The Effect of GDP per Capita On the Happiness
of Nations.
(Does money buy happiness?)
XheilanSalieski1 - s2332752Supervisor: Dr. V. Angelini University of Groningen, the Netherlands
20 January 2017
Abstract: The goal of this Master (MSc) thesis is to empirically contribute to the existing
field of happiness economics, which over the past several decades has developed rapidly. This paper shows that GDP per capita has a positive effect on well-being and that macroeconomic variables enter significantly and their signs accordingly with economic theory when regressing on the happiness of nations. Furthermore, the paper finds evidence that the political system is of importance for one’s happiness. Additionally, there is statistical support that level and change in GDP affects well-being in this research.
Keywords: Happiness economics, GDP per capita,
JEL code:E14, E22, M29
1MSc. Student at the University of Groningen, the Netherlands – Faculty of Economics and Business (FEB).
2 1. Introduction
3
4 The outline of this paper is organized as follows. It begins with a thorough discussion of the empirical literature on happiness research. Secondly, a discussion of international data on reported well-being levels of sixteen European countries from the Euro Barometer Data Surveys will be given. Thirdly, an elaboration of the methodology is given to answer the central research question. Fourth, the econometric relationship between, subjective well-being or happiness on GDP per capita, macroeconomic variables, and microeconomic characteristics. The next sections than present the results, robustness checks and calculates the marginal rate of substitution to put a value on economic downturns and inflation. Additionally, this paper also investigates the role of the political system on the happiness of nations. Whether the political system a respondent lives in leads to higher happiness levels among countries. I find statistically evidence for this. Lastly, I conclude.
2. Empirical Literature
The outline of this section considers the determinants of happiness. More specifically, happiness depends on three factors: demographic and personal characteristics, economic, and political factors (Frey and Stutzer, 2002). Where the former belongs to microeconomic and the latter on macroeconomics determinants.
MICROECONOMIC FACTORS
5
MACROECONOMIC FACTORS
Unemployment makes people very unhappy. Unemployment reduces well-being substantially more than other factors such as divorce (Clark and Oswald, 1994). This result is called the pure effect of being jobless whereby indirect effects and income are kept constant. Moreover, unemployment makes people very unhappy, over and above the concomitant income loss (Oswald, 1997). Here the problem of reverse causality regarding unemployment and happiness has been addressed as well. This is done with longitudinal data. The causation indeed runs in the opposite direction meaning that unhappy people do not perform well on the work floor. Additionally, general unemployment leads to a decrease in happiness because employed people may feel bad for the unemployed or consider the fear of being unemployed in the future (Di Tella et al, 2001). Furthermore, when a big group of people share a particular fate a mitigating effect occurs. More specifically, when unemployment hits an individual and this person can evaluate its situation relative to other people it becomes less unhappy compared to a situation when it is alone (Clarks et al, 1994). The loss from being unemployed equals the coefficient on being unemployed in a life-satisfaction micro regression (Blanchflower et all, 2014). Unemployment lowers happiness of the unemployed but also the happiness of everyone else (Blanchflower et all, 2014). Finally, there is a causal link between European unemployment and the generosity of the welfare system, particularly on the unemployment benefit system (Di Tella et al, 2003). The benchmark paper shows that countries with more generous benefit systems are happier. In addition, they fail to find evidence that the welfare states of Europe reduced the incentive for people to work, by making life to easy for the people that do not work (Di Tella et al, 2003). This component will is used in the happiness specification as an macroeconomic variable for this research.
6
7
a higher income does not lead to higher experienced happiness nor to a relief of unhappiness or stress (Kahneman and Deato, 2010).
In the literature there are several reasons why higher income does not translate into higher happiness. The key reasons are briefly introduced. Firstly, individuals compare themselves relative to others with respect to income, status, consumption and utility. Therefore, absolute income levels do not matter but relative ones (Frey and Stutzer, 2002). In order to make a statement it is important to know with which people compare themselves. The conclusion is such that the higher the income of the reference group, the less satisfied people with their job (Clark and Oswald, 1994). Secondly, income inequality affects happiness. There is a negative effect of inequality on happiness in Europe because upward social mobility is not as large compared to the United States. For example, being low on the income distribution is not seen as affecting future income (Alesina et al, 2001).
The relationship between income and happiness holds. This has been proven to be evident in the literature. Moreover when other macroeconomic variables are included to investigate the correlation between GDP and happiness the relationship still holds. Namely, when controlled for unemployment, inflation, and the degree of welfare state (Di Tella, MacCulloch and Oswald, 2003). Furthermore, the costs of recessions are calculated to identify its relationship with happiness. The recession lead to a substantial increase in well-being costs on society as many people became unemployed. When an individual becomes unemployed it approximately loses 200 dollar. However, for an individual who loses its job during the recession the real actual cost is approximately 3800 dollar (Di Tella et al, 2003). Furthermore, it has been proven with data for 12 European countries of the European Union and the United States that money buys happiness, but that the effect is small and sometimes statistically insignificant (Oswald, 1997).
8
Trade-off between Inflation and Unemployment
As discussed, inflation and unemployment lower well-being. Yet, unemployment depresses well-being more than inflation does. A one percentage point increase in the unemployment rate lowers well-being by more than five times as much as a one percentage point increase in the inflation rate (Blanchflower et al, 2014).
The kind of political systems is another condition for people’s happiness. Research shows that political, economic, and personal freedom are strongly correlated with happiness (Veenhoven, 2000). When differences in income per capita are controlled the results show that economic freedom (proxy variable for democracy) increases happiness in poor countries with a low general education level, while political freedom is strongly correlated with subjective well-being in rich countries (Veenhoven, 2000). Institutional factors in the form of direct democracy via referenda and initiatives, and of federal structure systematically raise self-reported individual well-being (Frey and Stutzer, 1999). The positive effect can be attributed to utility arising from political participation and the fact that municipalities are closer to relevant information regarding residents’ preferences, respectively. In more detail, an increase in the index of direct democracy by one point raises the share of persons indicating very high satisfaction with life by 2.8 percentage points Frey and Stutzer, 1999). The effect of federalism is as follows. Compared to a point less in the autonomous index, the share of persons indicating very high happiness increases by 3.3 percentage points.
9 3. Data
Subject well-being data and the microeconomic characteristics data from the respondents are taken form the Euro Barometer data Surveys. Briefly, the Euro Barometer Surveys is conducted on behalf of the European Commission. The objective of the survey is to get an indication of well-being, norms and social values underlying the governing conduct and social and political institutions across European countries2 .
In different European countries each year people are interviewed about these matters. Among others, there are two questions that are of interest for this paper. The first one is "Taking all things together, how would you say things are these days – would you say you're very happy, fairly happy, or not too happy these days?''. Where ''Don't know" and "No answer" categories are left out and not studied. The data was then recoded so that the answers correspond to the following numerical values: (1) ‘not too happy’, (2) ‘pretty happy’, and (3) ‘very happy’. The second one is "On the whole, are you very satisfied, fairly satisfied, not very satisfied, or not at all satisfied with the life you lead?". Once again, the "Don't know" and "No answer" categories are left out and not studied here. Moreover, also here the data answers were recoded as follows: (1) ‘not at all satisfied’, (2) ‘not very satisfied’, (3) fairly satisfied, and (4) very satisfied. Additionally, The questionnaire also asks people’s evaluations of their own democracies. These results have been used to generate a variable named satisfaction democracy in order to evaluate the role of the political system on individuals’ happiness.
The sampling of the survey was based on a random selection of individuals. Furthermore, the survey reports the answers of 646,552 individuals across 32 years on the life satisfaction question. But, the happiness question covers 137,599 individuals. This paper focusses mainly on life satisfaction data due to a longer availability regarding the period of time. Namely, from 1973 until 2002 instead of just 1986, which is the case with the happiness data. The sample size consists of sixteen European countries: France, Belgium, the Netherlands, Germany, Italy, Luxembourg, Denmark, Ireland, United Kingdom, Greece, Spain, Portugal, Norway, Finland, Sweden and Austria. Compared to the benchmark paper, this paper is the first one to extent the period of investigation. Namely, from the period 1973-2002.
The macroeconomic variables such as the unemployment rate, inflation rate, GDP per capita, and unemployment Benefits as percentage of GDP come primarily from the OECD
10
dataset. If the OECD had limitations in the data in terms of the availability of the period needed than Worlbank Indicators data is used. Nevertheless, this is done consistently as the definitions of the variables were checked and hence similar in both data sets. Additionally, the values of the variables were compared both from the OECD and WorlBank Indiactors dataset.
The Euro Barometer dataset also provides a number of background variables at the individual level. The ones that are used as controls in this study are well known in the happiness literature. They are known to impact the individual level of subjective well-being, and include age, gender, race, educational level, marital status, and personal unemployment.
Table 1 - Life Satisfaction in Europe: 1973-2002
Marital Status Reported Life
Satisfaction All (%) Unemployed (%) Married (%) Divorced (%)
Very satisfied 27.49 16.08 29.09 19.05
Fairly satisfied 55.04 46.78 54.80 54.03
Not very satisfied 13.33 25.64 12.33 20.20
Not at all satisfied 4.13 11.50 3.77 6.73
Sex Income Quartiles
Reported Life
Satisfaction Male (%) Female (%) 1st (lowest) 2nd 3rd 4th
Very satisfied 26.94 28.01 21.94 24.70 27.92 34.45
Fairly satisfied 55.69 54.43 51.50 55.09 57.15 55.48
Not very satisfied 13.15 13.51 19.19 15.49 11.89 8.17
Not at all satisfied 4.22 4.05 7.37 4.72 3.04 1.90
* Based on 646,552 observations. All numbers are expressed as percentages.
The well-being data and some microeconomic characteristics are presented in tables 1 and 2. Table 1 presents a cross-tabulation of life satisfaction for Europe in the period 1973-2002. Table 2 summarizes the happiness response for Europe in the period 1975-1986.
In general the scores in table 1 and table 2 show that well-being scores are particularly skewed towards the top of the given the answer distribution. It can be said that most of the respondents seem to answer positive. On average the individuals say that they are pretty happy and fairly (very) satisfied in Europe.
11
comes from the little period of data availability of these independent variables. The variables are education to age when the individual is 19 years or older, one child, two children, and three children or more. On top of that, when running the ordered Probit regression for happiness the following countries are dropped out as well: Norway, Finland, Sweden, and Austria. There are no observations for these countries.
* Based on 134,607 observations. All numbers are expressed as percentages.
In the appendix you will find this for each individual European country. Independent of the country where the respondent lives this papers obtains evidence that the same personal characteristics appear to correlate with reported well-being (life satisfaction).
Table 3 - Life Satisfaction Equations (Ordered Probits) for Europe: 1973 - 2002
Independent Variable Coefficient Standard Error
Unemployed -0.547*** 0.006 Self-employed 0.026*** 0.006 Retired -0.031*** 0.005 Home -0.025*** 0.005 School 0.115*** 0.009 Male -0.065*** 0.003 Age -0.027*** 0.001
Age² 2.9e-4*** 5.73e-6
Income Quartile:
2nd -0.001*** 0.004
3rd 0.118*** 0.004
4th (Highest) 0.298*** 0.004
Table 2 - Happiness in Europe: 1975-1986
Marital Status Reported Life Satisfaction All (%) Unemployed (%) Married (%) Divorced (%) Very happy 24.03 16.07 26.55 12.83 Pretty happy 57.97 51.51 57.91 55.40
Not too happy 18.00 32.43 15.54 31.77
Sex Income Quartiles
Reported Life
Satisfaction Male (%) Female (%) 1st (lowest) 2nd 3rd 4th
Very happy 22.41 25.54 18.56 21.10 25.38 30.98
Pretty happy 59.76 56.30 55.35 57.57 60.21 58.53
12 Education to age: 15-18 years -0.008*** 0.003 ≥19 years 0.025*** 0.009 Still studying 0.053*** 0.008 Marital status: Married 0.157*** 0.004 Divorced -0.248*** 0.008 Seperated -0.341*** 0.135 Widowed -0.169*** 6.9e-4 Number of children: 1 -0.018*** 0.006 2 -0.011*** 0.006 ≥3 -0.038*** 0.003 Country: Belgium 0.508*** 0.007 Netherlands 0.866*** 0.007 Germany 0.184*** 0.006 Italy -0.753*** 0.007 Luxembourg 0.757*** 0.009 Denmark 1.268*** 0.007 United Kingdom 0.531*** 0.007 Greece -0.297*** 0.007 Spain 0.188 *** 0.008 Portugal -0.282*** 0.008 Ireland 0.597*** 0.006 Norway 0.896 *** 0.013 Finland 0.554*** 0.012 Sweden 0.883*** 0.012 Austria 0.555*** 0.011 *** = 1%, ** = 5%. *= 10% level.
*Ordered Probit Regression. Number of observations = 646,552. Log likelihood = -644275,29. Cut1 = -2.04, Cut2 = -1.15, Cut3 = 0.56. The regression includes year dummies from 1973-2002. The base country is France. Question dependent variable: “On the whole, are you very satisfied, fairly satisfied, not very satisfied, or not at all satisfied with the life you lead?”
Dependent variable: reported life satisfaction.
13
Table 3 - Life Satisfaction Equations (Ordered Probit) for Europe: 1973 - 2002
Independent Variable Coefficient Standard Error
Unemployed 0.398*** 0.006 Self-employed 0.034*** 0.006 Retired 0.055*** 0.005 Home 0.060*** 0.005 School -0.076*** 0.009 Male -0.031*** 0.003 Age -0.031*** 0.001
Age² 3.1e-4*** 5.73e-6
Income Quartile: 2nd 0.013*** 0.008 3rd -0.176*** 0.009 4th (Highest) 0.292*** 0.01 Education to age: 15-18 years 0.009** 0.007 ≥19 years 0.039* 0.086 Still studying 0.049 0.15 Marital status: Married 0.222*** 0.009 Divorced -0.331*** 0.024 Seperated -0.450*** 0.035 Widowed -0.235*** 0.016 Country: Belgium 0.596*** 0.014 Netherlands 0.803*** 0.014 Germany 0.118*** 0.013 Italy -0.356*** 0.009 Luxembourg 0.341*** 0.007 Denmark 0.669*** 0.001 Ireland 0.521*** 1.3e-5 United Kingdom 0.372*** 0.012 Greece -0.464*** 0.017 Spain 0.249*** 0.023 Portugal -0.069*** 0.023 *** = 1%, ** = 5%, * = 10% level.
*Ordered Probit Regression. Number of observations = 134,150 Log likelihood = -120081,11. Cut1 = -1,20, Cut2 = 0.57. The regression includes year
dummies from 1975-1986. The base country is France. Question dependent variable: “Taken all things together, how would you say you are these days – would you say you’re very happy, fairly happy, or not too happy these days?” Dependent variable: reported happiness.
4. Empirical Strategy (Methodology)
14
which the respondent belongs, and the average income per capita in the country. The happiness regression takes the following form:
HAPPYjit = αGDPit + ∑ Personal Charjit + ɛi + Ωt + µjit , (1)
where HAPPYjit are the well-being levels by respondent j in country i in year t, and GDPit is the
gross domestic product per capita in that country (measured in constant 1985 US Dollars). Further, this variables takes log values. Then we get Personal Char.jit which is a vector of personal
characteristics of the individual. This vector in turn includes income quartile, whether the person is employed or unemployed, age, marital status, gender, education, and number of children. Furthermore, a later section in this paper checks econometric specifications with microeconomics controls such as gender and age due to the fact that many of the personal characteristics are potentially endogenous.
This regression also includes year fixed effect Ωt and country fixed effect ɛi. The first one
captures global shocks that are common to all the countries in each year and the second ensures institutional and cultural influences to be constant on reported happiness levels. In this paper there is no country-specific correction as the variable GDP per capita already takes variations into account. Next, the categorical nature of the data set leads to the use of ordered Probit models. A standard errors robustness check has been conducted. The variances are almost identical and therefore there is no evidence to correct the standard errors. Furthermore, this paper studies different lengths of lag, includes time dummies, and change-in-GDP variables. This in order to tackle the problem of trend which might potentially exist in the variable GDP. Yet, the above mentioned issues are not technical ones. Also this paper tries to estimate the effects of changes in GDP on reported well-being.
If indeed income per capita affects happiness, then a micro econometric specification designed to value other macroeconomic influences can be estimated. This regression has the following form:
HAPPYjit = αGDPit + ∑ Personal Charjit + βUnemplit + λMacroit + ɛi + Ωt + µjit , (2)
where UNEMPLOYit is the unemployment rate in country i in year t, and MACROit is a vector of
other macroeconomic variables such as INFLATIONit the rate of change of consumer prices in
nation i and year t and BENEFITit which stands for unemployment benefits, which is here defined
15
Table 5 - Summary Statistics for 16 European Countries
Statistic Obs Mean Std. Dev. Min Max
Reported life satisfaction 646.552 3.059 0.755 1 4
GDP per capita (1985 U.S.$) 338.169 22755 7734 8959 53357
∆GDP per capita 338.169 626 1925 -5844 7422
Unemployment Benefits (%GDP) 338.169 1.721 1.098 0.276 5.266
Inflation rate 338.169 3.267 2.742 -0.136 19.743
Unemployment rate 338.169 8.051 3.924 1.5 24.4
In order to explore potential problems regarding simultaneity in this paper exogenous personal controls are used; age and gender. Additionally, macroeconomic variables measured with a time lag are used. Furthermore, this paper also examines whether the respondents in these countries are satisfied with the developments of their democracy, as most of them joined the EU and therefore are obligated to adhere to a democratic system. As happiness is also related to the political system you live in. So the institutional part. The regression for this is specified as follows:
HAPPYjit = αGDPit + ∑ Personal Charjit + βsDemojit + ɛi + Ωt + µjit , (3)
where sDemojit is the satisfaction democracy variable reported by individual j in country i in
year t. This is a question included in the survey which asks people: “How satisfied are you with the democracy in your country?”. Respondents can answer from a scale 1 to 4. Where the numbers have the same meaning as in the life satisfaction question. Understanding what
the coefficients of ordered probit regressions mean is not something which is straightforward (Di Tella, 2003). The calculation of the benchmark paper can be seen in the footnote below3.
3 There are 3 cut points for a Probit regression. E.g. call them a, b and c. If an individual happiness score is equal
to H, then “very happy” (top) is P(“very happy”) = F(H – c), where F(..) is the standard cumulative normal distribution. Interpretation coefficients: ∆independent variable leads to ∆H the change in probability of calling oneself “very happy” will be ∆Prob(“very happy”) = F(H + ∆H – c) – F(H – c).
Table 6 - Correlation Coefficients for 16 European Countries
16
In this paper marginal effects will be used to interpret the magnitude of the coefficient. An ordered probit with j alternatives will have j sets of marginal effects. Each unit increase in the independent variable increases/decreases the probability of selecting alternative j by the marginal effect expressed as a percentage.
Lastly, table 6 sets out the means and standard deviations for the macroeconomic variables and table 7 contains correlation coefficients for these variables.
5. The Effect of GDP on Happiness
The first hypothesis to be tested is whether GDP per capita has an effect on the well-being feelings of citizens. Later in this paper a value on the effect of booms and busts on happiness levels will be attached. This will be done by calculating the marginal rate of substitution between GDP and unemployment. Section six elaborates on this. Table 7 presents ordered probit regressions for life satisfaction (happiness) equations focused on GDP and a variety of lagged lengths. The columns 1 to 6 regresses life satisfaction on the country’s GDP per capita and the set of personal characteristics of the respondent. Furthermore, the variable GDP is taken in log. The time period of the data is extended, 1973-2002, and covers sixteen European countries compared to the reference paper (Di Tella, 2003) which includes twelve countries and covers the period 1975-1992. Moreover, the regressions control for country and year fixed effects. The coefficient of GDP reflects the effect of an absolute increase on individual happiness. This is because the regressions control the quantile to which the respondent’s family income belongs (income quartile). This research shows that there is evidence of a significant and positive effect of GDP per capita on the well-being of people.
Table 6 Life satisfaction and GDP per capita, Ordered Probit Equation and Marginal Effects
Dependent variable: Life satisfaction
(1) Not at all satisfied -0.00463
(2) Not very satisfied -0.0087
(3) Fairly satisfied 0.0169
(4) Very satisfied 0.0171
17
And an $1000 extra GDP per capita (1985 dollars) lowers the proportion in the bottom category (1) which is “not at all satisfied” with life by 0.46 percentage points. This is 0.70 percentage points in the reference paper (Di Tella et al, 2003). To check robustness and understand the dynamics, columns (2) and (3) of table 7 give the results with GDP lagged levels. As you can see in column (2) of table 7 the coefficient of GDP per capita in the happiness equation is reduced a little. Hence column (2) shows a good and well-determined GDP effect. However, this effects weakens in column (3) but the GDP coefficient remains positive. According to the benchmark paper (Di Tella et al, 2003) a possible solution to counter this problem that the GDP regressor provides an unpersuasive estimator for the effect of national income on well-being is to focus on the growth rate in income. But, before providing these results, column (4) of table 7 presents a set of variables for GDP per capita, lagged GDP , and twice lagged GDP.
18
probably requires a much larger time period. This can be an idea to investigate for future research.
* Standard Errors in parentheses. Significance: *** = 1%, ** = 5% and * = 10% level. Cut points are -1.49, -0.61, 1.11 for reg. (1); -0.70, -0.18, 1.90 for reg. (2); -1.25, -0.37, 1.35 for reg. (3); -1.25, -0.36, 1.35 for reg. (4); -2.05, -1.17, 0.54 for reg. (5); -2.5, -1.60, 0.20 for reg. (6). Log is taken for GDP. Dependent variable: reported life satisfaction.
Table 7 Life Satisfaction and GDP per Capita, Ordered Probit Regressions, Europe from 1975-2002
Independent Variable [1] [2] [3] [4] [5] [6] GDP per capita 0.576 (0.003)*** 0.155 (0.014)** GDP per capita (-1) 0.540 (0.008)*** 0.883 (0.071)** GDP per capita (-2) 0.430 (0.003)*** -0.960 (0.068)** ∆GDP per capita 0.018 (0.002)** ∆GDP per capita (-1) 0.172 (0.012)*** Personal Characteristics Unemployed -0.551 (0.006)*** -0.551 (0.006)*** -0.551 (0.006)*** -0.546 (0.006)*** -0.534 (0.008)*** -0.235 (0.106)** Self-employed 0.023 (0.006)*** 0.023 (0.006)*** 0.023 (0.006)*** 0.029 (0.006)*** 0.027 (0.008)*** 0.273 (0.124)** Retired -0.039 (0.006)*** -0.038 (0.005)*** -0.038 (0.005)*** -0.032 (0.006)*** -0.036 (0.007)*** 0.050 (0.091) Home -0.027 (0.005)*** -0.026 (0.005)*** -0.026 (0.005)*** -0.018 (0.005)*** -0.026 (0.006)*** 0.121 (0.070)* School 0.097 (0.008)*** 0.099 (0.009)*** 0.099 (0.009)*** 0.100 (0.009)*** 0.124 (0.010)*** 0.183 (0.112) Male -0.065 (0.003)*** -0.065 (0.003)*** -0.065 (0.003)*** -0.064 (0.003)*** -0.064 (0.004)*** -0.082 (0.019)*** Age -0.028 (0.001)*** -0.027 (0.005)*** -0.027 (0.005)*** -0.064 (0.003)*** -0.027 (6e-3)*** -0.033 (0.004)*** Age² 3.0e-4 (5.78e-06)*** 2.9e-4 (5.75e-6)*** 2.9e-4 (5.75e-6)*** 2.9e-4 (6e-6)*** 2.8e-4 (6e-6)*** 4e-4 (4.3e-5)***
Income Quartile: 2nd 0.003 (0.004)*** 0.003 (0.004) 0.003 (0.004) 0.006 (0.003)*** -0.001 (0.005) 0.220 (0.021)*** 3rd 0.119 (0.004)*** 0.119 (0.004)*** 0.119 (0.004)*** 0.127 (0.004)*** 0.111 (0.005)*** 0.431 (0.074)*** 4th (Highest) 0.293 (0.004)*** 0.295 (0.004)*** 0.295 (0.004)*** 0.230 (0.004)*** 0.296 (0.005)*** 0.636 (0.100)*** Education to age: 15-18 years -0.009 (0.003)*** -0.010 (0.003)*** -0.009 (0.003)*** -0.010 (0.003)*** -0.004 (0.004) 0.124 (0.021)*** ≥19 years 0.021 (0.009)** 0.026 (0.009)*** 0.026 (0.009)*** 0.013 (0.009)*** 0.023 (0.009)*** 0.074 (0.116) Still studying 0.063 (0.008)*** 0.060 (0.008)*** 0.060 (0.008)*** 0.061 (0.008)*** 0.054 (0.009)*** 0.101 (0.043)** Marital status: Married 0.144 (0.006)*** 0.161 (0.004)*** 0.161 (0.004)*** 0.158 (0.004)*** 0.145 (0.005)*** 0.089 (0.028)*** Divorced -0.266 (0.009)*** -0.248 (0.008)*** -0.248 (0.008)*** -0.255 (0.008)*** -0.258 (0.010)*** -0.262 (0.005)*** Seperated -0.363 (0.014)*** -0.344(0.013)*** -0.344(0.013)*** -0.344 (0.014)*** -0.371 (0.016)*** 0.132 (0.221) Widowed -0.183 (0.008)*** -0.166 (0.007)*** -0.166 (0.007)*** -0.165 (0.007)*** -0.179 (0.008)*** -0.094 (0.103) Number of children: 1 -0.020 (0.006)*** -0.018 (0.006)*** -0.018 (0.006)*** -0.018 (0.005)*** -0.018 (0.007)*** 0.037 (0.100) 2 -0.0147 (0.006)*** -0.013 (0.006)*** -0.127 (0.006)** -0.132 (0.006)** -0.011 (0.008)* 0.046 (0.104) ≥3 -0.004 (0.004) -0.011 (0.004)** -0.011 (0.004)** 2.1e-3 (0.004) -0.034 (0.004)*** -0.229 (0.060)***
Country fixed effects Yes Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes Yes
Pseudo-R² 0.08 0.08 0.08 0.08 0.08 0.08
19
6. The Macroeconomics of Happiness (3x Macro Variables)
Income correlates with happiness. Now the question that remains is whether this correlation holds if macroeconomic variable are added. Basically, the previous analysis is repeated with the inclusion of the unemployment rate, the inflation rate, and unemployment benefits. It does not remove the correlation between happiness and GDP. As can be seen in column (1) of table 8 the macroeconomic variables enter the ordered probit regression with the signs as expected. Unemployment benefits has a positive relationship with happiness, unemployment and inflation rate enter negatively so an increase in these variables has negative impact on happiness. Further, all are statistically significant.
Table 8 - Life Satisfaction and Macroeconomic Variables, Ordered Probit Regressions, Europe: 1975-2002
Independent Variable [1] [2] [3] [4] GDP per capita 0.618 (0.007)*** 0.534 (0.024)** 0.776 (0.009)* GDP per capita (-1) 0.085 (0.024)* GDP per capita (-2) -0.080 (0.024) ∆GDP per capita 0.31 (0.002)*** 0.771 (0.002)** Unemployment Benefits (%GDP) 0.150 (0.001)*** 0.151 (0.002)*** 0.144 (0.002)*** 0.161 (0.002)***
Unemployment rate -0.025 (5.6e-3)*** -0.030 (0.001)*** -0.042 (0.001)*** -0.023 (0.001)***
Inflation rate -0.024 (8.4e-3)*** -0.024 (0.001)*** -0.069 (0.001)*** -0.018 (0.001)***
Personal Characteristics Yes Yes Yes Yes
Country fixed effects Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes
Pseudo R² 0.08 0.08 0.08 0.09
Number of Observations 664,552 664,552 664,552 664,552
* Standard Errors in parentheses. Significance: *** = 1%, ** = 5% and * = 10% level. Cut points are 3.9, 2.9, 1.2 for reg. (1); 2.5, 1.6, 0.19 for reg. (2); 3.1, -2.2, -0.5 for reg. (3); -4.6, -3.3, -1.6 for reg. (4). Log is taken for GDP.
Dependent variable: reported life satisfaction.
20
Table 9 - Life satisfaction Regressions and Exogeneity, Ordered Probit Regressions, Europe: 1975-2002.
Independent Variable [1] [2] [3] GDP per capita (-1) 0.571 (0.007)*** 0.638 (0.007)*** GDP per capita (-2) -0.181 (0.011)* GDP per capita (-3) 0.578 (0.007) ∆GDP per capita (-1) 0.015 (0.003)*** Unemployment Benefits (%GDP)(-1) 0.136 (0.002)*** -0.078 (0.010)*** Unemployment rate (-1) -0.025 (0.006)*** -0.002 (0.002)*** Inflation rate (-1) -0.020 (0.001)*** -0.012 (0.001)** Personal Characteristics: Male -0.002 (0.004)** -0.008 (0.007)** -0.010 (0.001)** Age -0.019 (0.006)*** -0.014 (0.001)*** -0.020 (0.001)***
Age² 1.7e-3(1.2e-5)*** 1.4e-4 (1.2e-5)*** 1.8e-3 (7.9e-6)***
Country fixed effects Yes Yes Yes
Year fixed effects Yes Yes Yes
Pseudo R² 0.07 0.07 0.07
Number of Observations 664,552 664,552 664,552
* Standard Errors in parentheses. Significance: *** = 1%, ** = 5% and * = 10% level. Cut points are 5.9, 4.2, 3.3 for reg. (1); 4.1, 1.4, 3.2 for reg. (2); 2.1, -1.3, 0.5 for reg. (3). Log is taken for GDP. Dependent variable: reported life satisfaction.
7. Costs of Recessions
This paper tries to attach a value on booms and busts (recessions), therefore, it compares the marginal effect of income and unemployment on happiness. Or put differently, the marginal rate of substitution between GDP and unemployment. This has been conducted in a similar approach in a previous research (Di Tella et al, 2003). In order to make these calculations a yardstick of the unemployment rate needs to be taken. In this paper an unemployment rate equal to 1.5 percentage points will be taken as the base paper. The authors chose this number by taking the average of the eleven full business cycles in the United States since WOⅡ and divide this by two to get the average unemployment deviation (Di Tella et al, 2003).
21
compensation for the GDP decline4. This is 330 dollars in the reference paper (Di Tella et al,
2003). So the amount of 556 dollars needs to be paid to the average citizen not only the ones that become unemployed. Additionally, the same calculations can be made when using the growth rate of GDP per capita. To keep life satisfaction constant, the average citizen’s income in these economies must be increased by 6.4% in the event of a recession which increases the unemployment rate with 1.5 percentage points5. This is 3.2% in the reference paper (Di Tella et al, 2003).
7.1 The full cost to society
The above calculations do not take into account the cost to society. They underestimate the full cost to society. These regression hold constant the microeconomic personal cost of being unemployed. From table 8 in column (1) an increase in the unemployment rate from 0 to 1.5% leads to a cost of 0.0004 (0.025*0.015) in utility. This is for the average person whether this person is employed or unemployed. A person that becomes unemployed experiences a loss of 0.55 in utility. This is unreported and named after the heading personal characteristics. However it is similar to the number in column (1) of table 7. The full social costs of an increase of 1.5 percentage points in the rate of unemployment in happiness units is the sum of the two: it is (0.55*0.015) + (0.025*0.015) = 0.0083 + 0.0004 = 0.0087. This is approximately 141 dollars. The actual loss for an individual that loses its job during a downturn is approximately 8906 dollars ((0.55 + 0.0004)/0.0000618).
Overall, the regressions in table 8 and the above calculations show that high unemployment in the economy is even bad for people that do have a job. People fear if the pool of unemployed increases in the economy because they might become overworked and stressed (Clark and Oswald, 1994). Also an taxation effect might be a good argument. If the pool of unemployed people increases this means that people have to pay more tax to fund the unemployment benefits (Helliwel, 2001). In conclusion, unemployment carries a large well-being costs which is not implied by the drop in income alone.
4 0.015*(0.025/0.0000618). The factor 0.015 comes from the yardstick. It is an assumption that a recession leads 1.5 percentage points to unemployment. 0.025 is the unemployment rate coefficient in table 8 column (1). In the same table 0.618 is the coefficient on GDP per capita scaled by a factor 10.000.
22
Yet, we can also put a value on the cost of inflation. Put differently, a calculation of the marginal rate of substitution between GDP and inflation can be made.
To keep life satisfaction constant an individual needs to receive a compensation of approximately 38 dollars (0.01* (0.024/0.0000618)) for each one percentage point rise in inflation. This is 70 dollars in the reference paper (Di Tella et al, 2003).
8. Happiness Evidence on the Role of the Political System
Table 10 shows that the coefficient on satisfaction democracy is positively correlated with happiness levels. Further, the coefficients are statistically significant at the one percent level. The sign is positive which implies that the political system is another condition for people’s happiness. In this case the political system is a democracy as the European countries in the sample are part of the European Union, which requires a country to be a democracy. So, living in a democracy contributes to citizens’ happiness levels and hence makes them satisfied about the democratic political system. Further, the same conclusions holds for the lagged variables just like the previous tables. When calculating the marginal effects regarding life satisfaction and satisfaction democracy the share of respondents that indicate to be very satisfied with their lives (4) increases by approximately 2.9 percentage points.
* Standard Errors in parentheses. Significance: *** = 1%, ** = 5% and * = 10% level. Cut points are -1.9, -1.05, 0.69 for reg. (1); -1.71, -0.81, 0.92 for reg. (2); -1.9, -0.97, 0.76 for reg.(3); -2.4, -1.5, 0.2 for reg. (4). Log is taken for GDP.
Dependent variable: reported life satisfaction.
Table 10 – Life Satisfaction regression and satisfaction democracy [Ordered Probit Regressions], Europe: 1975-2002.
Independent Variable [1] [2] [3] [4] Satisfaction democracy 0.116 (0.001)*** 0.115 (0.001)*** 0.116(0.001)*** 0.110*** GDP per capita 0.222 (0.016)*** 0.218 (0.17)*** GDP per capita (-1) 0.753 (0.074)*** 1.274 (0.096)** GDP per capita (-2) -0.896 (0.071)*** -1.114 (0.231)* ∆GDP per capita 0.038 (0.178) 0.005 (0.002)**
Personal Characteristics Yes Yes Yes
Country fixed effects Yes Yes Yes
Year fixed effects Yes Yes Yes
Pseudo R² 0.09 0.09
23
9. Conclusions
24 REFERENCES
Alesina, Alberto, Rafael Di Tella, and Robert MacCulloch (2001) “Happiness and Inequality: Are Europeans and Americans Different?”, Journal of Public Economics. NBER working paper no. 8198.
Bruno, S & A. Strutzer (2002), The economics of happiness, World Economics, 3 (1), 1-17. Bruno, S & A. Strutzer (1999), Happiness, Economy and Institutions, The Economic Journal, 110(466), 918-938.
Blanchflower, David G and Andrew J. Oswald (1999) “Well-being over Time in Britain and the USA”, Journal of Public Economics.
Blanchflower, David G., Beil, D., Montagnoli, A and Mirko Moro (2014) The Happiness Trade-Off between Unemployment and Inflation, Journal of Money, Credit and Banking, 46(2), 118-137.
Clark, Andrew and Andrew J. Oswald (1994) “Unhappiness and Unemployment”, Economic Journal, 104, 648-659.
Diener, Edward (1984) Subjective Well-being, Psychological Bulletin, 93, 542-575.
Di Tella, Rafael, Robert MacCulloch and Adrew J. Oswald (2001) Preferences over Inflation and Unemployment: Evidence from Happiness Surveys. American Economic Review, 91(1), 335-342.
Di Tella, R, Robert J. MacCulloch and Adrew J. Oswald (2003), The macroeconomics of happiness, The Review of Economics and Statistics, 85(4), 809-827.
Easterly, William (1999) “Life during growth”, Journal of Economic Growth, 4(3), 239-276. Easterlin, Richard A. (2001) “Income and happiness: toward an unified theory”, Economic Journal, 111(473), 465-484.
Easterlin, Richard (1974) Does Economic Growth improve the Human Lot?, Nations and Households in Economic Growth: Essays in Honor of Moses Abramowitz.
Easterlin, Richard (1995) Will raising the Incomes of All increase the Happiness of All?, Journal of Economic Behavior and Organization, 27(1), 35-48.
Frey, Bruno S. and AloisStutzer (2000) Happiness, Economy, and Institutions, Economic Journal, 110, 918-938.
Frey, Bruno S. and AloisStutzer (2002) What can Economists Learn from Happiness Research? Journal of Economic Literature, XL(2), 402-436.
Gardner, Jonathan and Andrew J. Oswald (2001) Does Money Buy Happiness? A Longitudinal Study using Data on Windfalls, Warwick University.
25
Mota, G.L. and Paulo Trigo Pereira (2008) Happiness, Economic Well-being, Social Capital and the Quality of Institutions, School of Economics and Management, Technical University of Lisbon, working paper no. 40.
Oswald, Andrew J. (1997) Happiness and Economic Performance, Economic Journal, 107, 1815-1831.
Sandvik, E., Diener E. and Seidlitz, L (1993) Subjective Well-Being: The Convergence and Stability of Self-Report and Non-Self-Report Measures, Journal of Personality, 61(3), 317-342.
Veenhoven, Ruut (2000) Freedom & happiness: an comparative study in fourth four nations in the early 1990s, Culture and Subjective well-being, 257-288, Cambridge, MA: MIT press. Winkelmann, Liliana and Rainer Winkelmann (1998) Why Are the Unemployed so Unhappy?, Economica, 65(257), 1-15.
APPENDIX
DATA SOURCES
A. Euro-Barometer Data Surveys: B. OECD
C. WorldBank Indicators
DATA DEFINITIONS
A. Reported life satisfaction: The answer to the Euro-Barometer Survey question that aks: "On the whole, are you very satisfied, fairly satisfied, not very satisfied or not at all satisfied with the life you lead?"
B. Reported happiness: The answer to the Euro-Barometer Survey question that asks: "Taken all together, how would you say things are these days - would you say that you are very happy, pretty happy, or not too happy?"
C. ALL the economic Variables
GDP = GDP per capita with constant US prices. From: Worldbank Indicators.
Unemployment benefits = public unemployment spending is defined as expenditure on cash benefits for people to compensate for unemployment. This indicator is measured in percentage of GDP. From the OECD data set.
Unemployment rate = the standardize unemployment rate from the OECD and Worldbank indicators data set.
26
* Sometimes data on the economic variables has been merged from two different data sources. This has been applied consistent in the sample by looking at the definition of the indicator (due to lack of data availability in one data set).
A. TABLES
Table A-1 Life Satisfaction Equations (Ordered Probits) in European Nations: 1973 - 2002
Independent Variable France Belgium Netherlands Germany
Unemployed -0.356 (0.020) -0.484 (0.020) -0.563 (0.024) -0.785(0.016) Self-employed 0.009 (0.247) -0.100 (0.019) 0.445 (0.031) 0.087 (0.021) Retired 0.207 (0.019) -0.083 (0.019) 0.005 (0.029) -0.057 (0015) Home 0.036 (0.016) -0.013 (0.017) -0.045 (0.015) 0.120 (0.014) School 0.142 (0.030) - 0.113(0.032) 0.100 (0.031) 0.163 (0.029) Male -0.045 (0.010) -0.045 (0.011) -0.180 (0.012) -0.017 (0.008) Age -0.033 (0.001) -0.025 (0.001) - 0.039 (0.002) -0.018 (0.001)
Age² 3.5e-4 (0.002) 2.6e-4 (1.9e-5) 4.2e-4 (2.1e-5) 2.1e-4 (1.6e-5) Income Quartile: 2nd -0.165 (0.013) -0.001 (0.012) -0.036 (0.015) -0.007 (0.011) 3rd 0.158 (0.012) 0.148(0.015) 0.121 (0.014) 0.113 (0.012) 4th (Highest) 0.379 (0.015) 0.225 (0.021) -0.355 (0.015) 0.326 (0.011) Education to age: 15-18 years -0.035 (0.010) -0.045 (0.010) 0.006 (0.100) -0.03 (0.010) ≥19 years 0.010 (0.035) 0.037 (0.028) 0.118 (0.034) 0.109 (0.025) Still studying 0.192 (0.026) 0.133 (0.029) 0.001 (0.026) 0.027 (0.025) Marital status: Married 0.107 (0.013) 0.215 (0.014) 0.218 (0.017) 0.136 (0.012) Divorced -0.213 (0.026) -0.279 (0.027) -0.366 (0.03) -0.264 (0.019) Seperated -0.273 (0.046) -0.395 (0.043) -0.655 (0.08) -0.362 (0.042) Widowed -0.170 (0.025) -0.222 (0.023) -0.340 (0.028) - 0.082 (0.019) Number of children: 1 -0.015 (0.020) -0.067 (0.021) 0.017 (0.024) - 0.069 (0.015) 2 0.006 (0.033) -0.053 (0.024) 0.015 (0.023) -0.035 (0.019) ≥3 -0.044 (0.012) -0.013 (0.011) -0.078 (0.012) 0.064 (0.009)
Table A-1 (CONTINUED)
Independent Variable Italy Luxembourg Denmark Ireland
Unemployed -0.547 (0.023) 0.622 (0.068) -0.401 (0.023) -0.764 (0.019) Self-employed 0.007 (0.016) 0.037 (0.040) 0.054 (0.030) 0.096 (0.027) Retired 0.030 (0.018) 0.074 (0.031) -0.126 (0.024) - 0.039 (0.022) Home 0.065 (0.015) 0.040 (0.025) -0.062 (0.025) -0.139 (0.015) School 0.277 (0.026) 0.192 (0.051) -0.025 (0.030) 0.075 (0.031) Male 0.009 (0.018) -0.079 (0.018) -0.131 (0.011) -0.175 (0.012) Age 0.026 (0,001) -0.017 (0.003) -0.026 (0.001) -0.023 (0.002)
27 Still studying 0.030 (0.023) 0.085 (0.043) 0.099 (0.028) 0.0102 (0.029) Marital status: Married 0.177 (0.014) -0.089 (0.050) 0.191 (0.014) 0.138 (0.013) Divorced -0.077 (0.046) 0.204 (0.024) -0.251 (0.026) 0.044 (0.087) Seperated -0.256 (0.043) -0.220 (0.048) - 0.314 (0.055) -0.465 (0.042) Widowed -0.161 (0.023) -0.341 (0.077) -0.119 (0.025) -0.169 (0.023) Number of children: 1 0.014 (0.019) -0.316 (0.030) -0.014 (0.023) -0.241 (0.021) 2 0.024 (0.025) 0.025 (0.034) -0.003 (0.025) - 0.051 (0.021) ≥3 0.231 (0.114) -0.117 (0.018) -0.123 (0.013) - 0.071 (0.011)
Table A-1 (CONTINUED)
Independent Variable U.K. Greece Spain Portugal
Unemployed -0.554 (0.017) -0.239 (0.028) -0.408 (0.025) -0.509 (0.033) Self-employed 0.031 (0.022) 0.057 (0.018) 0.016 (0.023) 0.102 (0.024) Retired -0.026 (0.016) 0.072 (0.021) 0.039 (0.025) -0.092 (0.023) Home -0.066 (0.012) 0.062 (0.017) - 0.022 (0.021) -0.007 (0.020) School 0.116 (0.032) 0.198 (0.032) -0.011 (0.036) 0.330 (0.046) Male -0.114 (0.009) 0.006 (0.011) -0.027 (0.014) 0.053 (0.013) Age -0.021 (0.002) -0.032 (0.002) -0.037 (0.002) -0.035 (0.002)
Age² 2.6e-3 (1.6e-4) 2.9e-3 (2.3e-4) 3.6e-3 (2.3e-4) 3.1e-3 (2.4e-4)
Income Quartile: 2nd -0.071 (0.012) 0.094 (0.013) -0.002 (0.016) 0.0367 (0.017) 3rd 0.057 (0.012) 0.199 (0.016) 0.042 (0.017) 0.108 (0.019) 4th (Highest) 0.250 (0.012) 0.364 (0.017) 0.172 (0.018) 0.242 (0.018) Education to age: 15-18 years -0.020 (0.010) 0.055 (0.012) 0.004 (0.015) 0.051 (0.017) ≥19 years - 0.018 (0.029) -0.108 (0.034) -0.132 (0.036) -0.072 (0.028) Still studying 0.065 (0.029) 0.044 (0.031) 0.128 (0.036) -0.083 (0.014) Marital status: Married 0.141 (0.012) 0.149 (0.017) 0.175 (0.018) 0.079 (0.019) Divorced -0.355 (0.023) -0.201 (0.047) -0.030 (0.064) -0.168 (0.049) Seperated -0.361 (0.034) -0.403 (0.105) - 0.153 (0.049) -0.144 (0.070) Widowed -0.217 (0.020) -0.142 (0.028) - 0.140 (0.029) -0.182 (0.029) Number of children: 1 -0.033 (0.018) -0.005 (0.028) -0.004 (0.019) - 0.013 (0.020) 2 -0.033 (0.019) -0.001 (0.021) -0.022 (0.024) -0.051 (0.026) ≥3 - 0.032 (0.010) 0.042 (0.012) 0.071 (0.014) 0.006 (0.015)
Table A-1 (CONTINUED)
Independent Variable Norway Finland Sweden Austria
Unemployed 0.551 (0.056) -0.449 (0.041) -0.435 (0.051) -0.607 (0.061) Self-employed -0.032 (0.062) 0.066 (0.060) 0.182 (0.060) -0.057 (0.050) Retired -0.402 (0.052) -0.035 (0.039) -0.0761 (0.044) 0.019 (0.041) Home -0.110 (0.056) -0.233 (0.052) 0.250 (0.109) 0.097 (0.038) School -0.021 (0.054) 0.164 (0.107) -0.071 (0.107) -0.190 (0.142) Male -0.070 (0.025) -0.195 (0.022) -0.081 (0.023) -0.053 (0.023) Age -0.031 (0.005) -0.048 (0.004) -0.046 (0.004) -0.017 (0.004)
28 Income Quartile: 2nd -0.030 (0.037) 0.163 (0.029) 0.149 (0.029) -0.030 (0.029) 3rd 0.056 (0.037) 0.242 (0.038) 0.338 (0.036) 0.079 (0.002) 4th (Highest) 0.275 (0.041) 0.422 (0.049) 0.520 (0.051) 0.311 (0.044) Education to age: 15-18 years -0.135 (0.031) 0.133 (0.030) -0.052 (0.027) -0.056 (0.024) ≥19 years -0.049 (0.039) 0.037 (0.045) 0.133 (0.048) 0.206 (0.045) Still studying -0.089 (0.061) -0.196 (0.109) -0.022 (0.110) 0.264 (0.146) Marital status: Married 0.161 (0.031) 0.276 (0.031) 0.314 (0.031) 0.149 (0.032) Divorced -0.152 (0.062) -0.123 (0.041) -0.163 (0.049) -0.142 (0.046) Seperated - 0.547 (0.116) -0.047 (0.128) -0.261 (0.093) -0.203 (0.091) Widowed -0.044 (0.064) 0.063 (0.051) -0.119 (0.056) -0.173 (0.052) Number of children: 1 0.071 (0.037) -0.011 (0.041) 0.012 (0.043) -0.055 (0.038) 2 0.076 (0.042) -0.072 (0.046) 0.015 (0.046) -0.17 (0.004) ≥3 0.037 (0.038) -0.016 (0.025) -0.013 (0.025) 0.061 (0.025)
Table A-2 Means and Standard Deviations for European Life Satisfaction Regression: 1975-2002
Variable Mean Standard Deviation
Dependent Variable:
Reported life satisfaction 3.058 0.755
29 Table A-3 Means and Standard Deviations for European Happiness Regression:
1975-2002
Variable
Mean Standard Deviation Dependent Variable:
Reported life satisfaction 2.060 0.645
