1. Stéfan Lollivier 20/02/2014 Satisfaction and quality of life SILC longitudinal training

2. 20/02/2014 SILC longitudinal Quality of life : an economic topic During the 70, works of the school of Leyden (van Praag..) Perception of inequalities Equivalence scales More recently : Effect on well-being of the macro situation (short term growth, unemployment, …) Analysis of poverty (Ravallion,..) One of the three recommendations of the Stiglitz commission Quality of life and capabilities Measurement of objective and subjective indicators of well- being

3. 20/02/2014 SILC longitudinal Quality of life : a political concern 2008 : OCDE initiates a project for the measurement of well-being 2008 : the European Council endorses the principle of an economic growth more in adequacy with the needs of society 2008 : the French president calls for the Stiglitz report 2009 : der Bundestag sets up a commission to implement economic indicators to measure well-being 2010 : the Prime minister of UK orders to ONS a statistical framework to measure well-being

4. 20/02/2014 SILC longitudinal Dimensions of quality of life We first consider dimensions first related to capabilities Material living conditions Health Education Productive activities (psychosocial risks) Governance and individual rights Leisure and social relationships Environment Economic and individual safety Satisfaction is analysed in a second step Another side in the literature Outcome rather capability

5. 20/02/2014 SILC longitudinal FR-SILC and well-being : a two-years panel A possibility to follow the same people Avoids difficulties related to the heterogeneity of preferences Underscores statistical causalities rather than correlations A large survey 21186 adults in 2010, 22288 in 2011 14967 and 15406 people responding to the question on welfare (no proxy) A little older, more women, fewer students, more retirees, fewer children Same proportion of people living with partners, same profiles of diploma, same equivalent income 10897 adults both in 2010 and 2011 Due to the renewal of the panel Same composition than cross sections

6. 20/02/2014 SILC longitudinal Computation of synthetic indicators for every dimension of the quality of life Items of privation are gathered by dimension and then aggregated using a usual score method. This assumes that Items provide consistent information on the underlying dimension The scoring method eliminates measurement errors coming from data collection The individual position in the dimension worsen with the accumulation of elementary privations Beyond a certain number privations, the individual is considered in trouble in the dimension The indicator or poor quality of life equals 1 when the score exceeds a given threshold This threshold is chosen in order that about 10% of people have difficulties in a given dimension (analogy with poverty in living conditions)

7. 20/02/2014 SILC longitudinal Example : differences by income

8. 20/02/2014 SILC longitudinal Example : differences according to age

9. 20/02/2014 SILC longitudinal Example : differences with urbanization

10. 20/02/2014 SILC longitudinal Satisfaction

11. 20/02/2014 SILC longitudinal Influence of quality of life indexes on satisfaction CS 2010CS 2011Pooled panel data EstimWaldEstimWaldEstimWald Year (dummy) -0,031,1 Indexes of quality of life Material deprivations -1,04370,9-1,08420,5-1,1573,8 Psychosocial risks -0,4691,2-0,58147,9-0,56190,6 Economic insecurity -0,3119,6-0,2615,5-0,2823,5 Health difficulties -0,71387,6-0,75449,4-0,71567,3 Weak social ties -1,611218,8-1,761470,5-1,681890,4 Physical insecurity -0,1919,6-0,1512,5-0,1926,7 Diff related to the environment -0,227,1-0,258,6-0,3320,1

12. 20/02/2014 SILC longitudinal Econometrics of qualititative variables with panel data : models Individual unobserved heterogeneity is usually taken into account by an individual effect : Yit=Xit*b+Ui+Wit With F(Wit/Xi;Ui)=F(Wit) The true question is about the exogeneity of Ui conditional on Xi (Lechner, Lollivier, Magnac ; The Econometrics of Panel Data) rather to decide if Ui is a fixed or a random effect (only parametric restrictions). Fixed effect : no parametric assumption on the distribution of U, but implicit correlation between Ui and Xi Random effect : can be parameterized as Ui=Xi*c+Vi, or as in the Mundlak style Ui=Xi.*c+Vi

13. 20/02/2014 SILC longitudinal Econometrics of qualititative variables with panel data : Estimation of static models No assumption on the distribution of Ui : conditional logit Loss of efficiency Only binary variables (high satisfaction and low satisfaction) Assumption on the distribution of Ui : augmented probit Yit=Xit*b+Xi.*c+Vi+Wit Exogeneity F(Vi+Wi/Xi)=F(Vi+Wi) The pooled probit estimator is consistent and normal PMV or moment The matrix variance of this estimator can be computed with bootstrap methods.

14. 20/02/2014 SILC longitudinal Panel estimators : augmented probit EstimWald Year (dummy) -0,0733,3 Quality of life indicators Material deprivations -0,52284,7 Psychosocial risks -0,30201,8 Economic insecurity -0,1314,2 Health difficulties -0,36322,8 Weak social ties -0,841212,8 Physical insecurity -0,0819,2 Environmental difficulties -0,139,9

15. 20/02/2014 SILC longitudinal Effect on satisfaction of changes in quality of life Increase welfareDecrease welfare EstimWaldEstimWald Indicators of quality of life in material deprivations 0,010,00,5822,0 Out material deprivations 0,9461,2-0,030,0 In psychosocial risks -0,273,50,3810,5 Out psychosocial risks 0,295,8-0,070,2 In economic insecurity 0,030,00,090,3 Out economic insecurity 0,334,2-0,221,5 In health difficulties -0,111,10,4021,7 Out health difficulties 0,299,00,131,8 In weak social ties -0,4914,21,13187,5 Out weak social ties 1,03142,1-0,4110,4 In physical insecurity -0,060,20,224,5 Out physical insecurity 0,030,1-0,100,8 Worsening environment -0,110,20,374,2 Improving environment 0,130,50,100,3

16. 20/02/2014 SILC longitudinal Effect on satisfaction of changes of other variables Increase welfare Decrease welfare EstimWaldEstimWald In unemployment 0,010,00,7313,7 Out unemployment 0,7518,3-0,281,7 With a parter 0,8224,6-0,030,0 Leave her partner 0,010,00,242,9 Increase of at least one decile of eq income 0,2513,10,010,0 Decrease of at least one decile of eq income 0,070,90,133,8

17. 20/02/2014 SILC longitudinal Interest of panel data for this study Specific contribution of the longitudinal dimension Specific doubts about exogeneity of variables Confirmation of the role played by certain dimensions of QoL Material deprivation Social ties Psychosocial risks Physical health Uncertainty about other dimensions But only two dates for this study

18. 20/02/2014 SILC longitudinal Interest of panel data in the study of income and living conditions Time variability of income Transitory vs permanent monetary poverty Between 2003 and 2006, 22% of people have experienced poverty at least one year 5.3% to 5.9% of people remain poor 5% to 6.3% leave poverty 3.9% to 4.8% enter poverty Dynamics of poverty and quality of life What dimension comes first ?

19. 20/02/2014 SILC longitudinal Thank you for your attention