Should we Measure Regional Subjective Well-Being?

Transcription

1 OECD, June 11th 2012 Should we Measure Regional Subjective Well-Being? Claudia Senik Paris School of Economics and University Paris-Sorbonne

2 Differences in regional well-being? Compensating differentials (Adam Smith) Mobility of individuals + Homogeneity of individual preferences + Amenities: ü People will move to places with better local amenities until the price of housing rises and the labor wage decreases so as to equalize net utility across regions è The prices of housing and labor should compensate fully the welfare effect of regional amenities and externalities. è So in principle, well-being should be the same in all points of any integrated geography with free and costless mobility. 2

3 Compensating differences Arbitrage: intrinsically nicer places should pay lower wages in equilibrium ü Roback 1982, Hoehn et al. 1987, Blomquist et al. 1988, and Krupka and Donaldson 2007 ü In equilibrium, wages and prices must adjust to equalize utility across places. Assumptions (perfect competition): ü sufficiently low mobility costs ü sufficiently accurate levels of information about what it would be like to live in another region (transparency) ü sufficiently accurate perception of costs, benefits and risks of amenities/ externalities 3

4 Valuing local public goods or amenities Local governments may still want to evaluate the well-being associated to a particular local public good or amenity ü Public goods Health, hospitals, schools, roads, green parks, pollution ü Quality of local governance Local democracy, local efficiency ü Local social capital Associations, non-market interactions, local culture, trust ü Criminality, inequality, poverty ü Weather, temperature, rainfall 4

5 Valuing local public goods or amenities 3 methods: ü Hedonic pricing ü Contingent valuation ü Subjective self-declared Happiness (Life Satisfaction) 5

6 1) Hedonic Prices Revealed preferences method Assumptions: Rationality + Perfect information + Instantaneous price adjustment + No obstacle to free mobility of individuals accross diffferent areas If the public good is complementary to private goods such as housing, information on public good demand is then embedded in the prices of the private goods (compensating difference in the price of this private good) To estimate Hedonic prices, regress housing prices on amenities or externalities, include in the regression all the relevant dimensions of the amenity + all associated market goods prices ü Retrieve the coefficients on amenities 6

7 Example: hedonic price of air quality (SO2) Regional differences in clean air can arise from different natural conditions or different policy choices or different economic activity that also have an impact on wages and housing prices. Important to take into account all the trade-offs otherwise biased estimates of hedonic prices. 7

8 Drawback of Hedonic Prices Market Imperfections ü Moving costs coefficients on positive amenity biased downward The benefits of the amenities must compensate people from cost of housing prices and wages + cost of moving Or, moving costs too high less mobility no adjustment of housing prices and wages no equalization of utility across areas. The benefits of the amenity are only incompletely capitalised in house prices. Bounded rationality, cognitive limitations Individual s behaviour in private markets is governed by perceived rather than objective gains, costs and risk. Example: pollution. To the extent that people are ignorant about pollution levels and effects, these effects are not reflected in private markets. Research suggests that hedonic estimates indeed substantially underestimate the benefits of clean air or other local amenities. 8

9 2) Stated Preferences Contingent valuation and choice modeling: direct method for eliciting individuals preferences. ü Asking people to compute their willingness to pay for amenities, public good, etc. e.g. improvements in air quality Drawbacks: ü strategic behavior of respondents ü lack of consideration of budget constraints ü difficult to handle all trade-offs among several substitutes 9

10 3) Life satisfaction Self-reported life satisfaction: proxy for underlying latent utility ü Happiness: If you were to consider your life in general these days, how happy or unhappy would you say you are, on the whole? 1. Not at all happy; 2. Not very happy; 3. Fairly happy; 4. Very happy ü Life Satisfaction: All things considered, how satisfied are you with your life as a whole these days? 1(dissatisfied) 10 (very satisfied) 10

11 New data sources with subjectives attitudes National Household Surveys ü General Social Survey (USA, ) ü German Socio-Economic Panel Study (GSOEP, ), ü British Household Panel Survey (BHPS, ), ü Australian Household Panel Survey (HILDA, ) ü Russian Longitudinal Monitoring Survey (RLMS, ) ü Ukrainian Longitudinal Monitoring Survey (ULMS) ü Korean Labor & Income Panel Study (KLIPS) ü 11

12 International Surveys ü World Values Survey (97 countries, ) ü European Social Survey (32 countries, ) ü Gallup World Poll (150 countries) ü Life in Transition Survey (29 countries, 2006, 2010) ü Latino-Barometer, ü Euro-barometer ü International Social Survey Program 12

13 Matching datasets Household surveys Fiscal data Census and other geo-localized datasets Study local differences in well-being 13

14 Why use self-declared happiness? Proxy of utility Useful when no direct relation can be traced between individual preferences and individual actions Non-market interactions: social interactions, externalities ü Examples: inflation, unemployment, income inequality, pollution à there is no price one can pay to buy more or less income inequality on a market 14

15 A synthetic measure of well-being Think of well-being as resulting from different sources: income, health, education, children, marriage, housing etc. Life satisfaction is a synthetic measure of the overall net effect of all these factors on individual well-being. Estimate an average individual utility function 15

16 Is subjective satisfaction reliable? Stability of estimated welfare functions in time and space Cross-rating: with friends and family, or the person who administered the interview IRM, other correlates with physiological symptoms of well-being: ü heart rate and blood pressure, digestive disorders and headaches, coronary heart disease and strokes. ü physical measures of brain activity, measured using electrodes on the scalp or scanners Subjective data predicts future outcomes, i.e. People mean what they say ü Job satisfaction is a good predictor of future job quits ü Well-being scores predict the length of life ü Satisfaction measures predict future marital break-up 16

17 Valuing local amenities with Life Satisfaction Regress LS on the public good of interest, income and other covariates Use the coefficients on the public good and income to calculate utility constant trade-off ratios (Marginal Rates of Substitution) ü Obtain the monetary value of non-marketed goods LS-based approach shares analogies with both revealed preferences and stated preferences methods ü MRS calculated indirectly ü A proxy for utility is directly observable and no reliance on private market transaction necessary to estimate the implicit price of amenities 17

18 Advantage of LS / HP: Does not require the assumption of perfect information, rationality, perfect mobility, and equilibrium on private markets. / Stated preferences: LS can capture the pure existence value of amenities, whereas stated preferences elicits the value of incremental changes. / Stated preferences: LS avoids problems of strategic behavior, lack of consideration of budget constraints and trade-offs among several substitutes. 18

19 Complementarity of HP and LS Change in the level of amenity: ü Direct effect on utility ü Indirect effect via housing prices and wages If equilibrium conditions hold: ü Regression of LS on Amenity, not controlling for housing or wage prices there should be not effect ü Regression of LS on Amenity, controlling for housing prices and wage allows capturing the utility of amenities If equilibrium conditions do not hold: ü regression of LS on Amenity, not controlling for housing or wage prices: statistically significant coefficient on amenities ü Total value of amenities should be calculated with HP and LS (which evaluates the residual term, not captured by HP because not capitalized in prices) 19

20 Complementarity of HP and LS Ferreira and Moro (2010) ü Use HP and LS to value local amenities in Ireland ü Show that there is no equalization of utility (LS) across Irish regions ü Warmer climatic conditions and air quality have a positive effect on wellbeing not fully capitalized Van Praag and Baarsma (2005) ü HP + LS to evaluate the value of noise around Amsterdam airport Luechinger (2009) ü Air quality (SO2 concentration) across American states 20

21 Complementarity of HP and LS - continued The LS approach has also been used to value (the residual effects) of: ü climate (Becchetti et al., 2007; Frijters and van Praag, 1998; Rehdanz and Maddison, 2005), ü urban air pollution (Welsch, 2002, 2006) ü sulphur emissions (Di Tella and MacCulloch,2007), ü urban regeneration schemes (Dolan and Metcalfe, 2008), ü terrorism (Frey et al., 2009) ü flood hazards (Luechinger and Raschky, 2009) 21

22 Simon Luechinger (The Economic Journal, 2009) Valuing local air quality using LS Valuing air quality using life satisfaction ü combining individual-level panel and high resolution SO2 data ü LS, rental prices of houses: GSOEP Simultaneity problems: ü While technical progress and air quality regulations are important reasons for improvements in air quality, local economic downturns and declining industrial production are other likely candidates ü These simultaneous developments have a countervailing effect on life satisfaction and rents. ü To avoid simultaneity problems, instrument exploiting the natural experiment created by the mandated scrubber installation at power plants, with wind directions dividing counties into treatment and control groups. 22

23 Luechinger (2009) Finding: SO2 concentration negatively affects life satisfaction. This indicates that the effects of SO2 are incompletely capitalized in private markets. At most 16% of the total effect of air quality is capitalized in the housing market. Explanations: Migration costs Incomplete information (on pollution levels and risks) Under-estimation of the marginal effect of income 23

24 Use local SWB to construct indices of market imperfections, market incompleteness Following the idea of compensating differentials If markets are complete, income should perfectly predict life satisfaction, and the coefficients on the characteristics of goods (e.g. housing size) should not be significant. Wherever these coefficients are statistically significant, this might indicate that (some) markets are incomplete. Empirical test of the economic arbitrage theory: state-by-state equality of wellbeing for a person of given characteristics? Regional measures of market imperfections ü On which markets and in which regions is capitalization complete/ incomplete? 24

25 Back to air pollution If all effects of air pollution are correctly perceived and the equilibrium condition holds, air pollution should not be systematically related to life satisfaction. But Luechinger (2009) finds that it does SO2 concentration negatively affects life satisfaction. => the effects of SO2 are incompletely capitalized in private markets. Thus, the life satisfaction approach also allows to directly test the fundamental assumptions of the hedonic method (the HP equilibrium conditions) 25

26 Well-being across America Andrew J. Oswald and Stephen Wu (2010) Study life satisfaction and mental health across the states of the USA Sample of 1.3 million citizens ü Behavioral Risk Factor Surveillance System, Life satisfaction (How happy do you feel?, scale -4, very dissatisfied very satisfied) Objective quality of life of American states, calculated by Gabriel, Mattey and Wascer (2003) hedonic housing and labour prices ü Precipitation, temperature, wind speed, sunshine, coastal land, hazardous waste sites, commuting time, violent crime, student-teacher ratio, local taxes, local spending on education, on railways, cost of living. 26

27 Oswald and Wu (Science, 2010) Find no correlation between states regression-adjusted well-being and their income per capita lower wages in nicer places => equal satisfaction No differences in well-being across regions when one runs an estimate of wellbeing without controlling for price of real estate. When control for people s incomes, satisfaction with life is lower in richer states just as compensating-differentials theory would predict. Correlation of 0.6 between objective (price of house rents) and subjective wellbeing, net of income, and other individual circumstances, across regions. 27

28 Oswald and Wu (Science, 2010) Positive correlation between hedonic prices (of housing and labor) and Life Satisfaction net of income and other socio-demographic features. 28

29 Oswald and Wu No correlation between income per capita (wages) and Life Satisfaction net of sociodemographic features (but not income) 29

30 Negative correlation between income per capita (wages) and average Life Satisfaction net of personal income and socio-demographic features 30

31 OLS Estimates of Happiness European Social Survey, Other controls: age, age square, gender, marital status, employment status, migration status. Robust standard errors clustered by region. 31

32 Frey et Stuzer (2002) The value of local democracy Survey of 6000 inhabitants in Switzerland in Life satisfaction (1-10). ü Federal structure of Switzerland. Large competences of cantons. ü Popular referendum initiatives (votations) to change the constitution of the canton, its laws, forbid new laws or new budgetary expenditures. ü Access to these rights varies across cantons. Number of signatures necessary to trigger a referendum, deadline for obtaining signatures, etc. Index of direct local democracy (1 to 6). ü Participation in initiatives and referenda is reserved to citizens with Swiss nationality. ü But foreigners benefit from the outcome of elections and referenda. Estimates: the marginal effect of direct democracy on subjective well-being is 3.4% for Swiss citizens, against 1% for foreigners. à Pure procedural utility: 2/3 of the utility gain from access to direct democracy in one s area of residence à Underlying assumption: low mobility across cantons 32

33 Conclusion LS measures useful for 3 purposes: ü Evaluating regional differences in well-being ü Evaluating local amenities ü Evaluating the competitiveness of markets 33