Willingness to pay for programmable heating thermostats: The effects of information and the role of incentives
How to cite
Kaestner, Kathrin. 2023. "Willingness to pay for programmable heating thermostats: The effects of information and the role of incentives." AEA RCT Registry. AEARCTR-0011637. https://doi.org/10.1257/rct.11637-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
The aim of this incentivized survey experiment is to elicit households’ willingness to pay (WTP) for programmable heating thermostats. Digital technologies are seen as a promising way to mitigate climate change by reducing energy demand, especially by combining technology with behavioral interventions such as real-time feedback or social comparisons.
The heating sector of private households is an important lever to reduce GHG emissions. While energy retrofits (e.g. facade insulation) promise substantial energy savings, they are often very costly, not only in terms of pecuniary costs but also other costs, such as time, noise and dirt during construction. Moreover, insulation is often not attractive for landlords to conduct and can only in rare cases be conducted by tenants in multiple apartment buildings. In addition to saving energy, programmable heating thermostats are designed to increase consumer utility by reducing the amount of changes and adjustments required by users. Despite these seemingly attractive attributes, as of 2022, only 8% of households in Germany have installed programmable thermostats (Statista, 2022).
Given this low penetration of programmable thermostats among German households, I aim to understand households’ valuation for programmable thermostats by eliciting the WTP in an incentivized survey experiment. The goal of the experiment is twofold: First, I investigate whether the low of share of programmable thermostats among private households is a reflection of well-informed preferences or ignorance. To this end, I test the effect of two different types of information on the WTP. Second, I want to study the role of using a lottery to make the experiment incentive compatible and consequential (this has especially been discussed in the context of public goods, see e.g. Vossler et al., 2012; Carson et al., 2014) for elicited WTP values. As researchers are often financially restricted when it comes to ensuring that choices in an experiment are incentive compatible and consequential, using a lottery to select only a certain percentage of respondents seems as a good compromise. However, if elicited preferences/WTP values differed depending on the probability of being selected, this would be important to consider when comparing results from experiments with different winning probabilities in incentivized settings.
Intervention
Participants will be randomly divided into a control or one of two treatments receiving additional information about programmable thermostats. All participants receive some general information about programmable thermostats. In one of the treatment groups, participants will additionally receive information about the energy and cost savings potential of programmable thermostats. In the other treatment group, participants will additionally receive information about the emissions reductions potential of programmable thermostats.
On the same level of intervention, all participants who do not opt out of the incentivized experiment will be randomly divided into a low or high winning probability group, where they will learn about their chance of being selected to actually receive one of their choices made in the experiment. These differences will also be made in the control group. Participants who opt out of the incentivized experiment will randomly be divided into a control or one of the two information provision groups without any winning probabilities.
Experimental design
With this study, I aim to investigate households’ WTP for programmable heating thermostats. In addition, I am interested in whether information on the potential cost savings for households or information on the potential emission reductions affects the WTP and whether this effect differs between the two types of information. Lastly, I also aim to study the role of the winning probability for the elicited WTP values. To this end, I designed a randomized, incentivized experiment with a lottery and integrate it in the questionnaire of the third survey wave of the GHHP.
Baseline data on household characteristics, demographic information, building characteristics, heating systems and energy efficiency of the building stem from the first, second or third survey wave of the GHHP, i.e. from 2021, 2022 or 2023. This information is usually only asked once, depending on the time participants enter into the GHHP. However, when participants move between two survey waves, they are asked to state their building characteristics again. Other information, like current energy cost, past and planned energy retrofit activities and socio-economic characteristics are asked to participants in each survey wave.
At the beginning of the experimental part, I first ask all respondents about their type of main thermostat that they use to control the heating temperature in the majority of their rooms. Here, households can for instance already state whether they own a programmable
thermostat. As all households, even households that have already installed programmable thermostats, will be asked about their WTP, this serves as an important control variable. Afterwards, all respondents will receive some basic background information about programmable heating thermostats to ensure that every respondent has some basic knowledge and understands what type of thermostat the subsequent questions refer to. Following a few questions about respondents' level of informativeness and beliefs about programmable thermostats, respondents will be informed that on the following survey pages, they have the chance to win a voucher or one programmable thermostat with the previously described characteristics. Here, respondents have the chance to opt out of this lottery.
Respondents who stay in the lottery will then be randomly divided into the six groups described in the intervention, whereas respondents who opt out will be randomly divided into three groups. Each group receives specific information (see Intervention) about programmable thermostats and will subsequently be presented with a multiple price list (MPL) consisting of ten choices, each between a voucher of a certain euro value and a programmable heating thermostat. At the outset of the experiment, I explain to participants who did not opt out of the lottery that if they are randomly drawn as part of the lottery with the respective winning probability, one of their ten choices between a voucher or a programmable thermostat will be randomly chosen and actually paid out such that participants have an incentive to state their true WTP.
Design details
Cited by
No citation data available for this registration yet.