Effects of the Basic Pension Reform on Consumption and Labor Supply of the Elderly^†

I. Introduction

Population aging is a growing global challenge, shifting from a primarily high-income issue to a worldwide phenomenon (Wilmoth et al., 2023).1 A majority of aging populations will reside in low- and middle-income countries (WHO, 2024), where underdeveloped pension systems pose significant risks to the well-being of the elderly (Bloom et al., 2015). Consequently, many governments are increasingly introducing public pension programs to ensure income security and to address the persistent concern of elderly poverty.

South Korea faces this challenge acutely, with one of the highest elderly poverty rates among OECD countries (OECD, 2022). To address this issue, the government introduced the Basic Pension in 2008 – a non-contributory cash transfer targeting the bottom 70% of the elderly – and significantly expanded benefit levels in 2014. This reform creates a unique opportunity to examine the causal effects of expanded old-age cash transfers in a context with a limited safety net for the elderly.

Theoretical models predict that cash transfers reduce the labor supply (Moffitt, 2002; Blau and Robins, 1986; Blundell and MaCurdy, 1999), yet the empirical evidence remains mixed, with some studies finding reduced labor force participation (Fetter and Lockwood, 2018; Ranchhod, 2006; Unnikrishnan and Imai, 2020) and others finding little to no impact (Banerjee et al., 2017; Koh and Yang, 2021; Sung and Lee, 2018).2 These conflicting findings underscore the importance of evaluating how specific pension reforms, particularly those with moderate benefit levels, influence the behavior of the elderly in contexts where the impact on consumption is also uncertain.

Building on this literature, this paper examines the effects of the 2014 expansion of South Korea’s elderly cash transfer program on consumption and the labor supply among the elderly. Using data from the Korea Welfare Panel Study (KoWePS) from 2009 to 2018 and implementing a difference-in-differences design, we find evidence of a modest increase in household consumption. Turning to labor supply, we find no significant impact on the extensive margin, measured by labor force participation and employment, but a significant reduction in full-time employment and hours worked among the employed. A triple-difference strategy (DDD) reveals no heterogeneous effects by socioeconomic status among men, but shows reduced labor force participation and employment among low-educated women. These findings suggest that the expansion of pension benefits may have alleviated financial constraints for this subgroup, allowing them to reduce their labor supply.

Our paper connects to the broader literature on the effects of pensions on consumption and labor supply. The early body of work examines how pension transfers influence consumption. While some studies generally find that old-age pensions increase consumption (Fan, 2010; Chen, 2017; Unnikrishnan and Imai, 2020), magnitudes vary considerably, with more limited consumption responses documented in the Korean context (Koh and Yang, 2021; Ku et al., 2024).3 Our study builds on this literature by examining the effects of Korea’s Basic Pension benefit expansion, which increased transfers to roughly 12% of monthly household income,4 providing new evidence on how a relatively modest but targeted pension affects both consumption and labor supply among the elderly.

This study is also related to the literature examining the effects of cash transfers on the labor supply of the elderly. Studies find sizable reductions in labor force participation across several contexts – approximately 8.4 and 12.6 percentage points for men and women in South Africa (Ranchhod, 2006), 13% lower employment likelihood in India (Unnikrishnan and Imai, 2020), and an 8.5% labor supply decline in the United States (Fetter and Lockwood, 2018) – though the effects vary depending on the generosity of the program, the institutional design, and certain recipient characteristics, such as gender (Kaushal, 2014). We contribute to this strand in the literature by examining the Korean Basic Pension, known for comparatively modest transfers relative to recipients’ incomes. While existing studies consistently document large reductions in labor force participation, we find little evidence of such extensive margin effects. Instead, pension eligibility appears to reduce the labor supply along the intensive margin. This suggests that the degree of program generosity may shape not only the magnitude but also the margin along which labor supply responses occur.

There has been an increase in the number of studies focusing on the effects of the basic pension in South Korea on elderly outcomes, mostly written in Korean (Sung and Lee, 2018; Kim and Kim, 2021; Kim et al., 2022; Jin, 2023; Park et al., 2024). A recurring limitation in most of these cases, however, is that identification typically compares pension recipients to non-recipients within the elderly population. Given that receipt status is related to income and asset levels, such comparisons may reflect underlying differences in economic conditions rather than effects of the pension itself. Many studies also rely on narrow pre- and post-policy windows, limiting pre-trend testing and the detection of longer-run effects. We address these concerns by exploiting the age-based eligibility cutoff for the Basic Pension in a difference-in-differences design using data from 2009 to 2018. Relying on eligibility variation rather than actual receipt mitigates selection bias and yields cleaner causal evidence. We also go beyond extensive margin effects to examine full-time employment and hours worked. Because our estimates reflect intent-to-treat effects based on age eligibility rather than actual receipt, we further examine heterogeneous effects by education level and income group to better approximate treatment effects on actual recipients.

The remainder of this paper is organized as follows. Section 2 describes the background and outlines the details of the Basic Pension reform. Section 3 reviews the relevant literature. Section 4 presents the data sources, details the sample construction process, and defines key variables. Section 5 outlines our empirical strategy. Section 6 presents the empirical results on labor supply and consumption. Section 7 summarizes the findings and concludes the paper.

II. Background

South Korea is undergoing one of the fastest demographic transitions toward an aging society. While the OECD average elderly population ratio increased from 12.81% in 2000 to 18.63% in 2024, South Korea’s rose from 7.22% to 19.27%, surpassing the OECD average (World Bank, 2024). The proportion of people aged 65 or older is expected to reach 44% by 2050 (The Lancet Regional Health–Western Pacific, 2023). This demographic shift has been accompanied by persistently high poverty rates among the elderly, averaging 44.08% between 2011 and 2022, a rate among the highest in the OECD for most of these years (OECD, 2022).

FIGURE 1.

POPULATION AGED 65+ (2000–2024)

Note: Population ages 65 and above (% of total population).

Source: World Bank Open Data (2024).

To address these challenges, South Korea introduced the National Pension System (NPS) in 1988, which is a defined benefit system (Lee and Shin, 2024) with initial coverage for workplaces employing ten or more full-time employees (National Pension Service, 2025), requiring at least ten years of contribution (Ministry of Government Legislation, 2024). Due to the gradual roll out to cover smaller workplaces, many older adults at the time had limited opportunities to enroll. Consequently, they reached old age without pension coverage. Early enrollees also tended to have short contribution periods, resulting in inadequate benefits.

In light of the persistent poverty rate among the elderly and the limited coverage of the NPS, the government enacted the Basic Old-Age Pension Act on April 25, 2007, which came into effect on January 1, 2008. The Basic Old-Age Pension is a non-contributory scheme in which the government provides a monthly cash benefit to the bottom 70% of the elderly population. This program aims to promote income security and enhance welfare in old age. However, the Basic Old-Age Pension, which had offered a maximum monthly benefit of 84,000 KRW (US$58),5 was criticized for failing to provide adequate support for the elderly (Lee et al., 2018; Kim and Jung, 2012).

In response, the government implemented a major reform in 2014 to better support low-income older adults (Kang et al., 2022; Kim and Kim, 2021; Lee et al., 2018). The reform significantly raised the maximum monthly benefit from 99,100 KRW (US$68) to 200,000 KRW (US$138) and renamed the program the Basic Pension. As a result, the budget increased significantly from 210 billion KRW (US$145 million) allocated for the Basic Old-Age Pension in the first half of 2014 to 470 billion KRW (US$324 million) for the Basic Pension in the second half of the same year (Ministry of Health and Welfare, 2015a; 2015b).

Eligibility for both the Basic Old-Age Pension and Basic Pension was limited to individuals aged 65 or older, with recognized income levels below annually adjusted thresholds.6 Figure 2 presents the eligibility thresholds for households with one eligible elderly member and those with two eligible elderly members. The thresholds are determined each year by the Minister of Health and Welfare based on the amount of the previous year adjusted for the national consumer price inflation rate as published by Statistics Korea.

FIGURE 2.

BASIC PENSION INCOME MONTHLY THRESHOLDS (2011–2025)

Note: The figure displays the annual income threshold used to determine eligibility for the Basic Pension, set by the Ministry of Health and Welfare (MOHW). All monetary values are in Korean won (KRW). US$1 is equal to 1,449 KRW as of February 7, 2025.

Source: Data are based on annual guidelines published by the MOHW.

Table 1 presents the annual changes in the maximum pension benefit7 for single-person households, reflecting the transition from the Basic Old-Age Pension to the Basic Pension scheme. The benefit amount is adjusted annually based on the Consumer Price Index (CPI) inflation rate, and it has increased every year except in 2021. To account for differences in living costs between single-person and two-person (typically a couple) households and to avoid overlapping support, the Basic Pension benefit is reduced by 20% for each recipient when both spouses are eligible (Ministry of Health and Welfare, 2015b). As a result, households with two eligible elderly members receive a combined benefit equal to 160% of the single eligible elderly household amount. For example, in 2014, eligible individuals in single-person households received 200,000 KRW (US$138) per month, while each eligible individual in a corresponding two-person household received 160,000 KRW (US$110), due to the 20% reduction applied to each recipient, making the combined total 320,000 KRW (US$221).

TABLE 1

CHANGES IN THE MAXIMUM PENSION AMOUNTS

Note: This table reports statutory maximum monthly benefit amounts. All monetary values are in Korean won (KRW). The combined amount for couples is calculated as 160% of the single-person amount, reflecting a 20% reduction per eligible spouse. The maximum benefit of 300,000 KRW was not uniformly applied to all recipients upon its introduction. The government initially extended the 300,000 KRW benefit to recipients in the bottom 20% of the income distribution in April of 2019, expanded it to the bottom 40% in January of 2020, and applied it to all eligible recipients (bottom 70%) starting in January of 2021.

Source: Data for 2008–2020 are reproduced from Kim and Kim (2021); data for 2021–2025 are based on annual guidelines published by the Ministry of Health and Welfare (MOHW).

III. Data

We use data from the Korea Welfare Panel Study (KoWePS) to estimate the causal impact of the Basic Pension program. KoWePS is an annual longitudinal survey that has been conducted jointly by the Korea Institute for Health and Social Affairs (KIHASA) and Seoul National University annually since 2006. It collects detailed information on household characteristics, income, welfare benefits, and the economically active status of households, making it a valuable dataset for analyzing labor supply decisions and household expenditure patterns and assessing the impact of the Basic Pension. KoWePS also records whether the respondents receive the Basic Pension. Given its focus on welfare, KoWePS utilizes a stratified sampling design that oversamples low-income households. To maintain national representativeness for the elderly population, our analysis incorporates the longitudinal sampling weights provided with the dataset to adjust for this oversampling.

We use the 4th–13th waves (2009–2018) of KoWePS. We begin our sample in 2009 to avoid contamination from the introduction of the Basic Old-Age Pension in 2008. We conclude the sample in 2018 to avoid the subsequent policy shifts in 2019 and 2020, during which benefit amounts began to vary based on the recipient’s income classification. We restrict our analysis to individuals aged between 55 and 74, defining the 55-64 age group as the control group and the 65-74 group as the treatment since age 65 is the eligibility threshold for the Basic Pension.

Our main outcome variables include household expenditure, which is adjusted for inflation, current employment status, labor force participation status, self-employment status (among the employed), full-time work status (among the employed) status, and work hours (among the employed).

Table 2 presents summary statistics, separately for the treatment and control groups in the analysis sample. In terms of household expenditure, the control group spends an average of 3,956,800 KRW (US$2,730), whereas the treatment group spends 2,481,000 KRW (US$1,712). The expenditure measure includes a broader range of expenses, such as housing-related costs (including rent), tax and social security contributions, and various consumption categories (e.g., food, healthcare, education, leisure, and transportation).

TABLE 2

SUMMARY STATISTICS

Note: 1) Only wage workers were asked about their full-time status; 2) Metropolitan region refers to the Seoul Metropolitan Area (Seoul, Incheon, and Gyeonggi-do); 3) All expenditures are presented as 10,000 KRW (US$7).

The labor force participation rate declines from 64% to 37% between the control and treatment groups, with the employment likelihood displaying a similar pattern. The likelihood of being self-employed among the employed is 28% for the control group and 41% for the treatment group. This pattern is consistent with the possibility that older individuals who wish to continue working may face greater difficulty obtaining wage employment and therefore turn to self-employment. In South Korea, limited opportunities for wage employment have been identified as a factor contributing to entry into self-employment (Lee, 2020). Among wage workers, the proportion of full-time workers is 89% for the control group, whereas it is 73% for the treatment group. Work hours are higher for the control group, with weekly work hours being about 38 hours and close to 29 hours for the control and treatment groups, respectively.

We control for additional covariates, including age, gender, household size, education level, marital status, and metropolitan residence. When comparing the proportion of males between the control and treatment groups, the former shows 48% and the latter 45%, consistent with the fact that women typically outlive men. We also control for asset ownership, specifically whether the individual belongs to a household that owns a home and/or a car.

IV. Empirical Strategy

Leveraging the variation in Basic Pension eligibility, where individuals become eligible for the old-age pension program starting from the month they turn 65, we employ the following difference-in-differences design:

Here, Y_it is the outcome variable, in this case labor market outcomes and consumption, and Treat_i is an indicator variable equal to 1 if the individual is aged 65 or older. Post_t is a post-pension reform indicator that equals 1 if the year is 2015 or later. Although the Basic Pension reform was enacted in July of 2014, the KoWePS survey is administered annually during the first half of each year. Because the 2014 survey wave was fielded prior to the reform’s implementation, we classify 2014 as belonging to the pre-reform period. Our coefficient of interest is β₃, which estimates the effect of the policy change on the outcome variable. The control vector X_it consists of household size, age, educational background, marital status, metropolitan residence status, and house and car ownership status. Individual fixed effects θ_i are included to capture individual unobservables that may be correlated with pension receipt status and outcomes. λ_t, year fixed effects, are also included. We use standard errors clustered at the individual level.

A key identifying assumption when using the difference-in-differences strategy is the parallel trend assumption. This assumption posits that in the absence of any policy intervention, the trends in outcome variables for both the treatment and control groups would have evolved similarly over time. To test this assumption, we conduct an event study style analysis. These results are presented in Appendix Figure A1.8 All outcome variables satisfy the parallel trends assumption. The treatment and control groups exhibited no statistically significant pre-reform differences that could confound post-reform trends, lending credibility to our analysis.

The literature suggests that responses to cash transfers vary by socioeconomic status (Barrientos, 2003; Fetter and Lockwood, 2018; Kaushal, 2014; Koh and Yang, 2021). Given that the target of the Basic Pension was the bottom 70% of the elderly, the marginal benefits and subsequent behavioral responses may differ depending on the economic circumstances of each individual. To examine such heterogeneity, we use education level as a proxy for individuals’ economic circumstances. We estimate a triple-difference (DDD) model that takes advantage of the variation in educational attainment among the elderly. Specifically, we define individuals whose highest level of education is middle school or below as the low-education group and examine whether their labor market responses and consumption behaviors display heterogeneity. We estimate the following regression:

This model builds on the baseline difference-in-differences specification by incorporating a triple-interaction term that captures variation by education level. The key coefficient of interest, β₇, captures the differential treatment effect for individuals with lower levels of education. All regressions include individual fixed effects (θ_i), time fixed effects (λ_t), and the same set of covariates X_it included in the DD specification.

V. Analysis and Discussion

A. Consumption

We begin by examining the impact of the Basic Pension on household consumption, measured by household expenditure. Table 3 reports the estimated effects on consumption behavior. Consistent with its aim to enhance the financial stability of older adults, the pension reform led to a statistically significant increase in household expenditure levels at the 10 percent level (column 1). The full-sample estimate suggests an increase of approximately 12,087 KRW following the near doubling of the pension transfer—from 99,100 KRW (US$68) to 200,000 KRW (US$138) as a result of the reform. Despite the near doubling of pension benefits, household expenditure increased by only about 12% of the additional pension income, indicating that the consumption response was relatively limited. As prior research suggests that a recipient’s gender affects resource allocation, 9 we also examine the effect separately by gender. The coefficient estimates for males (column 2) and females (column 3) are both positive and comparable in magnitude to the full-sample estimate, though statistical significance is no longer maintained. The results remain insignificant for the log of consumption (columns 4–6). This relatively modest consumption response is consistent with the evidence in Appendix Table A1 showing that public transfers partially crowded out private ones.

TABLE 3

EFFECTS OF THE BASIC PENSION REFORM ON HOUSEHOLD EXPENDITURE

Note: Individual fixed effects and year fixed effects are included in all estimations. Standard errors are clustered at the individual level. *** p<0.01, ** p<0.05, * p<0.1.

To further investigate, we restrict the sample to individuals aged 58–71 (Appendix Table A2). In this narrower window, we find a stronger and more pronounced increase in expenditure. This suggests that our baseline full-sample results provide a conservative estimate of the reform’s impact, as the consumption response appears concentrated among those closer to the eligibility threshold.

While we treat 2014 as the pre-reform year given that the survey data were collected in the first half of the year, the announced reform may have induced anticipatory behavioral responses prior to its implementation in July of that year. Such anticipation could lead to an underestimation of the true effect. To address this concern, we re-estimate the main specification while excluding the year 2014 for robustness. These results, reported in Appendix Table A3, are consistent with the findings of the main sample.

B. Labor Supply

Next, we explore the effect of the Basic Pension reform on elderly labor market outcomes. Table 4 shows that the increase in benefits from the Basic Pension reform had no impact on overall labor force participation (column 1) or the likelihood of employment (column 2). These findings suggest that pension eligibility does not lead to substantial labor market withdrawal among the elderly at the extensive margin or create significant labor disincentives. This result is consistent with prior findings using South Korean data showing no significant impact of the Basic Pension (Jin, 2023) or the Basic Old-Age Pension (Koh and Yang, 2021) on overall employment rates.10 We also find that the modest increase in pension benefits has no effect on the likelihood of self-employment among employed individuals (column 3), where self-employment serves as a proxy for low-quality jobs compared to wage employment. On the other hand, in column 4, the coefficient on the interaction term for full-time employment (among wage workers) is negative and significant. The estimated treatment effect indicates that individuals in the treatment group are 9 percentage points less likely to work full-time, relative to the baseline mean of 0.73 for the treatment group (a decline of approximately 12 percent from the mean). This is reflected in the impact of the basic pension reducing work hours by approximately three hours per week among those who are employed in column 5 of Table 4.11

TABLE 4

EFFECTS OF THE BASIC PENSION REFORM ON LABOR SUPPLY

Note: Individual fixed effects and year fixed effects are included in all estimations. Standard errors are clustered at the individual level. *** p<0.01, ** p<0.05, * p<0.1.

We also present the results from the subgroup analysis by gender in Table 5, which are broadly consistent with the full-sample findings. The pension reform shows no significant impact on labor force participation (column 1), employment likelihood (column 2), or the likelihood of self-employment (column 3) for either males or females. However, the male and female subsamples both exhibit declines in work intensity levels. Among those affected by the reform, males are 9.1 percentage points less likely to be employed full time, while females show a larger decline of 7.6 percentage points (column 4). Lastly, the reform reduced weekly working hours among the employed by 2.96 hours for men (column 5), though the reduction of 1.1 hours for women is not statistically significant. These findings suggest that the Basic Pension reform may have contributed to reduced work intensity, potentially reflecting a reduced financial need to remain in more demanding or time-intensive jobs.

TABLE 5

EFFECTS OF THE BASIC PENSION REFORM ON LABOR SUPPLY BY GENDER

Note: Individual fixed effects and year fixed effects are included in all estimations. Standard errors are clustered at the individual level. *** p<0.01, ** p<0.05, * p<0.1.

Although increases in unearned income typically discourage the supply of labor (Imbens et al., 2001), we find limited effects, with responses primarily at the intensive margin. The modest consumption response suggests that the (perceived) net financial gains were not substantial enough to induce significant changes in the labor supply, particularly for those whose incomes were low enough to be eligible for the transfer. It should also be noted that in 2013, prior to the reform, only 36.65% of individuals aged 65–74 in the treatment group were employed. This relatively low baseline employment rate implies more limited room for labor supply adjustments compared to younger age groups with higher employment levels.

We show that our results are robust to several sensitivity checks. Given that the difference-in-differences approach relies on the control group being a valid counterfactual relative to the treated group, we restrict the sample to a narrower age range of elderly individuals aged 58-71. In this case, the control group includes individuals aged 58-64, while the treatment group comprises those aged 65-71. As shown in Appendix Table A4, the estimated effects remain consistent with our main results. Although no confounding events coincided with the pension reform, we further test the robustness of our results by estimating the models over a shorter sample period. These results are presented in Figure A2. The results remain consistent across the 2009-2018, 2010-2017, and 2011-2016 windows, suggesting that our findings are not sensitive to the choice of sample period.12 In the male and female subsample analysis, statistical significance tends to diminish with a shorter sample period, but the direction of the effects remains consistent.

C. Heterogeneity Analysis by Education Level

The Basic Pension reform was designed to target economically disadvantaged elderly individuals. It is therefore likely that the impact would vary by the socioeconomic status of the recipients as found in previous studies (Baird et al., 2018; Barrientos, 2003; Fetter and Lockwood, 2018; Kaushal, 2014). We use educational attainment as a proxy for socioeconomic status. Within the treatment group (ages 65–74) in our analysis sample, 69% of individuals have a middle school education or less (see Table 2).

On average individuals with middle school education or less, have lower household expenditure, less likely to be a homeowner, and are also less likely to own a car, suggesting that education level is a suitable proxy for socioeconomic status in Appendix Table A6. Specifically, household expenditure is substantially higher for the higher-educated group (3,344,000 KRW (US$2,307)) compared to the lower educated group (2,091,100 KRW (US$1,443)). There also seem to be differences in gender composition in that 34% of individuals with a middle school education or less are male, whereas 71% of those with more than a middle school education are male.

Table 6 presents the results from the triple-difference estimation of Equation (2). For the full sample (Panel A), the estimated effects on consumption outcomes are statistically insignificant in both levels and logarithmic terms (columns 1–2), indicating no significant heterogeneity in the consumption response by education level. In contrast, labor market outcomes display heterogeneity. The reform led to a reduction in labor force participation and employment among the low-educated relative to their higher-educated counterparts, as indicated by the statistically significant estimates in columns 3 and 4. However, no such differential impact is observed among the employed, where estimates for self-employment, full-time status, and weekly working hours remain statistically insignificant (columns 5–7).

TABLE 6

EFFECTS OF THE BASIC PENSION REFORM ON LABOR SUPPLY AND CONSUMPTION BY GENDER

Note: Individual fixed effects, year fixed effects and control variables are included in all estimations. Standard errors are clustered at the individual level. *** p<0.01, ** p<0.05, * p<0.1.

Next, we examine gender-specific effects in Panels B and C. For males (Panel B), the triple interaction coefficients show no statistically significant effects on either household consumption or labor market outcomes. While these estimates are directionally consistent with the findings for women, their lack of statistical significance makes it difficult to distinguish whether the responses among men truly do not vary by education level or if the analysis lacks the statistical power to detect such differences within the smaller male subsample. For females (Panel C), the estimates similarly indicate no differential effects by education on consumption outcomes (columns 1–2). For labor market outcomes, however, we find significant heterogeneity by education at the extensive margin. Specifically, relative to their higher-educated counterparts, low-educated women experienced a 15.9 percentage point larger decline in labor force participation and a 15.7 percentage point larger decline in employment (columns 3–4). No differential effects by education are observed for self-employment, full-time status and work hours.

D. Heterogeneity by Income Group

While educational attainment serves as a proxy for long-term socioeconomic status, it may not fully capture a household’s current financial circumstances. To more directly assess the role of economic vulnerability and potential liquidity constraints, we examine heterogeneity by income group. We define ‘low-income’ households as those with earnings below 60 percent of the median household income12 and re-estimate the main specification for these subgroups. This approach allows us to examine whether the reform’s impact is concentrated among households that face binding liquidity constraints and may thus respond differently to the increase in pension income.

Table 7 presents the results of the heterogeneity analysis by income group. In column 1, while non-low-income households exhibit a significant increase in household expenditure (Panel A), low-income households show no such response (Panel B). Regarding labor supply, neither income group exhibits significant effects at the extensive margin (columns 3–4). However, we observe distinct patterns of adjustment in work intensity. Among non-low-income households, the reform led to a significant reduction in both the likelihood of full-time employment and total weekly working hours (columns 6–7), particularly for men. While the estimates for women in the same income bracket are not statistically significant, the coefficients carry the same signs and show comparable magnitudes, suggesting a consistent directional response. In contrast, for low-income households, full-time work status remains unaffected, and the reduction in labor supply is limited to a smaller decline in working hours observed only among men (column 7). These results suggest that while a reduction in work intensity is present across the sample, the ability to transition out of full-time employment is concentrated among elderly with higher current incomes.

TABLE 7

EFFECTS OF THE BASIC PENSION REFORM ON LABOR SUPPLY AND CONSUMPTION BY GENDER

Note: Each cell represents the estimated coefficient of the interaction term (Treat × Post) from separate regressions. All models include individual fixed effects, year fixed effects, and control variables. Standard errors are clustered at the individual level. *** p<0.01, ** p<0.05, * p<0.1.

VI. Conclusion

Using a difference-in-differences approach with data from the Korea Welfare Panel Study (KoWePS), this study examines the effects of South Korea’s 2014 Basic Pension reform on consumption and labor supply among the elderly. We find that while the expansion of public transfers was partially offset by a decline in private transfers, the reform still generated a modest but statistically significant increase in household expenditure, equivalent to roughly 12 percent of the benefit hike. Turning to labor supply outcomes, we find no significant effect on labor force participation or employment. Instead, we find evidence that the reform reduced work intensity, with significant declines in full-time employment and weekly working hours among the employed.

These average effects, however, mask important heterogeneity across socioeconomic groups. Using educational attainment as a proxy, we find that the labor supply responses among men were statistically indistinguishable across education levels. In contrast, low-educated women experienced significantly larger reductions in labor force participation and employment relative to their higher-educated counterparts, suggesting that the extensive-margin response was concentrated among the most socioeconomically vulnerable women. Further heterogeneity by income group reveals important differences in how households adjusted consumption and labor supply behavior. Among non-low-income households, the reform generated a significant increase in household expenditure and reductions in full-time employment and working hours among the employed, particularly among men. Among low-income households, in contrast, we find little evidence of changes in either consumption or labor supply, aside from a reduction in working hours among men.

Taken together, these findings suggest that the Basic Pension provided meaningful income relief for some older adults with sufficient economic flexibility to reduce their labor supply, while remaining insufficient to enable a broader withdrawal from work among elderly individuals facing persistent economic hardship. More broadly, our results highlight that even a near-universal pension expansion can generate highly uneven behavioral responses across socioeconomic groups, underscoring the importance of considering heterogeneity when evaluating the welfare and labor market effects of old-age income-support programs.