Empirical well-being measurement

Transcription

1 Empirical well-being measurement On composite indicators, life satisfaction, and equivalent income Koen Decancq (University of Antwerp) ISPRA September 2014

2 Motivation Measuring well-being is a central issue in (social) policy making. Well-being is multidimensional (Stiglitz et al. 2009). New questions: Can we construct an (operational) multidimensional well-being measure? Is there a single silver bullet? Does the choice between the measures matters empirically?

3 Roadmap Part 1. Measuring well-being on a crossroads Building blocks for a well-being measure Principles for a well-being measure An inconvenient result Part 2. Three well-being measures Composite well-being index Life satisfaction Equivalent incomes Part 3. Illustrations The worst off in Germany Beyond GDP: Social Progress in Europe

4 Roadmap Part 1. Measuring well-being on a crossroads Building blocks for a well-being measure Principles for a well-being measure An inconvenient result Part 2. Three well-being measures Composite well-being index Life satisfaction Equivalent incomes Part 3. Illustrations The worst off in Germany Beyond GDP: Social Progress in Europe

5 Building blocks of a well-being measure 1. The outcome vector

6 Building blocks of a well-being measure income health

7 Building blocks of a well-being measure 1. The outcome vector 2. Informed opinion on the good life (aka preference ordering )

8 Building blocks of a well-being measure income health

9 Building blocks of a well-being measure income health

10 Building blocks of a well-being measure 1. The outcome vector 2. Informed opinion on the good life (aka preference ordering ) 3. Satisfaction function (so that )

11 Building blocks of a well-being measure 1. The outcome vector 2. Informed opinion on the good life (aka preference ordering ) 3. Satisfaction function (so that ) A well-being measure:

12 A first well-being measure (the non-starter) We split the outcome vector income non-income Where is again a vector A first (familiar) well-being measure: Resource fetishism (Sen, 1985). We need a multidimensional measure

13 Principles for a well-being measure income health

14 Principles for a well-being measure income health

15 Principles for a well-being measure income health

16 An inconvenient result income health

17 An inconvenient result income health

18 An inconvenient result Trouble in paradise! income health

19 An inconvenient result Based on this little graph we find a deep (and inconvenient) result: As soon as people disagree on the good life, no well-being measure satisfies both principles income health And we have to choose

20 An inconvenient result

21 Roadmap Part 1. Measuring well-being on a crossroads Building blocks for a well-being measure Principles for a well-being measure An inconvenient result Part 2. Three well-being measures Composite well-being index Life satisfaction Equivalent incomes Part 3. Illustrations The worst off in Germany Beyond GDP: Social Progress in Europe

22 Route 1. Use a common view on the good life Based on this little graph we find a deep and annoying result: As soon as people disagree on the good life, no well-being measure satisfies both principles income health

23 Route 1. Use a common view on the good life income health

24 Route 1. Use a common view on the good life income health

25 Route 1. Use a common view on the good life income health

26 Route 1. Use a common view on the good life income health

27 Route 1. Use a common view on the good life A composite index of well-being A popular mathematical structure Degree of substitutability

28 Route 1. Use a common view on the good life A composite index of well-being A popular mathematical structure Degree of substitutability Transformation function

29 Route 1. Use a common view on the good life A composite index of well-being A popular mathematical structure Degree of substitutability Transformation function Weighting scheme

30 Route 1. Use a common view on the good life How to set the weights? Three main approaches 1. Data-driven Depend only on information on outcomes BUT: Hume s guillotine 2. Normative Depend only on the common opinion on the good life BUT: the opinion of whom? 3. Hybrid Depend on both

31 Route 1. Use a common view on the good life Data: LEVO (2010) (n=1552)

32 Route 1. Use a common view on the good life Data: LEVO (2010) (n=1552)

33 Route 1. Use a common view on the good life Example: target the multidimensional poor individuals (i.e. below 60% of median well-being) Multidimensional poverty always sometimes never Who are the worst-off?

34 Route 1. Use a common view on the good life... those with a stake in the outcomes will almost certainly be in a better position to determine what weights to apply than the analyst calibrating a measure of poverty.

35 Back to the cross road Based on this little graph we find a deep (and inconvenient) result: As soon as people disagree on the good life, no well-being measure satisfies both principles income health And take the other route

36 Route 2. Use life satisfaction Why don t we ask the individuals themselves? Subjective Well-Being (SWB) Affects (happiness) Cognitive valuations (life satisfaction)

37 Route 2. Use life satisfaction Why don t we ask the individuals themselves? Subjective Well-Being (SWB) Affects (happiness) Cognitive valuations (life satisfaction) Are the opinions of individuals (preferences) respected?

38 Route 2. Use life satisfaction Under the consistency assumption the preferences of the concerned individuals are respected in intra-personal comparisons What about interpersonal comparisons? A more attractive (useful) principle:

39 Route 2. Use life satisfaction income 5 for Iris 9 for John 3 for Iris 7 for John health SWB does not fulfil the Same Preference Principle SWB does not fulfil the Dominance Principle

40 Route 3. Equivalent Incomes Is there a third route? A measure that satisfies Same Preference Principle Back to the trouble maker: Let s weaken the Dominance Principle income health

41 Route 3. Equivalent Incomes Is there a third route? A measure that satisfies Same Preference Principle Back to the trouble maker: Let s weaken the Dominance Principle income health

42 Route 3. Equivalent Incomes Is there a third route? A measure that satisfies Same Preference Principle Back to the trouble maker: Let s weaken the Dominance Principle income health

43 Route 3. Equivalent Incomes Is there a third route? A measure that satisfies Same Preference Principle Back to the trouble maker: Let s weaken the Dominance Principle income health

44 Route 3. Equivalent Incomes There is a measure that satisfies And that is Equivalent Income Developed in 70s by Samuelson and others Revitalized recently by Fleurbaey and others

45 Route 3. Equivalent Incomes Equivalent income = the hypothetical income that -- if combined with a reference value on all non-income dimensions -- would place the individual in a situation that she finds equally good as her initial situation

46 Route 3. Equivalent Incomes income Ann Bob Perfect health health

47 Route 3. Equivalent Incomes income Ann Bob A Perfect health health

48 Route 3. Equivalent Incomes income Ann Bob Equivalent income A A Perfect health health

49 Route 3. Equivalent Incomes income Ann Equivalent income B Equivalent income A Bob B A Perfect health health

50 Route 3. Equivalent Incomes Equivalent incomes Additional information is neccessary on: The reference values: an ethical question, hence room for debate. The preferences of the individuals (see next part).

51 Three well-being measures In sum Life Sat. Equi. Inc Comp. Index

52 Outline Part 1. Measuring well-being on a crossroad Building blocks for a well-being measure Principles for a well-being measure An inconvenient result Part 2. Three well-being measures Composite well-being index Life satisfaction Equivalent incomes Part 3. Illustrations The worst off in Germany Beyond GDP: Social Progress in Europe

53 The worst off in Germany Data: GSOEP for 2010 (13,900 individuals) Three dimensions in Income Health Unemployment

54 The worst off in Germany Preferences are not observed! Stated preference approach (Fleurbaey Schokkaert, 2013) Revealed preference approach (Decoster Haan, 2014) Life Satisfaction approach (Fleurbaey et al. 2009) Starting point: canonical life satisfaction regression Sophistications Heterogeneity in coefficients Decreasing marginal returns

55 The worst off in Germany

56 The worst off in Germany

57 The worst off in Germany

58 The worst off in Germany Income We observe household equivalized disposable income Composite index PCA weights (0.41; 0.24; 0.35) Life satisfaction Observed Equivalent income Computed, with best values as reference

59

60

61

62

63

64

65

66 The worst off in Germany

67 The worst off in Germany

68 The worst off in Germany

69 Outline Part 1. Measuring well-being on a crossroad Building blocks for a well-being measure Principles for a well-being measure An inconvenient result Part 2. Three well-being measures Composite well-being index Life satisfaction Equivalent incomes Part 3. Illustrations The worst off in Germany Beyond GDP: Social Progress in Europe

70 Beyond GDP: Social Progress in Europe Recent interest in going beyond GDP (At least) two problems with GDP growth as measure of social progress: 1. GDP is not based on an attractive measure of individual well-being (income) 2. Average is blind for distribution What happens if other dimensions are included? (equivalent income) What happens if the distribution is included? (penalty for inequality)

71 Beyond GDP: Social Progress in Europe European Social Survey, 2008 and countries: 15 EU-members, Switzerland, Norway, the Russian Federation. About 52,000 individual observations. Dimensions:

72 Beyond GDP: Social Progress in Europe Estimating preferences: c c

73 Beyond GDP: Social Progress in Europe Estimating preferences: c

74 Beyond GDP: Social Progress in Europe Income Equivalent income Happiness (Norway, Swits.) (Norway, Swits.) (Denmark, Swits.) Germany (6) 3272 (10) 7.26 (9) Denmark (7) 6915 (4) 8.35 (1) France (10) 3604 (9) 6.34 (15) Spain (11) 3245 (11) 7.30 (8) Greece (13) 2547 (12) 5.71 (17) (Russia, Estonia) (Russia, Hungary) (Greece, Russia)

75 Beyond GDP: Social Progress in Europe Income Equivalent income Happiness (Norway, Swits.) (Norway, Swits.) (Denmark, Swits.) Germany (6) 3272 (10) 7.26 (9) Denmark (7) 6915 (4) 8.35 (1) France (10) 3604 (9) 6.34 (15) Spain (11) 3245 (11) 7.30 (8) Greece (13) 2547 (12) 5.71 (17) (Russia, Estonia) (Russia, Hungary) (Greece, Russia)

76 Beyond GDP: Social Progress in Europe) Income Equivalent income Happiness (Norway, Swits.) (Norway, Swits.) (Denmark, Swits.) Germany (6) 3272 (10) 7.26 (9) Denmark (7) 6915 (4) 8.35 (1) France (10) 3604 (9) 6.34 (15) Spain (11) 3245 (11) 7.30 (8) Greece (13) 2547 (12) 5.71 (17) (Russia, Estonia) (Russia, Hungary) (Greece, Russia)

77 Beyond GDP: Social Progress in Europe Income ( = 0) Income ( = 5) Equivalent income ( = 5) (NO, CH) (NO, SE) (NO, CH) United Kingdom (5) (9) 281 (7) Germany (6) (7) 180 (10) Denmark (7) (5) 595 (4) Belgium (8) (6) 390 (5) Spain (11) 8668 (13) 156 (11) Greece (13) 7716 (14) 115 (12) Czech Republic (14) 8983 (11) 89 (14) (RU, EE) (EE, RU) (RU, HU)

78 Beyond GDP: Social Progress in Europe income growth ( =0) (CH, PL) Equivalent income growth ( =5) (CH, RU) Switserland % (1) % (1) Germany % (3) % (9) Belgium % (4) % (3) Denmark % (8) -4,64% (10) Spain % (11) -11,19% (17) Greece % (17) -21,72% (18) Estonia % (18) -9,29% (15) (GR, EE) (ES, GR)

79 Conclusion Can we construct an (operational) multidimensional well-being measure?

80 Conclusion Can we construct an (operational) multidimensional well-being measure? YES Is there a single silver bullet?

81 Conclusion Can we construct an (operational) multidimensional well-being measure? YES Is there a single silver bullet? NO Does the choice between the measures matters empircally?

82 Conclusion Can we construct an (operational) multidimensional well-being measure? YES Is there a single silver bullet? NO Does the choice between the measures matters empircally? YES Different measures take a different position on what are the most appealing principles. This is a value judgment. Let s be explicit about these value judgments, so that they are open to public scrutiny

83 References On well-being measurement and respect for preferences: Decancq K., Fleurbaey M. and Schokkaert E. (2014) Inequality, income, and well-being. In Handbook on Income Distribution vol. 2 (Atkinson and Bourguignon) Fleurbaey M., Schokkaert E. and Decancq K., (2009) What good is happiness? CORE DP 2009/17. On composite indicators and weights: Decancq, Koen and María Ana Lugo (2013) Weights in Multidimensional Indices of Well-Being: An Overview, Econometric Reviews, 32(1), Decancq, Koen, Luc Van Ootegem and Elsy Verhofstadt (2013) What if we voted on the weights of a multidimensional well-being index? An illustration with Flemish data, Fiscal Studies, 34(3), On illustrations: Decancq K. and Neumann D. (2015) Does the choice of well-being measure matter empirically? An illustration with German data in: Oxford handbook on well-being and public policy (Adler and Fleurbaey). Decancq K., Schokkaert E. (2013) Beyond GDP: Measuring social progress in Europe. Euroforum paper 4