Relevance of Economic Preferences in Electricity Consumption

Titelmasterformat durch Klicken bearbeiten WINPEC Seminar Waseda University, Tokyo 16.05.2018 On the relevance of economic preferences, values, norms, and socio-demographics for electricity consumption Andreas Ziegler University of Kassel Germany

Titelmasterformat durch Klicken bearbeiten 1. Background

Titelmasterformat durch Klicken bearbeiten Energy production and demand Generation of energy strongly contributes to total CO2 emissions and thus climate change if fossil fuels are used or leads to nuclear waste (besides danger of horrible acci- dents) in case of nuclear energy Besides switching to alternative energy sources (e.g. wind energy, solar thermal energy), another direction to limit problems is reduction of energy production and demand Not only firms, but also households strongly contribute to total energy demand (e.g. household sector in Germany consumes more than 25% of total energy and also more than quarter of total electricity) Individual reduction of electricity use is important direction of environmental and especially climate protection

Titelmasterformat durch Klicken bearbeiten Reduction of electricity consumption In addition to command and control regulations (e.g. pro- hibition of traditional light bulbs in EU) and common price based policies (e.g. electricity taxes), voluntary programs encouraging reduction of electricity consumption are in- creasingly popular Examples are specific public information campaigns for electricity conservation or also private initiatives such as German electricity savings check ( Stromspar-Check , www.stromspar-check.de) or electricity saving initiative ( Stromsparinitiative , www.die-stromsparinitiative.de) However, necessary condition for effective programs is knowledge of individual factors that determine electricity use

Titelmasterformat durch Klicken bearbeiten Literature review I Previous studies have already empirically analyzed diffe- rent groups of determinants for electricity use Dwelling characteristics are very important: Dwelling size has strong positive effect on total electricity consumption (e.g. Chong, 2012, Costa and Khan, 2013) Brounen et al. (2012) additionally reveal that dwelling size is positively correlated with electricity consumption per ca- pita (in contrast, number of rooms has negative effect) Another important group of variables are socio-demogra- phics: Brounen et al. (2012) show that household income and age of household head are positively correlated with electricity consumption per capita

Titelmasterformat durch Klicken bearbeiten Literature review II In contrast, Costa and Khan (2013) report negative effect of age of household head on total electricity use Costa and Khan (2013) especially reveal strong effects of political identification and thus individual values, i.e. left- wing households consume less electricity Recently, another group of variables, i.e. economic prefe- rences, are considered as determinants for total energy expenditures Volland (2017) shows that risk-taking preferences are po- sitively correlated with energy expenditures, while trust has negative effect Fischbacher et al. (2015) additionally reveal that future oriented homeowners have lower energy costs

Titelmasterformat durch Klicken bearbeiten Economic preferences I Many studies show that economic preferences (especially risk and time preferences, trust, social preferences, posi- tive and negative reciprocity, e.g. Falk et al., 2016, 2018) generally play important role for individual behavior Risk preferences affect behaviors and outcomes like buy- ing stocks, housing ownership, or patterns of occupational choice (e.g. overview in Dohmen et al., 2012) Time preferences affect long-term outcomes like income or employment (e.g. Golsteyn et al., 2014) Trust plays important role for buying stocks (e.g. Guiso et al., 2008) (and is also relevant for several macroeconomic variables like GDP growth, inflation, or volume of trade between countries, e.g. overview in Fehr, 2009)

Titelmasterformat durch Klicken bearbeiten Economic preferences II Reciprocal preferences are relevant for employee con- cerns like employee representations (e.g. Jirhahn and Lange, 2015) With respect to environmentally relevant behavior, Fisch- bacher et al. (2015) show that risk taking preferences are relevant for renovation decisions, while patience affects probability to live in energy efficient homes Other studies examine importance of risk and time prefe- rences for energy efficient or energy saving measures (e.g. Epper et al., 2011, Newell and Siikam ki, 2015) Kotchen and Moore (2007) show relevance of social pre- ferences for participation in green electricity tariff

Titelmasterformat durch Klicken bearbeiten Contribution of study I Empirical analysis of determinants of electricity use Basis: Unique data from representative large web-based survey among more than 3700 German households Part of research project Energio , funded by German Fe- deral Ministry of Education and Research (co-operation partners in Germany: ifo Institute, University of Hamburg, Institute of Energy Economics at University of Cologne) Econometric analysis includes all relevant variable groups considered so far to avoid distorted estimation results due to omitted variable biases Analysis of dwelling characteristics, socio-demographics, and individual values (e.g. political identification, environ- mental preferences, religious affiliation)

Titelmasterformat durch Klicken bearbeiten Contribution of study II Also psychological motives like feelings of warm glow and social norms from society and social environment are considered, which e.g. have strong effects on individual climate protection activities (e.g. Schwirplies and Ziegler, 2016) Especially six key economic preferences (i.e. social and time preferences, trust, risk preferences, positive and ne- gative reciprocity) are examined Latter four factors are measured by experimentally valida- ted questions Additional inclusion of artefactual field experiments for measurement of social preferences (based on dictator game) and time preferences (i.e. patience)

Titelmasterformat durch Klicken bearbeiten 2. Data and variables

Titelmasterformat durch Klicken bearbeiten Data Basis: Large-scale computer-based survey among 3705 German citizens Survey was carried out in June and July 2016 by German market research company Psyma Population: Citizens who are responsible for electricity bills and choice of electricity providers and tariffs (mostly heads of household) Stratified sample in terms of age, gender, place of resi- dence, religious affiliation Questionnaires comprised several groups of questions including personal values and economic preferences, de- tails about electricity consumption and costs, and choice experiment for different electricity tariffs

Titelmasterformat durch Klicken bearbeiten Dependent variable It is difficult to collect reliable data on electricity consump- tion in surveys due to poor knowledge of households Participants from panel of market research company Psy- ma were asked to pick up last annual electricity bill and to indicate annual overall electricity consumption (besides electricity costs, provider, and tariff) Analysis of (logarithmized) total electricity consumption (in kilowatt hours) and equivalent electricity consumption (in line with equivalent income in official statistics, first adult in household is weighted with factor one, children with 0.3, and other older household members with 0.5) OLS estimation of linear regression models, heteroske- dasticity-robust z-statistics

Titelmasterformat durch Klicken bearbeiten Social and time preferences Social and time preferences were identified by two incen- tivized artefactual field experiments For both experiments respondents were informed that about 1% of them are randomly selected to be paid, res- pectively Respondents were also informed that both experiments are completely independent Furthermore, respondents were informed that winners are immediately notified after survey and that Euro amount is credited in bonus points on account as member of Psyma panel

Titelmasterformat durch Klicken bearbeiten Social preferences I Experiment is based on standard dictator game, where generosity is costly Respondents were informed that they can divide amount of 100 Euro with another randomly selected respondent in case that they belong to winner in lottery Choice table:

Titelmasterformat durch Klicken bearbeiten Social preferences II Frequencies of payment amounts (in Euro) for other res- pondents: Variable social preferences : Amount allocated to other respondents divided by 100

Titelmasterformat durch Klicken bearbeiten Time preferences I Respondents were presented choice table with 12 deci- sion situations and asked to make decision in every row Choice table:

Titelmasterformat durch Klicken bearbeiten Time preferences II Higher amount at which respondent switches to payment in seven months implies higher discount rate and thus lower patience Variable patience : Minimum discount factor calculated as ratio between 80 Euro and value at which respondent chooses amount in seven months for first time Distribution of discount factors:

Titelmasterformat durch Klicken bearbeiten Risk preferences Based on survey question from German Socio-Economic Panel (SOEP) that was experimentally validated and shown to be reliable indicator for willingness to take risks Respondents were asked about general willingness to take risks on five-stage scale (instead of eleven-stage scale in SOEP) with response categories not at all willing to take risks , rather not willing to take risks , undeci- ded , rather willing to take risks , and very willing to take risks , for which integers from one to five are assigned On basis of ordinal variable, dummy variable risk prefe- rences takes value one for two highest values of ordered variable

Titelmasterformat durch Klicken bearbeiten Trust attitudes Variable is also based on experimentally validated survey questions from SOEP and thus on three statements: In general, one can trust people Nowadays one cannot rely on anyone When dealing with strangers, it is better to be careful before you trust them Respondents were asked how strongly they agree with statements on symmetric scale with five ordered response categories from totally disagree to totally agree , for which increasing integers from one to five for first item and decreasing integers from five to one for two latter items are assigned Variable trust : Sum of values of three dummy variables on basis of two highest values of ordinal variables, res- pectively

Titelmasterformat durch Klicken bearbeiten Positive and negative reciprocity Variables are in line with survey questions from SOEP and thus based on six statements: If someone does me a favor I am ready to return it I am particularly trying to help someone who has helped me before I am willing to pay costs to help someone who has helped me before If I am faced with a great injustice, I will avenge myself at next opportunity If someone puts me in a difficult position, I'll do same with him If someone insults me, I will also be offensive to him Five ordered response categories vary again between to- tally disagree and totally agree , for which integers from one to five are assigned Variables positive reciprocity and negative reciprocity : Sum of values of three dummy variables, respectively, on basis of two highest values of ordinal variables

Titelmasterformat durch Klicken bearbeiten Descriptive statistics

Titelmasterformat durch Klicken bearbeiten Political identification Instead of simple one-dimensional indicator for right-wing or left-wing political identification, four statements are con- sidered: I identify myself with conservatively oriented politics I identify myself with liberally oriented politics I identify myself with socially oriented politics I identify myself with ecologically oriented politics Five ordered response categories vary again between to- tally disagree and totally agree , for which integers from one to five are assigned Dummy variables conservative , liberal , social , and ecological on basis of two highest values of ordinal vari- ables

Titelmasterformat durch Klicken bearbeiten Environmental preferences Environmental preferences on basis of NEP is index on basis of six standardized statements from social and be- havioral science Humans have the right to modify the natural environment to suit their needs Humans are severely abusing the planet Plants and animals have the same right to exist as humans Nature is strong enough to cope with the impacts of modern industrial nations Humans were meant to rule over the rest of nature The balance of nature is very delicate and easily upset Five ordered response categories vary again between to- tally disagree and totally agree , for which integers from one to five are assigned Variable NEP : Sum of values of six dummy variables on basis of two highest values of ordinal variables

Titelmasterformat durch Klicken bearbeiten Religious and other social norms Four dummy variables for religious affiliation: Catholic , Protestant , other religious affiliation , and no religious affiliation (base category) Dummy variable warm glow for perceived good feelings from contribution to environmental protection (on basis of five ordered response categories, which vary again bet- ween totally disagree and totally agree) Similarly, dummy variable expectation society for per- ceived expectation of society to contribute to environmen- tal protection Similarly, dummy variable contribution environment for perceived contributions of friends, family, and colleagues to environmental protection

Titelmasterformat durch Klicken bearbeiten Descriptive statistics

Titelmasterformat durch Klicken bearbeiten Socio-demographics Age (measured in years) Dummy variable female if respondent is woman Dummy variable university for high education Dummy variable Eastern Germany if respondent lives in new Eastern federal states of Germany including Berlin Household size is number of persons in household of respondent Dummy variable goodhealth if respondent describes own health as rather or very good on five-stage scale from very bad to very good Dummy variable high income if household income of respondent is higher than median (i.e. at least 2500 Euro)

Titelmasterformat durch Klicken bearbeiten Descriptive statistics

Titelmasterformat durch Klicken bearbeiten Dwelling characteristics Four dummy variables for housing status with respect to rental/ownership and apartment/house: Rental apart- ment , rental house , own apartment , and own house (base category) Logdwelling size (measured in square meters) Three dummy variables for type of domestic heating: No heating with electricity , partly heating with electricity , completely heating with electricity (base category) Three dummy variables for type of water heating: No wa- ter heating with electricity , partly water heating with elec- tricity , completely water heating with electricity (base category)

Titelmasterformat durch Klicken bearbeiten Descriptive statistics

Titelmasterformat durch Klicken bearbeiten 3. Econometric results

Titelmasterformat durch Klicken bearbeiten Summary Patience and conservative and social policy identification have significantly negative effect on electricity use Persons who are younger, are female, have university de- gree, are from Eastern Germany, and have good health consume significantly less electricity High income is significantly positively correlated with total electricity consumption Household size has significantly positive effect on total electricity use, but significantly negative effect on equiva- lent electricity use Households with larger dwelling sizes, living in houses, and heating with electricity consume significantly more electricity

Titelmasterformat durch Klicken bearbeiten Economic preferences

Titelmasterformat durch Klicken bearbeiten Values and psychological motives

Titelmasterformat durch Klicken bearbeiten Socio-demographics

Titelmasterformat durch Klicken bearbeiten Dwelling characteristics

Titelmasterformat durch Klicken bearbeiten 4. Discussion and conclusions

Titelmasterformat durch Klicken bearbeiten Economic preferences I Patient persons consume significantly less electricity, pos- sibly due to more investments in energy efficient technolo- gies, e.g. appliances Other economic preferences have no significant effect on electricity use in Germany In contrast, economic preferences are even more relevant for alternative individual behavior and attitudes Individual change of electricity tariffs: While patience has significantly positive effect on general changes of electrici- ty tariffs, social preferences and trust have significantly positive effect on specific change to green electricity ta- riffs

Titelmasterformat durch Klicken bearbeiten Economic preferences II Environmental preferences: Social preferences and posi- tive reciprocity are significantly positively correlated with NEP Trust and negative reciprocity are significantly negatively correlated with NEP To avoid omitted variable biases, results generally imply that economic preferences should be considered in eco- nometric models for explanation of individual behavior and attitudes, especially when environmental preferences are included

Titelmasterformat durch Klicken bearbeiten Values and psychological motives With exception of conservative and social policy identifica- tion, no variable has significant effect on electricity con- sumption Especially environmentally related values (ecological po- licy identification, NEP, warm glow) and environmentally related norms (expectation of society, contribution of en- vironment) are obviously not very relevant Results imply that electricity consumption is not strongly associated with environmental protection in Germany, which is in contrast to other countries like USA In contrast, ecological policy identification has e.g. signifi- cant effect on demand for green electricity tariffs

Titelmasterformat durch Klicken bearbeiten Socio-demographics All socio-demographic variables have significant effects on electricity consumption Higher electricity use of older persons with bad health could be due to longer stays at home Lower electricity use of well-educated persons could be due to more knowledge about energy efficient technolo- gies Different estimated effects of household size on total and equivalent electricity consumption clearly reveal savings, e.g. due to less appliances (e.g. refrigerator) per capita in larger households

Titelmasterformat durch Klicken bearbeiten Dwelling characteristics Estimation results for dwelling size suggest non-linear ef- fects since dwelling size has no significant effect if it is in- cluded without logarithm Estimation results reveal that ownership of dwellings only play minor role for electricity consumption In contrast, living in houses leads to significantly more electricity use than living in apartments, possibly due to more appliances in houses (e.g. for own gardens) Estimated positive effect of water heating with electricity on electricity consumption is higher than estimated effect of general heating with electricity

Titelmasterformat durch Klicken bearbeiten Implications Empirical results suggest several directions e.g. for policy to decrease electricity consumption For example, high importance of patience and education suggests information about economic gains in future for investments in energy efficient technologies Significant effects of additional socio-demographic variab- les and of some dwelling characteristics suggest targeted information campaigns or other initiatives for specific po- pulation groups (e.g. orientation towards males or per- sons living in houses has higher potential for decrease of electricity use) Methodologically, empirical analysis contributes to pre- vious studies that transfer laboratory experiments to field

Titelmasterformat durch Klicken bearbeiten Direction for further research Robustness checks with alternative model specifications Theoretical foundation of estimation results Analysis of relationship between general education and energy literacy with respect to electricity consumption Comparison of estimation results with results in other countries beyond USA, e.g. Japan?

Titelmasterformat durch Klicken bearbeiten Thank you