Household preferences for private versus public subsidies for new heating systems: insights from a multi-country discrete choice experiment

ABSTRACT

This paper employs demographically representative discrete choice experiments (DCEs) with owner occupiers in Poland, Sweden, and the United Kingdom (UK) to estimate the effects of subsidies, heating cost savings, installation time (reflecting ‘hassle costs’) and warranty length on owner occupiersʻ propensity to invest in a new heating system. In particular, the paper explores whether owner occupiers value subsidies received from public funding sources differently than subsidies received from private funding sources. The results from estimating mixed logit models suggest that respondents not only value subsidies for new heating systems because they decrease the net price, but they also value receiving a subsidy per se. For participants from Sweden (but not from Poland and the UK), this non-monetary value was found to be higher for subsidies offered by a public than by a private funding source. The results for heating cost savings in the three countries imply implicit discount rates between about 11 and 13%. We further find that respondents in Poland dislike longer installation times, and that respondents in all three countries value longer warranty times.

KEYWORDS:

• Energy efficiency
• heating systems
• choice experiment
• subsidies

JEL CLASSIFICATION:

• D12
• H31
• Q48

Acknowledgements

We are grateful to three anonymous reviewers for their insightful comments and suggestions. We are also thankful for the technical advice and feedback received by Wolfgang Eichhammer and Barbara Schlomann (Fraunhofer Institute for Systems and Innovation Research, Karlsruhe, Germany), Eftim Popovski and Jan Steinbach (Institute of Resources Efficiency and Energy Strategies, Karlsruhe, Germany) and Andreas Müller (Energy Economics Group, TU Vienna, Austria).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Becoming ‘climate neutralʻ means avoiding greenhouse gas emissions as far as possible, and to compensate any remaining emissions through emissions reduction projects that reduce or absorb greenhouse gas emissions elsewhere.

2 For example, heating degree days (HDD) in Sweden are typically above 5000, in Poland above 3500 and in the UK around 2100 (using 15.5°C as the reference temperature). In comparison, HDD in Greece, Italy and Spain typically range between 1500 and 1800. HDD are a measure of how much (in °C), and for how long (in days), outside temperature is below a base temperature (here 15.5°C). That is, heating demand in Sweden is about three times as high as in Greece.

3 This is in contrast for instance to Nishijima et al. (2019) who combine findings from a DCE model for air conditioner replacement decisions with an input-output model to evaluate economic efficiency of Japanʻs appliance replacement program.

4 Only participants who completed the entire questionnaire received a monetary reward equivalent to a few Euros in form of Norstat Coins. Norstat panel members can cash out their Coins as soon as they reach 200 Coins (about €20).

5 This completion rate seems satisfactory, especially since the survey is rather long. Of course, we cannot rule out a potential selection bias, yet the direction of such a bias on our findings is not clear.

6 This set-up abstracts from the fact that owner occupiers have a strong tendency to replace their heating system with the same technology as the existing one (e.g. Michelsen and Madlener 2012, 2016) even when there are no differences in transition costs, comfort, and certainty of cost savings (Lang et al. 2021). The findings by Lang et al. (2021) imply that subsidies have to be substantially larger if they involve a change in the heating technology/fuel type, ceteris paribus. This aspect, however, should not affect owner occupier preferences for funding sources. Using very similar instructions and attributes, Schleich et al. (2021) found no evidence that the fuel type participants had in mind when responding to the DCE affected their preference for owning versus renting their new heating system.

7 To adjust the monetary amounts shown to the participants by country, we used the following exchange rates to the Euro (used as reference currency): Poland €1 = PLN3; Sweden €1 = SEK10, and UK €1 = £1. The actual exchange rates for PLN and SEK were approximately €1 = PLN4.3and €1 = SEK10.3, respectively, at the time our survey was fielded. The purchasing power parity (PPP) indicator between Poland and the UK, however, was much lower: about 2.2 in 2018 (Eurostat 2018). We thus chose an exchange rate of €1 = PLN3 to ensure that amounts were similar in terms of purchasing power across countries.

8 The levels for the attributes heating bill and warranty were taken from Achtnicht (Achtnicht 2011) and Achtnicht and Madlener (2014). The upper and lower levels for purchase price were chosen based on a screening of market prices for heating systems in the three countries. For the purchase price we rely on the ranges provided by Ruokamo (Ruokamo 2016), Rouvinen and Matero (2013), Alberini and Bigano (2015) and Fleiter et al. (2016) which includes detailed technology assessment of residential heating systems for the EU distinguished by technology, size and country. The subsidy levels correspond to those typically observed in government or utility-sponsored subsidy programs and are similar to those employed by Alberini and Bigano (2015) or Olsthoorn et al. (2017). Values for installation were provided by technology and sector experts of Fraunhofer Institute for Systems and Innovation Research and the Technical University of Vienna.

9 The survey item read: “Approximately, how much was the total annual cost for heating your primary residence in the last year? If you do not know the exact figure, please estimate”.

10 We considered the self-reported heating costs to be unreasonable when the difference between reported and estimated annual heating costs was larger than €750. Using the data on dwelling size and heating costs, we calculate the specific heating costs at about 10 €/m² in Poland and about 13 €/m² in Sweden and in the UK. These figures are similar to the 11 €/m² employed by Achtnicht (2011) and Achtnicht and Madlener (2014) for their sample of homeowners in Germany. We report in the robustness check section that our findings are not sensitive to using €750 as the cutoff value.

11 As we will discuss in the robustness check section, our main results also hold if we estimate the model using the percentages of heating cost savings rather than the monetary values.

12 We also estimated a standard conditional logit model. For all three countries, we find that the conditional logit model violates the IIA assumption.

13 To calculate the implicit discount rate, we use the formula developed by Min et al. (2014, p. 43). Based on Hausman (1979), they show that the implicit discount rate r can be computed using the capital recovery factor for a given heating lifetime q: ${\hat{\beta }}_{netprice}={\hat{\beta }}_{savings}\frac{r{(1+r)}^q}{{(1+r)}^q-1}.$ ${\hat{\beta }}_{netprice}$ and ${\hat{\beta }}_{savings}$ are the estimated parameters of net_price and savings from Table 2.

14 Preferences are assumed to be linear in all attributes. A higher subsidy would accentuate the differences between the policy scenarios and the baseline scenario but otherwise lead to similar results.

15 Since probabilities add up to 1, the choice probabilities for Option A are not reported.

16 Note that our data allow us to study the effect of subsidies on the choice of heating system conditional on the homeowner replacing the heating system. Our data do not allow us to study the effect of subsidies on a homeownerʻs decision to replace or not their heating system.

17 The findings of these analyses are available upon request.

Additional information

Funding

This research received funding by the European Union’s Horizon 2020 Framework Programme under the project CHEETAH - CHanging Energy Efficient Technology Adoption in Households (Grant agreement ID: 723716)].