Transcript Government expenditure

The Fifteenth Dubrovnik
Economic Conference
Organized by the Croatian National Bank
Investigating Macroeconomic
Determinants of Happiness in
Transition Countries: How Important
is Government Expenditure?
by L. M. Perovic and S. Golem
Discussant: K. Zigic CERGE-EI Prague,
Czech Republic
Summary of the Paper
• Objective:
– Analysis of the role of macro factors and especially
government expenditures for happiness
• Methodology:
– Based on survey data
– Ordered logit econometric analysis of dataset
constructed from survey respondent-level data on
happiness and social characteristics and macroeconomic
variables
• Conclusions
– “Government expenditure significantly and non-linearly
influences happiness in transition countries”
Main Variables Used
Variable
Happiness
Government
expenditure
Source
Description
Taking all thing together,
would you say you are: 1:
World Values Survey
not at all happy, 2: not very
Waves 3, 4 and 5
happy; 3: quite happy and 4:
very happy.
General government final
World Development
consumption expenditure as
Indicators, World
a percentage of GDP
Bank (2008)
(annual %)
Measuring the Quality of Life: Overview
• GDP – very bad measure of the quality of life (but often the only one readily
available)
• Measures derived from GDP – e.g. GDP adjusted for leisure and pollution
– Hicks, Kaldor, and others (Measure of Economic Welfare, Net Economic Welfare)
• Non-economic indicators – social, health indicators
– Infant mortality, life expectancy
• Happiness
– J. Bentham – utility in terms of psychological states
– “Feeling good – enjoying life and feeling it is wonderful.” Layard (2003)
– Can we measure it by GDP?
• In spite of economic growth people in USA, Japan, UK do not seem to be happier than 50 years ago.
Hedonic adaptation?
• But Deaton (2007) finds relationship between life satisfaction and per capita GDP
– Can we measure it by surveys?
• D. Kahneman, A. Kruger: Day Reconstruction Method
– Or borrow methods from neuro-science? – D. Prelec: positive feelings correspond to
brain activity in the left side of the pre-frontal cortex, somewhat above and in front of
the ear.
– Not only tangible wealth matters, but also our ability to appreciate and enjoy life.
• We should also have teachers that will teach us the ways to happiness – R. Layard (2007)
Easterlin (2003)
• Analysis of happiness that attempts to bridge the gap
between psychology (set theory) and economics (revealed
preferences)
– Focuses on aspirations/social comparison and hedonic adaptation
– Differentiates between nonpecuniary factors (marriage, divorce,
health) with lasting effects and pecuniary factors with possibly
only temporary effects (due to hedonic adaptation and social
comparison)
• Traces happiness of identical individuals in time
• Two of Easterlin’s conclusions relevant to the discussed paper:
– “Policies to improve health or facilitate more time with one’s
family are consistent with greater happiness”
– “Increase in income … does not bring with it a lasting increase in
happiness…”
“Happiness” Variable In the Discussed Paper
• Data from World Value Survey used
– Not too many details about the data provided in the paper
• Methodology and quality of the data on happiness (World Value Survey)
should be analyzed in more detail in the paper as it is crucial for the whole
analysis
• Possible problems with the data
– E.g. Croatia seems to have been included in two waves (1996,
1999). Is this representative enough?
– While the data on happiness are on “respondent level”, from the
description on WVS web it seems that the same set of
respondents is not tracked during different waves of the survey
(i.e. random sampling in each wave is used). Is this true?
– If yes, then this could be a problem
• As e.g. Easterlin (2003) discusses, life-cycle and “hedonic adaptation” do
matter a lot in analysis of happiness
• The data are rather a sequence of cross-sections than a real panel
• Description of equation 1 (indexes) can be misleading
Sampling used in Croatia (1999 Survey)
• Source: WVS website
• Use was made of two-stage probability sampling.
– At the first stage 63 locations were selected from a list of all
towns/villages in Croatia that was sorted by administrative
districts and degree of urbanisation.
• In each location 16 interviews (systematic selection) were conducted.
Some minor corrections were made to adjust for urban
representation of each country. More than one sampling point were
made in larger cities (e.g. Zagreb had 10 sampling points).
– At the second stage respondents were selected randomly
within the household (using the Throdal and Carter method balancing gender and age) from a list of addresses in each
location.
• According to description, the same methodology used in
1996. But how about the stability of the sample and
ability to track the original individuals?
Feeling of Happiness: Croatia in 1996
Source: World Value Survey
Feeling of Happiness: Croatia in 1999
Source: World Value Survey
Other Comments on Variables Used
• There seem to be substantial (extreme) variability in some of
the variables (Table 2), especially in inflation
– It seems that there may be some outliers
– Did you test robustness of your results with respect to the role of
the outliers
• E.g. – if it is the case – by omitting observations for a country (countries?)
that acts as an outlier with respect to inflation?
• On the other hand, variability in happiness variable is fairly
low – as expected, most respondents were between 2 and 3
on the 4 grade scale
• This is not unusual, but may lead to question concerning
quality of relationship between very variable macroeconomic
variables and the happiness variable
– No additional information about the properties of the resulting
estimates (i.e. no other than t-tests) were given
– Do you have more details on fit/predictive powers of the model or
some specification tests?
Table 2: Summary Statistics
Variable
Happiness
Sex
Age
Education
Marital
status
Employme
nt status
Income
scale
Gen. govt.
exp.
Inflation
Unemploy
ment
GDP per
capita
Obs
30057
30820
30803
30649
Mean
2.72
1.52
44.87
4.64
Std. Dev.
0.71
0.49
16.85
2.16
Min
1
1
17
1
Max
4
2
101
9
30689
2.48
2.08
1
6
30543
3.14
2.17
1
8
26468
4.52
2.50
1
10
30828
17.72
47.81
5.87
192.27
5.69
0.54
27.78
1058.37
30828
12.79
6.66
5.8
34.5
30828
10520.81
4422.48
3631.99
23010
30828
Who is Happy? Average Happiness 1995-2005
How much people enjoy their life-as-a-whole on scale 0 to 10
Top
> 7,7
Middle range
± 6,0
Bottom
<4
Denmark
8,2
Phillipines
6,4
Armenia
3,7
Switzerland
8,1
India
6,2
Ukraine
3,6
Austria
8,0
Iran
6,0
Moldova
3,5
Iceland
7,8
5,9
Zimbabwe
3,3
Finland
7,7
5,8
Tanzania
3,2
Poland
South Korea
USA – 17th place, Czech R. 40-43rd place out of 95 countries
Results based on surveys
Source: World Database of Happiness, Ruut Veenhoven, Erasmus Univ. Rotterdam
Life Satisfaction and GDP p. Capita
Source: Deaton (2008)
Interesting Unaswered Question
• Authors assume inverted u-shaped relationship
between government expenditures and
happiness
– E.g. „we hypothesize that there is a “useful” amount
of government expenditures that positively
influences happiness…”
– Find that marginal effects correspond with
expectations (Table 4 and 5) and are significant
• Logical question:
– have you tried to derive the “optimal” levels of
government spending for the analyzed countries?
Conclusion: Final Comment
• Interesting paper based on a fairly wide dataset. However, consistency and
properties of the dataset needs better description and clarification
• Approach standard in analysis based on probit/logit regressions with similar
types of survey data and macroeconomic characteristics but adds share of
government expenditures to the set of variables
– Compare e.g. Di Tella, MacCulloch, Oswald (2003) that represent the standard
form of similar studies
• Survey could focus more on literature from the frontier segment between
psychology and economics – such as e.g. Easterlin (2003)
• More attention should be paid to the variables describing happiness, it has
to be made clear whether the same individuals were or were not tracked in
different waves
• Econometrics: more details on the results and tests of the results should be
provided
• Analysis of results: results can be “dangerous” for fiscal stability if
interpreted literally
– It may be useful to try to experiment with searching “optimal” level of
government spending – after taking care about interactions between spending
and other variables in the model
References
• A. Deaton: Income, Health, and Well-Being Around the World: Evidence
From the Gallup World Poll. Journal of Economic Perspectives —
Volume 22, Number 2 — Spring 2008
• B.S. Frey, A. Stutzer: Happiness Research: State and Prospects. Review
of Social Economy, June 2005
• R. Layard: Happiness: Has Social Science Have a Clue? Lionel Robbins
Memorial Lecture, LSE, March 2003
• Easterlin R.A. (2003): Explaining happiness. Proceedings of the National
Academy of Sciences of the United States of America, September 16,
2003
• Di Tella R., MacCulloch R.J., Oswald A.J. (2003): The Macroeconomics of
happiness. The Review of Economics and Statistics, November 2003, 85
(4): 809-827
• World Database of Happiness: http://worlddatabaseofhappiness.eur.nl/
• World Values Survey
http://www.worldvaluessurvey.org/