Happiness across the life span: Evidence from urban Pakistan Khadija Shams a and Alexander Hendrik Kadow b a Dept. of Economics, Shaheed Benazir Bhutto Women University, Peshawar, Pakistan; email: kshams2008@gmail.com b Dept. of Economics, University of Frankfurt, Frankfurt, Germany; email: alexander@kadow.eu September 28-30, 2017 (Innsbruck, Austria)
Outline 1 Motivation and Contribution 2 Literature review 3 Data 4 Descriptive statistics of the SWB metric 5 Happiness index and average monthly income across the provinces 6 Baseline results 7 Specification error test 8 Weighting matrix 9 Conclusion and policy implications
Motivation Pakistan is the 6 th most populated country in the world. Pakistan is typically characterised by poverty, income inequality and low literacy rate particularly among females. While happiness studies provide some empirical evidence for industrialised countries, there are few studies on developing economies. The present research will contribute to the existing literature and may motivate future research on other parts of the developing economies In particular, our results allow for comparisons between drivers of happiness in rural and urban areas.
Contribution So: Research Question Does happiness vary across the life span/life course? identify the tipping/turning points to SWB across the life span (econometric analysis) determine average happiness across the regions (provinces) measure the impact of each driver on SWB (regression analysis)
A brief look at the literature I Blanchflower (NBER Working Paper No. w14318, 2008): Based on US and European panel data, reports SWB is U-shaped in age. Stutzer and Frey (Happiness and Economics, 2002): argue that well-being from an economic perspective tends to be particularly driven by: age, health and monetary factors. Easterlin (Journal of Economic Psychology, 2006): Based on US data, find evidence for health and financial well-being to follow a U-shaped pattern in age. One factor for both health and financial satisfaction to increase as individuals become older are perhaps the compulsory social insurance programmes typically in place. Shams (Journal of Happiness Studies, 2016):
A brief look at the literature II Based on rural Pakistan data, reports financial well-being to be U-shaped in age. naturally, age-specific turning points are likely to differ across countries/regions and over time. Perz (Population Research and Policy Review, 2001); Sutherland et al. (Poblacin y Salud en Mesoamrica, 2004); Shams (Social Indicators Research, 2014) and Shams (Journal of Happiness Studies, 2016): favor the idea of children being considered as an economic security/safety against the scarce resources in agricultural economies, for instance, Brazilian Amazon, Guatemala s Peten (an agricultural frontier) and rural Pakistan, respectively. Blanchflower (NBER Working Paper No. w14318, 2008) and Tella et al. (Review of Economics and Statistics, 2003): find a negative impact of children on household s happiness.
A brief look at the literature III Stutzer and Frey (Journal of Socio-Economics, 2006); Shams (Social Indicators Research, 2014) and Shams (Journal of Happiness Studies, 2016): find a positive impact of having children on household s happiness. Clark (PSE Working Papers n-2006-35, 2006) and Clark et al. (Journal of Economic Literature, 2008): find a null/no effect of children on household s happiness.
Data I collected data by means of household survey (2016) total sample size N = 6000 random selection of 6000 households within 8 predetermined major cities across Pakistan s provinces (stratified sampling) based on population figures of the provinces and the sampled cities, we assigned weights: Punjab (= 50%), i.e. 1500 households from Lahore, 600 from Faisalabad, 450 from Rawalpindi, 300 from Multan and 150 from Islamabad Sind (= 43%), i.e. 2600 households from Karachi KPK (= 4%), i.e. 240 households from Peshawar Baluchistan (= 3%), i.e. 160 households from Quetta The number of cities from each province has been chosen in line with population figures of the given provinces. In all cases, the corresponding provincial capital has been included in the sample.
Data II detailed questionnaire advantage: differentiated picture in terms of SWB and household characteristics limitation: static perspective (do not consider changes over time)
Descriptive statistics of the SWB metric I Data set allows for distinguishing among 4 different levels of household s happiness (with its existing SES), in an ordinal scale ranking from 1 to 4, such that 1 stands for not at all happy, 2 for less than happy, 3 for rather happy and 4 for fully happy.
Descriptive statistics of the SWB metric II SWB happiness with socio-economic status (1-4) Mean 2.23 Standard deviation 1.17 Frequency of value: 1 40% 2 16% 3 25% 4 19% Table: Descriptive statistics of the subjective well-being metric. Source: Survey 2016.
Descriptive statistics of the SWB metric III More than half of the respondents lie on the lower end of the happiness scale i.e. being not at all happy or less than happy (which is typical for many developing countries where most of the respondents have lower happiness index) In contrast, Angeles (2009) did a similar study on British households and found that more than three quarters of the respondents lied on above average satisfaction level
Happiness index and average monthly income across the provinces I Average happiness Punjab Sind KPK Baluchistan with Socio-Economic Status (SES) 2.36 2.21 2.13 1.56 Table: Happiness index across the provinces. Note: N = 6000. Source: Survey 2016. Average Punjab Sind KPK Baluchistan monthly income in Pakistani Rupee (PKR) 42,641.22 35,297.15 35,116.67 22,075.00 Table: Average monthly income across the provinces. Note: N = 6000. Source: Survey 2016. hadija Shams a and Alexander Hendrik Kadow b ( a September 28-30, 2017(Innsbruck, Austria)
Happiness index and average monthly income across the provinces II average happiness increases with average monthly income across the provinces or in other words living in richer areas ensures greater happiness
Baseline results Ordered probit regression Number of obs = 6000 Wald χ 2 (18) = 318.88 Prob > χ 2 = 0.0000 Pseudo R 2 = 0.4513 Log pseudolikelihood = -435.4678 Dependent variable: SWB Independent Coef. Robust Variable Std. Err. Male 0.6012* 0.3199 Age -0.3414*** 0.0611 AgeSquared 0.0043*** 0.0007 Years of education 0.1338*** 0.0262 Unemployed -0.2738* 0.1566 Log of household s income 2.7208*** 0.2192 No. of children: Being childless Reference Group Child 1 2.3898*** 0.3828 Child 2 0.5767*** 0.1181 Child 3 (or more) 0.0525 0.2010 Marital status: Married couple 0.2935* 0.1757 Health index: 4 0.5233** 0.2693 3 0.3982* 0.2383 2 0.3346 0.2722 1 0.0989 0.1247 0 Reference Group
Baseline results (continued...) Dependent variable: SWB Independent Coef. Robust Variable Std. Err. Region: Punjab 0.2974* 0.1778 Sind 0.2574* 0.1540 KPK 0.0794 0.1163 Baluchistan Reference Group Religious belief: Non-Muslim -0.0513 0.1182 /cut1 19.6866 2.3606 /cut2 20.6818 2.3956 /cut3 22.2772 2.4630 Table: Baseline results. *,**,*** indicate 10%, 5% and 1% levels of statistical significance, respectively. the positive and negative coefficients indicates that with one unit increase in the independent variable(s), it s more likely to be in the higher categories (i.e. 3 and 4) and lower categories (i.e. 1 and 2) of SWB, respectively. Similar to the results for rural China (Knight et al., 2009), the US or many EU countries (Blanchflower, 2008), we establish a U-shaped pattern between age and happiness with a statistical turning point of 39.70 years of age that is approximately forty years. Naturally, age-specific turning points are likely to differ across countries/regions and over time.
Specification error test Specification error test Number of obs = 6000 Wald χ 2 (2) = 221.52 Prob > χ 2 = 0.0000 Pseudo R 2 = 0.4517 Log pseudolikelihood = -435.19531 Dependent variable: SWB Independent coef. Robust Variable Std. Err. -hat 1.7256* [0.052] 0.8881 -hatsq -0.0174 [0.416] 0.0214 /cut1 27.1896 9.2166 /cut2 28.1960 9.2420 /cut3 29.7803 9.2191 Table: Specification error test: Baseline Model. Notes: *,**,*** indicate 10%, 5% and 1% levels of statistical significance, respectively; p-values are given in square brackets. the specification error test is found to be statistically insignificant which shows that there is no omitted variable bias in the baseline model
Weighting matrix Census 1998 Survey 2016 Province/City Total Urban Sample pweights pweights-normalised Population (in k) Population (in k) Population [(UP) j /(SP) j ] [(pw) j /Σ(pw) j ] (k/j) (TP) j (UP) j (SP) j (pw) j (pw) j Punjab 74,426 23,548 3000 Lahore 5,443 4,485 1500 2.990 0.164 Faisalabad 2,009 858 600 1.430 0.079 Rawalpindi 1,410 750 450 1.667 0.092 Multan 1,197 505 300 1.683 0.092 Islamabad (Capital) 529 348 150 2.320 0.127 Sind 30,440 14,840 2600 Karachi 9,339 9,122 2600 3.508 0.193 KPK 17,744 2,994 240 Peshawar 983 477 240 1.988 0.109 Baluchistan 6,566 1,569 160 Quetta 565 420 160 2.625 0.144 - - - - 1.000 Table: The weighting scheme. The last column reports the corresponding pweights for the sample households. Our regression analysis is based on that weighting scheme
Conclusion and policy implications I Our results support furthermore the notion of an U-shaped pattern which characterizes the age-happiness profile Crucially and unlike advanced economies, the study finds evidence for SWB to increase in the number of children These findings may be connected: Children are seen as insurance mechanism for parents in general and particularly in their advanced stages of their life course On a related note, this paper may also add to our understanding of the relatively high birth rates and the practice of early marriages in many developing countries On a more general note, our results confirm several findings commonly established in happiness literature: SWB is higher among married couples, educated, employed, healthy and relatively richer households
Conclusion and policy implications II Similarly, our findings are in line with Easterlin (2001) who argues that life satisfaction rises with average income or living in a financially stable area within a country In contrast to evidence that exists predominantly for developed countries, our finding reveals happiness to be higher among males compared to females. However, one could expect such result for a male dominant society in Pakistan. Policymakers should hence put more emphasis on social insurance programs in fighting rising old age dependency ratios: try to give high coverage to old age and poor population in social security schemes strengthen institutional and organizational structure of pension schemes bring unorganized/informal sector under the social security or pension net
Conclusion and policy implications III expand private pension and social welfare schemes In particular, it appears that policies which foster educational attainment are key to ensure support for well-being on a national level. This may over time also lower the burden on future cohorts of young people in terms of expectations and dependency of the elders.
Thanks Thanks for listening! Any comments?
References I Angeles, L. (2009). Do children make us happier?, Discussion Paper Series University of Glasgow. Blanchflower, D. (2008). International evidence on well-being, NBER Working Paper. Clark, A. E. (2006). Born to be mild? Cohort effects don t explain why well-being is U-shaped in age, Working Paper 35 Paris-Jourdan Sciences Economiques. Clark, A. E., Frijters, P. and Schields, M. A. (2008). Relative income, happiness and utility: an explanation for the Easterlin paradox and other puzzles, Journal of Economic Literature 46(1): 95 144. Easterlin, R. (2001). Income and happiness: Towards a unified theory, Economic Journal 111: 465 84.
References II Easterlin, R. (2006). Life cycle happiness and its sources: Intersections of psychology, economics, and demography, Journal of Economic Psychology 27: 463 482. Frey, B. and Stutzer, A. (2002). Happiness and Economics, Princeton University Press. Perz, S. (2001). Household demographic factors as life cycle determinants of land use in the amazon, Population Research and Policy Review 20(3): 159 186. Shams, K. (2014). Determinants of subjective well-being and poverty in rural pakistan: A micro-level study, Social Indicators Research 119(3): 1755 1773. Shams, K. (2016). Developments in the measurement of subjective well-being and poverty: An economic perspective., Journal of Happiness Studies 17(6): 2213 2236.
References III Stutzer, A. and Frey, B. S. (2006). Does marriage make people happy or do happy people get married?, The journal of Socio-Economics 35: 326 347. Sutherland, E., Carr, D. and Curtis, S. (2004). Fertility and the environment in a natural resource dependent economy: Evidence from petn, guatemala, Poblacin y Salud en Mesoamrica 2(1): 1 14. Tella, R. D., MacCulloch, R. J. and Oswald, A. J. (2003). The macroeconomics of happiness, Review of Economics and Statistics 85: 809 827.