Targeting and Mistargeting of Family Policies in High-Income Pacific Asian Societies: A Review of Financial Incentives

Abstract

Very low fertility rates can be found in many high-income Pacific Asian societies, such as Hong Kong, South Korea, Japan, Singapore, and Taiwan. Governments in these territories have already taken pronatalist policies but with only modest effects, especially when measured by overall total fertility rate. Mistargeting has been cited as a potential explanation for this impact. To explore this notion in greater depth, we first identify the potential target groups that are most influential in changing the TFR for the five societies, based on a stochastic model and fertility elasticity analyses. Then we examine the targeting of current pronatalist policies, especially financial incentives and marriage policies. The analyses show that marriage rates, especially among women aged 25–29 are the most influential factor in shaping contemporary TFRs. Third and higher order births are insignificant in changing the fertility trajectories for all the five places. Besides, there are also territory-specific patterns. For Hong Kong, Taiwan and Singapore, first births (especially among women aged 30–34) are the second most influential factor; for South Korea, second births (especially among women aged 30–34) actually play a very important role, next only to marriage; for Japan, first- and second births are much less influential while marriage is an overwhelmingly essential factor of fertility. Furthermore, the review of financial incentives in these places reveals the mismatch between the targeting suggested by our analysis and the targeting implied by current policy measures. The mistargeting, piecemeal measures and the low level of financial support may be partly responsible for the ineffectiveness of the governmental action.

Appendix 1

As shown in Fig. 1, a woman at age n may be in any of the five states: “unmarried with zero child”, “married with zero child”, “married with one child”, “married with two children”, or “married with three children”. These five states are denoted as $U\left( 0 \right)$, $M\left( 0 \right)$, $M\left( 1 \right)$, $M\left( 2 \right)$, and $M\left( 3 \right)$, respectively. The stationary probability vector of these five states at age n (i.e., the distribution of a hypothetical cohort of women at age n) is denoted by:

$$\pi_{\text{n}} = \left( {\pi_{{{\text{u}},{\text{n}}}} \left( 0 \right),\pi_{{{\text{m}},{\text{n}}}} \left( 0 \right),\pi_{{{\text{m}},{\text{n}}}} \left( 1 \right),\pi_{{{\text{m}},{\text{n}}}} \left( 2 \right),\pi_{{{\text{m}},{\text{n}}}} \left( 3 \right)} \right).$$

Initially, all women of a hypothetical cohort are in state $U\left( 0 \right)$, and so the stationary probability vector at the beginning is $\pi_{15} = \left( {1,0,0,0,0} \right)$. This means that all women are unmarried and with no children at age 15.

The matrix equation below is used to specify the dynamics in Fig. 1, as a hypothetical cohort of women move from age n to n + 1,

$$\left( {\begin{array}{*{20}c} {\pi_{{{\text{u}},{\text{n}} + 1}} \left( 0 \right)} \\ {\pi_{{{\text{m}},{\text{n}} + 1}} \left( 0 \right)} \\ {\pi_{{{\text{m}},{\text{n}} + 1}} \left( 1 \right)} \\ {\pi_{{{\text{m}},{\text{n}} + 1}} \left( 2 \right)} \\ {\pi_{{{\text{m}},{\text{n}} + 1}} \left( 3 \right)} \\ \end{array} } \right) = T_{\text{n}} \left( {\begin{array}{*{20}c} {\pi_{{{\text{u}},{\text{n}}}} \left( 0 \right)} \\ {\pi_{{{\text{m}},{\text{n}}}} \left( 0 \right)} \\ {\pi_{{{\text{m}},{\text{n}}}} \left( 1 \right)} \\ {\pi_{{{\text{m}},{\text{n}}}} \left( 2 \right)} \\ {\pi_{{{\text{m}},{\text{n}}}} \left( 3 \right)} \\ \end{array} } \right).$$

Here, $T_{\text{n}}$, the transition matrix at age n, is given by,

$$T_{\text{n}} = \left( {\begin{array}{*{20}c} {1 - m_{\text{n}} } & 0 & 0 & 0 & 0 \\ {m_{\text{n}} } & {1 - p_{{{\text{m}},{\text{n}}}} \left( 1 \right)} & 0 & 0 & 0 \\ 0 & {p_{{{\text{m}},{\text{n}}}} \left( 1 \right)} & {1 - p_{{{\text{m}},{\text{n}}}} \left( 2 \right)} & 0 & 0 \\ 0 & 0 & {p_{{{\text{m}},{\text{n}}}} \left( 2 \right)} & {1 - p_{{{\text{m}},{\text{n}}}} \left( 3 \right)} & 0 \\ 0 & 0 & 0 & {p_{{{\text{m}},{\text{n}}}} \left( 3 \right)} & 1 \\ \end{array} } \right)$$

The TFR measures the expected number of children a hypothetical cohort of women would have, if they were subject to the ASFRs of a given year through their lifetime. Here, by assuming that the fertility rate of women aged 50 and over is negligible, TFR can be computed from the following equation:

$$TFR = 0 \times \pi_{{{\text{u}},50}} \left( 0 \right) + 0 \times \pi_{{{\text{m}},50}} \left( 0 \right) + 1 \times \pi_{{{\text{m}},50}} \left( 1 \right) + 2 \times \pi_{{{\text{m}},50}} \left( 2 \right) + 3 \times \pi_{{{\text{m}},50}} \left( 3 \right)$$

$$= \left( {\begin{array}{*{20}l} 0 \hfill & 0 \hfill & 1 \hfill & 2 \hfill & 3 \hfill \\ \end{array} } \right)T_{49} T_{48} \cdots T_{15} \left( {\begin{array}{*{20}l} 1 \hfill \\ 0 \hfill \\ 0 \hfill \\ 0 \hfill \\ 0 \hfill \\ \end{array} } \right)$$

To estimate the TFRs, the ASMRs and APSFRs were used as realistic values for the 28 parameters (i.e., $m_{\text{n}}$, $p_{{{\text{m}},{\text{n}}}} (1$), $p_{{{\text{m}},{\text{n}}}} \left( 2 \right)$, and $p_{{{\text{m}},{\text{n}}}} \left( 3 \right)$ for 75-year age groups, covering ages from 15 to 49). Since we only modeled transitions up to parity 3, to reduce potential underestimation, we used births of parity 3 and higher parities as the numerator while the number of married women with 2 children as the denominator to compute the ASPFR for parity 3.