Alternative measures of well-being

Transcription

1 Alternative measures of well-being Marco Mira d Ercole marco.mira@oecd.org OECD Social Policy Division Joint work of the OECD Economics Department, the OECD Employment, Labour and Social Affairs Directorate and the OECD Statistics Directorate 1 Background In the 1920s: discussions on measuring income as the foundations of the NA systems were laid down. Marriage of convenience between macro and welfare? In the 1970s: discussions on environmental and social limits to growth (e.g. Tobin and Nordhaus) In recent years: concerns on broader measures of wellbeing within discussions on sustainable development 2 1

2 Motivation of OECD interest Economic perspective: In the context of the OECD annual assessment of the impact of structural reforms on performance: Is GDP per capita an adequate measure of well-being? Social perspective: What light do social indicators bring to an assessment of living conditions (beyond that provided by GDP pc)? Statistical perspective: Can we do better than the current proliferation of different sets of diverse indicators? 3 Content of the paper Focus on measures rather than their policy implications. Measures drawn from four different approaches 1. SNA-based measures of economic resources 2. Expanded monetary measures (monetary equivalents) 3. Objective social indicators (non monetary) 4. Subjective measures of happiness and life-satisfaction I will use this structure in my presentation. 4 2

3 1. SNA based measures of economic resources Two parts Measures for the economy as a whole Measures for the household sector 5 Measures for the economy as a whole Premise: SNA-measures of income depart from the theoretical definition (what can be consumed today without compromising consumption opportunities tomorrow). Two main reasons: Relations with the rest of the world (GNI) Net income transfers from abroad (for levels of GNI) Terms of trade effects (for volume changes in GNI) Effect of consumption of fixed capital (NNI) -- but SNA asset boundary excludes several assets affected by production 6 3

4 NNI per capita, per cent 27-Apr-07 Gaps relative to the US when considering net income flows and capital consumption Ireland, 2003 Czech Rep NNI pc 0.73 NNI pc 0.39 GNI pc 0.74 GNI pc 0.44 GDP pc 0.89 GDP pc Volume growth of GDP and NNI: similar in most countries with some notable exceptions GDP and NNI per capita growth, average annual growth, SVK 4.0 FIN GBR GDP and NNI per capita growth, average annual growth, AUS DNK ISL CAN SWE GRC ESP USA AUT NLD FRA ITA BEL MEX CHE DEU GDP per capita, per cent 8 4

5 Measures for the household sector From the perspective of assessing well-being, focus should be on households and consumption Three SNA measures of consumption possibilities of individuals: Household disposable income Household final consumption expenditure Household actual consumption and disposable income (i.e. including free services) 9 Gaps relative to US when moving from economy-wide to household measures France, 2003 Sweden, 2003 AHDY pc 0.83 AHDY pc 0.75 HDY pc 0.64 HDY pc 0.52 Acons pc 0.78 Acons pc 0.72 Fcons pc 0.58 Fcons pc 0.49 GDP pc 0.76 GDP pc

6 Household disposable income per capita Private consumption per capita 27-Apr-07 Starker differences when looking at growth rates Real annual growth in household s disposable income, final consumption expenditure and real GDP per capita, GDP and household disposable income per capita GDP and final consumption expenditure per capita PRT NOR GBR USA CZE FRA AUS FIN NLD NZL DEU AUT SWE CHE DNK CAN GRC BEL MEX ITA TUR JPN HUN SVK POL KOR GDP per capita POL SVK KOR CZE HUN NOR GBR AUS FIN CAN USA ISL NZL ESP LUX FRA GRC NLD SWE ITA PRT AUT DEU DNK CHE MEX JPN BEL TUR GDP per capita IRL 11 Summing-up Economy-wide measures in SNA are closely related to each other with significant differences limited to some countries There are larger differences between household and economy wide-measures in terms of both levels and growth rates 12 6

7 2. Expanded monetary measures ( monetary equivalents ) What is included in utility functions is arbitrary but with practical implications To account for non-market factors, long tradition of satellite accounts (e.g recommendations from the US National Research Council of the Academies of Sciences) In the paper, only calculation for some factors (leisure-time of workers, living arrangements, income distribution) Illustrative calculations, subject to arbitrary assumptions; no attempt to compute cumulative effects 13 A. Leisure time of workers large differences in annual working hours and fall since 1970s KOR POL CZE GRE HUN MEX USA AUS NZL ICE JPN CAN SPA FIN UKG SLV POR IRL LUX ITA SW AUT BEL DNK FRA GER NLD NOR HUN MEX POL NZL AUT GRE CZE AUT GER USA POR LUX SW CAN SLV UKG FIN ITA SPA BEL ICE DNK NO JPN KOR FRA IRL NLD

8 Approach Adjustment limited to leisure time of workers (to avoid assumptions on the voluntary nature of time not worked of people without paid jobs) Quantity measured as the difference between each worker s time endowment and annual working hours, common adjustments for time devoted to personal care and work-related activities Prices based on three valuations: GDP per hour worker, hourly compensation, and half of hourly compensation. 15 Evidence: smaller gaps relative to the US after valuing leisuretime in some Continental European countries Gaps in levels, relative to the US, in leisure-adjusted GDP per capita, POR ITA GER JPN UKG FRA Leisure valued at half of hourly compensation Leisure valued at GDP per hour worked SWE CAN NLD DNK NOR Leisure valued at hourly compen GDP per capita 16 8

9 stronger growth in several EU countries than in the US CAN DNK FRA GER JPN NDL NOR PRT SWE GBR USA Leis at GDP ph Leis. at hour earn. GDP pc B. Changes in living arrangements Long tradition within studies of households living conditions As households get smaller (more people living alone, lower fertility rates), economies of scale in consumption are lost (greater needs for a given economies resources) You can capture the effect of living arrangements by equivalising income: we apply a standard adjustment for household size (square root elasticity) Different adjustment affect levels of resources for population subgroups but are less important for changes 18 9

10 Mean equivalised income Mean equivalised income 27-Apr-07 Evidence: some significant differences in growth of household disposable income after accounting for living arrangements in some countries Real annual change of per capita household disposable income and adjustments for changes in household size, selected OECD countries GBR DEU 2 AUTFRA SWE USA NLD 1 DNK FIN AUS CAN JPN ITA Mean non-equivalised income POL NOR 3 GBR MEX FRA SWE USA TUR 2 NZL NLD FIN PRT AUT CZE AUS ESP 1 CAN DEU DNK 0 GRC -1 0 ITA JPN -1 Mean non-equivalised income 19 C. Income inequality Growth and inequality typically opposed to each other: unwarranted Focus on average income is only a convention (which weight people by their income) One way to show the impact of inequality is through the concept of equally distributed income W N 1 N i y i Calculations under 3 hypothesis on the degree of aversion to inequalities (higher a elasticity implies a lower weight to higher income) 20 10

11 Evidence: significant impact on levels of household disposable income, smaller in terms of rankings... Levels of equally-distributed household disposable income for different values of the coefficient of aversion to inequality, Coeff. of aversion to inequality of 0 Ceoff. of aversion to inequality of 1 Coeff. of aversion to inequality of 10 GDP per capita Turkey Mexico Poland Hungary Czech Republic Portugal Greece New Zealand 0 Spain Italy Germany Japan Finland Sweden France Australia United Kingdom Canada Netherlands Denmark Austria Switzerland United States Norway 21 Other ways of bringing together growth and distribution Trends in real equivalised household income by decile Average annual change mid-1980s to mid-1990s Average annual change mid-1990s to 2000 Bottom 2 deciles Middle 6 deciles Top 2 deciles Average Bottom 2 deciles Middle 6 deciles Top 2 deciles Average Australia Belgium Canada Czech Republic Denmark Finland France Germany Greece Hungary Ireland Italy Japan Luxembourg Mexico Netherlands New Zealand Norway Poland Portugal Spain Sweden Switzerland Turkey United Kingdom United States OECD

12 Summing up Accounting for a range of non-markets factors lead to some significant differences in levels of countries well-being relative to GDP per capita Differences in terms of growth rates are large based on extreme assumptions on valuation but for some countries also evident for other valuations Objective social indicators (non monetary) Measuring well-being is not the same as measuring economic performance. From the perspective of wellbeing, economic growth is a means rather than an end What do we know about the relation between GDP per capita and a range of social outcomes? Analysis based on a subset of OECD social indicators 24 12

13 To address this question.. We have selected measures of social outcomes (rather than inputs ) in 4 different fields fields. 16 indicators, chosen with a view to provide a comprehensive assessment of well-being: Self-sufficiency (employment rate, share of population in jobless households, avg. years of schooling, students test scores) Equity (Gini income inequality, relative income poverty for total population and for children, gender wage gap). Health (life-expectancy and HALE at birth, infant mortality, potential years of life lost) Social cohesion (participation in voluntary groups, victimisation rate, share of convicted adults, suicides) 25 correlations between social indicators and GDP pc Levels Self-sufficiency Equity Health Social cohesion 80% 60% 40% 20% 0% -20% Changes 80% 60% 40% 20% 0% -20% -40% Total employment rates (2003) Average years of schooling (2002) People in jobless households* (2000) Mean student preformance (2000) Income Inequality* (2000) Relative poverty rate* (2000) Child poverty* (2000) Gender wage gap* (1999) Healthy life expectancy at birth (2002) Life expectancy at birth, total (2002) Infant mortality* (2002) Potential number of years lost* (2002) Volunteering (2000) Victimisation rate* (2000) Convicted adults* (2000) Suicide rate* (2002) 26 13

14 Main patterns Prevalence of significant positive correlations in levels There are exceptions (e.g. countries with higher GDP pc have more crime, suicides, imprisonment) Correlations are not high, even weaker when excluding countries with GDP pc < 25,000 USD Nor relation in terms of changes Conclusion: social indicators provide additional beyond that provided by GDP per capita. You ignore it at your own peril. 27 Can we do better? Composite indicators Richness and limits of social indicators. One way to overcome these limits is through composite indicators.. Popular approach but several difficulties: How to select indicators? How to express them on a single metric? How to weight them? Some of these problems are more difficult than others 28 14

15 Sensitivity to weighting system: significant differences in economic and social performances for some countries are independent of weights used Median value and confidence interval of a composite index based on selected social indicators in OECD countries and GDP per capita Mexico Average random composite index (RCI) (based on randomly assigned weights for each indicator) Poland Hungary GDP per capita, normalised score Czech Rep. Portugal Greece New Zealand Spain Italy Germany Finland 90% confidence interval for RCI Japan Sweden France United Kingdom Australia RCI, median value Austria Netherlands Denmark Canada Switzerland United States 29 Summing up Objective social indicators provide additional information on dimensions of well-being that go beyond what is provided by conventional economic aggregates Difficult to get a parsimonious representation of well-being (levels and changes) in different countries Composite indicators are one way to respond to this challenge, but they mix apples and oranges. Our way forward will be by using small set of headline indicators 30 15

16 4. Subjective measures of happiness and lifesatisfaction Long standing US tradition of psychological studies More recently, rapid grow in economic research (from 1 article every 2 months, to 1 per week since 2000) Some scepticism (sensitivity to whimsical circumstances) but also several validation studies Main conclusion: robust enough to be taken seriously 31 Different concepts and approaches Different dimensions of well-being: life chances or actual results; well-being of the individual or of (whole) society. Life-satisfaction refers to actual results for the individual Different dimensions of life-satisfaction: satisfaction with life as a whole or with life domains (work, family, mental and physical health); passing or enduring Different concepts (positive and negative affects, i.e. moods and emotions; stress; happiness; life-satisfaction) and different measuring tools methods (recall methods, diary and methods to assess people s experience on line 32 16

17 Scores Percentages 27-Apr-07 Sketch of selected results from existing literature Aggregate measures of life-satisfaction (by country) Individual measures, with focus on determinants 33 Aggregate measures: 90% of respondents satisfied with their life in ⅔ of OECD countries 10 9 Mean life-satisfaction score (left-axis) Mean Happiness score (left-axis) Share of very/fairly happy people (right-axis) Turkey Hungary Slovakia Korea Poland Japan Greece France Portugal Spain Czech Republic Italy United Kingdom Australia Belgium Germany Sweden United States Norway Canada Finland Luxembourg Netherlands Austria Iceland Mexico Switzerland Ireland Denmark

18 Life-satisfaction, average score Life-satisfaction, average score 27-Apr-07 Focus on the relation of life-satisfaction and income: in cross section (Inglehart).. 9 GDP per capita and mean level of satisfaction, 2000 All countries OECD countries GDP per capita, current US $ TUR MEX CZE POL SLK HUN NLD CAN DEN BEL CHE AUT ISL IRL FIN SPA NOR NZL SWE GBR AUS USA ITA GER PRT FRA GRC JPN GDP per capita, current US $ 35 and time-series (Easterlin) Mean Satisfaction, index GDP per capita, index France

19 Limits of aggregate measures Not all agree on cardinal comparisons (i.e. that an individual scoring 8 is twice as happy as one scoring 4) Values for most OECD countries are in a very narrow range, and different surveys / concepts may lead to different ranking We don t have a satisfactory account of differences in life-satisfaction across OECD countries and over time Mean Satisfaction, index GDP per capita, deviation from trend Conversely, aggregate life-satisfaction and objective data (suicides, hypertension) tell consistent stories. Best use of national indicators of life-satisfaction as complement of economic and social indicators (as in SaG) 37 Evidence on determinants of lifesatisfaction for individuals: key results Subjective well-being depends on both individual factors and features of the community and society where individuals live. Empirical models provide estimates of the trade-off between various determinants (e.g. the extra income needed to compensate for loosing a job) Difference in personal income of respondents explain little of the variance in subjective well-being across individuals, and less than a range of other non-financial factors

20 Relation between personal income and individual s life-satisfaction Difference in life-satisfaction across individuals are not proportional to their income (if cardinality accepted). When tracking people over time, changes in income do not bring similar changes in their subjective well-being Personal well-being depends on the direction of income changes (change in happiness twice as large for income gains as for income losses) Individuals (fully?) adapt to higher income while this may differ for other determinants (e.g. leisure time) Individual compare their income with that of their reference group Overall: Several consistent findings. Conceptual objections, but also potential policy implications. Partial measures, need for more systematic approach. 39 References S.E.M. Working paper No. 33. Alternative Measures of Well-Being ( ) S.E.M. Working paper No. 22. Income Distribution and Poverty in OECD Countries in the Second Half of the 1990s ( ) OECD (2007), Society at a Glance: OECD Social Indicators ( ) 40 20