Household preferences of hybrid home heating systems – A choice experiment application

Energy Policy 95 (2016) 224–237

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

Household preferences of hybrid home heating systems – A choice experiment application Enni Ruokamo a,b,n a b

Finnish Environment Institute, Paavo Havaksen tie 3, 90570 Oulu, Finland Department of Economics, Oulu Business School, University of Oulu, Pentti Kaiteran katu 1, 90014 Oulu, Finland

H I G H L I G H T S








Study of hybrid heating where supplementary and main heating systems are combined. Choice experiment is applied to study the determinants of hybrid heating adoption. Hybrid heating appears to be generally accepted among households. Households exhibit differing attitudes toward hybrid heating. Policy makers should not underestimate the potential of hybrid heating.

art ic l e i nf o

a b s t r a c t

Article history: Received 1 October 2015 Received in revised form 18 March 2016 Accepted 11 April 2016 Available online 12 May 2016

The residential heating sector presents considerable energy savings potential, as numerous heating solutions for reducing electricity consumption and utilizing renewable energy sources are available in the market. The aim of this paper is to examine determinants of household heating system choices and to use this information for policy planning purposes. This paper investigates residential homeowner attitudes regarding innovative hybrid home heating systems (HHHS) with choice experiment. Heating system scenarios are designed to represent the most relevant primary and supplementary heating alternatives currently available in Finland. The choice sets include six main heating alternatives (district heat, solid wood, wood pellet, electric storage heating, ground heat pump and exhaust air heat pump) that are described by five attributes (supplementary heating systems, investment costs, operating costs, comfort of use and environmental friendliness). The results imply that HHHSs generally appear to be accepted among households; however, several factors affect perceptions of these technologies. The results reveal differing household attitudes toward the main heating alternatives and show that such views are affected by socio-demographic characteristics (age, living environment, education, etc.). The results suggest that households view supplementary heating systems (especially solar-based) favorably. The other attributes studied also play a significant role in decision making. & 2016 Elsevier Ltd. All rights reserved.

JEL classification: C25 D12 Q40 Q48 Q55 Keywords: Hybrid home heating system Heating Choice experiment Discrete choice

1. Introduction Sustainable energy usage is a key element and driver of the modern world. This decade, the European Union (EU) has sought to tackle three key objectives known as the ‘20–20–20′ targets, which emphasize the reduction of greenhouse gas emissions, growth in renewable energy usage and improvements of energy efficiency. Residential energy demand plays an essential role in achieving these targets. Households consume a quarter of all

n Corresponding author at: Finnish Environment Institute, Paavo Havaksen tie 3, 90570 Oulu, Finland E-mail address: enni.ruokamo@ymparisto.fi

http://dx.doi.org/10.1016/j.enpol.2016.04.017 0301-4215/& 2016 Elsevier Ltd. All rights reserved.

energy consumed in the EU (European Commission, 2013). More specifically, a large fraction of this energy consumption is attributed to the residential heating sector, which focuses on heating household spaces and water (Pardo et al., 2013). The residential heating sector presents considerable energy savings potential. Several technical heating solutions that utilize renewable energy sources and/or that reduce energy consumption levels are available in the market. These energy efficient heating solutions serve as relevant and cost-efficient alternatives in all climate conditions around the world. However, households have been slow to switch to heating systems of superior environmental performance (see Connolly et al. (2013)). A major share of residential energy consumption is used for heating, especially in cold climate conditions. In Finland, over 80% of

E. Ruokamo / Energy Policy 95 (2016) 224–237

annual household energy consumption is dedicated to heating household spaces and water (Statistics Finland, 2012b). Additionally, approximately 50% of Finns live in privately owned detached houses (Statistics Finland, 2012a). The Finnish residential heating sector thus presents the capacity to meet renewable energy targets and to reduce overall energy demand levels and greenhouse gas emissions.1 Vihola and Heljo (2012) detected trends in heating methods used in Finnish buildings between 2000 and 2012. They discovered that the residential heating sector and the technologies that are used are changing. They identified a rapid decline in oil and direct electric heating usage alongside an increase in ground heat pump usage. While it has traditionally been common to rely primarily on one heating system, households are now beginning to utilize a combination of complementary heating technologies as a result of technological advances. This emerging trend warrants our study of innovative hybrid home heating systems (HHHS) and of household views of these technologies. Such analyses are needed, as knowledge of household preferences facilitates the promotion of HHHSs. In combining more than one energy source for household space and water heating, HHHSs serve as an alternative to traditional heating systems (e.g., oil, gas or direct electric heating). HHHSs use a supplementary heating system alongside a primary heating system and can utilize several sources of renewable energy to generate heat (e.g., solar, solid wood, wood pellet and ground heat) as well as outside air and exhaust air.2 Generally speaking, hybrid heating is flexible, cost effective and environmentally friendly for its users. HHHSs can also offer further protection from unpredictable fuel cost increases, as such heating systems do not rely on a single form of technology or fuel source. Moreover, HHHSs are easily adjustable. For example it is feasible to add supplementary heating technologies to central heating system. HHHSs can also be operated via automatized control systems and can thus automatically use the most efficient fuel source available. While hybrid heating systems are growing more popular among households, there are only few studies (see Michelsen and Madlener (2013) and Scarpa and Willis (2010)) on determinants of household HHHS adoption that have simultaneously considered the effects of primary and supplementary heating systems on decision making processes. Previous studies have mainly focused on socio-demographic and motivational factors that influence the adoption of various heating technologies (see Section 2). Furthermore, previous studies have largely examined house renewal activities (both refurbishment and renovating activities), whereas the preferences of individuals who are building new detached houses have not been examined thoroughly (see Section 2). The heating literature lacks a thorough investigation of HHHSs. Most studies have focused on individual heating alternatives; in turn, the hybrid nature of space heating has been given little or no attention. This paper addresses this gap. We used a stated preference (SP) method referred to as the choice experiment (CE) method (see Hensher et al., (2015)) to analyze individual preferences of HHHSs. The CE method is a widely used quantitative statistical approach that is employed to analyze individuals’ discrete choices (see Adamowicz et al. (1994); Boxall et al. (1996); Phillips (2012); Viney et al. (2002)). The method has two important features: it allows one to examine hypothetical heating scenarios and to identify trade-offs between different heating alternative attributes.3 Using this method, individuals were presented 1 Finland's goal is to achieve the EU's “20–20–20″ targets and to further increase its share of renewable energy use to 38% by 2020. 2 We mean by exhaust air the waste air leaving the house. 3 With other techniques, we can only study events that have already occurred. Additionally, when we compare the Contingent Valuation (CV) method with the CE method, the latter can be used to identify trade-offs between different attributes. CE studies are conducted to examine an individual's response to changes in attributes of the chosen situation and in the chosen situation as a whole.

225

with a hypothetical setting involving energy efficient heating alternatives and were asked to select their preferred alternative among a predetermined choice set. Each heating alternative was described by a number of attributes and attribute levels. Thus, individuals implicitly made trade-offs between attribute levels related to different heating alternatives presented in choice sets. This study has several objectives. First, it presents general information on household attitudes and perceptions of HHHSs. Second, it examines heating mode choices that households make when presented with various scenarios that involve currently used heating technologies. Moreover, this paper investigates the role that attributes play when households select one type of heating system over another. This study specifically focuses on the hybrid nature of each heating mode. The third goal is to explain patterns of preference heterogeneity among households. Different socio-demographic and behavioral household characteristics are expected to play a significant role in explaining household heating system choices. Finally, an account of how the study results may help facilitate the development of a greener residential heating sector is presented. To examine these issues, the paper is organized as follows. After introducing previous studies related to household heating system choices, the survey design is presented in Section 3. The results are presented in Section 4 and the main findings are discussed in Section 5. Section 6 concludes with policy implications.

2. Previous studies Numerous studies in the field of energy economics have examined household heating system choices. Michelsen and Madlener (2013) divided these empirical studies into three categories based on the nature of preference information used. In this paper, we follow their method and update it. The first category focuses on household-specific data (e.g., socio-demographic, housing or geographic characteristics) by relating such characteristics to heating system choices and energy demand (see Braun (2010), Dubin and McFadden (1984), Nesbakken (2001) and Vaage (2000)). Dubin and McFadden (1984) investigated U.S. household residential energy demand and developed a modeling approach that has been later utilized in many studies (Braun, 2010; Nesbakken, 2001; Vaage, 2000). Vaage (2000) and Nesbakken (2001) examined Norwegian households’ heating mode choices and energy consumption. These studies showed that the electricity and fuel prices have a significant impact on the choice of heating system. The analyses also revealed a high degree of heterogeneity among households. Braun (2010) focused on the determinants of the heating mode choices in Germany. The results implied that regional effects and dwelling features are important for heating system choices. The second category includes empirical data on real adoption actions and on planned decisions that focus on behavioral aspects of heating system adoption (see Bjørnstad (2012), Decker et al. (2010), Decker and Menrad (2015), Mahapatra and Gustavsson (2008, 2009, 2010), Michelsen and Madlener (2012, 2013, 2016) and Sopha et al. (2010)). Sopha et al. (2010) studied Norwegian households’ perceptions of electric heating, heat pumps and wood pellet heating, whereas Bjørnstad (2012) examined levels of investment satisfaction among Norwegian households that use heat pumps and pellet stoves and that have participated in a subsidy program. The latter study showed that economic factors (electricity prices) affect investment satisfaction levels, but importantly, households also value their investments based on multiple dimensions (e.g., heat comfort and technical service availability levels). Mahapatra and Gustavsson (2008, 2009, 2010) investigated the adoption of heating systems, such as district heat systems, heat pumps and wood pellet boilers, among Swedish households. The findings of these studies implied that economic and system reliability factors are deemed the most

226

E. Ruokamo / Energy Policy 95 (2016) 224–237

Table 1 Categorization of CE studies concerning heating system choices. Study

Investigated attributes

Investigated heating systems and/or choice context

Rouvinen and Matero, 2013

Investment costs Operating costs CO2 emissions Fine particle emissions Required own work

Wood pellet boiler Solid wood fired boiler District heat Electricity Ground heat pump Oil boiler

Achtnicht, 2011

Acquisition costs Annual energy-saving potential Payback period CO2 savings Opinion of an energy adviser Public and/or private funding Period of guarantee

Choice between a modern heating system and an improved thermal insulation for home

Scarpa and Willis, 2010; Willis et al. 2011

CE #1: Capital cost Energy bill per month Maintenance cost Recommendation Contract length Inconvenience of the system CE #2: Micro-generation technology Capital cost Maintenance cost Source of recommendation Energy saved by the technology

CE #1: Replacing existing combi-gas boiler with a new system such as ground source heat pump or with a biomass boiler

important among respondents, while environmental factors are often considered less important. By contrast, Decker et al. (2010) showed that environmental factors considerably influence heating decisions made by German households. Additionally, they found that costs, required fuel reserves, information on a given system and public subsidies affect heating system adoption choices. Decker and Menrad (2015) examined the buying behavior of German households with respect to four heating systems (gas heating, oil heating, wood pellet heating and heat pumps) and analyzed the main factors influencing the purchasing choice. Their results suggested that ecological attitudes differentiate the households heating system choices and that economic aspects are more important in the case of more traditional technologies (e.g. oil heating). Michelsen and Madlener (2012, 2013, 2016) studied homeowner preferences for certain heating systems in Germany. Michelsen and Madlener (2012) found clear preference heterogeneity patterns among households for determinants of heating system adoption. Determinants for replacing a heating system in an existing home and for selecting a heating system for a newly built home were found to differ considerably. In the case of newly built houses, heating system preferences were significantly driven by environmental benefits, usage difficulty factors, cost factors and recommendations, and less evidence was found for socio-demographic, household and spatial factor effects on decisions. Michelsen and Madlener (2013) showed that numerous adopters are driven by existing habits and by perceptions regarding the convenience of a given heating system. In the most recent study, Michelsen and Madlener (2016) investigated drivers and barriers behind homeowners’ decisions to switch from fossil fuels to renewable energy sources to heat their homes. They found that lower dependency on fossil fuels and knowledge work as drivers and that old habits and the perceived difficulties in the heating system functionality work as barriers. The third class of research employs SP methods (including contingent valuation (CV) and CE methods) to investigate preferences related to certain attributes of heating systems in both real and hypothetical choice settings (see Achtnicht (2011), Claudy

CE #2: Solar power Solar water Wind turbine

et al. (2011), Rouvinen and Matero (2013), Scarpa and Willis (2010) and Willis et al. (2011)). Table 1 presents a summary of heating technologies and attributes examined in previous CE studies. Using a CE method, Scarpa and Willis (2010) investigated household willingness-to-pay (WTP) levels in relation to energy efficient heating technologies in the United Kingdom. The results indicated that while renewable energy adoption is highly valued by households, this enthusiasm does not account for the higher investment costs of micro-generation technologies. Utilizing the same data, Willis et al. (2011) focused on future renewable energy supply expansion and on potential effects of the aging population. The paper examined whether renewable energy technologies are less likely to be adopted in aging households. The findings suggested that aging households are in some cases more sensitive to investment costs and less likely to adopt supplementary systems compared to the rest of the population. Achtnicht (2011) conducted a CE study on energy retrofits for existing houses in Germany. The results showed that while environmental benefits significantly affect heating system selection patterns, they play no role in terms of insulation choices. Claudy et al. (2011) employed a CV method to elicit Irish household WTPs for micro wind turbines, wood pellet boilers, solar panels and solar water heaters. Their results indicated that the WTP value varies significantly between the four technologies, and it is especially influenced by differing respondent views of these technologies. Finally, one study utilized SP methodologies to investigate heating system choices in Finland. Rouvinen and Matero (2013) conducted a CE study on how different heating system attributes affect private homeowner heating system choices after completing renovations when heterogeneity in homeowner preferences is allowed. Their results showed that investment costs have the greatest effects on heating system choices. However, the other attributes also had notable effects to varying degrees across different heating systems.

E. Ruokamo / Energy Policy 95 (2016) 224–237

3. Survey design 3.1. Theoretical framework of choice experiments The CE technique is an application of the characteristics-based theory of value (Lancaster, 1966) combined with random utility theory (Thurstone, 1927). Assume that a decision maker n can choose among J heating alternatives. Levels of utility relating to each heating alternative j, as evaluated by each individual n, is represented in the following general form

Unj = Vnj + εnj ,

(1)

where Vnj is a deterministic observable component and εnj is a random idiosyncratic error.4 When individual chooses heating alternative j it is assumed that this gives the highest utility among J heating alternatives. For heating system j, the deterministic part of the utility Vj is as follows

Vj = ASCj + δ′ (ASCj⋅Sn) + β′ j Xj .

(2)

In Eq. (2), Xj is a vector that contains attributes of heating alternative and βj are the corresponding system specific coefficients.

Sn contains individual characteristics and ASCj is an alternative specific constant that allows for an intrinsic preference for the heating choice alternative itself. The estimated ASC plays an important role in the case of a labeled CE, as it captures the mean effect of unobserved factors for each main heating alternative. In this study, not everything can be described based on chosen attributes (see Section 3.3), as numerous intangible heating system features (e.g., reputation, first-hand experience and space needs) affect decisions that are captured only by the ASC. McFadden (1974) related the theoretical random utility model to statistical discrete choice models and to the Conditional Logit (CL) model in particular. The CL model is limited in that it generates homogeneous average taste parameter estimates and only makes sense under the independence of irrelevant alternatives (IIA) property. In turn, the Mixed Logit (MXL) model has become a frequently used specification (see Ben Akiva et al. (1997), Revelt and Train (1998) and Train (2009)), as it avoids the IIA property while taking into account preference heterogeneity. In the MXL, the deterministic part of the utility function has an additional random taste term γnj , whose distribution over individuals and alternatives depends on underlying parameters and observed data related to each alternative and individual. In our analysis, we used the CL model as a starting point and then utilized the MXL model to account for taste variations among respondents. We treated the ASCs as random and assigned normal distributions to them. 3.2. Questionnaire development and sampling The survey instrument was carefully developed and tested. The design of the survey instrument was started by identifying possible factors affecting individuals’ heating mode choices based on previous literature (see Section 2).5 Generally speaking, the amount of examined alternatives and attributes is rather limited in CE studies as individuals cannot consider too many of them at the same time. Hence, we began discussions with experts (building 4 The idiosyncratic error is assumed to be independently and identically distributed (IID) with extreme value one (EV1) distribution. 5 We identified the following factors: investment costs, operating costs, payback period, CO2 and fine particle emissions (and other environmental aspects), comfort of use factors, automatized control systems and recommendation by peers. As we wanted to focus on HHHSs, one of the attributes was fixed to be supplementary heating systems since the beginning.

227

authorities, civil engineers and researchers) to determine the most relevant main and supplementary heating technologies available today and the most important attributes with what we could describe these technologies in a realistic way. We had two pilot rounds before executing the final survey. The first pilot survey was executed in September of 2013. Data collection was conducted by interviewing 12 individuals who had recently been issued a building site. These interviews were important in deciding the most relevant attributes and corresponding attribute levels (see Section 3.3). At this point we also tested how to present the investigated attributes in a meaningful and understandable way. In the second broader pilot survey, pilot questionnaires were mailed to 400 Finnish households, wherein heating system choices were considered of interest. In this study, the objective is to examine preferences of individuals who are potentially making heating system decisions today. Correspondingly, the relevant population for this kind of investigation includes all individuals who are building or planning to build a new detached house as well as homeowners who are replacing or planning to replace their existing heating system (i.e. all adult Finns potentially belong to this group). As the heating technologies examined in this study (see Section 3.3) are especially suitable for new detached houses, we decided to use a group of homeowners living in new detached houses and individuals building one as a proxy for potential decision makers. We believe that the preferences of new homeowners and housebuilders should reflect relatively well the preferences of individuals who are potentially making heating system choices today. The sample of second pilot survey was drawn from the Population Information System of Finland, and it included 200 randomly selected households who had built a detached house after 2012 and 200 randomly selected households issued a building license after 2012. We received 78 responses to the second pilot yielding a response rate of 19.5%. The response rate was higher among households who had already finished building their new detached houses than among households who were still building or just starting to build their houses (23.5% vs. 15.5%). In order to receive higher amount of responses we decided to concentrate only on the former group in the final study. The final survey took place in August of 2014 and was executed via mailed questionnaire. Two thousand Finns were selected from the Population Information System of Finland. This sample was randomly drawn from a group of homeowners whose new detached houses had been finished between January of 2012 and May of 2014. We received a total of 432 completed questionnaires resulting in a response rate of 21.6%. This response rate is acceptable, as we employed only one mailing round and sent no reminder letter. The second pilot survey proved critical when we implemented experimental designs to reduce the number of choice profiles shown to respondents in the final study.6 For the final survey, we created 36 choice tasks and blocked these to six questionnaire versions with the help of Ngene 1.1.1. We used the Bayesian efficient D-optimal design in the Conditional Logit framework. Generally speaking, efficient designs are intended to identify designs that are statistically as efficient as possible in terms of predicted standard errors and parameter estimates (see Carlsson and Martinsson (2003)). In efficient designs, parameter prior values are assumed to be known and fixed. However, there is always some 6 In our case, the number of possible choice task profiles outsized as the number of possible choice profiles was (4  4  4  2  2)5(4  4  2  2) E7.0369  1013. Five of the labeled choice alternatives were defined using five attributes. Three had four levels, and two had two levels. The remaining choice alternative was defined using five attributes. Two had four levels, two had two levels and one had only one possible level.

228

E. Ruokamo / Energy Policy 95 (2016) 224–237

uncertainty surrounding true parameter values. To take this uncertainty into consideration, we used Bayesian efficient designs, which make use of random priors rather than fixed priors (see Ferrini and Scarpa (2007), Scarpa and Rose (2008) and Vermeulen et al. (2011)). Thus, when we employed the Bayesian efficient design for the final survey, prior parameter values were based on the priors obtained through the second pilot survey.7 3.3. Labeled choice alternatives and attributes The most recently introduced Finnish building regulations instruct to use water-operated heat distribution systems rather than direct electric heaters in new detached houses of over 120 m2. Hence, we focused on water-operated heating alternatives that cover all relevant, currently used heating systems. The following six main heating alternatives were selected: district heating, solid wood fired boiler, wood pellet boiler, electric storage heating, ground heat pump (i.e. ground source heat pump) and exhaust air heat pump (see Appendix A for descriptions).8 In this study, the labeled CE method was the only way to represent realistic heating choice scenarios for the respondents, as each main heating alternative has label-specific attribute levels (see Table 2). For instance, the realistic investment cost of district heat varies between 6000 and 10,500 euros, whereas this cost varies between 13,000 and 22,000 euros for ground heat pumps. Additionally, labels are recommended, as we estimate alternative specific coefficients (see Hensher et al. (2015)). The main heating systems were described using the following five attributes (shown in Table 2): supplementary heating systems, investment costs, operating costs, comfort of use and environmental friendliness. Supplementary heating system attribute had three alternatives: solar panel/solar water heaters, water-circulating fireplaces and outside air heat pumps. Both solar panel/solar water heaters and outside air heat pumps are well-known supplementary heating systems, and the water-circulating fireplace system is a less familiar alternative (see Appendix A for descriptions). Furthermore, a supplementary heating system is not required, as all of the main heating alternatives can function independently. The supplementary heating system attribute takes varying levels for all of the main heating alternatives, though not for district heating system.9 Note that the presence of a supplementary heating system is taken into account when planning cost attribute levels by increasing the highest levels for investment costs and by decreasing the lowest levels for annual operating costs. The comfort of use and environmental friendliness are categorical variables with three associated levels: satisfactory, good and excellent. To help respondents discriminate between the meanings of these three levels, we utilized both text and emoticons.10 The comfort of use and environmental friendliness 7 We conducted exactly the same kind of CL model with the pilot data as we were using in the design. We used dummy-coding for the categorical supplementary heating system variable and the rest were treated as continuous in the design. 8 We excluded oil heating systems from the analysis, as they do not represent a relevant alternative for new detached houses. While water-operated heating systems called air-to-water heat pumps also exist, they can be combined with exhaust air heat pumps. 9 It is typically inefficient to use supplementary heating systems alongside district heating. 10 Discussions in the first pilot round indicated that using categorical adjectives (satisfactory, good and excellent) instead of actual CO2 emission levels or actual maintenance hours were preferred. Our test group reported that adjectives made the interpretation of the choice tasks easier and more understandable. We also want to mention that even though the environmental friendliness and comfort of use attributes are described with adjectives (causing subjective interpretations), they should preserve the same ordering among individuals.

attribute levels differed across the main heating alternatives. The comfort of use attribute level varied from satisfactory to good in the case of solid wood and wood pellet heating alternatives, as these two wood-based heating systems require more maintenance than the other main heating alternatives. Regarding environmental friendliness levels, good and excellent levels were employed for ground heat, district heat, solid wood and wood pellet heating systems. This division structure was based on energy efficiency levels, emission levels and on relative electricity requirements involved when operating a heating system.11 The CE method was employed using six hypothetical choice tasks (see the example in Fig. 1) presented to each respondent. The respondents were presented with the following description relating to choice situations: “The following six choice situations are descriptions of different heating systems. Please imagine that you are choosing a heating system for a new 150 m2 detached house with a heat distribution system that uses water to transfer heat. The total annual heating energy consumption level of this house is approximately 16,000 kWh (including heating for household water). It is assumed that the detached house includes a fireplace for supplementary heating purposes. In every choice situation, compare the heating alternatives presented and select the best alternative. Make a choice even if some alternatives present some inconsistencies. Answer each choice situation as a new and separate situation.” Between choice situations, the respondents were reminded of the hypothetical setting described above. In 2012, the average size of a new detached house in Finland was 144.7 m2 (Statistics Finland, 2013). Based on such conditions, the given building floor area (150 m2) serves as a good approximation for an average detached house. The presented assumption regarding fireplaces is justified, as most new detached houses include one.12 Hence, a common fireplace is not informative enough to be included for the supplementary heating alternatives. In addition to the choice tasks, the questionnaire included questions on the respondents’ attitudes and awareness of HHHSs, and respondents’ socio-demographic characteristics were recorded.

4. Results We begin this section by presenting descriptive statistics of the data. We then present observations regarding overall views of HHHSs. In the last section, we focus on the CE results. 4.1. Respondents’ descriptive statistics Table 3 presents the respondents’ descriptive statistics and compares the descriptive statistics of the collected sample (N ¼423) with known corresponding statistics of the random sample (N ¼ 2000) drawn from the Population Information System of Finland. The collected sample was representative of the original random sample regarding those variables that were available. Hence, the results are representative for the studied population i.e. 11 While district heat is occasionally generated by burning fossil fuels, district heat plants in Finland are increasingly being powered by biomass. Additionally, district heat plants are usually combined heat and power plants, thus offering efficiency advantages. 12 In Finland, it is common to find more than one heating system in a detached house. At a minimum, most Finns typically have a fireplace for supplementary and/ or back-up heating purposes. However, the number of households using more innovative supplementary heating technologies (e.g., outside air heat pumps and solar water heaters) is increasing (Vihola and Heljo, 2012).

E. Ruokamo / Energy Policy 95 (2016) 224–237

229

Table 2 Attributes and levels. Attribute

Description

Levels

Supplementary heating system

Supplementary heating system is used alongside a main heating system.

District heat: no supplementary heating systems Others: Level 1: no supplementary heating systems Level 2: solar panel and solar water heater Level 3: water-circulating fireplace Level 4: outside air heat pump

Investment cost (€)

Investment costs include costs associated with the heating device and installation as well as space requirements.

District heat: 6000€, 7500€, 9000€, 10,500€ Solid wood fired: 4500€, 7000€, 9500€, 12,000€ Wood pellet: 8000€, 11,000€, 14,000€, 17,000€ Electric storage heating: 6000€, 8500€, 11,000€, 13,500€ Ground heat: 13,000€, 16,000€, 19,000€, 22,000€ Exhaust air heating pump: 7000€, 9000€, 11,000€, 13,000€

Operating cost (€/year)

Operating costs include heating system's annual electricity/fuel consumption and maintenance costs.

District heat: 800€, 1000€, 1200€, 1400€ Solid wood fired: 600€, 850€, 1100€, 1350€ Wood pellet: 750€, 950€, 1150€, 1350€ Electric storage heating: 1050€, 1350€, 1650€, 1950€ Ground heat: 500€, 650€, 800€, 950€ Exhaust air heating pump: 800€, 1000€, 1200€, 1400€

Comfort of use

Solid wood fired and wood pellet: satisfactory, good Comfort of use describes the required work to ensure faultless operation of a heating system, e.g., cleaning and adjusting the District heat, electric storage heating, ground heat and exhaust air heating pump: good, excellent device and adding fuel.

Environmental friendliness

Environmental friendliness describes the ecological facts associated with each available heating system.

District heat, solid wood fired, wood pellet and ground heat: good, excellent Electric storage heating and exhaust air heating pump: satisfactory, good

As a reminder: the heating system is chosen for a new 150 m2 detached house Ground heat

Exhaust air heating pump

Solid wood fired

Wood pellet

Electric storage heating

District heat

Solar panel and solar water heater

Water circulating fireplace

No supplementary heating systems

Outside air heat pump

Watercirculating fireplace

No supplementary heating systems

16000

7000

7000

17000

8500

9000

Operating cost (€/year)

650

1400

1100

1350

1350

800

Comfort of use

Good

Excellent

Satisfactory

Satisfactory

Good

Excellent

Environmental friendliness

Excellent

Satisfactory

Excellent

Good

Good

Good

CHOICE TASK 1

Supplementary heating system Investment cost (€)

I CHOOSE:

Choose the best alternative by ticking one of the above boxes. Fig. 1. Example of a choice task.

homeowners living in new detached houses.13 We did not have information about income, education, forest ownership or living environment for individuals in the original random sample. Even though the studied population included only homeowners living in new detached houses, the results should be generalizable to some extent for all individuals who are making or potentially making heating system choices today. 4.2. General views of hybrid home heating systems It is essential to determine household views of HHHSs in order to obtain a more thorough understanding of both the roles and potential of hybrid heating systems. The questionnaire included several claims in regards to hybrid heating. Fig. 2 presents 13 Note that this random sample included homeowners whose new detached house was completed between January 2012 and May 2014. In 2012, a total of 12,000 detached houses were completed in Finland, whereas the corresponding numbers in 2013 and 2014 were 11,000 and 9,000 (Statistics Finland, 2016).

responses to these claims measured on a five-point Likert scale. A “do not know” option was included to allow respondents to state if they had no opinion or if they had not thought about a particular issue. This alternative option turned out to be essential, as respondents used it fairly often in response to some of the claims. We also asked the respondents to evaluate their level of HHHS knowledge, with the average result being neither high nor low.14 The results reveal that HHHSs appeared to be generally accepted among respondents. HHHSs were also perceived to be ecological, as most of the respondents reported that they could reduce their carbon footprint by using HHHSs. On average, the respondents believed that HHHSs could reduce annual heating costs and increase property resale values. However, HHHS investment costs were often found to be too high. The respondents 14 A positive correlation was found between having limited background knowledge on HHHSs and stating “do not know” in response to the presented statements.

230

E. Ruokamo / Energy Policy 95 (2016) 224–237

Table 3 Descriptive statistics of the respondents and the corresponding random sample. Respondents Random sample Sample size

432

2000

Average 42.6 3.3 Percent

Average 40.5 3.5 Percent

Gender Female Male

25.8 74.2

26.5 73.5

Household’s income (gross, €/month) o4000 4000–5999 6000–7999 8000–9999 410000

17.7 34.0 29.8 10.2 8.3

NA NA NA NA NA

7.7 37.3

NA NA

33.3 21.7

NA NA

Socio-demographic characteristics Age (years) Household size

Education Basic education Matriculation examination or/and vocational degree Polytechnic degree University degree

Living environment Sparsely populated area or small population 36.0 center Town or city 64.0

NA

Forest owner Yes No

NA NA

28.8 71.2

NA

NA ¼ Not available

had mixed views regarding how much maintenance HHHSs require. They also reported a need for relevant information on HHHSs. On the other hand, HHHSs were not viewed as fully trustworthy, and several of the respondents noted that operating such heating systems may require higher than average levels of expertise. Finally, the results show that the respondents did not hold strong opinions or did not have enough background information to express views on claims concerning the need for automatized control systems and concerning the ability to adjust HHHSs by replacing or adding supplementary heating systems. In addition to the HHHS-specific statements, respondents were presented with general heating-related claims. When asked about the development of future energy prices, almost 80% of the respondents noted that they expected electricity prices to increase and that electric heating costs should therefore significantly increase in the forthcoming years. Nearly half of the respondents had received training on energy efficient heating and building solutions, and approximately 20% had used web-based tools for comparing heating solutions. 4.3. Determinants of heating mode choice Before discussing the estimated CE results in detail, we present interesting results regarding the respondents’ main heating choices, which are shown in Table 4. The results reveal ground heating as the most popular system, followed by district heating. Both solid wood heating and exhaust air heating pump systems accounted for approximately 10% of the responses, whereas electric storage heating and wood pellet boiler systems were the least favored alternatives. A full list of determinants (the ASCs are presented in Table 4) of the respondents’ heating choices is presented in Table 5. We

excluded 13 individuals from the sample as they had answered fewer than 3 out of 6 choice tasks or had a missing value for some explanatory variable. The dataset was ultimately composed of 2504 choices for 419 respondents. District heat levels were selected to work as reference categories, as district heat attribute levels overlapped with other label attribute levels. As noted before (see Section 3.1), the ASC played an important role in this study, as it captures intrinsic factors of main heating alternatives (labels) that are not covered by the chosen attributes. On the other hand, labeling alternatives may significantly affect how respondents make their choices. Many studies have highlighted that individuals may have preferences for the label beyond just the attributes of the investigated goods or services (BekkerGrob et al., 2010; Blamey et al., 2000; Czajkowski and Hanley, 2009; Doherty et al., 2013; Shen and Saijo, 2009) and the use of labels may play a significant role in reaching choice outcomes (e.g. reducing the attention respondents give to the attributes). In this study, 80 respondents chose the same main heating alternative in every choice task and stated that it was the best heating alternative in the follow-up question. Hence, we excluded all attributes from the analysis for these respondents, and these respondents’ utility functions were composed of ASCs only. This was done to ensure variation in the examined attributes. The respondents who chose the same alternative in every choice task did not make any trade-offs between investigated attributes and their levels. The main factor affecting their decisions was the labeled main heating alternative.15 The CL and MXL models were estimated using Nlogit5. The results are presented in Table 6, where the CL model is shown in the three middle columns, and the MXL model is presented in the last three columns. The CL model had a reasonable overall fit (0.27) measured as McFadden's Pseudo R2, whereas the MXL model fit improved from 0.27 to 0.39. The estimated MXL model was based on 3000 Halton intelligent draws. The number of estimated parameters was particularly high in this study. Thus, we treated only ASCs as random variables, while the rest of the parameters were treated as non-random variables. We assigned normal distributions to random ASCs. We believe that our use of normal distributions is justified, as preferences for main heating alternatives may have both positive and negative domains. Coefficients for operating costs (OPE_GRO, OPE_EXH, OPE_WOO, OPE_PEL, OPE_ELE, OPE_DIS) and investment costs (INV_GRO, INV_EXH, INV_WOO, INV_PEL, INV_ELE, INV_DIS) presented expected signs and were highly statistically significant in both the CL and MXL models. As operating and investment costs increase, the probability of choosing a system declines and utility levels decrease. To control for the relatively large number of parameters involved, we organized the comfort of use (COM) and environmental friendliness (ENV) variables into label groups (common attribute levels inside each group) and estimated corresponding label group-specific joint parameters (see note in Table 6). Coefficients COMþ, COM-, ENV þ and ENV- presented their expected signs and were highly statistically significant in the CL and MXL models. The COMþ coefficient measures change from level good to excellent for the associated heating system group. 15 Alternatively, labeling effects can be taken into account by employing a discrete mixture modeling approach that probabilistically determines what proportion of the respondents are only influenced by labels when making their decisions (see Doherty et al. (2013)). Doherty et al. (2013) treated these label based choices by composing the utility function of the ASC only. In our case, it was reasonable to exclude attributes without utilizing this modeling approach, because we had access to additional information on the respondents’ decision rules from the follow-up questions. In the follow-up questions the respondents were asked the reason why they had chosen the same main heating system in every choice task. If the respondent stated that the chosen main heating system was truly the best one in every choice task, we excluded all attributes from the analysis for this individual.

E. Ruokamo / Energy Policy 95 (2016) 224–237

0%

231

20%

40%

60%

Disagree

Strongly disagree

80%

100%

HHHSs are suitable for heating detached houses Annual heating costs can be reduced with the help of HHHSs One can reduce one's carbon footprint by using an HHHS to heat a detached house There is not enough relevant infromation available on HHHSs HHHSs increase the resale value of a detached house HHHS usage requires more than average know-how

HHHSs exhibit too high investment costs

HHHSs present high levels of operational reliability

HHHSs are adjustable

One should not use an HHHS without an automatized control system HHHS maintanance is more intensive on average

Strongly agree

Agree

Neither agree nor disagree

Do not know

Number of respondents = 412. Fig. 2. Views of hybrid heating.

Table 4 Main heating system choice. Chosen alternative

Notation

Percent

Ground heat District heat Solid wood Exhaust air heating pump Electric storage heating Wood pellet

GRO DIS WOO EXH ELE PEL

42.4 23.6 11.7 10.7 5.6 3.9

Note that the share of missing answers covers the remaining 2.2%.

Table 5 Definition of explanatory variables. Variable

Notation

Type

Operating cost Investment cost Comfort of use Environmental friendliness Solar panel/solar water heater Water-circulating fireplace Outside air heat pump High monthly gross income Age Higher education degree Forest owner Living in a city

OPE INV COMþ /ENV þ /SOL WAT OUT HIN AGE HED FOR CIT

Continuous Continuous Categorical Categorical Categorical Categorical Categorical Categorical Continuous Categorical Categorical Categorical

Therefore, when a coefficient has a positive sign, the described change increases the probability of selecting an alternative in that group. Correspondingly, the COM- coefficient measures change from level good to satisfactory in the associated heating system

group. When this coefficient is negative, the change decreases the probability of selecting an alternative in that group. The ENV þ and ENV- coefficients were interpreted in a similar way. Examined supplementary heating systems (SOL, WAT and OUT) were compared against a level with no supplementary heating systems. According to the results, the probability of supplementary heating system selection alongside main heating system use varied between alternatives. Approximately half of the coefficients were statistically significant. This was expected, as not every combination of main and supplementary heating systems is equally feasible. Nevertheless, the results of the CL and MXL models indicate that the probability of ground heat, exhaust air heat pump and wood pellet alternative selection increases when the solar panel/solar water heater combination is available, whereas the probability of exhaust air heat pump, wood pellet and electric storage heating alternative selection increases when outside air heating pumps are available. Water-circulating fireplace solely increased the choice probability of ground heat alternative in the CL model. Accounting for systematic preference heterogeneity in the MXL model improved the overall performance of supplementary heating system attributes in the case of the solid wood alternative. In the MXL model, the choice probability of solid wood alternative selection was positively affected by solar panel/ solar water heater and water-circulating fireplace alternatives. 4.3.1. Influence of individual characteristics Differences in the mean coefficients of ASCs in the MXL model suggest that, on average, the respondents preferred ground heat and district heat systems over exhaust air heating pump, solid wood, wood pellet and electric storage heating systems with respect to other aspects not presented in the choice tasks. Additionally, high levels of heterogeneity were found among the respondents with

232

E. Ruokamo / Energy Policy 95 (2016) 224–237

Table 6 Results of the Conditional Logit and Mixed Logit models. VARIABLES

Ground heat ASC_GRO Std. Dev. ASC_GRO ASC_GRO*HIN ASC_GRO*CIT INV_GRO OPE_GRO SOL_GRO WAT_GRO OUT_GRO COMþ _GROa ENV þ _GROc Exhaust air heat pump ASC_EXH Std. Dev. ASC_EXH ASC_EXH*HED INV_EXH OPE_EXH SOL_EXH WAT_EXH OUT_EXH COMþ _EXHa ENV-_EXHd Solid wood ASC_WOO Std. Dev. ASC_WOO ASC_WOO*HED ASC_WOO*FOR ASC_WOO*CIT INV_WOO OPE_WOO SOL_WOO WAT_WOO OUT_WOO COM-_WOOb ENV þ _WOOc Wood pellet ASC_PEL Std. Dev. ASC_PEL ASC_PEL*CIT INV_PEL OPE_PEL SOL_PEL WAT_PEL OUT_PEL COM-_PELb ENV þ _PELc

CL

MXL

Coeff.

Std. err.

WTP

Coeff.

Std. err.

1.477***

0.124

6463

 0.229***  3.480*** 0.327** 0.290** 0.095 0.221*** 0.386***

0.013 0.248 0.143 0.142 0.144 0.059 0.060

1430 1270 (415) 968 1688

2.325*** 2.050*** 0.673***  0.627**  0.338***  5.405*** 0.465** 0.429** 0.129 0.337*** 0.560***

0.367 0.137 0.252 0.264 0.020 0.362 0.189 0.184 0.201 0.075 0.080

 1.138***

0.220

 6303

 0.181***  3.010*** 0.937*** 0.188 0.656*** 0.221***  0.597***

0.031 0.301 0.232 0.271 0.223 0.059 0.141

5190 (1039) 3635 1226  3306

 1.647*** 1.531***  0.512**  0.264***  4.048*** 0.920*** 0.185 0.633** 0.337***  0.780***

0.404 0.165 0.253 0.040 0.387 0.276 0.326 0.271 0.075 0.173

 0.788***

0.196

-3380

 0.233***  2.976*** 0.174 0.126  0.058  0.681*** 0.386***

0.022 0.245 0.194 0.208 0.197 0.126 0.060

(747) (540) (  250)  2921 1655

 1.883*** 2.330***  1.166*** 0.662*  1.711***  0.277***  3.762*** 1.025*** 0.646** 0.363  1.108*** 0.560***

0.623 0.227 0.349 0.378 0.372 0.034 0.384 0.283 0.285 0.276 0.185 0.080

 3.545***

0.529

-55297

0.034 0.503 0.322 0.373 0.379 0.126 0.060

24,479 (6743) 10,580  10622 6018

0.795 0.284 0.362 0.041 0.662 0.420 0.417 0.472 0.185 0.080

 40049

 0.064*  3.160*** 1.569*** 0.432 0.678*  0.681*** 0.386***

 4.430*** 1.752***  1.056***  0.111***  4.450*** 2.074*** 0.514 1.006**  1.108*** 0.560*** -6.748*** 1.672*** 0.023*  0.211***  1.459*** 0.416 0.162 1.322*** 0.337***  0.780***

0.866 0.228 0.013 0.059 0.481 0.350 0.322 0.386 0.075 0.173

 32024

 0.292***  4.670*** 0.337*** 0.560*** 2504  2753.821  4486.566 0.39

0.032 0.272 0.075 0.080

Electric storage heating ASC_ELE Std. Dev. ASC_ELE ASC_ELE*AGE INV_ELE OPE_ELE SOL_ELE WAT_ELE OUT_ELE COMþ _ELEa ENV-_ELEd

 3.628***

0.456

 22007

 0.165***  1.018** 0.434 0.173 1.001*** 0.221***  0.597***

0.051 0.407 0.276 0.252 0.284 0.059 0.141

(2633) (1052) 6072 1343  3621

District heat INV_DIS OPE_DIS COMþ _DISa ENV þ _DISc Number of observations Log likelihood Log likelihood (0) Mc Fadden Pseudo R2

 0.182***  3.350*** 0.221*** 0.386*** 2504  3269.051  4486.566 0.27

0.025 0.203 0.059 0.060

Note: ***,**,*- Significance at 1%, 5%, 10% level. a

Coefficient is common for GRO, EXH, ELE and DIS alternatives. Coefficient is common for WOO and PEL alternatives. Coefficient is common for GRO, WOO, PEL and DIS alternatives. d Coefficient is common for EXH and ELE alternatives. b c

1217 2122

WTP

6874

1375 1269 (382) 996 1655 -6239

3484 (699) 2399 1276  2953  6806

3705 2334 (1311)  4003 2024

18,747 (4642) 9098  10013 5062

(1972) (767) 6273 1599  3700

1154 1918

E. Ruokamo / Energy Policy 95 (2016) 224–237

respect to ASC values, as the magnitudes of coefficients for standard deviations were greater than the corresponding mean coefficients. To explain this heterogeneity between respondents, we introduced an interaction between random ASCs and other covariates.16 The interaction between household gross income (HIN) and ASC_GRO was statistically significant, denoting that variations in the marginal utility of the ASC_GRO can be explained by differences in respondent income levels. In particular, the choice probability of ground heat was greater among high-income households (household monthly income 46000 €). The interaction between respondent age (AGE) and ASC_ELE variables was statistically significant, further implying that higher age levels increase the probability of electric storage heating selection. The statistically significant and negative interaction between the higher education degree (HED) and ASC_EXH variables suggests that highly educated individuals are less likely to opt for exhaust air heat pump systems than individuals with less education. The same relationship was found between higher education levels and solid wood fired heating selection. The household forest ownership (FOR) variable was a statistically significant interaction variable that positively affected solid wood heating selection patterns. The residential living environment (CIT) interaction was found to have a statistically significant effect on ASC_GRO, ASC_WOO and ASC_PEL. The results suggest that individuals living in sparsely populated areas are more likely to select ground heat, solid wood fired and wood pellet alternative systems than individuals living in towns and cities. In explaining preference heterogeneity, we also tested several other intuitively relevant factors that were not included in the presented models,17 as associated coefficients were not found to be significant. The findings suggest that the gender of the respondent does not explain heating mode decisions. In addition, the respondents who had received training on energy efficient heating solutions did not make significantly different choices from the respondents without training. The use of web-based comparison tools for heating systems appeared to be irrelevant in this sample. 4.3.2. Willingness-to-pay and marginal effects Marginal WTP values were calculated from βk /β€ , where βk and β€ are the parameters for the non-cost and investment cost attributes, respectively. Note that in order to obtain meaningful marginal WTP measures, both attributes used in the calculation must be statistically significant (Hensher et al., 2015).18 As shown in Table 6, WTP values varied significantly between the investigated variables. Generally speaking, interpretations of WTP values are not straightforward in terms of absolute values in labeled CEs.19 However, they can be used to describe relative levels of attribute importance for each main heating alternative. In terms of supplementary heating system types, WTP levels for solar panel/solar 16 Note that in the reported CL model, we did not include explanatory variables that interact with the ASCs. However, the results of the CL model with interactions complement those of the MXL model presented in Table 6. 17 Note that we have also tested if respondent's actual main heating mode choice was reflected in his/her hypothetical choices. We introduced an interaction between random ASC and existing system (assuming that respondent was satisfied with it), and this turned out to be a significant factor in our choice model. Respondents’ actual heating system choices were reflected in the hypothetical choices as follows: those individuals who were satisfied with their existing main heating system were more likely to choose the same system also in the hypothetical choice tasks. The authors are leaving thorough investigation of this topic for future research due to space limits and scope of this paper. 18 The WTP values are calculated for dummy-coded categorical variables only. 19 The unreasonably low marginal WTP values of ASC_PEL and ASC_ELE may be attributed to the fact that the respondents were strongly influenced by alternative names (labeling effects, see Doherty et al. (2013)). Hence, the investment cost coefficients (INV_PEL and INV_ELE) were likely underestimated, and this is directly reflected in the larger WTP values found.

233

water heater combinations were consistently higher than those for the other supplementary alternatives. Moreover, the respondents appeared to be more sensitive to good-to-satisfactory changes than to good-to-excellent changes with regard to both comfort of use and environmental friendliness attributes. Direct and cross marginal effects are presented in Tables 7 and, 8. These effects were determined by averaging the specific individual results. Marginal effects reflect changes in main heating alternative choice probability levels ascribed to a unit change (1000 €) for a cost attribute, ceteris paribus. The calculated direct marginal effects denoted that ground and district heating systems were both relatively more sensitive to increases in their own investment costs than the other main heating alternatives. Generally speaking, a unit change in the operating cost had a stronger effect on choice probabilities than a unit change in the investment cost. A change in operating costs had a strong effect on district heat (62%) selection probability levels, followed by exhaust air heat pump (35%) and ground heat (34%) systems. Cross marginal effects indicate that the closest substitute to ground heat was district heat, whereas for district heat, the substitution pattern was not as obvious. Additionally, direct marginal effects with respect to investment costs of wood pellet and electric storage heating systems imply the existence of a labeling effect, as the calculated coefficients were clearly lower for these two than for the other systems, wherein labeling effects were likely not as significant.

5. Discussion The results revealed that ground heating and district heating were favored over the other studied main heating alternatives. This finding is in line with the results of earlier studies (Mahapatra and Gustavsson, 2008, 2010; Rouvinen and Matero, 2013). District heat serves as a common source of space heating for new detached houses in Finland. However, the most popular heating technology source is the ground heat pump system. Other heat pump technologies (e.g., exhaust air heat pumps) have also become prominent in the residential heating market (Motiva, 2012, 2015). There are several possible reasons for the popularity of district heat and heat pump technologies. Our results indicate that both comfort of use and environmental friendliness factors are important for choosing these systems. Additionally, the perceived reliability of these technologies has likely been enhanced due to household learning effects, as the systems are highly popular among households. However, the results show that district heat is relatively sensitive to increases in operating costs, whereas ground heat is relatively sensitive to increases in investment costs. Solid wood fired heating was the third most popular main heating alternative examined in this study. Finnish selection trends are not fully in line with this, as the popularity of solid wood heating has decreased (see Vihola and Heljo (2012)). Nevertheless, wood is often used as fuel for supplementary heating purposes.20 Our results support this conclusion, as some of the households were in favor of using a water-circulating fireplace as a supplementary heating source. Wood is a local renewable energy source, and it is thus easily accessed by forest owners and households living in rural areas. Wood also serves a convenient heat source during potential power cuts, as other heating systems are often supported by electricity. According to our findings, wood pellet boilers and electric storage heaters were the least favored main heating alternatives. Rouvinen and Matero (2013) predicted an increase in the use of wood pellets from 2009 to 2020 by simulating demand for wood 20

Finns typically own a fireplace for supplementary and/or back up heating.

234

E. Ruokamo / Energy Policy 95 (2016) 224–237

Table 7 Direct and cross marginal effects with respect to investment costs. w.r.t. INV GRO EXH WOO PEL ELE DIS

GRO

EXH

WOO

PEL

ELE

DIS

 0.5157 0.0338 0.0364 0.0077 0.0173 0.1108

0.1744  0.2231 0.0299 0.0073 0.0153 0.0934

0.1696 0.0286  0.1750 0.0074 0.0110 0.0670

0.1372 0.0248 0.0193  0.0921 0.0129 0.0554

0.1560 0.0279 0.0233 0.0085  0.1098 0.0609

0.3282 0.0652 0.0455 0.0135 0.0251  0.3299

Note that the calculated marginal effects are all significant at 1% level. Direct marginal effects are bolded. Table 8 Direct and cross marginal effects with respect to operating cost. w.r.t. OPE GRO GRO EXH WOO PEL ELE DIS

 0.3355 0.0554 0.0541 0.0260 0.0190 0.2131

EXH

WOO

PEL

ELE

DIS

0.1234  0.3543 0.0443 0.0226 0.0168 0.2077

0.1111 0.0470  0.2760 0.0225 0.0127 0.1356

0.0773 0.0352 0.0397  0.2552 0.0118 0.1354

0.0903 0.0432 0.0328 0.0234  0.1294 0.1423

0.2351 0.1137 0.0770 0.0415 0.0291  0.6176

Note that the calculated marginal effects are all significant at 1% level. Direct marginal effects are bolded.

pellet systems resulting from heating system renovations and new building construction. By contrast, both our findings and selection trends for the Finnish population (see Motiva (2015)) show a decline in the popularity of wood pellet heating. Previous studies have shown that wood pellet boilers are viewed as difficult to operate and as less functionally reliable (Claudy et al. (2011), Mahapatra and Gustavsson (2008), Michelsen and Madlener (2016) and Sopha et al. (2010)). Our results suggest that wood pellet alternative choices were strongly affected by labels rather than by attributes. Thus, wood pellet heating systems likely present more intangible factors than other heating systems. As for electric storage heating systems, our results complement the real conditions of new home builders in Finland (see Motiva (2015)). High annual operating costs associated with electric storage heating have likely rendered this system less popular (Mahapatra and Gustavsson, 2008). Uncertainty regarding future energy prices may also affect household preferences for electric storage heating. Our results support this conclusion, as most of the respondents expect an increase in electricity prices in forthcoming years along with increases in the operating costs of electric heating. As the hybrid nature of heating systems was examined in this paper, a discussion of respondent preferences for supplementary heating systems is pivotal. Our results show that the respondents favorably view combined solar panel and solar water heater systems and outside air heat pumps. Interestingly, the results reveal persisting views and habits regarding suitable supplementary heating alternatives for electric storage heating. Only outside air heat pumps positively affected electric storage heating choices, whereas the other main heating alternatives were supported by at least two of the examined supplementary heating alternatives. This may be attributed to the fact that electric storage heating is viewed as a very close substitute for direct electric heating. Additionally, the respondents might not recognize the full potential of the accumulating features of the former alternative. Earlier studies support our positive findings in regards to solar-based heating (Claudy et al., 2011; Scarpa and Willis, 2010) and outside air heat pump systems (Bjørnstad, 2012). Water-circulating fireplaces, as the less familiar supplementary heating alternative, did not affect the respondents’ decisions as much as the other supplementary heating alternatives. This may be due to the fact that water-circulating fireplaces require wood as

fuel and storage space for this fuel. Additionally, water-circulating fireplaces can be more time-consuming to operate than other supplementary heating alternatives. In addition, the common fireplace may be too close a substitute to this alternative. We could not conduct a detailed comparison between our results on watercirculating fireplaces and previous findings, as household preferences for this alternative have not yet been studied. When considering the remaining attributes, increasing investment and operating costs reduced the probability of corresponding heating system selection, as expected. These findings support the results of previous studies (Mahapatra and Gustavsson, 2008; Rouvinen and Matero, 2013; Willis et al., 2011). The comfort of use variable emerged as a highly significant factor that affected the heating system decisions of the respondents, and especially when comfort of use levels declined from good to satisfactory. In earlier studies, however, perceptions of this attribute have been shown to vary widely. While Michelsen and Madlener (2012) and Willis et al. (2011) presented similar results to ours, Rouvinen and Matero (2013) showed that the system's prominence was heating system-related, whereas Mahapatra and Gustavsson (2008) stated that the comfort of use variable was less of relevance to decision making outcomes. Finally, the respondents did consider environmental aspects when making decisions, even though ecological differences between heating alternatives were quite minor in the case of HHHSs. By contrast, Mahapatra and Gustavsson (2008) found their respondents to ascribe low priority to ecological factors. On the other hand, more recent studies have shown that environmental factors matter for heating system decisions (see Achtnicht, 2011; Rouvinen and Matero, 2013). However, differences among these findings can be partly due to a fact that environmental and comfort of use aspects have been presented and described to the respondents in differing ways. Socio-demographic variables serve as important determinants of household heating system decisions. We found that living environment was often correlated with heating system choices. Individuals living in rural areas are not as restricted by heating system space requirements than those living in towns and cities. Hence, solid wood, wood pellet and ground heat systems were more popular in rural areas. Additionally, preferences for district heating often depend on site-specific circumstances, as households positioned close to a district heating network can utilize this heating alternative, whereas households in rural areas lack such opportunities (see Mahapatra and Gustavsson (2008)). Second, our results show that older individuals were more willing to adopt electric storage heating systems. This may be attributed to the fact that electric storage heating systems are easy to operate and serve as a more favorable alternative for small households (see Braun (2010)). Additionally, older individuals may be slower to change their behaviors with regard to the adoption of innovative heating technologies (Willis et al., 2011). Third, a higher level of education decreased the probability of exhaust air heat pump and solid wood heating selections. For the latter alternative, this may relate to fuel acquisition and maintenance work requirements of system operation. Not surprisingly, solid wood heating selection was

E. Ruokamo / Energy Policy 95 (2016) 224–237

positively affected by forest ownership, and ground heating (which is associated with relatively high investment costs) selection was correlated with higher income levels. Similar results were presented by Rouvinen and Matero (2013). There exist also some limitations and possible biases that should be addressed when discussing the results of this paper. First, we cannot generalize our results to all homeowners in Finland as our sample was comprised of new detached house owners. However, we believe that this study can give suggestive results over all households who are potentially making heating system decisions today. We collected the responses from individuals who were familiar with the investigated topic in order to draw reliable and informative results. Comparing relatively new heating technologies might have been a cognitive burden for general public and this would have likely led to low response rate and biased estimates. We also decided to exclude homeowners of older detached houses from the analysis, because some of the investigated heating technologies would have not been suitable for older houses. Second, a possible bias may be caused by the restricted set of the investigated main and supplementary heating systems. In general, CEs cannot be conducted with too many labels and attributes, as then respondents would be facing daunting choice tasks. This might, in turn, lead the respondents using heuristics and rule of thumbs to simplify the decision task. To avoid this we concentrated only on those main heating technologies that were relatively new and were supported by a water-operated heat distribution system. Thus, we excluded conventional oil heating and direct electric heating from the analysis. We also restricted some supplementary heating alternatives (e.g., active chimneys and convection fireplaces) from the analysis due to limitation in the number of the respective attribute's levels. Nevertheless, the most important heating technologies were covered in this study. Third, we cannot conclude how respondents perceived solar panels and solar water heaters separately, because they were presented as a combination in the hypothetical choice tasks. It is possible that some respondents considered only one of these technologies while answering. In addition, our study may suffer from biases that are specific for survey based methods such as CEs (see Street and Burgess (2007)). Even though the questionnaire was assigned and sent to the oldest household member, we cannot be entirely sure that the responses were made by one single respondent and not jointly at household level. We acknowledge that joint decisions concerning major household level investments such as heating system choices might have a role. Additionally, respondents might have had incentives to justify their real-life choices in the hypothetical choice tasks (e.g. they might have justified their actual heating mode choice). There is also a risk of misspecification if the respondents did not perceive the information (e.g. attributes and labels) in the way that was intended by the researcher. We have tried to minimize the effect of the latter by testing the survey instrument carefully.

6. Conclusion and policy implications The purpose of this study was to provide information on household attitudes, perceptions and preferences in regards to innovative hybrid home heating systems that utilize supplementary heating solutions along with a main heating system. Our first goal was to examine general household views of different HHHS features. Furthermore, in using the CE method, we aimed to more specifically determine how selected attributes influence household heating system choices. We explicitly considered the hybrid nature of decision making by including supplementary heating as an attribute. We also accounted for preference heterogeneity

235

among households and elaborated on how their characteristics explain taste variations. A number of policy implications can be derived from this study when we consider what policy makers should do to encourage individuals to increase their energy efficiency levels and the share of renewables used for heating. One collection of solutions pertains to subsidies and taxes. It would be beneficial for policy makers to subsidize supplementary heating alternatives, as households appear to view them favorably, though many of these systems have yet to reach their market potential. In particular, solar-based heating should be subsidized, as it is the most favored supplementary heating alternative. Ongoing trends also support supplementary heating alternative subsidization. Due to the recent tightening of energy efficiency regulations and the accelerated construction of low-energy houses, the competitiveness of electric heating has increased. Supplementary heating systems serve as good complementary systems, especially in combination with electric heating, and they can therefore reduce overall energy consumption of the heating solution. Furthermore, overall taxation and subsidy planners should consider the fact that the investigated heating alternatives differ considerably in terms of direct and cross marginal effects of investment and operating costs. It appears that policies that target operating costs may be more effective, as households are more sensitive to changes in operating costs than to changes in investment costs. However, as investment cost subsidization is likely simpler than operating cost subsidization and/or taxation, our findings suggest that investment cost subsidies for heat pump technologies and district heating are likely to be effective. Our empirical analyses illustrate the importance of careful policy targeting, as socio-demographic characteristics clearly affect household heating system decisions. In particular, the living environment (city vs. rural area) of a household plays a key role in heating choices, and policies should take this finding into account. Additionally, environmental features of heating systems should be leveraged in policy planning strategies, as households appear to consider them. Policy makers should cite environmental factors when implementing different policies and when promoting heating systems. Furthermore, policy makers should focus on increasing the efficiency of district heating networks while paying greater attention to fuel used in district heat plants, as district heat is popular among new home builders. Moreover, the results indicate that households hold positive views of HHHSs. From a policy standpoint, this suggests that HHHSs must be taken carefully into account, as higher HHHS utilization will increase the overall energy efficiency of the residential heating sector and render households more heat selfsufficient. Realizing this potential will involve disseminating information on HHHSs. According to previous studies (see Islam and Meade (2013) and Pepermans (2014)), technological awareness significantly affects the adoption probability of innovative technologies, highlighting the need for effective information sharing and education mechanisms. Our results in this area are mixed. Surprisingly, the respondents who had received training on energy efficient heating solutions did not make choices that significantly differed from those of the respondents without training. This also applied to the use of web-based heating system comparison tools, as they also seem to be irrelevant to decision making processes. These findings may be partly explained by technical regulations and instructions listed in the National Building Code of Finland that guide individuals who are building new detached houses.21 Building authorities have been adopting these demands and have 21 There is legislation that specifically focuses on the energy efficiency of buildings.

236

E. Ruokamo / Energy Policy 95 (2016) 224–237

supervised and instructed individuals who are building new detached houses. Our results suggest that careful compliance with the given guidelines has positively affected household heating system decisions. Nevertheless, there remains a need for informative and objective heating system consultation, as most of the respondents claimed that there is not enough valid information available on HHHSs. While some households are growing more familiar with these technologies, the development of information provisions is of vital importance. While this study was conducted in Finland, the findings can be used as guidelines for applying policies in other countries. Finland's residential heating sector can serve as an example to other countries as more energy efficient structures are developed. Proper legislation, clear guidelines, high-quality building supervision and information sharing serve as key elements in achieving a greener residential heating sector. In general, the results indicate that the policy instruments targeting the residential heating sector must be designed so that essential determinants of heating system adoption and taste variations between households are taken into account. Furthermore, policy makers should not underestimate the potential of hybrid heating, as the residential heating sector is likely to move towards utilizing combinations of different heating alternatives, specifically HHHSs. As energy performance of buildings is becoming more important, HHHSs can provide notable energy efficiency gains. Flexible HHHSs are suitable solutions for smart buildings where the heating, electricity and ventilation systems are all integrated. Finally, even though households have positive views of HHHSs in general, not all combinations of main and supplementary heating systems are equally likely to be chosen. Lastly, we want to highlight some areas for future research. One possible research topic would be to investigate household preferences of HHHSs in other countries than Finland. It would be important to determine the potential of hybrid heating and investigate in detail what are the barriers that are slowing down the adoption of these solutions. Additionally, it would be interesting to investigate how homeowners’ actual heating system choices are reflected in their hypothetical choices. From methodological viewpoint in choice experiments, modeling labeling effect explicitly is an important task for future research.

Acknowledgements The author is thankful for valuable comments and suggestions provided by Artti Juutinen and Rauli Svento. Furthermore, the research assistance of Santtu Karhinen and discussions with the City of Oulu's building authorities are acknowledged. Comments from the two anonymous referees improved the paper considerably. Funding from the Academy of Finland Strategic Research Council project BC-DC (AKA292854), the Finnish Cultural Foundation, Maj and Tor Nessling Foundation, Tauno Tönning Foundation and Martti Ahtisaari Institute is also gratefully acknowledged.

Appendix A. Main and supplementary heating systems District heat District heat is usually generated in a combined heat and power production (CHP) plant. Heat is transferred to the end user via water circulation in the district heat network. In a detached house, heat is directed into rooms via water radiators or underfloor heating networks. District heating is very low-maintenance for the end user.

Solid wood and wood pellet heating Solid wood heating involves the use of wood chips, chopped firewood or firewood to generate heat. Wood pellets are used as heating fuel made from compressed wood. Heat from burning wood in boilers is typically stored in a hot water accumulator, from which warm water is directed further into the heat network. Wood chips, chopped firewood and firewood are typically fed into the stoves manually. Pellet stoves are typically more automated, though little maintenance is required weekly. All wood-based heating systems require storage space. Electric storage heating Electric storage heating systems use water to circulate heat around a household. A heat generating apparatus consists of a water accumulator combined with electric resistors. The electric storage heating system is very user friendly. Ground heat pump A ground heat pump is a central heating system that pumps heat from the ground. Ground heat pumps require electricity access to function. Ground heat systems are generally easy to use. Exhaust air heat pump An exhaust air heat pump extracts heat from the exhaust air of a building and transfers this heat to the supply air and water-operated heat distribution system. The exhaust air heat pump requires electricity access to function and is user friendly. Solar panel and solar water heater Solar roof panels convert sunlight into electric energy, whereas thermal roof panels use sunlight to heat water. Outputs are determined by panel areas, efficiency levels and sunlight levels. These two systems are often used in combination. Water-circulating fireplace A water-circulating fireplace circulates heating water internally and pumps this water further into a boiler. Using a heat exchanger, up to 50% of the energy stored in the fireplace's soapstone area is transferred to a water system. Firewood is fed into the fireplace manually. Outside air heat pump An outside air heat pump is a system that transfers heat from outside a building to inside a building, or vice versa. Heating and cooling is accomplished by pumping a refrigerant through the indoor and outdoor coils of the heat pump. Outside air heat pumps require electricity access to function and are user friendly.

References Achtnicht, M., 2011. Do environmental benefits matter? Evidence from a choice experiment among houseowners in Germany. Ecol. Econ. 70, 2191–2200. Adamowicz, W., Louviere, J., Williams, M., 1994. Combining revealed and stated preference methods for valuing environmental amenities. J. Environ. Econ. Manag. 26, 271–292. Ben Akiva, M., McFadden, D., Abe, M., Böckenholt, U., Bolduc, D., Gopinath, D., Morikawa, T., Ramaswamy, V., Rao, V., Revelt, D., Steinberg, D., 1997. Modeling methods for discrete choice analysis. Mark. Lett. 8, 272–286. Bekker-Grob, E.D., Hol, L., Donkers, B., van Dam, L., Habbema, J., van Leerdam, M., Kuipers, E., Essink-Bot, M., Steyerberg, E., 2010. Labeled versus unlabeled

E. Ruokamo / Energy Policy 95 (2016) 224–237

discrete choice experiments in health economics: an application to colorectal cancer screening. Value Health 13, 315–323. Bjørnstad, E., 2012. Diffusion of renewable heating technologies in households. Experiences from the Norwegian Household Subsidy Programme. Energy Policy 48, 148–158. Blamey, R.K., Bennett, J.W., Louviere, J.J., Morrison, M.D., Rolfe, J.C., 2000. A test of policy labels in environmental choice modelling studies. Ecol. Econ. 32, 269–286. Boxall, P.C., Adamowicz, W.L., Swait, J., Williams, M., Louviere, J., 1996. A comparison of stated preference methods for environmental valuation. Ecol. Econ. 18, 243–253. Braun, F.G., 2010. Determinants of households’ space heating type: a discrete choice analysis for German households. Energy Policy 38, 5493–5503. Carlsson, F., Martinsson, P., 2003. Design techniques for stated preference methods in health economics. Health Econ. 12, 281–294. Claudy, M.C., Michelsen, C., O’Driscoll, A., 2011. The diffusion of microgeneration technologies – assessing the influence of perceived product characteristics on home owners. Energ. Policy 39, 1459–1469. Connolly, D., Mathiesen, B.V., Østergaard, P.A., Möller, B., Nielsen, S., Lund, H., Trier, D., 2013. Heat Roadmap Europe, Vol. 2: Second Pre-Study for the EU27. Department of Development and Planning, Aalborg University. Available from: 〈http://vbn.aau.dk/files/77342092/Heat_Roadmap_Europe_Pre_Study_II_ May_2013.pdf〉 (accessed 21.01.15). Czajkowski, M., Hanley, N., 2009. Using labels to investigate scope effects in stated preference methods. Environ. Resour. Econ. 44, 521–535. Decker, T.A., Zapilko, M., Menrad, K., 2010. Purchasing Behavior Related to Heating Systems in Germany with Special Consideration of Consumers’ Ecological Attitudes. University of Weihenstephan, Triesdorf, Straubing Centre of Science, Conference Papers, (no. 137596. Available from: 〈http://EconPapers.repec.org/ RePEc:ags:uwtscp:137596〉 (accessed 05.03.14). Decker, T., Menrad, K., 2015. House owners’ perception and factors influencing their choice of specific heating systems in Germany. Energ. Policy 85, 150–161. Doherty, E., Campbell, D., Hynes, S., van Rensburg, T.M., 2013. Examining labelling effects within discrete choice experiments: an application to recreational site choice. J. Environ. Manag. 125, 94–104. Dubin, J.A., McFadden, D.L., 1984. An econometric analysis of residential electric appliance holdings and consumption. Econometrica 52, 345–362. European Commission, 2013. EU Energy in Figures. Statistical Pocketbook 2013. Available from: 〈http://ec.europa.eu/energy/publications/doc/2013_pocket book.pdf〉 (accessed 12.12.14). Ferrini, S., Scarpa, R., 2007. Designs with a priori information for nonmarket valuation with choice experiments: a Monte Carlo study. J. Environ. Econ. Manag. 53, 342–363. Hensher, D.A., Rose, J.M., Greene, W.H., 2015. Applied Choice Analysis, a Primer, second ed. Cambridge University Press,, Cambridge. Islam, T., Meade, N., 2013. The impact of attribute preferences on adoption timing: the case of photo-Voltaic (PV) solar cells for household electricity generation. Energy Policy 55, 512–530. Lancaster, K.J., 1966. A new approach to consumer theory. J. Polit. Econ. 74, 132–157. Mahapatra, K., Gustavsson, L., 2008. An adopter-centric approach to analyze the diffusion patterns of innovative residential heating systems in Sweden. Energy Policy 36, 577–590. Mahapatra, K., Gustavsson, L., 2009. Influencing Swedish homeowners to adopt district heating system. Appl. Energ. 86, 144–154. Mahapatra, K., Gustavsson, L., 2010. Adoption of innovative heating systems – needs and attitudes of Swedish homeowners. Energy Effic. 3, 1–18. McFadden, D., 1974. Conditional logit analysis of qualitative choice behavior. In: Zarembka, P. (Ed.), Frontiers in Econometrics. Academic Press, Zürich, Switzerland, pp. 105–142. Michelsen, C.C., Madlener, R., 2012. Homeowners’ preferences for adopting innovative residential heating systems: a discrete choice analysis for Germany. Energy Econ. 34, 1274–1283. Michelsen, C.C., Madlener, R., 2013. Motivational factors influencing the homeowners’ decisions between residential heating systems: an empirical analysis for Germany. Energy Policy 57, 221–233.

237

Michelsen, C.C., Madlener, R., 2016. Switching from fossil fuels to renewables in residential heating systems: An empirical study of homeowners’ decisions in Germany. Energy Policy 89, 95–105. Motiva, 2012. Pientalon Lämmitysjärjestelmät [Heating System Alternatives for Detached Houses]. Motiva Ltd, Helsinki. Available from: 〈http://www.motiva.fi/ files/7201/Pientalon_lammitysjarjestelmat_2012.pdf〉 (accessed 10.06.15). Motiva, 2015. Lämmitysjärjestelmien Osuudet Uusissa Pientaloissa 2006 – 2014 [Heating System Shares of New Residential Buildings 2006 – 2014]. Available from: (〈http://www.motiva.fi/rakentaminen/lammitysjarjestelman_valinta〉 (accessed 10.06.15). Nesbakken, R., 2001. Energy consumption for space heating: a discrete-continuous approach. Scand. J. Econ. 103, 165–184. Pardo, N., Vatopoulos, K., Riekkola, A.K., Perez, A., 2013. Methodology to estimate the energy flows of the European Union heating and cooling market. Energy 52, 339–352. Pepermans, G., 2014. Valuing smart meters. Energy Econ. 45, 280–294. Phillips, Y., 2012. Landlords versus tenants: Information asymmetry and mismatched preferences for home energy efficiency. Energy Policy 45, 112–121. Revelt, D., Train, K., 1998. Mixed logit with repeated choices: households’ choices of appliance efficiency level. Rev. Econ. Stat. 80, 647–657. Rouvinen, S., Matero, J., 2013. Stated preferences of Finnish private homeowners for residential heating systems: a discrete choice experiment. Biomass Bioenergy 57, 22–32. Scarpa, R., Rose, J.M., 2008. Designs efficiency for nonmarket valuation with choice modelling: how to measure it, what to report and why. Aust. J. Agr. Resour. Econ. 52, 253–282. Scarpa, R., Willis, K., 2010. Willingness-to-pay for renewable energy: primary and discretionary choice of British households. Energ. Econ. 32, 129–136. Shen, J., Saijo, T., 2009. Does an energy efficient label alter consumers’ purchasing decisions? A latent class approach based on a stated choice experiment in Shanghai. J. Environ. Manag. 90, 3561–3573. Sopha, B.M., Klöckner, C.A., Skjevrak, G., Hertwich, E.G., 2010. Norwegian households’ perception of wood pellet stove compared to air-to-air heat pump and electric heating. Energy Policy 38, 3744–3753. Statistics Finland, 2012a. Official Statistics of Finland (OSF): Dwellings and Housing Conditions. Statistics Finland, Helsinki. Available from: 〈http://tilastokeskus.fi/ til/asas/tup_en.html〉 (accessed 24.09.14). Statistics Finland, 2012b. Statistics: Energy Consumption in Households. Statistics Finland, Helsinki. Available from: 〈http://tilastokeskus.fi/til/asen/2012/asen_ 2012_2013–11–13_tau_001_en.html〉 (accessed 24.09.14). Statistics Finland, 2013. Construction and Housing. Yearbook 2012. Statistics Finland, Helsinki. Available from: 〈http://www.stat.fi/tup/julkaisut/tiedostot/julk aisuluettelo/yrak_raas_201200_2012_7676_net_p2.pdf〉 (accessed 10.12.14). Statistics Finland), 2016. Official Statistics of Finland (OSF): Dwellings and housing conditions. Statistics Finland, Helsinki. Available from: 〈http://www.stat.fi/til/ asas/tie_en.html〉 (accessed 02.03.16). Street, D.J., Burgess, L., 2007. The Construction of Optimal Stated Choice Experiments: Theory and Methods. Wiley,, New York. Thurstone, L.L., 1927. A law of comparative judgement. Psychol. Rev. 4, 273–286. Train, K., 2009. Discrete Choice Methods with Simulation, second ed. Cambridge University Press,, Cambridge. Vaage, K., 2000. Heating technology and energy use: a discrete/continuous choice approach to Norwegian household energy demand. Energ. Econ. 22, 649–666. Vermeulen, B., Goos, P., Scarpa, R., Vandebroek, M., 2011. Bayesian conjoint choice designs for measuring willingness to pay. Environ. Resour. Econ. 48, 129–149. Vihola, J., Heljo, J., 2012. Lämmitystapojen Kehitys 2000–2012 [Development of the Heating Systems 2000–2012]. Tampereen teknillinen yliopisto. Rakennustekniikan laitos. Rakennustuotanto ja -talous. Raportti 10. Tampere. Viney, R., Lancsar, E., Louviere, J., 2002. Discrete choice experiments to measure consumer preferences for health and healthcare. Expert Rev. Pharm. Outcomes Res. 2, 319–326. Willis, K., Scarpa, R., Gilroy, R., Hamza, N., 2011. Renewable energy adoption in an ageing population: heterogeneity in preferences for micro-generation technology adoption. Energy Policy 39, 6021–6029.