TX_1~AT/TX_2~AT International Journal of Energy Economics and Policy | Vol 7 • Issue 6 • 201778 International Journal of Energy Economics and Policy ISSN: 2146-4553 available at http: www.econjournals.com International Journal of Energy Economics and Policy, 2017, 7(6), 78-84. Japanese Households’ Energy Saving Behaviors Toward Social Risks by Conjoint Analysis#1 Shin Kinoshita* Faculty of Economics, Ryukoku University, 67 Tsukamoto-cho, Fukakusa, Fushimi-ku, Kyoto, Japan. *Email: skinoshita@econ.ryukoku.ac.jp ABSTRACT Aftermath the earthquake in March 2011, Japanese face the drastic changes in energy environment. We have been concerned about electric power shortage because nuclear power plants cease their operation. In Tokyo, people suffered from the planned power outage and even in Osaka people has been often required to save electricity use at the peak of demand. In case the shortage of electricity, we should reduce electricity use. In addition, we should promote renewable energy sources such as solar and wind power instead of nuclear power and fossil fuel such as oil and coal to reduce greenhouse gas emissions. I analyze the factors that households reduce electricity use by conjoint analysis. I find that households reduce electricity use if monthly electricity rates increase and if they recognize the possibility of outages. It might be effective to announce the possibility of outages or instability of supply to households as a nudge. Keywords: Energy Saving, Conjoint Analysis, Outages JEL Classifications: C25, L94, Q48 # This study was aided by research funding from Ryukoku University, Japan. 1. INTRODUCTION Since the Great East Japan Earthquake in March 2011, Japanese people have faced drastic changes to the energy environment. The nuclear meltdown at Fukushima was an accident of epic proportions and it is now difficult to operate nuclear power plants in the country due to shifts in public and political opinion away from supporting nuclear power. Since the earthquake, Japan has relied on natural gas (liquefied natural gas [LNG]). However, this is problematic because of greenhouse gas emissions and the increasing international momentum and resolve to mitigate anthropogenic climate change. After the earthquake, households served by the Tokyo Electric Power Company experienced planned power outages. Even households covered by Kansai Electric Power Company have often been required to curb electricity use at the peak of demand in summer and winter while nuclear power plants remain out of operation pending the outcome of inspections. In Japan, energy saving is an important topic because it can help prevent electric power shortages and climate change. The promotion of renewable energy sources such as solar and wind power is similarly important, instead of favoring and focusing on nuclear power and fossil fuels such as oil and coal. Energy saving among households is important because the CO2 emissions of households have slightly increased and the share of residential electric power in household energy consumption is large1. In fiscal year 2011, energy consumption in household sector is more than 2.8 times compared with 1973. The share of electricity in household sector is 50% in fiscal year 2011. There are some approaches to prompt energy saving. Some local governments and electric power companies implement the demand response system to change the electricity demand of consumers. 1 The agency for natural resources and energy in the ministry of economy, trade and industry (2014) “White paper about energy in fiscal year 2013” Available from http://www.enecho.meti.go.jp/about/faq/001/ Kinoshita: Japanese Households’ Energy Saving Behaviors Toward Social Risks by Conjoint Analysis International Journal of Energy Economics and Policy | Vol 7 • Issue 6 • 2017 79 This system changes the patterns of consumers’ behavior through an electricity rate system and an incentive payment when wholesale prices are higher and the system reliability is lower. The demand response system is divided into two types: The electricity rate type and nega-watt deal type. The time-of-day rate system, the real-time rate system and the peak load rate system are the examples of the electricity rate type. The time-of-day rate system sets higher price when electricity use is tight. The real-time rate system changes electricity rates moment by moment in response to the balance of demand and supply. The peak load rate system changes electricity rates at the peak and at the off-peak. In the nega- watt deal, an electric power company contracts with customers and if customers reduce electricity use they gain rewards from the electric power company. The factors affecting households’ energy saving is analyzed using conjoint analysis. Conjoint analysis is one of the stated preference methods and, in this context, households determine whether to save electricity use under a hypothetical situation. Monthly electricity rates, CO2 emissions, the stability of electricity supply and energy sources are adopted as factors. In the short-term, energy saving is justified in terms of ensuring a better match between electricity demand and supply, while in the long-term energy saving is important to reduce greenhouse gas emissions and thus militate against climate change on the one hand while also dealing with security of supply concerns. Both types of energy saving are important. Energy shortages will continue unless nuclear power plants resume operations and renewable energy sources prevail. In the short-term, electricity use should be reduced to minimize the frequency and duration of sudden power outages. After the earthquake, in the Tokyo Electric Power Company’s domain, households experienced planned outages. On the other hand, in the area supplied by Kansai Electric Power Company, households have only been required to reduce electricity use. Realistically, the possibility of outages there is too small. However, if it indeed transpires that nuclear power plants do not resume operations in the future, this increases the salience of saving more electricity in case of energy shortages and planned or unplanned outages. Even if renewable energy sources prevail, electricity supply may be unstable in general because it depends on weather conditions. I analyze that households reduce electricity use if they recognize the possibility of outages and its duration lasts for many hours. Usually, households respond to relevant prices. If electricity rates rise, households will reduce electricity use. However, we want to know whether households save electricity use to avoid climate change and outages. CO2 emissions, the stability of electricity supply and energy sources are non-monetary factors. If they save electricity use for non-monetary reasons, this is an interesting finding. Clearly, outage costs are very high once such outages occur. Some households may save energy for non-monetary reasons that is social norm. Free-riding is also a serious problem in the context of long-term energy saving. Some people will not reduce electricity use on the assumption that others will reduce more electricity use or that any reduction on their part will be in consequential in overall, population level, terms. The relationship between energy saving behavior and preferences for energy sources is focused. It would be suggestive if households who object to nuclear power and support renewable energy sources save electricity use. This paper consists of the following sections. In Section 2, I introduce some related studies. In Section 3, I explain the conjoint analysis employed herein. In Section 4, I present results from a questionnaire. In Section 5, I delineate the econometric methods used to analyze the foregoing survey data and present estimation results in section 6. In section 7, I propose some policy implications. 2. RELATED LITERATURE After the Great East Japan Earthquake in March 2011, many studies about energy saving have been published in Japan. Tanaka and Ida (2013) analyze the energy saving behavior of households in Japan after the earthquake by conjoint analysis. They found that households in the Kanto area tended to reduce electricity use because they experienced planned outages after the earthquake. Mizobuchi and Takeuchi (2013) examine which monetary factors and non-monetary factors have more effects on energy saving for households by a field experiment. They find that monetary factors have more effects on households’ energy saving. Mizobuchi and Takeuchi (2012) use results from a field study to suggest that both economic and psychological factors affect energy saving in households. Among psychological factors, especially, social norms such as individual responsibility have great effects on energy saving. In a subsequent study, and by contrast, the same authors determined that monetary factors have a greater effect on energy saving in households compared to non-monetary factors. Ito et al. (2015) examined households’ energy saving behaviors by a field experiment in Keihanna smart city of Kyoto prefecture. They divide their sample households into three groups: (1) Economic incentive group, (2) moral suasion group and (3) control group. In the economic incentive group, the electricity rate is raised at the peak of demand. In the moral suasion group, households are only requested to save electricity use at the peak of demand. In the control group, households are not subjected to any interventions. That study revealed that households in the economic incentive group save more electricity than those in the moral suasion group. Mizobuchi and Takeuchi (2016) analyze repurchase and additional purchase of energy saving appliances among Japanese households. Households who purchase an energy saving air conditioner can save more electricity than households who do not make such a purchase. Households who purchase an additional energy saving air conditioner can save more electricity, while households who repurchase an energy saving air conditioner do not benefit from such savings. These foregoing studies all concern saving electricity in Japan after the earthquake. Some studies use field experiments but other studies use conjoint analysis. Poortinga, Steg, Vlek and Wiersma (2003) analyzed the effects of social and psychological factors on households’ energy saving by conjoint analysis. They estimate preferences for ways of saving energy such as turning off lights in unused rooms. They find that while households exhibit preferences Kinoshita: Japanese Households’ Energy Saving Behaviors Toward Social Risks by Conjoint Analysis International Journal of Energy Economics and Policy | Vol 7 • Issue 6 • 201780 for different ways of saving energy, they do not concomitantly reduce their electricity consumption. Some papers focus on the effects of nudges. Costa and Kahn (2013) find that nudges are effective. Therein, energy saving reports have effects on energy saving according to a field experiment; the ideology of individuals such as liberal or conservative also affects propensities to save energy. Allcott and Kessler (2015) estimate the consumer welfare implications of nudge effects via willingness to pay (WTP) and find that nudges do indeed increase consumer welfare. Newell and Siikamaki (2013) propose that proper information is effective for nudges. Davis and Metcalf (2014) suggest that good information leads to good consumer choices and contributes to energy saving. Some studies focus on the effects of social norms on energy saving. Allcott (2011) showed that price and non-price interventions have the same impact on households’ energy saving. Each household compares with their neighborhood. On the other hand, Arimura et al. (2014), focusing on the social interdependencies among individuals, find that social norms have little effect on households’ energy saving in Japan. From these studies, manipulation of monetary factors appears to be the most effective way to reduce household energy use. This paper focuses on both monetary and non-monetary factors. In terms of the latter, climate change, outages, and renewable energy sources are all considered. The social costs of outages and climate change are immense. The free-rider problem is also serious in energy saving contexts. Some households might not save electricity use while others do so disproportionately. Grosche and Vance (2009) estimate WTP for free-riding in Germany. When households acquire government subsidies for energy saving interventions, free-riding is induced. Free-riding is calculated by the excess of WTP over the actual cost. Some studies have examined the long-term effects of energy saving by households’ by analyzing electricity consumption data. Allcott and Rogers (2014) compared a treatment and control group where households received a report periodically or received no such report, respectively. In addition, they compare a group where households continue to receive reports and a group where households cease to do so 2 years later in order to analyze similarities/differences between groups in the short- and long-term. They find that even where households cease to receive reports, energy saving effects continue due to the habitual effects. Ayres et al. (2013) note that peer comparison with other households promotes energy saving at a lower cost. 3. CONJOINT ANALYSIS Households’ energy saving behavior is analyzed using a stated preference method, conjoint analysis.2 We can estimate the 2 Referred to Louviere et al. (2000), Kuriyama et al. (2005), Tsuge et al. (2011), Kuriyama et al. (2013) for conjoint analysis. preference of individuals for hypothetical goods or services which have several attributes using this technique. Households choose an option from a set of alternatives framed in terms of hypothetical goods or services. Conjoint analysis is adopted to examine households’ energy saving behavior under the hypothetical situations involving changes in monthly electricity rates, CO2 emissions and the possibility of outages. In conjoint analysis, we present profiles of goods or services which have several attributes to households. A profile which has too few attributes will not allow significant heterogeneity in preferences to be expressed, while a profile which has too many attributes places a cognitive burden on respondents. In general, we adopt five or six attributes. After we decide on attributes and their levels, we construct profiles using the orthogonal planning method to militate against multicollinearity. From various cards which we get through the orthogonal planning method, selecting cards, and their combinations, we construct profiles while removing unrealistic and dominant cards. SPSS conjoint version 17.0 is used to implement the orthogonal planning method. The contingent valuation method (CVM) is another popular stated preference method but it is not a choice experiment. We use CVM when we evaluate users’ values of non-marketable targets such as forests and coastal areas. The following alternatives are presented to households to analyze their behavior vis-à-vis energy saving. Households choose one of the alternatives under some hypothetical situations. • Alternative 1: Decrease by 10–20% (decrease a lot) • Alternative 2: Decrease by 5–10% (decrease a little) • Alternative 3: Unchanged • Alternative 4: Increase. An ordered logit model is used for estimation because these alternatives have a clear order. The following profile attributes are adopted: Monthly electricity rates, emissions of global greenhouse gases such as CO2, the possibility and duration of electric power outages, and the main energy sources which are used to generate electricity. 3.1. Monthly Electricity Rates Monthly electricity rates increase or decrease compared with current rates. The levels are −2000 JPY, −1500 JPY, −1000 JPY, −500 JPY, 0 JPY (unchanged), +500 JPY, +1000 JPY, +1500 JPY, +2000 JPY. Electricity rate are related with energy sources. Electricity generated by nuclear power might be cheaper. Electricity generated by thermal power might be higher due to the volatile fuel prices and foreign exchange rates. Electricity generated by renewable energy sources might be higher due to the feed-in tariff. The electricity rates are monetary factors. 3.2. CO2 Emissions CO2 emissions will increase or decrease in 2030 compared with 1990 which is the benchmark year of the Kyoto Protocol. The levels are −20%, −10%, 0% (unchanged), +10% and +20%. CO2 emissions are related with energy sources. Nuclear power might reduce CO2 emissions. Coal and LNG might increase emissions while renewables might reduce emissions. Households who reduce their electricity use for concerns about climate change may be anchoring Kinoshita: Japanese Households’ Energy Saving Behaviors Toward Social Risks by Conjoint Analysis International Journal of Energy Economics and Policy | Vol 7 • Issue 6 • 2017 81 to a social norm. CO2 emissions are non-monetary factors. We note a free-rider problem. Some households think that they do not need to save electricity use to reduce CO2 emissions if others are doing so. 3.3. The Stability of Electricity Supply The possibility of outages and their duration are presented to households. Sometimes households should reduce electricity use to avoid sudden outages. I examine whether households save electricity use or not when they recognize the possibility of outages: 1. An outage occurs once in a year and lasts for one hour or more (60 min) 2. An outage occurs once in a year and lasts for half an hour (30 min) 3. An outage occurs once in a year and lasts for a few minutes (3 min) 4. An outage occurs once in a year and lasts for a few seconds (0.05 min) 5. No outage (electricity is always supplied constantly). Stability of supply or lack there of, is related with energy sources. In the case of nuclear power, an outage could be caused due to accidents at nuclear power plants. If nuclear power is not used, a planned outage might be experienced at the peak of demand in summer and winter to avoid one or more unplanned outages. If renewable energy sources are used, electricity supply might be unstable due to weather conditions such as short daylight hours or insufficient wind. Households who save electricity use because they are concerned about outages may be adhering to a social norm. If households save electricity use by the notice of outages, it might be effective for local governments or electric power companies to announce the possibility of outages or instability of electric power to households as a nudge. Outage is one of the non-monetary factors. 3.4. Energy Sources An energy supply source set consisting of nuclear power, coal, natural gas (LNG), solar, and wind power are supposed. Households use one of these energy sources. Each energy source is represented by a dummy variable. Coal is the base category. Energy sources are non-monetary factors. Each energy source has some risks. The possible risks associated with each energy source are presented to households. Households recognize these risks when they choose energy sources. Households who prefer renewable energy sources might save their electricity use because they recognize that their electricity supply is unstable due to weather conditions. The possible risks with each energy source are as follows. 1. Nuclear power: Accidents in nuclear power plants 2. Coal: Climate change because of CO2 emissions 3. LNG: Volatile and rising electricity rates 4. Renewable energy: The possibility of outages and instability of electricity supply due to weather conditions. The levels of each variable are summarized in Table 1. An example profile is shown in Table 2. Households choose one of the alternatives about energy saving under the conditions of the profiles. They answer with respect to ten choice questions. Each question has various levels of attributes. To use various profile configurations, households were divided into two groups and each group was asked to answer with respect to ten profiles. Data were collected online via a web-based questionnaire utilizing the services of the Rakuten Research company. The total sample size is 800 households: n = 400 in Kanto3 and n = 400 Kansai4. Data were collected in October 16, 2015. 4. QUESTIONNAIRE RESULTS In this section, I illustrate the results of the questionnaire. Table 3 provides attributes of the sample households. 5. ECONOMETRIC ANALYSIS An ordered logit model is used because alternatives about energy saving have a clear order5. The regression model is * i i iy x= β + ε∑ (1) * iy is the potential utility level of a household based on a random utility model, ε is the error term and β denotes parameters. The mechanisms as to how household choices dictate values of the dependent variable can be summarized as follows: yi =1 if c−1 10,000 75 (9.4) Educational background Junior high school, high school 193 (24.1) Technical school, junior college 170 (21.3) University, graduate school 437 (54.6) Family composition Single 147 (18.4) Two 205 (25.6) Three 197 (24.6) Four 179 (22.4) Fve 58 (7.3) Six or more 14 (1.8) Dwelling type Detached house (including two households’ house) 410 (51.3) Collective housing (condominium, apartment, housing complex etc.) 372 (46.5) Company housing, dormitory housing etc. 18 (2.3) Sex Male 523(65.4) Female 277 (34.6) Age (years) 20–29 38 (4.8) 30–39 150 (18.8) 40–49 240(30.0) 50–59 240 (30.0) 60 and above 132 (16.5) Mean 48.01 Minimum 21 Maximum 69 Kinoshita: Japanese Households’ Energy Saving Behaviors Toward Social Risks by Conjoint Analysis International Journal of Energy Economics and Policy | Vol 7 • Issue 6 • 2017 83 of consciousness to save electricity use is “very conscious: 1,” “a little conscious: 2,” “unchanged: 3” and “not conscious: 4.” The variable of support for renewable energy sources is a dummy variable. If a household chooses renewable energy sources such as solar, wind, geothermal heat, biomass, tidal, and wave power in the first or second priority, I assign 1, else 0. The variable scale of consciousness of CO2 reduction is “reduce more: 1,” “reduce a little: 2,” “no need to reduce: 3” and “increase: 4.” The coefficients of consciousness to save electricity and consciousness for CO2 are positive and significant at the 1% level. Naturally, households who have high consciousness to reduce electricity indeed do so under any conditions regarding the possibility of outages and energy sources. And households who have high consciousness for CO2 reduce electricity. However, the coefficient of support for renewable energy sources is negative and insignificant; thus households who support renewable energy sources do not appear to be motivated to reduce their electricity use. 7. CONCLUSIONS AND POLICY IMPLICATIONS I analyze the conditions under which households reduce, or do not reduce, their electricity use by conjoint analysis. It is necessary to reduce residential electricity consumption. Future electric power shortages are possible and plausible unless nuclear power plants resume operations and renewable energy sources prevail. Moreover, to reduce CO2 emissions and thus mitigate climate change, residential energy demand reduction is important in the long run. We find that households save electricity use when monthly electricity rates and the possibility of outages increase. Households are thus responding to a monetary stimulus in the case of the former and a non-monetary stimulus in the case of the latter, which can be conceived from a social norm perspective. However, CO2 emissions and renewable energy sources are not the conditions to induce households to reduce their electricity consumption. From the results, we support the electricity rate system which raises prices at peak times to reduce demand accordingly. If households are aware of the possibility of outages, they will reduce their electricity consumption. Thus it might be effective for local governments or electric power companies to announce the possibility of outages or instability of supply to households as a nudge. Since the 2011 earthquake there has been a shift in knowledge about and preferences for different energy sources among Japanese people. Some people are against nuclear power and for renewable energy resources. The relationship between preferences for renewable energy resources and energy saving behavior is an interesting topic. Generally, electricity supplied from renewable energy resources is unstable because of its dependence upon the requisite weather conditions. Long-term energy saving is important Table 5: Descriptive statistics Alternative Electricity rate CO2 Outage Nuclear Coal LNG Solar Wind 1 Average 531.217 0.546 15.186 0.174 0.212 0.268 0.207 0.138 Standard deviation 1458.631 14.909 21.753 0.379 0.409 0.443 0.405 0.345 2 Average 428.774 0.316 13.611 0.174 0.221 0.259 0.205 0.141 Standard deviation 1395.248 14.576 21.165 0.379 0.415 0.438 0.404 0.348 3 Average 61.708 −1.930 14.116 0.255 0.172 0.231 0.177 0.166 Standard deviation 1396.903 14.256 22.284 0.436 0.377 0.421 0.382 0.372 4 Average 231.481 −4.074 17.274 0.130 0.130 0.185 0.352 0.204 Sandard deviation 1436.931 13.349 23.853 0.337 0.337 0.389 0.479 0.404 Table 6: Estimation results Variable Coefficient Standard error Z-value P-value Electricity rate −0.000197 0.000017 −11.33 0 CO2 −0.003646 0.003198 −1.14 0.25 Outage −0.002903 0.001079 −2.69 0.01 Nuclear 0.105617 0.111215 0.95 0.34 LNG 0.117357 0.062346 1.88 0.06 Solar 0.246804 0.10591 2.33 0.02 Wind 0.442355 0.109648 4.03 0 Cut1 −1.115342 0.066002 Cut2 0.636091 0.065097 Cut3 3.688323 0.092296 Log likelihood−9285.7997 Pseudo R2 0.0107 Table 7: Estimation results (including households’ attributes) Variable Coefficient Standard error Z-value P-value Electricity rate −0.000197 0.000017 −11.33 0 CO2 −0.003625 0.0032 −1.13 0.26 Outage −0.002882 0.00108 −2.67 0.01 Nuclear 0.107942 0.111253 0.97 0.33 LNG 0.117277 0.062378 1.88 0.06 Solar 0.246738 0.105965 2.33 0.02 Wind 0.443412 0.109717 4.04 0 Income 0.00037 0.014185 0.03 0.98 Education −0.039796 0.026232 −1.52 0.13 Family members −0.034945 0.018054 −1.94 0.05 Detached house −0.485616 0.1384 −3.51 0 Condominium −0.319477 0.137544 −2.32 0.02 Kanto 0.037566 0.041916 0.9 0.37 Cut1 −1.686646 0.170928 Cut2 0.071095 0.169804 Cut3 3.129201 0.181399 Log likelihood−9267.5176 Pseudo R2 0.0126 Kinoshita: Japanese Households’ Energy Saving Behaviors Toward Social Risks by Conjoint Analysis International Journal of Energy Economics and Policy | Vol 7 • Issue 6 • 201784 to avoid climate change and to prepare energy shortages. In this paper, households respond to temporal outages caused by the tightness of electricity demand. However, they do not respond to CO2 emissions. The system should be considered to promote long-term energy saving for households. 8. ACKNOWLEDGMENTS This study was aided by research funding from Ryukoku University, Japan. This paper was presented in the academic meeting of the Japanese Economic Association in September 2016 and the 15th international conference of the Japan Economic Policy Association in October 2015. All errors are mine. I thank for discussants of academic meeting: Prof. Kenichi Mizobuchi and Prof. Satoru Hashimoto. REFERENCES Allcott, H. (2011), Social norms and energy conservation. Journal of Public Economics, 95, 1082-1095. Allcott, H., Kessler, JB. (2015), The Welfare Effects of Nudges: A Case Study of Energy use Social Comparisons, NBER Working Paper Series No. 21671. National Bureau of Economic Research. p1-74. Allcott, H., Rogers, T. (2014), The short-run and long-run effects of behavioral interventions: Experimental evidence from energy conservation. American Economic Review, 104(10), 3003-3037. Arimura, T.H., Katayama, H., Sakudo M. (2014), Do social norms matter to energy saving behavior? Endogenous Social and Correlated Effects, TCER Working Paper Series, E-76, Tokyo Center for Economic Research. The Agency for Natural Resources and Energy in the Ministry of Economy, Trade and Industry. (2014), White Paper about Energy in Fiscal Year 2013. Available from: http://www.enecho.meti.go.jp/ about/faq/001 Ayres, I., Raseman, S., Alice, S. (2013), Evidence from two large field experiments that peer comparison feedback can reduce residential energy usage. Journal of Law Economics and Organization, 29, 992-1022. Costa, D.L., Mattew, E.K. (2013), Energy conservation “Nudges” and environmentalist ideology: Evidence from a randomized residential electricity field experiment. Journal of the European Economic Association, 11, 680-702. Davis L.W., Metcalf, G.E. (2014), Does better information lead to better choices? Evidence from Energy-Efficiency Labels, NBER working paper series 20720, National Bureau of Economic Research. p1-51. Greene, W.H., Hensher, D.A. (2010), Modeling Ordered Choices. Cambridge: Cambridge University. Grosche, P., Vance, C. (2009), Willingness to pay for energy conservation and free-ridership on subsidization: Evidence from Germany. The Energy Journal, 30(2), 135-153. Ito, K., Ida, T., Tanaka, M. (2015), The Persistence of Moral Suasion and Economic Incentives: Field Experimental Evidence from Energy Demand, NBER Working Paper Series No. 20910. p1-44. Kuriyama, K., Shoji, Y., edited (2005), Economic Evaluation of Environment and Tourism. Tokyo: Keisou-Shobou. Kuriyama, K., Tsuge, T., Shoji, Y. (2012), Introduction of Economic Evaluation of Environment for Beginners. Tokyo: Keisou Shobou. Louviere, J.J., Hensher, D.A., Swait, J.D. (2000), Stated Choice Methods Analysis and Application. Cambridge: Cambridge University Press. Mizobuchi, K., Takeuchi, K. (2012), The Influences of Economic and Psychological Factors on Energy-Saving behavior: A Field Experiment in Matsuyama, Discussion Paper, No. 1206. Japan: Faculty of Economics, Kobe University. p1-39. Mizobuchi, K., Takeuchi, K. (2013), The influences of financial and non- financial factors on energy-saving behaviour: A field experiment in Japan. Energy Economics, 63, 775-787. Newell, R.G., Siikamaki, J.V. (2013), Nudging Energy Efficiency behavior: The Role of Information Labels, NBER Working Paper Series 19224, National Bureau of Economic Research. p1-41. Poortinga, W., Linda, S., Charles, V., Gerwin, W. (2003), Household preferences for energy-saving measures: A conjoint analysis. Journal of Economic Psychology, 24, 49-64. The Agency for Natural Resources and Energy in the Ministry of Economy, Trade and Industry. (2014), White Paper About Energy in Fiscal Year 2013. Tanaka, M., Ida, T. (2013), Voluntary electricity conservation of households after the great east Japan earthquake: A stated preference analysis. Energy Economics, 39, 296-304. Tsuge, T., Kuriyama, K., Mitani, Y. (2011), New Technique of Environmental Evaluation. Japanese: Keisou Shobou. Table 8: Estimation results (including consciousness variables) Variable Coefficient Standard error Z-value P-value Electricity rate –0.00021 0.0000182 –11.56 0 CO2 –0.00448 0.0033441 –1.34 0.18 Outage –0.003259 0.0011321 –2.88 0 Nuclear 0.1015609 0.1160521 0.88 0.38 LNG 0.1221245 0.0653314 1.87 0.06 Solar 0.263804 0.1108557 2.38 0.02 Wind 0.4559248 0.1145589 3.98 0 Consciousness for saving 0.476094 0.029571 16.1 0 Support renewable energy –0.035601 0.0461481 –0.77 0.44 Consciousness for CO2 0.5556902 0.0329335 16.87 0 Cut1 0.6237048 0.1028449 Cut2 2.549215 0.1070492 Cut3 5.732475 0.1331753 Log likelihood –8294.5046 Pseudo R2 0.0513