Son Preference, Parental Satisfaction, and Sex Ratio Transition Junji - - PDF document

1

Son Preference, Parental Satisfaction, and Sex Ratio Transition

Junji Kageyamaa, Department of Economics, Meikai University Risa Hagiwara, Department of Economics, Meikai University Kazuma Sato, Department of Economics, Takushoku University Eriko Teramura, Department of Economics, Meikai University April, 2017

Abstract

This study aims to understand the sources of son preference using satisfaction data in various domains of life. To do this, we use Korean panel data and apply regression

• analyses. The results show that sons better satisfy parents in the domains of household

income, relations with relatives, and social relations at the timing of birth, while no advantage is found for daughters. These results are consistent with the idea that parents expect sons to contribute to the family in earning income, financially support aged parents, and represent the family in relative and social networks. Namely, we can argue that such expectation causes the parent of a newborn boy to be more satisfied in related domains of life and is manifested as a preference for sons. These results are also consistent with the idea that socioeconomic changes occurring in the demographic transition eventually lower the sex ratio even in a country with son preference. Socioeconomic changes, such as the introduction of social security system, the trend toward the nuclear family, more equal gender roles, and more working opportunity for females, cause the traditionally expected roles of sons to be less valuable and, therefore, the son preference to be weaker. This lowers the sex ratio at birth without any change in fertility.

a Corresponding author. kagejun@me.com. This research is supported by Grant-in-Aid

for Scientific Research from JPSS in Japan (263880243, 17KT0037).

2

• 1. Introduction

This study aims to understand the sources of son preference using satisfaction data in various domains of life (See Van Praag et al. 2003; Easterlin 2006 and references therein for the domain-of-life approach). In particular, we examine whether the gender of children affects life satisfaction and domain-specific satisfaction in South Korea where culturally sons are preferred to daughters (See Das Gupta et al. 2003; Guilmoto 2009, 2012; Hesketh and Xing 2006 for reviews). The sex ratio at birth is naturally around 105 (105 boys per 100 girls), but, in South Korea, it hit 116.5 in 1991 and remained high in the 1990s and in the 2000s (Statistics Korea 2017). The effect of child gender on subjective well-being has been previously studied by Lee et. al. (2013) and Margolis and Myrskyla (2016) as far as we know. Lee et. al. (2013) used survey data conducted for the elderly (65 or older) in Kangwha County in South Korea, a rural island 50km from Seoul, and demonstrated that life satisfaction is highest when the parent has both sons and daughters. Margolis and Myrskyla (2016), on the other hand, used British and German longitudinal data and showed that the gender mix

• f children has little impact on parental satisfaction in European countries. However, no

study has yet to employ panel data to examine the impact of the gender of children on parental satisfaction in a country with son preference, nor turn to domain-specific satisfaction. The reason for using domain-specific satisfaction data is that these data “reveal” parents’ feelings. For example, if having a boy, and not a girl, raises parental satisfaction in the financial domain, we can infer that parents feel positively about having a boy in the financial domain and can consider it as a source for son preference. This study is the first study to use domain-specific satisfaction to analyze the sources of son preference. By identifying sources of son preference, we can also provide evidence for explaining the sex ratio transition (Guilmoto 2009). Even in countries where sons are traditionally preferred, the sex ratio does not stay high for long and eventually falls toward the natural level (See Allendorf 2012; Angrist 2002; Bhattacharjya et al. 2008; Bongaarts 2013; Chung and Das Gupta 2007; Das Gupta et al. 2003; Diamond-Smith and Bishai 2015; Ebenstein and Leung 2010; Edlund and Lee 2013; Lin 2009). We apply the obtained results to explain this phenomenon. The reminder of the paper is organized as follows. The next section explains data and methods. Section 3 presents results. Our main findings are that sons better satisfy

3 parents in the domains of household income, relations with relatives, and social relations at the timing of birth, while no advantage is found for daughters. These results are consistent with the idea that parents expect sons to contribute to the family in earning income, financially support aged parents, and represent the family in relative and social

• networks. Section 4 concludes.
• 2. Data and Methods

The data are taken form Korean Labor and Income Panel Survey, Waves 1-17 (Korean Labor Institute 2016). The survey was first conducted in 1998 for 13,321 individuals in 5,000 households. The survey tracks not only original households but also branch households, and added 1,415 households in Wave 12. As a result, Wave 17 covers 13,163 individuals in 6,738 households. The data set contains demographic and socio-economic characteristics, including the numbers of female and male children. The data set also includes overall life satisfaction and satisfaction in domains of household income, family relations, leisure activities, housing environment, relations with relatives (from Wave 3 onwards), and social relations (from Wave 3 onwards), all of which are coded between 1 (very dissatisfied) and 5 (very satisfied). This study employs Waves 2-17 for which the necessary data are available. We focus on married individuals since it is rare to have children without getting married in South Korea. In total, our data set contains 8,821 individuals and 74,523 observations. The descriptive statistics for key variables in the most recent wave appear in Table 1. [Table 1 around here] With this data set, we regress three models to examine the following relationships: (1) the gender of existing children and the probability of progressing to the next parity, (2) children and parental satisfaction, and, most importantly, (3) the gender of children and parental satisfaction. The aims of these models are: (1) to make sure that son preference translates into actual behavior, (2) to compare the impact of having children in South Korea with other countries, and (3) to test if son preference translates into parental satisfaction. For the second model, we include both the number of children and birth dummy

4 (having a child in the last year, including adoption) as explanatory variables. We use the birth dummy to separate the impact of birth (becoming a parent of a new child) from having children (being a parent). By doing this, we can distinguish the instantaneous and continuous impact of having children. However, to test the validity of the birth dummy, we regress the model with and without the birth dummy. We use the same strategy for the third model and employ the numbers of female/male children and female/male birth dummies. In particular, we expect that female/male birth dummies reveal parents’ feelings about having a female/male child. With respect to the model specification, we employ a pooled logit model for the first model. For the second and third models, we follow previous studies that treat satisfaction levels as cardinal and that apply the ordinary least squared (OLS) method with fixed individual effects (Ferrer-i-Carbonell and Frijters 2004; de Ree and Alessie 2011; Van Landeghem 2011; Frijters and Beatton 2012; Kassenboehmer and Haisken-DeNew 2012; Wunder et al. 2013). As Ferrer-i-Carbonell and Frijters (2004) has observed, “assuming ordinality or cardinality of happiness scores makes little difference, whilst allowing for fixed-effects does change results substantially.” For all regression models, we control for age, age-squared, household-size adjusted real income, standard income for measuring relative income (the average of household-size adjusted real income in the same year, in the same sex, in the same age cohort (5-year classification), and in the same district), job, ownership of the house, living district (15 districts), and the survey year, all of which are commonly used in the subjective well-being literature.

• 3. Results

3.1. Probability of Progressing to the Next Parity For this model, we focus on parents with either only daughters or only sons. We also exclude parents at age fifty and above since the chance of having a new child becomes very low. Table 2 presents the results. Equation (1-1) includes all the parents regardless of the number of existing children, and equations (1-2) and (1-3) are respectively restricted to parents with one existing child and parents with two existing children. [Table 2 around here]

5 The results point to the existence of son preference. In comparison to only sons, the odds ratio for only daughters exceeds 1 at the 1% level of significance in equations (1-1) and (1.3) and at the 10% level in equation (1-2). These results demonstrate that the probability of progressing to the next parity is higher for parents with only daughters than for parents with only sons. We also run the same regression model restricting the sample periods to the year 2011 onwards. In 2011, the sex ratio at birth fell below 106, the first time in the data set available since 1990, and has remained at a reasonable level thereafter. Equations (1-4) to (1-6) present the results. The odd ratio for only daughters becomes insignificant in all equations. In equation (1-6), however, the significance level is slightly above the 10% level. While we cannot reject that these results are at least partially due to the reduction in the sample size, they are consistent with the national trend, suggesting that son preference has become weaker in recent years. 3.2. Children and Satisfaction Table 3 presents the results for the impact of children on parental satisfaction. The number

• f children is aggregated to four for parents with four and above children.

There are three important findings. First, the inclusion of the birth dummy does not significantly affect the estimation, pointing to the validity of the birth dummy. Second, as shown in equation (2-1), the coefficient of the number of children is negative at the 1% level when life satisfaction is regressed. This is consistent with the findings in other

• countries. Third, the birth dummy is significantly positive at the 1% level for life

satisfaction and for the domains of household income, relations with relatives, and social relations, and at the 10% level for housing environment. These results present a sharp contrast to impacts of the number of children. Namely, becoming a parent of a new child provides satisfactory feelings in various domains in life, but these feelings do not last long. [Table 3 around here] 3.3. Gender of Children and Satisfaction Table 4 presents the results for the impact of the gender of children on parental satisfaction.

6 The numbers of female/male children are aggregated to three if parents have three and more children. Again, the results do not significantly change with and without the birth dummies. [Table 4 around here] With respect to the numbers of female/male children, we observe significant differences in the results. Focusing on the equations with birth dummies, we find that the coefficients for female and male children match only in the domains of household income and leisure activities, both becoming significantly negative at the 10% level. Only the number of daughters has a significant negative impact in the housing environment domain at the 5% level. On the other hand, only the number of sons has a negative impact in the domains of life, relations with relatives, and social relations, respectively, at the 1%, 5%, and 10% levels. As for birth dummies, the coefficient for the female birth dummy becomes significantly positive only for life satisfaction within the 10% range. The coefficient for the male birth dummy also becomes significantly positive for life satisfaction, and additionally, in the domains of household income, relations with relatives, and social relations. 3.4. Sources of Son Preference We now put these results together. First, with the birth of a boy, but not a girl, parents feel more satisfied in the domains of household income, relations with relatives, and social

• relations. On the other hand, we do not observe an advantage in the birth of a girl in any
• domains. These results are consistent with the idea that parents expect sons to contribute

to the family in earning income, financially support aged parents, and represent the family in relative and social networks. Namely, we can argue that such expectation causes the parent of a newborn boy to be more satisfied in the related domains of life and is manifested as son preference. However, the positive effects of the birth of a boy do not last long. In the regression equations, the number of male children does not have similar positive impacts

• n parental satisfaction, suggesting that the positive effects of having a boy is temporary.

This could be simply due to the change in preferences, or instead, could be because of

7

• verly high parental expectations. Perhaps, parents tend to expect too much from boys and

get disappointed in the long run since such high expectations are usually not met. The same thing can be said to life satisfaction: the number of male children has a negative impact on life satisfaction while the number of female children does not. Finally, looking at the household income domain, we observe that the negative impact of the number of male children is not as large as that of the number of female

• children. The coefficient is insignificant without the birth dummies. This could be because

sons still play an important role in earning income and financially supporting their old parents.

• 4. Concluding Remarks

This study addresses the sources of son preference using domain-specific satisfaction data in South Korea. The results point to the existence of son preference in the domains of household income, relations with relatives, and social relations. To put it another way, son preference would become weaker if the advantage of having sons in these domains vanishes. This is true not only in South Korea but also in

• ther countries in which parents have similar expectations toward sons.

These results provide an explanation for the sex ratio transition. In cultures with a son preference, economic development ignites the sex ratio transition by lowering fertility, which in turn raises the sex ratio at birth. Then, the concurrently on-going socioeconomic changes, such as the introduction of social security system, the trend toward the nuclear family, more equal gender roles, and more working opportunities for females, cause the traditionally expected roles of sons to be less valuable. In particular, the introduction of social security system makes not only fertility lower (Ehrlich and Kim 2007; Kageyama and Matsuura 2016; Zhang and Zhang 2004) but also sons less valuable. As a result, son preference weakens and the sex ratio at birth falls toward the natural level without any change in fertility. The cause is the same, but the rise and the fall of the sex ratio follow different mechanisms. References Allendorf, K. (2012) “Like daughter, like son? Fertility decline and the transformation of gender systems in the family,” Demographic research, Vol.27, pp.429-454. Angrist, J. (2002) “How do sex ratios affect marriage and labor markets? Evidence from America's second generation,” Quarterly Journal of Economics, Vol.117(3),

8 pp.997-1038. Bhattacharjya, D., Sudarshan, A., Tuljapurkar, S., Shachter, R., & Feldman, M. (2008) “How can economic schemes curtail the increasing sex ratio at birth in China?” Demographic research, Vol.19(54), pp.1831-1850. Bongaarts, J. (2013) “The implementation of preferences for male offspring,” Population and Development Review, Vol.39(2), pp.185-208. Chung, W., & Gupta, M. D. (2007) “The decline of son preference in South Korea: The roles of development and public policy,” Population and Development Review, Vol.33(4), pp.757-783. Das Gupta, M., Zhenghua, J., Bohua, L., Zhenming, X., Chung, W., & Hwa-Ok, B. (2003) “Why is son preference so persistent in East and South Asia? A cross-country study of China, India and the Republic of Korea,” The Journal of Development Studies, Vol.40(2), pp.153-187. de Ree, J., & Alessie, R. (2011). “Life satisfaction and age: Dealing with underidentification in age-period-cohort models,” Social Science & Medicine, Vol.73(1), pp.177-182. Diamond-Smith, N., & Bishai, D. (2015) “Evidence of self-correction of child sex ratios in India: A district-level analysis of child sex ratios from 1981 to 2011,” Demography, Vol.52(2), pp.641-666. Ebenstein, A., & Leung, S. (2010) “Son preference and access to social insurance: evidence from China's rural pension program,” Population and Development Review, Vol.36(1), pp.47-70. Edlund, L., & Lee, C. (2013) “Son preference, sex selection and economic development: The case of South Korea,” National Bureau of Economic Research, No.w18679. Ehrlich, I., & Kim, J. (2007). “Social security and demographic trends: Theory and evidence from the international experience,” Review of Economic Dynamics, Vol.10(1), pp.55–77. Ferrer-i-Carbonell, A. & Frijters, P. (2004) “How important is methodology for the estimates of the determinants of happiness?” Economic Journal, Vol.114(497), pp.641-659. Frijters, P., & Beatton, T. (2012) “The mystery of the U-shaped relationship between happiness and age,” Journal of Economic Behavior & Organization, Vol.82(2), pp.525-542. Guilmoto, Christophe Z. (2009) "The sex ratio transition in Asia," Population and Development Review , Vol.35(3), pp.519-549. Guilmoto, Christophe Z. (2012) Sex imbalances at birth: Trends, consequences and policy implications, Thailand: UNFPA, United Nation Population Fund of Asia and the Pacific Regional Office. Guilmoto, Christophe Z., & Duthé, G. (2013) “Masculinization of births in Eastern Europe,” Population & Societies, Vol.506, pp.1-4. Hesketh, T., & Xing, Z. W. (2006) “Abnormal sex ratios in human populations: causes and consequences,” Proceedings of the National Academy of Sciences, Vol.103(36), pp.13271-13275. Kageyama, Junji, and Matsuura, Tsukasa (2016) "The Financial Burden of Having

9 Children and Fertility Differentials Across Development and Life Stages: Evidence from Satisfaction Data," Journal of Happiness Studies, pp.1-26. Kassenboehmer, S. C., & Haisken-DeNew, J. P. (2012). “Heresy or enlightenment? The well-being age U-shape effect is flat,” Economics Letters, Vol.117(1), pp.235-238. Korea Labor Institute (2016) Korean Labor and Income Panel Survey, https://www.kli.re.kr/klips_eng/index.do. Lee, M., Kim, S.H., Ohrr, H., & Park, E.C. (2013) “Life Satisfaction of Korean Elderly Parents According to Offspring Gender: A Kangwha Cohort Study,” International Journal of Translation & Community Medicine, Vol.1, pp.1-5. Lin, Tin-chi (2009) "The decline of son preference and rise of gender indifference in Taiwan since 1990," Demographic research, Vol.20, pp.377-402. Margolis, R., & Myrskyla, M. (2016) “Children’s Sex and the Happiness of Parents,” European Journal of Population, Vol.32(3), pp.403-420. Statistics Korea (2017) Vital Statistics, Data Updated

• n:

2017-01-26, https://kosis.kr/eng/. Van Landeghem, B. (2012) “A test for the convexity of human well-being over the life cycle: Longitudinal evidence from a 20-year panel,” Journal of Economic Behavior & Organization, Vol.81(2), pp.571-582. World Bank (2016) World Development Indicators, http://data.worldbank.org. Wunder, C., Wiencierz, A., Schwarze, J., & Küchenhoff, H. (2013) “Well-being over the life span: Semiparametric evidence from British and German longitudinal data,” Review of Economics and Statistics, Vol.95(1), pp.154-167. Zhang, J., & Zhang, J. (2004) “How does social security affect economic growth? Evidence from cross-country data,” Journal of Population Economics, Vol.17(3), pp.473–500.

Table 1: Descriptive statistics

Average S.D. Min Max Satisfaction Overall life 3.43 0.56 1 5 Household income 2.97 0.68 1 5 Family relations 3.67 0.55 1 5 Leisure activities 3.20 0.68 1 5 Housing environment 3.50 0.62 1 5 Relations with relatives 3.51 0.56 1 5 Social relations 3.50 0.55 1 5 # Female children 1.16 1.01 9 # Male children 1.27 0.80 9 Age 57.2 12.1 24 91 Observations: 3,663

Table 2: Odds ratio for progressing to the next parity

(1-1) (1-2) (1-3) (1-4) (1-5) (1-6) Sample Periods All Periods All Periods All Periods 2011-2014 2011-2014 2011-2014 # Children Any One Two Any One Two Only Daughters 1.413*** 1.177* 4.034*** 0.921 0.743 3.188 (Ref: Only sons) (0.108) (0.105) (0.822) (0.241) (0.254) (2.269) Age 2.421*** 3.208*** 2.526*** 3.502*** 5.780*** 1.101 (0.251) (0.397) (0.711) (1.317) (2.757) (1.129) Age-squred 0.984*** 0.981*** 0.985*** 0.979*** 0.973*** 0.994 (0.00151) (0.00181) (0.00388) (0.00529) (0.00660) (0.0141) ln(Income) 0.890* 0.813*** 0.955 1.107 0.696 2.172 (0.0540) (0.0576) (0.144) (0.348) (0.248) (1.647) ln(Standard Income) 0.533** 0.523* 0.537 0.0608* 0.0487 0.0111 (0.153) (0.180) (0.326) (0.0905) (0.0949) (0.0468) Job Dummy 0.797*** 0.725*** 0.720* 0.629* 0.774 0.286* (0.0632) (0.0664) (0.141) (0.177) (0.273) (0.217) Own House Dummy 0.951 1.084 0.858 0.923 1.478 0.282* (0.0751) (0.0991) (0.164) (0.246) (0.515) (0.184) Observations 17,749 6,380 10,336 2,416 774 1,143 Odds ratio. Robust standard errors in parentheses. Living district and survey year are also controlled. *** p<0.01, ** p<0.05, * p<0.1

Table 3: Impacts of children on satisfaction

(2-1) (2-2) (2-3) (2-4) (2-5) (2-6) (2-7) (2-8) (2-9) (2-10) (2-11) (2-12) (2-13) (2-14) Domains Overall Life Household Income Family Relations Leisure Activities Living Environment Relations with Relatives Social Relations Overall Life Household Income Family Relations Leisure Activities Living Environment Relations with Relatives Social Relations # Children

• 0.0504***
• 0.0601***

0.0237

• 0.0869***
• 0.0599***
• 0.0321*
• 0.0259
• 0.0414**
• 0.0536***

0.0267

• 0.0889***
• 0.0555***
• 0.0253
• 0.0203

(0.0165) (0.0196) (0.0169) (0.0243) (0.0212) (0.0185) (0.0192) (0.0162) (0.0191) (0.0164) (0.0239) (0.0208) (0.0181) (0.0187) Birth Dummy 0.0838*** 0.0612*** 0.0279

• 0.0187

0.0413* 0.0621*** 0.0518*** (0.0189) (0.0216) (0.0183) (0.0233) (0.0223) (0.0210) (0.0200) Age 0.0131***

• 0.0117***
• 0.0263***

0.0542*** 0.0327***

• 0.00359

0.00251 0.0108***

• 0.0134***
• 0.0271***

0.0547*** 0.0315***

• 0.00520

0.00118 (0.00372) (0.00441) (0.00363) (0.00456) (0.00434) (0.00403) (0.00390) (0.00367) (0.00434) (0.00358) (0.00449) (0.00428) (0.00399) (0.00386) Age-squred 0.000113*** 0.000273*** 0.000107*** -0.000277*** -0.000174*** 5.96e-06

• 1.89e-05

0.000131*** 0.000287*** 0.000113*** -0.000281*** -0.000165*** 1.90e-05

• 8.01e-06

(3.30e-05) (3.94e-05) (3.28e-05) (4.00e-05) (3.83e-05) (3.64e-05) (3.49e-05) (3.26e-05) (3.89e-05) (3.24e-05) (3.95e-05) (3.78e-05) (3.60e-05) (3.46e-05) ln(Income) 0.108*** 0.212*** 0.0707*** 0.0847*** 0.0650*** 0.0558*** 0.0555*** 0.108*** 0.212*** 0.0707*** 0.0846*** 0.0651*** 0.0558*** 0.0556*** (0.00458) (0.00584) (0.00445) (0.00545) (0.00508) (0.00473) (0.00461) (0.00458) (0.00584) (0.00445) (0.00545) (0.00507) (0.00473) (0.00462) ln(Standard Income)

• 0.0423***
• 0.0166
• 0.0152
• 0.0343**
• 0.0215
• 0.0278**
• 0.0195
• 0.0413***
• 0.0159
• 0.0148
• 0.0345**
• 0.0210
• 0.0269**
• 0.0188

(0.0130) (0.0150) (0.0129) (0.0151) (0.0141) (0.0136) (0.0129) (0.0130) (0.0150) (0.0130) (0.0151) (0.0141) (0.0136) (0.0129) Job Dummy 0.0248*** 0.0743***

• 0.000677
• 0.142***
• 0.0408***

0.00663 0.0128 0.0242*** 0.0738***

• 0.000901
• 0.142***
• 0.0411***

0.00618 0.0124 (0.00850) (0.0101) (0.00872) (0.0109) (0.00970) (0.00895) (0.00871) (0.00851) (0.0101) (0.00872) (0.0109) (0.00970) (0.00895) (0.00871) Own House Dummy 0.0819*** 0.0792*** 0.0446*** 0.0765*** 0.211*** 0.0469*** 0.0572*** 0.0812*** 0.0787*** 0.0443*** 0.0766*** 0.210*** 0.0464*** 0.0568*** (0.00936) (0.0103) (0.00913) (0.0118) (0.0119) (0.00971) (0.00951) (0.00936) (0.0103) (0.00912) (0.0118) (0.0119) (0.00971) (0.00951) Observations 74,523 74,523 74,523 74,523 74,523 67,532 67,532 74,523 74,523 74,523 74,523 74,523 67,532 67,532 R-squared 0.058 0.078 0.016 0.049 0.034 0.008 0.009 0.057 0.078 0.016 0.049 0.034 0.007 0.009 # ID 8,821 8,821 8,821 8,821 8,821 7,931 7,931 8,821 8,821 8,821 8,821 8,821 7,931 7,931 Robust standard errors in parentheses. Living district and survey year are also controlled. *** p<0.01, ** p<0.05, * p<0.1

Table 4: Impacts of the gender of children on satisfaction

(3-1) (3-2) (3-3) (3-4) (3-5) (3-6) (3-7) (3-8) (3-9) (3-10) (3-11) (3-12) (3-13) (3-14) Domains Overall Life Household Income Family Relations Leisure Activities Living Environment Relations with Relatives Social Relations Overall Life Household Income Family Relations Leisure Activities Living Environment Relations with Relatives Social Relations # Female Children

• 0.0327
• 0.0739**

0.0325

• 0.0884**
• 0.0797***

0.0101 0.00553

• 0.0228
• 0.0691**

0.0356

• 0.0905**
• 0.0755**

0.0121 0.00625 (0.0251) (0.0299) (0.0252) (0.0371) (0.0309) (0.0271) (0.0275) (0.0243) (0.0290) (0.0243) (0.0359) (0.0302) (0.0261) (0.0267) # Male Children

• 0.0652***
• 0.0442*

0.0197

• 0.0891***
• 0.0432
• 0.0679**
• 0.0550*
• 0.0579***
• 0.0357

0.0223

• 0.0907***
• 0.0382
• 0.0568**
• 0.0446

(0.0230) (0.0265) (0.0251) (0.0342) (0.0318) (0.0277) (0.0285) (0.0223) (0.0258) (0.0242) (0.0334) (0.0307) (0.0267) (0.0275) Female Birth Dummy 0.0896*** 0.0491 0.0283

• 0.0190

0.0397 0.0309 0.0206 (0.0271) (0.0305) (0.0257) (0.0341) (0.0317) (0.0288) (0.0280) Male Brith Dummy 0.0740*** 0.0711** 0.0251

• 0.0156

0.0452 0.0834*** 0.0752*** (0.0250) (0.0286) (0.0249) (0.0311) (0.0287) (0.0289) (0.0272) Age 0.0131***

• 0.0118***
• 0.0265***

0.0543*** 0.0328***

• 0.00373

0.00247 0.0108***

• 0.0135***
• 0.0272***

0.0548*** 0.0315***

• 0.00524

0.00120 (0.00372) (0.00441) (0.00363) (0.00456) (0.00434) (0.00403) (0.00390) (0.00367) (0.00434) (0.00358) (0.00449) (0.00428) (0.00399) (0.00386) Age-squred 0.000113*** 0.000274*** 0.000108*** -0.000278*** -0.000175*** 7.02e-06

• 1.85e-05

0.000132*** 0.000287*** 0.000114*** -0.000282*** -0.000165*** 1.93e-05

• 8.23e-06

(3.30e-05) (3.94e-05) (3.28e-05) (4.00e-05) (3.83e-05) (3.63e-05) (3.49e-05) (3.26e-05) (3.89e-05) (3.24e-05) (3.95e-05) (3.78e-05) (3.60e-05) (3.46e-05) ln(Income) 0.108*** 0.212*** 0.0707*** 0.0846*** 0.0650*** 0.0557*** 0.0555*** 0.108*** 0.212*** 0.0707*** 0.0846*** 0.0651*** 0.0557*** 0.0555*** (0.00458) (0.00584) (0.00445) (0.00545) (0.00507) (0.00473) (0.00461) (0.00458) (0.00584) (0.00445) (0.00545) (0.00507) (0.00473) (0.00462) ln(Standard Income)

• 0.0423***
• 0.0166
• 0.0151
• 0.0344**
• 0.0216
• 0.0278**
• 0.0196
• 0.0413***
• 0.0159
• 0.0148
• 0.0346**
• 0.0211
• 0.0269**
• 0.0188

(0.0130) (0.0150) (0.0129) (0.0151) (0.0141) (0.0135) (0.0129) (0.0130) (0.0150) (0.0130) (0.0151) (0.0141) (0.0136) (0.0129) Job Dummy 0.0248*** 0.0743***

• 0.000655
• 0.142***
• 0.0408***

0.00649 0.0126 0.0242*** 0.0738***

• 0.000875
• 0.142***
• 0.0411***

0.00617 0.0124 (0.00850) (0.0101) (0.00872) (0.0109) (0.00970) (0.00895) (0.00871) (0.00851) (0.0101) (0.00872) (0.0109) (0.00970) (0.00895) (0.00871) Own House Dummy 0.0819*** 0.0793*** 0.0445*** 0.0765*** 0.211*** 0.0470*** 0.0573*** 0.0812*** 0.0787*** 0.0443*** 0.0767*** 0.210*** 0.0464*** 0.0568*** (0.00936) (0.0103) (0.00913) (0.0118) (0.0119) (0.00971) (0.00951) (0.00936) (0.0103) (0.00912) (0.0118) (0.0119) (0.00970) (0.00950) Observations 74,523 74,523 74,523 74,523 74,523 67,532 67,532 74,523 74,523 74,523 74,523 74,523 67,532 67,532 R-squared 0.058 0.078 0.016 0.050 0.034 0.008 0.009 0.057 0.078 0.016 0.049 0.034 0.008 0.009 # ID 8,821 8,821 8,821 8,821 8,821 7,931 7,931 8,821 8,821 8,821 8,821 8,821 7,931 7,931 Robust standard errors in parentheses. Living district and survey year are also controlled. *** p<0.01, ** p<0.05, * p<0.1