The Financial Burden of Having Children and Fertility Differentials Across Development and Life Stages: Evidence from Satisfaction Data

Examples of socioeconomic factors that are potentially responsible for the fertility decline include mortality decline (Ehrlich and Lui 1991; Kalemli-Ozcan 2002, 2003; Nerlove 1974; Soares 2005), technological progresses (Galor and Weil 1996, 2000), the empowerment of women (de la Croix and Vander Donckt 2010; Diebolt and Perrin 2013; Eswaran 2002; Komura 2013), the establishment of social security systems (Ehrlich and Kim 2007), and a decline in relative costs of leisure goods (Galindev 2011).

For example, in the Leibenstein’s framework, in which the benefits of children are separated into consumption, labor, and old-age security utilities, the importance of labor utility and, later, security utility decreases with economic development (Leibenstein 1957). Similarly, in the value-of-children approach pioneered by Hoffman and Hoffman (1973), in which the psychological benefits of children are separated into nine categories, including economic utility, the importance of specific values differs across culture, countries, and development stages (e.g., Bulatao 1979a, b; Kağitçibaşi 1982; Kağitçibaşi and Ataca 2015; Nauck and Klaus 2007). Especially, economic utility decreases with economic development. Furthermore, Campbell and Bedford (2009) showed that in traditional societies, where psychological barriers exist, such as non-evidence-based restrictive medical rules and misinformation that prevent women from restricting the number of children, the removal of these barriers lead to a reduction in fertility. This suggests that the cost of restricting the number of children is perceived larger in traditional societies than in industrialized societies. These studies demonstrate that the perception of the role and value of children changes with economic development and affects fertility behavior.

Here, we use happiness and life satisfaction interchangeably.

See, e.g., Hamermesh (2004) for the issues using subjective data for regression analyses.

As argued in Stanca (2012), “in the absence of longitudinal data or appropriate instruments, relatively little can be done to address endogeneity.” Given this constraint, Stanca (2012) added the ideal number of children to explanatory variables as a proxy for unobserved individual characteristics. Our approach would potentially be a promising method to handle endogeneity in the SWB studies and deserves further research.

Our preliminary analyses also could not reject the endogeneity of the ideal number of children for regressing financial satisfaction.

Indeed, unobserved individual heterogeneity discussed above is one prominent example of omitted variables.

For the details of the database, see Inglehart et al. (2000).

China is omitted as its one-child policy may influence the regression results in unexpected ways.

The outlier is Nigeria.

Exceptions are found in some years in Bangladesh, Taiwan, India, Vietnam, and Turkey.

The regression model can be presented as

(1) \(SFS = \alpha CHILD + {\mathbf{x}}\mathcal{\beta } + u\)

(2) \(CHILD = \gamma IDEAL + {\mathbf{x}}\mathcal{\delta } + v\)where \({\mathbf{x}}\) is a vector of control variables including a constant, α, \(\mathcal{\beta }\), γ, and \(\mathcal{\delta }\) are corresponding coefficients, and \(u\) and \(v\) are error terms.

Excluding respondents that scored financial and overall life satisfaction equally does not change the results in any meaningful manner. Neither does only using recent data (Waves 3–5).

It is worth noting that these results do not necessarily mean that children are net financial contributors to parents in an absolute monetary term even in high TFR countries.

With respect to control variables, we obtain three interesting results. First, the coefficients of income class are always significantly positive. This is consistent with expectation. Second, the sex dummy becomes significantly negative only in low and very low TFR countries, likely pointing to a substantial financial load among mothers in these countries. Third, the coefficients of education are significantly positive only in high TFR countries, demonstrating the importance of education as an economic activity in less developed countries. As for other variables, further research is required to fully understand their impacts.

Fertility has recently increased in several countries in which fertility used to be extremely low. This could potentially be due to the decline in the financial burden of parenthood in these countries, as suggested by the present study. However, the data set in this study contains data up to the year 2008, and thus the present findings are not directly applicable for examining the recent upward trend in fertility.

Instead of ln(SFS), we can also use the difference of FS and LS as the dependent variable. However, the results do not change in any meaningful way.