Happy for How Long?

Social capital, GDPand well-being over

time

Introduction

The (unanswered)question

Results

The long run

From the long to themedium-term

From the medium tothe short-term

Conclusion

38th Annual Meeting of the EEA, Boston

Happy for How Long? How Social Capital andGDP relate to Happiness over Time

Stefano Bartolini Francesco Sarracino

University of Siena CEPS/INSTEADand

CEPS/INSTEAD Population et Emploi

9th of March 2012

Stefano Bartolini, Francesco Sarracino Social capital, GDP and well-being over time 9th of March 2012 1 / 36


time

Introduction


Results

The long run



Conclusion

GDP and SWB over time

Much of the SWB literature focuses on the question:

“How far is general income growth (beyondincome levels already achieved) likely to

increase average happiness?” (Layard et al., 2009, p. 1)

This is a question about time series relationships



time

Introduction


Results

The long run



Conclusion

GDP and SWB over time

Much of the SWB literature focuses on the question:

“How far is general income growth (beyondincome levels already achieved) likely to

increase average happiness?” (Layard et al., 2009, p. 1)

This is a question about time series relationships

The answer:

SWB does not increase as income grows(the Easterlin Paradox)



time

Introduction


Results

The long run



Conclusion

Main explanations

Aspirations: after a while people adapt to their living standardby raising their aspirations. SWB is predicted by past income;

Social comparisons: people compare themselves with othersand set their aspirations accordingly. SWB is predicted byreference income.



time

Introduction


Results

The long run



Conclusion

Is GDP on the wane?

All-in-all:

Easterlin paradox +plausible explanatory theories +robust empirical evidence =

GDP is on the wane

we should dedicate to “something else” the attention and thepolicy efforts that contemporary societies pour into economicgrowth (i.e. social capital, social tolerance, political freedom,

religiosity and health).



time

Introduction


Results

The long run



Conclusion

Three new developments

1 SWB varies in the long run and it doesn’t vary in the sameway in every country (Stevenson and Wolfers, 2008; Inglehart, 2009;

Sarracino, 2011.)



time

Introduction


Results

The long run



Conclusion



Sarracino, 2011.)

2 average SWB is positively correlated with average income overtime (Stevenson and Wolfers, 2008; and Sacks et al., 2010.)



time

Introduction


Results

The long run



Conclusion



Sarracino, 2011.)

2 average SWB is positively correlated with average income overtime (Stevenson and Wolfers, 2008; and Sacks et al., 2010.)

3 before focusing on “something else” we need to be sure thatthis something else is not subject to adaptation and socialcomparisons as GDP (unemployment, marriage, divorce,widowhood, the birth of the first child, health, social capitaland religion).



time

Introduction


Results

The long run



Conclusion

or is GDP not on the wane?

trends of SWB show international variability +GDP is a good predictor of this variability +the alternatives to GDP have the same limitations =

GDP is not on the wane

we should not downsize the role of GDP as an indicator ofwell-being.



time

Introduction


Results

The long run



Conclusion

the Easterlin paradox: a matter of time

The paradox exists!

Stevenson et al. fail to distinguish between the short andlong-term relationship between SWB and GDP;

in the short term SWB and GDP are related, but unrelated inthe long run.

Easterlin et al. (2008, 2010)



time

Introduction


Results

The long run



Conclusion

the Easterlin paradox: a matter of time

The paradox exists!

Stevenson et al. fail to distinguish between the short andlong-term relationship between SWB and GDP;

in the short term SWB and GDP are related, but unrelated inthe long run.

Easterlin et al. (2008, 2010)

It’s a matter of time spans!

The contrast is not methodological: they use the same bivariatemethodology with the variations over time of SWB regressed onGDP variations.



time

Introduction


Results

The long run



Conclusion

The (unanswered) question

If not GDP, what does correlate with the variations of SWB overtime?



time

Introduction


Results

The long run



Conclusion



The aim of our work is to analyze the potential of the variations ofSC to predict the variations of well-being and compare it with thepredictive potential of the variations of GDP.



time

Introduction


Results

The long run



Conclusion




Since it’s a matter of time



time

Introduction


Results

The long run



Conclusion




Since it’s a matter of time

We compare this relationship in the:

Long-term Medium-term Short-term



time

Introduction


Results

The long run



Conclusion

the long-term



time

Introduction


Results

The long run



Conclusion

the Easterlin paradox: an example

"argentina"

"australia"

"austria"

"belgium"

"brazil"

"canada"

"chile" "china"

"denmark"

"finland"

"france"

"germany"

"iceland"

"ireland"

"italy" "japan"

"south korea"

"malta"

"mexico""netherlands""norway"

"portugal"

"south africa"

"spain"

"sweden"

"great britain"

"united states"

Y = −0.51 + −0.06 X N = 27

−2

−1

01

−2 0 2 4annual change of log GDP

annual change of life satisfaction Linear prediction

Figure: Life satisfaction and GDP over at least 15 years.



time

Introduction


Results

The long run



Conclusion

Life satisfaction and group membership over atleast 15 years

"argentina"

"australia"

"austria"

"belgium"

"brazil"

"canada"

"chile""china"

"denmark"

"finland"

"france"

"germany"

"iceland""ireland"

"italy" "japan"

"south korea"

"malta"

"mexico""netherlands""norway"

"portugal"

"south africa"

"spain"

"sweden"

"great britain"

"united states"

Y = −0.63 + 0.30 X***N = 27

−2

−1

01

−1 0 1 2 3annual change of membership

in at least 1 group or association

annual change of life satisfaction Linear prediction

Figure: Correlations among long-term trends of life satisfaction and ofgroup membership.



time

Introduction


Results

The long run



Conclusion

Happiness and group membership over at least 15years

"argentina"

"australia""austria"

"belgium"

"brazil"

"canada"

"chile"

"china"

"denmark"

"finland"

"france"

"germany"

"iceland""ireland"

"italy"

"japan"

"south korea""malta"

"mexico"

"netherlands""norway"

"portugal"

"south africa""spain"

"sweden"

"great britain""united states"

Y = −0.69 + 0.62 X***N = 27

−2

−1

01

2

−1 0 1 2 3annual change of membership

in at least 1 group or association

annual change of happiness Linear prediction

Figure: Correlations among long-term trends of happiness and of groupmembership.



time

Introduction


Results

The long run



Conclusion

Tri-variate regressions for long-term trends

Table: Trivariate regressions of long-term trends of proxies of subjectivewell-being over trends of SC and GDP (standardized variables).

happiness life satisfaction

membership in group or association 0.608∗∗ 0.330∗∗

(2.19) (3.58)

log GDP −0.0100 0.0447(−0.07) (0.35)

Constant −0.690∗∗∗ −0.634∗∗∗

(−3.88) (−6.87)

Observations 27 27Adjusted R

2 0.302 0.087

t statistics in parentheses∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.001



time

Introduction


Results

The long run



Conclusion

From the long to the medium-term

−→

What does happen if we shorten the time span? Will theresults about GDP and social capital change?



time

Introduction


Results

The long run



Conclusion

Happiness and GDP over the medium-term

ATBE

CHCZDEDK

EE

ES

FI

FRGB

GR

HUIE

IL

NL

NO

PL

PT

SE

SI

SK

TR

UA

Y = −0.00 + 0.307 X*N = 24

−4

−2

02

annu

al c

hang

e in

tren

d of

hap

pine

ss(s

tand

ardi

zed

valu

es)

−1 0 1 2 3 annual change in trend of log GDP

(standardized values)

trend of happiness Linear prediction

Figure: Correlations among medium-term trends of happiness and of thelog of GDP per capita.



time

Introduction


Results

The long run



Conclusion

Happiness and trust over the medium-term

ATBE

CHCZDEDK

EE

ES

FI

FRGB

GR

HUIE

IL

NL

NO

PL

PT

SE

SI

SK

TR

UA

Y = −0.00 + 0.810 X***N = 24

−4

−2

02

annu

al c

hang

e in

tren

d of

hap

pine

ss(s

tand

ardi

zed

valu

es)

−3 −2 −1 0 1 annual change in index of social trust


trend of happiness Linear prediction

Figure: Correlations among medium-term trends of happiness and of theindex of trust.



time

Introduction


Results

The long run



Conclusion

Tri-variate regressions for medium-term trends

Table: Trivariate regressions of trends of subjective well-being overchanges of the index of social trust and trends of GDP (standardizedvariables).

(1) (2)happiness life satisfaction

index of social trust 0.797∗∗∗ 0.731∗∗∗

(4.03) (8.06)

trend of log GDP 0.268∗∗ 0.323∗

(2.41) (2.02)

Constant −7.96e − 10 5.56e − 10(−0.00) (0.00)


2 0.702 0.630




time

Introduction


Results

The long run



Conclusion

From the medium to the short-term

−→

What does happen if we further shorten the period? Willthe results about GDP and social capital change?



time

Introduction


Results

The long run



Conclusion

Happiness and GDP over the short-term

ATAT

BE

BE

BE

CH CH

CH

CZ

DE

DE

DE

DKDK

DK

EE

EE

ES

ES

ESFI

FI

FI

FR

FRFR

GB

GB

GB

GR

HU

HU

HU

IE

IEIE

NL

NL

NL

NONONO

PL

PLPL

PT

PT

PTSE

SE

SE

SI

SI

SI

SK

SK

UA

UA

Y = −0.00 + 0.591 X***N = 58

−2

−1

01

23

annu

al c

hang

e in

cha

nges

in h

appi

ness

(sta

ndar

dize

d va

lues

)

−2 0 2 4annual change in changes in log GDP (2yrs)


changes in happiness Linear prediction

Figure: Correlations among short-term trends of happiness and of thelog of GDP per capita.



time

Introduction


Results

The long run



Conclusion

Happiness and trust over the short-term

ATATBE

BE

BE

CHCH

CH

CZ

DE

DE

DE

DKDK

DK

EE

EE

ES

ES

ESFI

FI

FI

FR

FRFR

GB

GB

GB

GR

HU

HU

HU

IE

IEIE

NL

NL

NL

NO NONO

PL

PLPL

PT

PT

PTSE

SE

SE

SI

SI

SI

SK

SK

UA

UA

Y = −0.00 + 0.308 X**N = 58

−2

−1

01

23

annu

al c

hang

e in

cha

nges

in h

appi

ness

(sta

ndar

dize

d va

lues

)

−3 −2 −1 0 1 2annual change in index of social trust


changes in happiness Linear prediction

Figure: Correlations among short-term trends of happiness and of theindex of trust.



time

Introduction


Results

The long run



Conclusion

Tri-variate regressions for short-term trends

Table: Trivariate regressions of trends of subjective well-being overchanges of the index of social trust and trends of GDP (standardizedvariables).


index of social trust 0.255∗∗ 0.214(2.18) (1.58)

changes in log GDP (2yrs) 0.568∗∗∗ 0.525∗∗∗

(4.69) (4.73)

Constant −3.27e − 09 1.76e − 09(−0.00) (0.00)


2 0.393 0.318




time

Introduction


Results

The long run



Conclusion

Medium and short run compared


index of social trust 2.713∗∗∗ 2.681∗∗∗

(4.03) (8.06)

log of GDP 1.209∗∗ 1.574∗

(2.41) (2.02)

Constant −0.121∗∗∗ −0.111∗∗

(−3.98) (−2.55)


2 0.702 0.630


Table: Medium-term


index of social trust 0.569∗∗ 0.563(2.18) (1.58)

log of GDP 2.092∗∗∗ 2.286∗∗∗

(4.69) (4.73)

Constant −0.108∗∗∗ −0.094∗∗∗

(−4.45) (−3.51)


2 0.393 0.318


Table: Short-term



time

Introduction


Results

The long run



Conclusion

Summarizing... GDP

GDP:

does not matter for SWB in the long run;

it begins to be important in the medium-term;

its importance increases as the period shortens:short run coefficients are greater and more significant thanmedium run coefficients.



time

Introduction


Results

The long run



Conclusion

Summarizing... GDP

GDP:

does not matter for SWB in the long run;

it begins to be important in the medium-term;

its importance increases as the period shortens:short run coefficients are greater and more significant thanmedium run coefficients.

Easterlin seems to be right: the time span matters!



time

Introduction


Results

The long run



Conclusion

Summarizing... social capital

Social capital:

is strongly correlated to SWB in the long and themedium-term;

this correlation tends to evaporate in the short run:the coefficients are much smaller and less significant than themedium-term ones.



time

Introduction


Results

The long run



Conclusion

Possible policy implications

at least for rich countries:

do we want a quick effect on happiness which tends to get lostas time goes by?We should persist being obsessed by economic growth;

do we want an effect on happiness which is slow and durable?Policies should target social capital.



time

Introduction


Results

The long run



ConclusionThanks for your kind attention!



time

Data

The countries included in our sample depends on the availability ofinternationally comparable time series on social capital variables.

WVS/EVS data

collected in 6 waves between 1981 and 2009

all available countries with at least 15 years and 3obs. in time

no transition economies

27 countries

169000 obs



time

list of associations mentioned in the WVS/EVS

Social Capital:Long-term: share of the population participating in at least onegroup or association.

Respondents were asked to mention whether they belonged or wereperforming unpaid voluntary work for any of the following groupsor associations:

social welfare service for elderly; animal rights;religious organization; professional associations;education, arts, music or cultural activities; youth work;labour unions; sports or recreation;political parties; women’s group;local political actions; peace movement;human rights; organization concerned with health;conservation, the environment, ecology; consumer groups;other groups.



time

Data

ESS data

collected in 4 waves between 2002 and 2008

every 2 years

all available countries

24 countries and 58 intervals

162000 obs



time

Our variables

Social Capital:

Medium and Short-term: index of social trust using the following 3items:

Most people can be trusted“Would you say that most people can be trusted?”

Most people are helpful“Would you say that most of the time people try to be helpful?”

Most people are unfair“Do you think that most people would try to take advantage of you if

they got the chance, or would they try to be fair?”

Given the similarities among these 3 questions, we group them toproxy one latent concept.



time

Three measures of well-being

Well-being can be proxied by:

happiness

life satisfaction

objective data: mental illnesses, suicides, alcoholism, drugsabuse, psychopharmaca, etc.

Happiness and life satisfaction are called subjective well-being(SWB)

Subjective well-being (SWB): individual’s evaluation of itsown life regarded as a whole (Helliwell, 2008)



time

Observing SWB (WVS/EVS)

Happiness:

“Taking all things together, would you say you are:”

1 “very happy”

2 “quite happy”

3 “not very happy”

4 “not at all happy”



time

Observing SWB (WVS/EVS)

Happiness:

“Taking all things together, would you say you are:”

1 “very happy”

2 “quite happy”

3 “not very happy”

4 “not at all happy”

Life satisfaction:

“All things considered, how satisfied are you with your life as awhole these days?” Answers range on a 1 to 10 points scale:

1 “dissatisfied”...

10 “satisfied”



time

Reliability of SWB

Reliability

data on SWB are:

consistent with more objective measures of well-being (heartrate, blood pressure, Duchenne smile, brain scanners, suicide)(Blanchflower and Oswald, 2008a; Van Reekum et al., 2007);

highly correlated with other proxies of SWB (Schwarz andStrack, 1999; Wanous and Hudy, 2001; Schimmack et al.,2009);

consistent with evaluations about the respondent’s happinessprovided by friends, relatives or clinical experts (Schneider andSchimmack, 2009; Kahneman and Krueger, 2006; Layard,2005).



time

China 1990 - 2007, an ongoing work...

SWB in China

500

1000

1500

2000

aver

age

GD

P p

.c. (

2000

US

$)

6.6

6.8

77.

27.

4av

erag

e lif

e sa

tisfa

ctio

n

1990 1995 2000 2005 2010year survey

average life satisfactionaverage GDP p.c. (2000 US $)



time

Evidence from within country studies: China

Explained part−

.15

−.1

−.0

50

.05

.1

finan

cial_s

at

freed

om_c

hoice

healt

h

pol_t

rust

soc_

cap

scale

of in

com

es

parameter estimate 90% conf. interval



time

Evidence from within country studies: China

Unexplained part−

.50

.51

1.5

finan

cial_s

at

freed

om_c

hoice

healt

h

pol_t

rust

soc_

cap

scale

of in

com

es

parameter estimate 90% conf. interval


Documents

Happy for How Long?