A. Data

To analyze the impact of marriage and childbirth on workers’ labor market outcomes using an event-study approach, it is essential to use data that (i) are representative of the working population; (ii) contain precise information on the timing of marriage and childbirth; (iii) include key labor market outcomes such as employment status, working hours, and labor income; and (iv) encompass a sufficiently long period both before and after the event. Most previous studies of child penalties—including those conducted in Sweden (Angelov et al., 2016), Denmark (Kleven, Landais, and Søgaard, 2019), Finland (Sieppi and Pehkonen, 2019), Austria (Kleven et al., 2024), and Germany (Lim and Duletzki, 2025)—rely on large-scale administrative data to meet these requirements.

In contrast, due to limited access to administrative data in Korea, most existing studies on marriage or child penalties rely on the Korean Labor and Income Panel Study (KLIPS), conducted by the Korea Labor Institute (Kwak, 2020; Yoo and Lee, 2020; Park, 2021; Kim and Hahn, 2022).8 KLIPS is a nationally representative panel survey of household members aged 15 and older in the general population, tracking respondents annually since 1998. It provides detailed information on marriage, childbirth, and labor market outcomes over more than two decades. While the sample size of the survey is not as large as that of the administrative datasets used in studies from other countries, it remains the best available source in Korea that satisfies the key requirements for this analysis.

Following previous studies, this paper estimates marriage and child penalties in Korea using the KLIPS data covering the years 1998 to 2021. Unlike the large-scale administrative datasets used in international research, KLIPS—as a household survey—has a limited sample size. This poses challenges in constructing a fully balanced panel, as relatively few respondents are observed over a sufficiently long period both before and after marriage or childbirth. Therefore, while allowing for some imbalance in the number of observations at each event time, it is essential to construct a sample in which the estimates are not sensitive to compositional changes across the event time.

Accordingly, the sample used to analyze marriage penalties and child penalties is constructed based on the following criteria. It includes women born in 1970 or later and men born in 1969 or later who experienced their first marriage between 1999 and 2018 and have at least six years of observation, including at least one year prior to marriage. The event-time window is defined as ranging from four years before to seven years after marriage (−4≤ t ≤7 ). Individuals with missing values for any labor market outcomes or other covariates included in the model are excluded. In addition, observations in the top 1 percent of the wage distribution are trimmed to preclude outliers that could make the estimates noisy.

The sample includes married women and men, both with and without children. All mothers, fathers, and their children are identified based on the household survey section of the KLIPS, which provides information on each household member’s gender, date of birth, and relationship to the household head. Married women and men for whom children are not observed in the sample are treated as childless.

B. Descriptive Statistics

Table 1 presents summary statistics for the sample, including all observations used to estimate marriage and child penalties. The sample comprises 16,892 individual-year observations from 1,725 individuals, with women accounting for 49.4% of the sample.9 The average age of individuals in the sample is 31.3, and the average ages at first marriage and first childbirth are 29.7 and 30.8, respectively. Among those who have ever had at least one child, the average number of children observed during the analysis period in the sample is 1.6. The percentage of individual-year observations of those who never had any children during the period of analysis in the sample is 12.2%.10

TABLE 1

DESCRIPTIVE STATISTICS

Note: Standard deviations (S.D.) are reported in parentheses. KRW denotes South Korean won. The sample is drawn from the Korean Labor and Income Panel Study (KLIPS) and includes women born in 1970 or later and men born in 1969 or later, who experienced their first marriage between 1999 and 2018, and for whom at least six years of observation exist, including at least one year prior to marriage. The sample comprises 16,892 individual-year observations from 1,725 individuals. For the self-employed—including own-account workers, employers, and unpaid family workers—labor income refers to business income. Hourly wage is calculated by dividing monthly labor income by total monthly hours worked.

† Age at first childbirth and total number of children are calculated conditional on having a child during the period of analysis in the sample.

‡ Weekly hours worked and hourly wages are calculated conditional on employment.

The labor market outcomes of interest in this study include monthly labor income, labor force participation,11 working hours, and wages. Labor income and wages are adjusted using the Consumer Price Index, with 2020 as the base year. Monthly labor income is measured based on an individual’s primary job for wage workers (regular, temporary, and daily employees). Similarly, labor income refers to business income for the self-employed (own-account workers, employers, and unpaid family workers). For individuals who are not employed and therefore do not generate income, labor income is recorded as zero. As such, labor income captures labor market performance along both the intensive margin (i.e., wages and hours worked) and the extensive margin (i.e., employment status).

According to Table 1, average monthly labor income is 1.11 million KRW for women and 2.54 million KRW for men. The large gender gap in labor income primarily reflects differences in employment status. In the sample, the employment rate is 56.2% for women and 89.4% for men, underscoring the substantial labor market exit of women following marriage and childbirth: the gender disparities in labor income, participation, and employment observed before the first marriage (Table A1) are not as large as the gaps in Table 1.

In contrast, gender gaps are relatively small along the intensive margin. Men work longer hours per week (49.4 hours) than women (43.6 hours), and the average hourly wage—calculated by dividing monthly labor income by total monthly hours worked—is 11,240 KRW for women and 13,990 KRW for men. It is important to note that monthly labor income and employment status are calculated for the full sample, while working hours and hourly wages are computed only for those employed at the time of observation.

In terms of employment type, the share of permanent employment is significantly lower among women (45.9%) than men (73.5%), suggesting that women’s labor market exit, as noted earlier, is primarily associated with the loss of regular jobs. Regarding educational attainment, there is no substantial gender difference, although the proportion of college graduates is slightly lower among women (38.2%) compared to men (42.4%).

A. Family Penalties

Previous studies of child penalties focused mainly on developed countries, where high-quality panel data are available. In these contexts, marriage penalties are typically negligible: female and male labor market outcomes evolve in parallel until the arrival of the first child. In contrast, Korea (Yoo and Lee, 2020) and other developing countries (Kleven, Landais, and Leite-Mariante, 2025) exhibit diverging pre-childbirth trends in female and male labor market outcomes. Motivated by this contrast, I begin by estimating family penalties, defined to encompass both marriage and child penalties.

Figure 1 visualizes gender-specific family penalties based on Model (2): the estimated coefficients of event-time dummies from Model (1), normalized by counterfactual outcomes in the absence of marriage and childbirth. The horizontal axis indicates the event time relative to marriage, while the vertical axes in Panels A through D display monthly labor income, labor force participation rate, working hours, and hourly wages, respectively. Red triangles and blue circles represent the estimated family penalties for women and men, respectively, with corresponding 95% confidence intervals also shown. Each estimate represents the change in the labor market outcome relative to the year immediately preceding marriage (i.e., t = −1), expressed as a percentage of the counterfactual outcome.

FIGURE 1.

FAMILY PENALTIES IN KOREA

Note: The figure presents family penalties defined in Model (2), which are the estimated coefficients of event-time dummy variables () divided by the counterfactual outcomes in the absence of marriage and childbirth (E[ | t]). Red triangles and blue circles represent the family penalties for women and men, respectively, with corresponding 95% confidence intervals also displayed. The long-run family penalty is measured as the average penalty five or more years after marriage ( t ≥ 5 ). The sample consists of 16,892 individual-year observations from 1,725 individuals: women born in 1970 or later and men born in 1969 or later, who married in or after 1999. For the self-employed—including own-account workers, employers, and unpaid family workers—labor income refers to business income. Hourly wages are calculated by dividing monthly labor income by total monthly hours worked. Weekly hours worked and hourly wages are calculated conditional on employment, while monthly labor income and labor force participation are calculated unconditional on employment status.

First, Panel A shows that women’s and men’s labor incomes follow a generally parallel trend prior to their first marriage. However, beginning in the year of marriage, women’s income declines sharply, with the drop continuing for three years after their marriage. This is followed by a period of stagnation lasting at least seven years post-marriage. These results suggest that in Korea, the family penalty for women is substantial and persistent over the long term: the long-run family penalty in labor income, measured as the average penalty five or more years after marriage ( t ≥ 5), is 49.3%. In contrast, men’s labor income continues to rise steadily after marriage, indicating that men do not experience a family penalty—and may even benefit from a marriage premium (Juhn and McCue, 2017; Krapf et al., 2020; Goldin et al., 2022). This result is both qualitatively and quantitatively consistent with previous findings on the female family penalty and male premium in Korea (Yoo and Lee, 2020).12

To examine the sources of the female family penalty in labor income, Panels B, C, and D illustrate family penalties in terms of labor force participation, working hours, and hourly wages, respectively. The results indicate that the most share of the penalty stems from the extensive margin—i.e., labor force participation. The long-run penalty suggests that women’s participation rate declines by 52.5% relative to the level observed in the year immediately preceding marriage.

In contrast, the penalties associated with working hours and wages are relatively small: the long-run family penalties in terms of hours worked and hourly wages are –4.8% and 1.4%, respectively. These findings suggest that Korea’s rigid labor market offers limited flexibility for workers to adjust their working hours or wages to accommodate household responsibilities and childcare. Consequently, women who bear the primary burden of domestic and caregiving duties are unable to achieve a work-family balance and are more likely to exit the labor market entirely. As a result, the family penalty manifests more strongly along the extensive margin—that is, through reduced employment.

B. Marriage and Child Penalties

The baseline specification presented in Figure 1 shows that women’s labor market outcomes decline substantially, possibly due to the burden of household responsibilities and childcare. A natural question, then, pertains to the degree to which the overall family penalty is attributable to marriage-related factors versus child-related factors. To address this, Figures 2 and 3 present the estimated marriage and child penalties, respectively, as calculated using Models (4) and (5).

FIGURE 2.

MARRIAGE PENALTIES IN KOREA

Note: The figure presents marriage penalties defined in Model (4), which are the estimated coefficients of event-time dummy variables () divided by the counterfactual outcomes in the absence of marriage and childbirth (E[ | t] ). Red triangles and blue circles represent the marriage penalties for women and men, respectively, with corresponding 95% confidence intervals also displayed. The long-run marriage penalty is measured as the average penalty five or more years after the first marriage ( t ≥ 5 ). The sample consists of 16,892 individual-year observations from 1,725 individuals: women born in 1970 or later and men born in 1969 or later, who married in or after 1999. For the self-employed—including own-account workers, employers, and unpaid family workers—labor income refers to business income. Hourly wages are calculated by dividing monthly labor income by total monthly hours worked. Weekly hours worked and hourly wages are calculated conditional on employment, while monthly labor income and labor force participation are calculated unconditional on employment status.

FIGURE 3.

CHILD PENALTIES IN KOREA

Note: The figure presents child penalties defined in Model (5), which are the estimated coefficients of event-time dummy variables () divided by the counterfactual outcomes in the absence of marriage and childbirth (E[ | c] ). Red triangles and blue circles represent the marriage penalties for women and men, respectively, with corresponding 95% confidence intervals also displayed. The long-run child penalty is measured as the average penalty five or more years after the first childbirth ( c ≥ 5 ). The sample consists of 16,892 individual-year observations from 1,725 individuals: women born in 1970 or later and men born in 1969 or later, who married in or after 1999. For the self-employed— including own-account workers, employers, and unpaid family workers—labor income refers to business income. Hourly wages are calculated by dividing monthly labor income by total monthly hours worked. Weekly hours worked and hourly wages are calculated conditional on employment, while monthly labor income and labor force participation are calculated unconditional on employment status.

Controlling for the effect of childbirth (as in Model 3), marriage penalties capture changes in labor market outcomes attributable to marriage-related factors that are not directly linked to childbearing. These may include women’s preferences between labor supply and leisure, specialization in household work due to the domestic division of labor between spouses, and social norms or labor market discrimination against married women.

The estimates of marriage penalties reported in Figure 2 are qualitatively similar to the family penalties shown in Figure 1. As with family penalties, women experience substantial marriage penalties in terms of labor income and labor force participation. For instance, the participation rate for women declines continuously for three years after marriage and does not recover thereafter. The long-run penalty is –26.7%, indicating that women’s participation decreases by 26.7% relative to the level in the year immediately preceding marriage. In contrast, the penalties in terms of working hours and wages are minimal.

In contrast to women, no marriage penalties are observed for men. Men do not reduce their labor force participation following marriage; in fact, their working hours increase. Their wages also rise gradually, possibly due to career accumulation and promotions. These contrasting patterns in marriage penalties for women and men are consistent with a domestic division of labor that is shaped by patriarchal social norms.

With limited variation in the time interval between first marriage and first childbirth, the marriage penalty estimates in Figure 2 may capture not only the effect of marriage but also part of the ex-ante effect of childbirth. Table A2 presents the distribution of the interval between the two events. The results indicate that most of the identifying variation in marriage penalties, particularly for long-term estimates, comes from comparing childless married women with mothers. To assess the robustness of Figure 2 across subsamples, I separately re-estimate marriage penalties for married women without children and for mothers whose time gap between the two events is at least zero, one, or two years. The results, illustrated in Figure A1, show that the long-term estimates are fairly robust, whereas the short-term estimates differ somewhat between childless married women and mothers. This short-term difference may reflect selection into childbirth or anticipatory childbirth-related factors such as pregnancy (Hwang and Yoo, 2025).

The child penalties estimated in Figure 3 are also qualitatively similar to the family (or marriage) penalties. As with family penalties, (i) the child penalty in terms of labor income is substantial for women but not for men, and (ii) the majority of the female penalty arises along the extensive margin—that is, reduced labor force participation—while there is no significant penalty along the intensive margin, such as working hours or wages.

However, the dynamic patterns of child penalties are rather different from those of marriage penalties. First, the decline in women’s labor income and participation is sharp and substantial immediately following childbirth, particularly at t = 0 or 1. This pattern has also been observed in prior studies focusing on Korea (Kwak, 2020; Park, 2021; Kim and Hahn, 2022; Hwang and Yoo, 2025) and other developed countries (e.g., Kleven et al., 2019). Second, the mothers’ participation rate gradually recovers beginning two years after their first childbirth, as many women return to the labor market during this time phase. As a result, the long-run penalty on participation is 20.4% of the counterfactual outcome. The gradual recovery in the long run aligns with the well-known M-shaped pattern of female labor force participation across age in Korea (e.g., Park, 2023). Combined with the fact that mothers’ participation drops immediately after childbirth, the gradual recovery in participation suggests that a primary driver of the child penalty is the burden of childcare borne by women.

Considering that the sample is unbalanced (see Section 3), the dynamic patterns of child penalties may be influenced by sample selection across event times. For example, mothers observed for six years after childbirth may differ in terms of their characteristics from those observed for only three years or less. To assess whether such differences affect the results in Figure 3, I separately re-estimate child penalties for mothers who have been last observed for six or more years, for five years, and for four years or fewer after their first childbirth. Figure A2 shows that the childpenalty estimates are highly robust to sample selection, indicating that the observed dynamic patterns are not driven by the unbalanced panel structure.

The estimated child penalties presented in Figure 3 are significantly smaller than those reported in previous studies on child penalties in Korea. In previous studies, child penalties in Korea were estimated without accounting for the presence of marriage penalties, resulting in significant non-parallel pre-trends by gender (Park, 2021; Kim and Hahn, 2022; Hwang and Yoo, 2025). As a result, existing estimates of child penalties may have been overstated, as they partially capture the effects of marriage given the strong correlation between marriage and childbirth. In contrast, as shown in Figure 3, the estimates in this study exhibit a relatively parallel trend prior to childbirth as this study controls for marriage-related effects unrelated to childbirth. Consequently, the child penalty estimates reported here are significantly smaller than those in previous studies. For example, the long-run penalties on labor income and labor force participation implied by the estimates presented in Figure 3 are 24.1% and 26.7%,13 respectively, whereas the corresponding estimates in Kim and Hahn (2022) are 66.2% and 35.4%.

Despite such quantitative differences, the finding that child penalties in Korea arise primarily along the extensive margin is consistent with existing evidence (Park, 2021; Kim and Hahn, 2022; Hwang and Yoo, 2025). Notably, the near-zero penalty along the intensive margin contrasts with findings from developed countries, where child penalties are also significant with regard to that dimension. For example, sizable child penalties for women are observed in both working hours and wages in Denmark (Kleven, Landais, and Søgaard, 2019), Austria (Kleven et al., 2024), and Germany (Lim and Duletzki, 2025). The fact that child penalties in Korea are concentrated on the extensive margin suggests the presence of Korea-specific factors driving these outcomes, which are not relevant in Western developed countries.

A potential key factor contributing to the concentration of female penalties along the extensive margin is Korea’s working environment, characterized by long and rigid working hours. Most permanent employees work around 40 hours per week, with limited opportunities to adjust their schedules flexibly (Figure A3). As a result, if affordable childcare is not available, one spouse must exit the labor market to provide care. In the context of prevailing patriarchal social norms, this burden disproportionately falls on mothers rather than fathers. In the absence of viable options to continue their careers with moderately reduced hours or wages, mothers are often left with a binary choice: work full time or exit the labor market entirely.

Existing empirical evidence from previous research on Korea also supports the hypothesis that child penalties are larger in less family-friendly work environments. For example, Lee, Lee, and Yang (2025) find that access to grandparental childcare increases the likelihood of mothers remaining in full-time employment. This suggests that in the absence of affordable childcare, full-time work becomes less feasible for mothers. Additionally, Kim and Hahn (2022) show that working mothers tend to transition into female-dominated occupations and industries after childbirth. This finding implies that child penalties can be greater for women in less supportive or flexible work environments.

C. Evolution of Marriage and Child Penalties

In this subsection, I examine whether the marriage and child penalties documented in Figures 2 and 3 have changed over time. This analysis is motivated by two main considerations. First, it allows for an assessment of whether family policies aimed at improving the work-family balance have effectively enhanced women’s labor market outcomes and, in turn, reduced child penalties. The family policy initiative, known as The Basic Plan for a Low Fertility and Aging Society, has been implemented in five-year intervals since 2006 and has expanded for the past two decades. It includes a wide range of measures, such as housing support for newlywed couples, services and cash benefits related to childbirth and childcare, expansion of childcare facilities and subsidies, and work-family balance policies such as parental leave and reduced working hours during the childcare period. If these policies have been effective, the magnitude of marriage and child penalties should have declined.

Second, the factors potentially contributing to marriage and child penalties may have evolved over the past decades. In particular, patriarchal social norms and discrimination against married women and mothers may have shifted in a more female- and family-friendly direction. Women’s educational attainment and labor force participation have increased substantially in the 21st century. Furthermore, evidence from social surveys suggests a growing public endorsement of gender equality in the division of domestic responsibilities between spouses (Figure A4).

To examine how marriage and child penalties have changed over time, I compare marriage and child penalties for individuals who married before 2010 with those for individuals who married in 2010 or later using the specification in Model (6). Tables A3 and A4 report descriptive statistics for each subsample, which reflect broad economic and demographic changes in the labor market over the past two decades: higher ages at first marriage and first childbirth, fewer children (and more childless couples), and improvements in women’s educational attainment and labor market outcomes.

Figures 4 and 5 present the changes in marriage and child penalties, which are calculated per Models (7) and (8), respectively. The estimation results in Figure 4 show that marriage penalties have decreased for women who married in the 2010s, while the marriage premium has also declined for men who married during the same time period. The marriage penalty for women in terms of labor income decreased over time by 4.5%p. Such a decrease is driven mainly by the increase in married women’s wages and moderately by the increase in labor force participation. In terms of working hours, women’s marriage penalty did not change significantly during the 2010s.

FIGURE 4.

CHANGES IN MARRIAGE PENALTIES

Note: The figure presents changes in marriage penalties during the 2010s, as defined in Model (7). These are the estimated coefficients of interaction terms between event-time dummy variables and indicators for the women or men who married in the 2010s () divided by the counterfactual outcomes in the absence of marriage and childbirth (E[ | t] ). Orange triangles and green circles represent the change in marriage penalties for women and men, respectively, with corresponding 95% confidence intervals also displayed. The change in marriage penalty is measured as the average change after the first marriage ( 0 ≤ t ≤ 5). The sample consists of 16,892 individual-year observations from 1,725 individuals: women born in 1970 or later and men born in 1969 or later, who married in or after 1999. For the self-employed—including own-account workers, employers, and unpaid family workers—labor income refers to business income. Hourly wages are calculated by dividing monthly labor income by total monthly hours worked. Weekly hours worked and hourly wages are calculated conditional on employment, while monthly labor income and labor force participation are calculated unconditional on employment status.

Figure 5 presents changes in child penalties over time. In contrast to the overall decrease in marriage penalties, child penalties for women increase significantly: women’s child penalty on labor income increased by 7.8%p. Most of the increase seems to be driven by lower wages of women who married in the 2010s: their penalty on wages is 13.7%p larger than women who married before 2010. No significant changes are observed in women’s participation and working hours over time. Notable changes are not found for child premiums for men either, only except the slight increases in wages and incomes during the four years after the birth of the first child.

The estimation results in Figure 5 are consistent with the findings of Hwang and Yoo (2025), which compare child penalties across birth cohorts of Korean women. Using large-scale administrative data, they provide precise estimates showing that child penalties have increased for more recent cohorts in terms of labor income, but not in terms of employment, which aligns with the patterns observed in Panels A and B of Figure 5. Building on this finding, Panels C and D indicate that the increase in labor income was driven by larger penalties in wages for recent cohorts rather than by changes in working hours.

The results in Figures 4 and 5 indicate that marriage penalties have eased over time, whereas child penalties have persisted. Compared to those who married before 2010s, married women who married more recently are more likely to remain in the labor market and earn higher wages, possibly reflecting their greater educational attainment and stronger preferences for career continuity. However, despite these gains, many women who married in the 2010s eventually leave the labor market after childbirth, forfeiting wages that exceed those of married women who married before 2010. This pattern suggests that while the potential drivers of marriage penalties, such as personal preferences and social norms pertaining to married women’s work, have become more supportive of female labor supply, the underlying factors behind child penalties have remained largely unchanged. Evidence from a national social survey supports this interpretation: the share of respondents who believe that housework is primarily the role of married women declined steadily throughout the 2010s (Figure A4), whereas the decline in the percentage agreeing that women should prioritize caregiving over employment was more gradual and then plateaued in the 2020s (Figure A5).

FIGURE 5.

CHANGES IN CHILD PENALTIES

Note: The figure presents changes in child penalties during the 2010s, as defined in Model (8). These are the estimated coefficients of interaction terms between event-time dummy variables and indicators for the women or men who married in the 2010s () divided by the counterfactual outcomes in the absence of marriage and childbirth (E[ | c] ). Orange triangles and green circles represent the change in child penalties for women and men, respectively, with corresponding 95% confidence intervals also displayed. The change in child penalty is measured as the average change after the first childbirth ( 0 ≤ c ≤ 5 ). The sample consists of 16,892 individual-year observations from 1,725 individuals: women born in 1970 or later and men born in 1969 or later, who married in or after 1999. For the self-employed—including own-account workers, employers, and unpaid family workers—labor income refers to business income. Hourly wages are calculated by dividing monthly labor income by total monthly hours worked. Weekly hours worked and hourly wages are calculated conditional on employment, while monthly labor income and labor force participation are calculated unconditional on employment status.

In addition to social norms, other potential factors may explain the evolution of child penalties. For instance, Hwang and Yoo (2025) provide evidence that the increase in child penalties (measured in terms of labor income) for more recent cohorts is primarily driven by (i) increased parental leave take-up and (ii) changes in selection into motherhood. Labor income among mothers has declined (at least in the short run) as relatively more women actually utilize parental leave. Furthermore, as women with lower incomes or weaker job security levels become less likely to have children, this changing selection into motherhood mechanically increases child penalties in labor income among recent cohorts.

Despite accounting for various potential factors, the increase in child penalties suggests that family policies implemented during the 2010s were not sufficiently effective in mitigating mothers’ disadvantages in labor market outcomes. In particular, the near-zero change in child penalties along both the extensive (labor force participation) and intensive (working hours) margins of the female labor supply indicates that mothers’ labor supply has remained persistently constrained over the past two decades, with little observable impact from family policies.

D. Family Penalties across Countries

Growing evidence from developed countries indicates that child penalties are closely correlated with social norms regarding gender roles within families with children. For example, child penalties are smaller for women whose mothers participated in the labor market (Kleven, Landais, and Søgaard, 2019) but are larger for women from regions or countries with more conservative gender norms in childcare (Kleven et al., 2019; Kleven, 2025). As discussed above, marriage penalties, and thus overall family penalties, may also be strongly correlated with prevailing gender norms.

To examine whether family penalties are also correlated with social norms, I compare the estimated family penalty on labor income in Korea (Figure 1, Panel A) with those of developed countries including Denmark, Sweden, the United Kingdom, the United States, Austria, and Germany. Because marriage penalties are negligible in these countries, their family penalties correspond to child penalties. To ensure cross-country comparability, I recalculate the family penalty following Kleven et al. (2019), defining it as the ratio of the gender difference in the estimated coefficients of event-time dummy variables () to women’s counterfactual outcomes in the absence of marriage and childbirth (E[ | t] ) in Models (1) and (2).

The results in Figure 6 show that Korea is not an outlier in this relationship: women’s family penalty in labor income is larger in countries where a higher share of people agree that mothers with young children should stay at home rather than participate in the labor market

As discussed earlier, long and rigid working hours are a potential key factor amplifying child penalties, and consequently family penalties, by concentrating these penalties along the extensive margin, i.e., female labor force participation. Therefore, I also examine whether a relationship similar to that observed in Figure 6 emerges between family penalties and the prevalence of flexible working-time arrangements. The measure of flexible working-time arrangements is based on the incidence of employees (aged 15–64) who have full or partial control over their work schedules, as calculated by the OECD (2021). As illustrated in Figure 7, women’s family penalties are smaller in countries where flexible working-time arrangements are more widespread. Notably, family penalties are highly correlated with the prevalence of flexible work among not only women (Panel A) but also men (Panel B). Given that countries with well-established flexible working systems tend to exhibit high utilization rates for both women and men (Figure A3), this finding is not surprising.

FIGURE 6.

FAMILY PENALTIES AND SOCIAL NORMS ON THE ROLE OF MOTHERS ACROSS COUNTRIES

Note: The figure presents the correlation between the long-run family penalty in labor income (for women) and social norms regarding gender roles within families with children. The family penalty in labor income is calculated as the difference in the estimated coefficients of event-time dummy variables between men and women () divided by women’s counterfactual outcomes in the absence of marriage and childbirth (E[ | t] ) in Models (1) and (2). The long-run family penalty is measured as the average penalty five or more years after marriage ( t ≥ 5). See Table 1 for detailed information about the sample used to estimate the family penalty in Korea.

Source: For Korea, the family penalty estimate of Korea is from Figure 1 and the estimate of social norms is calculated using the data from the 2023 Social Survey (Statistics Korea). For the other countries, both the family penalty estimates and social norm estimates are from Figure 4 of Kleven et al. (2019), where the estimates of social norms are calculated using the International Social Survey Program (ISSP).

FIGURE 7.

FAMILY PENALTIES (FOR WOMEN) AND FLEXIBLE WORKING TIME ARRANGEMENTS ACROSS COUNTRIES

Note: The figure presents the correlation between the long-run family penalty in labor income (for women) and flexible working time arrangements for women (Panel A) and men (Panel B). The family penalty in labor income is calculated as the difference in the estimated coefficients of event-time dummy variables between men and women () divided by women’s counterfactual outcomes in the absence of marriage and childbirth (E[ | t] ) in Models (1) and (2). The long-run family penalty is measured as the average penalty five or more years after marriage ( t ≥ 5). See Table 1 for detailed information about the sample used to estimate the family penalty in Korea. Flexible working time arrangements indicate the incidence rate of employees aged 15–64 who have flexible working time arrangements. Flexible working time arrangements describe a work arrangement in which employees have full or partial control over their schedules, including cases where adjustments are allowed within certain limits (e.g., flexitime).

Source: For the family penalties, the estimate for Korea is from Figure 1, while the estimates for the other countries are from Figure 4 of Kleven et al. (2019). The source of the estimates of flexible working time arrangements is the OECD (2021), of which the calculations are based on the 4th, 5th and 6th European Working Conditions Survey (EWCS); the 1st, 3rd and 5th Korean Working Conditions Survey (KWCS); and the American Working Conditions Survey (AWCS) 2015. Korean data are from 2017, while data for other countries are from 2015.