not only that the vouchers achieved high LPG adoption but also the reasons influencing this result, as well as frequent data collection, including monthly surveys, voucher usage tracking, and direct, continuous stove use monitoring through temperature sensors. However, some limitations should be noted. First, qualitative interviews were conducted by study fieldworkers. Participants may have been hesitant to mention any negative perceptions of the voucher scheme for fear of displeasing the staff. However, the likelihood that challenges were underreported is relatively low given that participants were informed that their responses would not affect their continued receipt of vouchers, and the trust developed between the interviewer and participant over the two years of the study. Second, it is possible that some vouchers were used and not reported, for example, in cases where distributors may have misplaced vouchers before returning them to the study for reimbursement. However, this indicates that voucher usage may have been higher than we documented. Although some voucher leakage or LPG sharing may have occurred, we believe it was minimal given the redemption procedures, monthly reinforcement by fieldworkers, and the alignment of LPG quantities provided with typical household fuel needs⁴. Third, stove use monitors on LPG stoves may have picked up cooking events on nearby traditional stoves, or vice versa. However, a previous analysis showed that concurrent use of the LPG and traditional stove in our study setting was rare.⁸ Thus, we expect that misclassification in stove use was low and balanced across stove types.

Additionally, the research study context in which participants came to trust fieldworkers over the one year of visits prior to beginning the voucher scheme may have enhanced their willingness to use the vouchers and comply with fieldworker encouragement to use LPG exclusively. Lastly, we recognize that economic conditions, including LPG prices, have changed since our study ended in 2020. However, a successful subsidy scheme consists of multiple components, including (1) the amount of the subsidy, (2) the mechanism through which subsidies are delivered, and (3) the infrastructure in which they operate. While fuel prices have fluctuated since our study was conducted, our findings on effective subsidy delivery mechanisms and the key supporting elements of the infrastructure required to facilitate exclusive LPG use remain highly relevant for informing current and future LPG subsidization programs.

5. Conclusion

Adoption of clean cooking technologies and abandonment of polluting fuels is essential to realize health, environmental, climate, and other benefits that clean fuels could afford. Our results indicate that a voucher-based system for delivering subsidized LPG is effective for achieving near-exclusive cleaner fuel adoption. However, other contextual factors must also be in place, such as a consistent LPG supply, choice of LPG providers, availability of home delivery, promotion of LPG use, access to cell phones and credit (or toll-free calling options), and access to a spare LPG tank. Achieving these conditions requires that governments recognize the threat of household air pollution and prioritize investment in cleaner fuels and energy infrastructure to reduce the burden.

Highlights.

• Voucher-based LPG subsidies achieved near-exclusive clean fuel adoption in Peru

• Reliable LPG infrastructure and behavioral motivation were also necessary

• Vouchers must sufficiently cover the cost of fuel needed for daily energy demands

• Findings can inform LPG subsidy design to realize the benefits of clean fuel use