## Can Healthy Aging Boost Labor Supply? Evidence from Korea

## Source details

**Canonical URL:** [Can Healthy Aging Boost Labor Supply? Evidence from Korea](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025260-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025260-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025260-source-pdf.pdf.json)

---

### Introduction and motivation
- Population aging driven by falling fertility rates and rising longevity; Asia aging rapidly (Figure 1).
- Japan has one of the highest shares of old-age population (above age 65); Korea and China projected among the fastest aging populations (UNWPP).
- Traditional macro view: aging ⇒ shrinking labor force, rising fiscal spending on pensions and healthcare, reduced labor income tax revenue; this view abstracts from health improvements as a driver of rising longevity.
- Global life expectancy increased by around 4.5 years over the last two decades; the additional years have been largely free from chronic diseases.
- In Asia, several countries have seen life expectancy and healthy life expectancy (HALE) rising faster than the world average (Figure 2).

### Research questions and contributions
- Core questions:
  - Whether individuals from more recent cohorts in Korea have been aging in better health compared with earlier birth cohorts.
  - Whether healthy aging has raised individuals’ ability and willingness to remain attached to the labor market.
- Focus: Korea (acute demographic change, rich longitudinal microdata); comparators: China, India, Japan, Malaysia, Thailand.
- Two main contributions:
  - Empirical evidence on healthy aging trends across multiple health measures for Korea and selected Asian peers.
  - Identification of the causal impact of health on labor market participation using a novel instrument—incidence of chronic diseases—as a proxy for exogenous health shocks.

### Data, sample coverage, and health measures
- Time period: Korea 2006-20; up to 2022 for other countries.
- Countries: Korea, China, India, Japan, Malaysia, Thailand (Figure 3).
- Around 200,000 observations for individuals aged 50-90; ~one-third from Korea.
- Data sources: CHARLS, HART, JSTAR, KLoSA, LASI, MARS and authors’ calculations; pre-harmonized via Gateway for Global Aging except Thailand (harmonized by authors).
- Health metrics:
  - Measured: maximum grip strength (kilograms).
  - Self-reported: overall health status (1-5 scale); ADLs; IADLs.
  - Constructed: first principal component of self-reported measures.
- Standardization: all health indicators standardized to mean zero and standard deviation one; higher values = better health.
- Data caveats: Grip strength missing for Japan and Thailand; IADLs (and principal component) missing for Malaysia and Thailand (Appendix Table 1).
- Additional variables: self-reported health behaviors and incidence of 17 chronic diseases; instrument = number of chronic diseases reported, expressed as a share of total listed.

### Empirical strategy and identification
- Healthy-aging trends: pooled OLS controlling for age and socioeconomic characteristics; Mundlak (1978) regressions (modified random effects) exploiting longitudinal data.
- Health → labor supply:
  - Correlational: OLS regressions of labor-market indicators on health indicators, controls Z_i,t (age, gender, education, log household wealth), and time trend or year fixed effects.
  - Causal: 2SLS instrumenting health with incidence of chronic diseases (assumed partially exogenous); controls include smoking, poor nutrition, physical inactivity, excessive alcohol use.
- Robustness: variants of the chronic disease instrument; controlling for year fixed effects.

### Main findings — healthy aging trends (Korea and peers)
- Broad-based healthy aging in Korea: steady improvements across successive birth cohorts.
- Preserved example statement: "On average, the grip strength of a 70-year-old individual in 2022 was approximately equivalent to that of a 60-year-old in 2006."
- Healthy aging evident across measured and self-reported indicators.
- Asia-5 (Excl. Korea): self-reported indicators show significant improvements (principal component, overall health, IADLs); no significant evidence for grip strength on average.
- Age-equivalent gains (Korea):
  - ADLs and IADLs: age-equivalent improvements of 8–13 years.
  - Self-reported overall health status: a 70-year-old in 2022 comparable to someone aged 64-68 in 2006.
  - Overall characterization: “the 70s are the new 60s” in Korea, with roughly a decade of improvement in age-equivalent health over the 17-year sample period.

### Labor force status patterns
- Two inactivity measures: participation in labor market; retired status.
- Korea: transition from activity into retirement as age increases (notably from 60 onward); small fraction move into non-retirement inactivity.
- Japan: more shift into non-retirement inactivity, especially before 65.
- Gender gaps: pronounced in Korea, even at ages 50-59, contributing to lower aggregate participation.

### Effect of health on labor market outcomes — main OLS and 2SLS estimates (Korea)
- OLS (coefficients on standardized health measures; dependent variables = Labor force participation (dummy) and Retired (dummy)):
  - Labor force participation:
    - Grip strength 0.056*** (0.007)
    - Self-reported PC 0.080*** (0.007)
    - Overall health status 0.072*** (0.006)
    - ADLs 0.059*** (0.005)
    - IADLs 0.073*** (0.006)
  - Retired:
    - Grip strength -0.042*** (0.006)
    - Self-reported PC -0.068*** (0.007)
    - Overall health status -0.065*** (0.005)
    - ADLs -0.048*** (0.006)
    - IADLs -0.062*** (0.007)
- 2SLS (IV) second-stage estimates (Korea):
  - Labor force participation:
    - Grip strength 1.013*** (0.144)
    - Self-reported PC 0.287*** (0.037)
    - Overall health status 0.196*** (0.020)
    - ADLs 0.493*** (0.091)
    - IADLs 0.559*** (0.095)
  - Retired:
    - Grip strength -0.938*** (0.124)
    - Self-reported PC -0.266*** (0.033)
    - Overall health status -0.181*** (0.021)
    - ADLs -0.456*** (0.077)
    - IADLs -0.518*** (0.083)
- IV first-stage coefficients (chronic disease → health measures):
  - Grip strength -0.326*** (0.050)
  - Self-reported PC -1.334*** (0.094)
  - Overall health status -2.021*** (0.052)
  - ADLs -0.751*** (0.108)
  - IADLs -0.666*** (0.089)
- Weak IV F-statistics listed: 4320, 2149, 6496, 485, 6 (Table 1); first-stage F-statistics exceed Stock and Yogo (2005) critical value of 10 for primary IVs.

### Effects of a 1-decade improvement in health (rescaled 2SLS; Korea)
- Labor force participation (1-decade improvement):
  - Grip strength 0.191*** (0.027)
  - Self-reported PC 0.025*** (0.003)
  - Overall health status 0.007*** (0.001)
  - ADLs 0.036*** (0.007)
  - IADLs 0.048*** (0.008)
- Retirement (1-decade improvement):
  - Grip strength -0.177*** (0.027)
  - Self-reported PC -0.024*** (0.003)
  - Overall health status -0.006*** (0.001)
  - ADLs -0.033*** (0.007)
  - IADLs -0.045*** (0.008)
- Text interpretation: "a 10-year gain in grip strength increases the probability of labor force participation by about 19 percentage points and decreases the probability of retirement by about 18 percentage points." Improvements in self-reported health over one decade increase labor force participation by about 2.5 percentage points.

### Heterogeneity of causal effects (2SLS — labor force participation)
- By age (Grip strength; Self-reported PC; Overall health; ADLs; IADLs):
  - Ages 50-59: 1.172***; 0.391***; 0.194***; 1.082***; 1.110***
  - Ages 60-69: 0.892***; 0.348***; 0.210***; 0.675***; 0.751***
  - Ages 70-79: 0.891***; 0.221***; 0.205***; 0.307***; 0.346***
  - Ages 80-89: 0.517***; 0.115***; 0.183***; 0.125***; 0.160***
- By gender:
  - Female: 0.584***; 0.248***; 0.155***; 0.449***; 0.529***
  - Male: 1.301***; 0.348***; 0.263***; 0.561***; 0.623***
- By education:
  - < Upper secondary: 0.926***; 0.301***; 0.218***; 0.503***; 0.540***
  - Upper secondary: 1.580***; 0.340***; 0.215***; 0.623***; 0.737***
  - Tertiary: 0.210; 0.107; 0.072; 0.164; 0.254
- By wealth:
  - Quintile 1 (lowest): 2.190**; 0.367***; 0.295***; 0.602***; 0.610***
  - Quintile 5 (highest): 0.800**; 0.219***; 0.122***; 0.437**; 0.600**
- Summary heterogeneity: larger effects for younger older groups (50s and 60s); effects for men almost twice as large as for women; stronger effects for lower-education and lower-wealth individuals.

### Robustness checks and alternative instruments
- Results robust to:
  - Restricting instrument to subset of chronic diseases plausibly exogenous to work (cataracts, cancer, stroke).
  - Weighting diseases by severity (weights from regression of PC on diseases).
  - Using a dummy for top 10 chronic diseases (Parkinson’s, Alzheimer’s, psychological disorders, stroke, kidney disease, lung disease, osteoporosis, arthritis, asthma, urinary incontinence).
  - Instrument definitions: total number of chronic diseases; dummy for any of the 17 chronic diseases.
  - Controlling for year fixed effects.
- Alternate IV specifications (examples reported in Appendix):
  - IV: chronic diseases unrelated to work — Labor force participation coefficients (Grip strength 0.738***; Self-reported PC 0.282***; Overall health 0.290***; ADLs 0.367***; IADLs 0.451***).
  - IV: severity-weighted chronic diseases — Grip strength 1.182***; Self-reported PC 0.250***; Overall health 0.222***; ADLs 0.401***; IADLs 0.344***.
  - IV: top-10-disease dummy — Grip strength 0.806***; Self-reported PC 0.231***; Overall health 0.190***; ADLs 0.345***; IADLs 0.406***.
- Instrument strength examples (Weak IV F-statistics):
  - Chronic diseases unrelated to work: 55.27; 283.7; 411.9; 109.3; 136.2
  - Severity-weighted: 59.23; 181.1; 1462; 59.14; 69.86
  - Top-10 dummy: 34.97; 121.7; 528.6; 47.02; 47.35
- Unemployment: instrumented health has no significant impact on probability of being unemployed conditional on participation (IV unemployment coefficients not significant as presented).

### Comparative evidence (Asia-5 excluding Korea)
- OLS: grip strength and self-rated overall health not statistically significant in comparator group.
- 2SLS (instrument: incidence of chronic disease) finds significant effects for all health indicators across China, India, Japan, Malaysia, Thailand.
- IV second-stage coefficients, Asia-5 (Excl. Korea):
  - Labor force participation:
    - Grip strength 0.260* (0.151)
    - Self-reported PC 0.101** (0.041)
    - Overall health 0.108*** (0.037)
    - ADLs 0.142** (0.059)
    - IADLs 0.149** (0.064)
  - Retired:
    - Grip strength -0.237** (0.105)
    - Self-reported PC -0.105*** (0.032)
    - Overall health -0.101*** (0.028)
    - ADLs -0.140*** (0.045)
    - IADLs -0.165*** (0.047)
- Magnitude: causal effects are substantially larger in Korea than in comparator countries (nearly twice as large for most indicators), except self-rated overall health where effects are similar.

### Aggregate contribution to Korea’s labor force participation (2006–20)
- Healthy aging measured by grip strength more than accounted for the total rise in elderly labor force participation (Figure 10).
- Healthy aging measured by self-reported PC contributed about 40 percent of the total rise in older workers’ labor force participation.
- Over 2006-20, healthy aging implied an increase in the labor supply of older individuals in Korea by around 24 percentage points (text reference).

### Mechanisms, interpretation, and limitations
- IV > OLS magnitudes indicate endogeneity bias in OLS and support a causal interpretation: better health ⇒ higher labor supply and lower retirement probability.
- Objective health measures (grip strength) show particularly strong effects in Korea.
- Diminishing effects at older ages (weaker for 70s and 80s) suggest constraints from skill obsolescence, pension incentives, and age discrimination.
- Positive health effects reflect stronger voluntary attachment to the labor market by healthier older individuals rather than primarily country-level policy or firm actions.

### Policy implications and recommendations
- Health promotion and prevention policies that reduce incidence of chronic disease can expand effective labor supply among older individuals. Potential measures include:
  - Expanding preventative healthcare (e.g., immunization and health screenings).
  - Expanding mental health services.
  - Targeting healthy behaviors: substance-abuse prevention, tobacco and junk food taxes, smoke-free regulations, promoting physical exercise.
- Pair health policies with labor-market measures to extend working lives:
  - Incentives to postpone retirement, phased retirement, reduced early-retirement benefits.
  - Lifelong reskilling and training.
  - Age-friendly workplaces and flexible work arrangements.
  - Combating age discrimination.
- Consider demand-side reforms to address seniority-based wage and promotion systems that create incentives for early retirement and discourage hiring older workers.

### Data availability and appendix notes
- Survey coverage (Appendix Table 1):
  - CHARLS (China): 4 waves (2011, 2013, 2015, 2018) — Grip strength, Overall health, ADLs, IADLs.
  - LASI (India): 1 wave (2017-19, 2021) — Grip strength, Overall health, ADLs, IADLs.
  - JSTAR (Japan): 3 waves (2007, 2009, 2011) — Overall health, ADLs, IADLs.
  - KLoSA (Korea): 8 waves (2006, 2008, 2010, 2012, 2014, 2016, 2018, 2020) — Grip strength, Overall health, ADLs, IADLs.
  - MARS (Malaysia): 1 wave (2018-19) — Grip strength, Overall health, ADLs.
  - HART (Thailand): 4 waves (2015, 2017, 2020, 2022) — Overall health, ADLs.
- OLS and Mundlak regression diagnostics and exact year-of-birth coefficients, R-squared, sample sizes, and a detailed set of robustness and heterogeneity tables reported in Appendix Table 1 and associated appendix tables and figures.

*Source: IMF Working Paper — Can Healthy Aging Boost Labor Supply? Evidence from Korea (content and figures/tables referenced within the document).*

### References .............................................................................................................

### Can Healthy Aging Boost Labor Supply? Evidence from Korea

### Introduction and Motivation
- Population aging is driven by falling fertility rates and rising longevity and is occurring rapidly in Asia (Figure 1).
- Japan has currently one of the highest shares of old-age population (above age 65); Korea and China are projected to be among the fastest aging populations in coming decades according to the United Nations World Population Projections (UNWPP).
- Traditional macro view: aging implies a shrinking labor force, rising fiscal spending on pensions and healthcare, and reduced labor income tax revenue; this view abstracts from health improvements as a driver of rising longevity.
- Global life expectancy has increased by around 4.5 years over the last two decades, and the additional years have been largely free from chronic diseases.
- In Asia, several countries have seen life expectancy and healthy life expectancy (HALE) rising faster than the world average (Figure 2).

### Research Questions and Contributions
- The paper studies:
  - Whether individuals from more recent cohorts in Korea have been aging in better health compared with earlier birth cohorts.
  - Whether healthy aging has raised individuals’ ability and willingness to remain attached to the labor market.
- Korea is the main focus because of its acute demographic change (notably low fertility and steep rise in life expectancy over the last two decades) and rich longitudinal microdata for older individuals.
- Comparator countries: China, India, Japan, Malaysia, and Thailand.
- Two main contributions:
  - Empirical evidence on healthy aging trends across multiple health measures for Korea and selected Asian peers.
  - Identification of the causal impact of health on labor market participation using a novel instrument—incidence of chronic diseases—as a proxy for exogenous health shocks.

### Data and Sample Coverage
- Time period for Korea: 2006-20; full sample includes up to 2022 for other countries.
- Countries included: Korea, China, India, Japan, Malaysia, Thailand (Figure 3).
- Around 200,000 observations for individuals aged 50-90 from household surveys focused on older populations, with about one-third of the sample drawn from Korea.
- Data sources: CHARLS, HART, JSTAR, KLoSA, LASI, MARS and authors’ calculations; data pre-harmonized via the Gateway for Global Aging except Thailand (harmonized by authors).
- Longitudinal tracking possible in all countries except India and Malaysia (data exist for only one wave).

### Health Indicators and Measurement
- Health metrics used:
  - Measured: maximum grip strength (kilograms).
  - Self-reported: overall health status (1-5 scale); ease of performing activities of daily living (ADLs); ease of performing instrumental activities of daily living (IADLs).
  - Constructed: first principal component of self-reported measures.
- All health indicators standardized to mean zero and standard deviation one; higher values indicate better health.
- Observed patterns:
  - Health outcomes tend to deteriorate with age (Figure 4).
  - Cross-country differences: average grip strength substantially higher in China than in India and Malaysia; average self-rated health higher in Thailand, Japan and Malaysia compared with China and Korea.
  - Cross-country variation is smaller for ADLs and IADLs.
- Data availability caveats:
  - Grip strength missing for Japan and Thailand.
  - IADLs (and the principal component) missing for Malaysia and Thailand (Appendix Table 1).
- Additional data:
  - Self-reported health behaviors and incidence of 17 chronic diseases; the instrument for exogenous health shocks is constructed as the number of chronic diseases an individual reports, expressed as a share of the total number listed in the survey.

### Empirical Strategy and Identification
- Estimation approaches for healthy aging trends:
  - Pooled OLS regression controlling for age and socioeconomic characteristics.
  - Mundlak (1978) regressions (modified random effects) exploiting longitudinal data.
- Health-to-labor-supply analysis:
  - Correlational analysis via OLS.
  - Causal identification via two-stage-least-squares (2SLS), instrumenting health with incidence of chronic diseases (assumed in part to be exogenous to lifestyle and socioeconomic determinants).
  - Controls include smoking, poor nutrition, physical inactivity, and excessive alcohol use.
- Robustness checks:
  - Different variants of the chronic disease instrument.
  - Controlling for year fixed effects to capture changes in labor markets and retirement policies.

### Main Findings: Healthy Aging Trends
- Broad-based healthy aging in Korea: steady improvements in health indicators across successive birth cohorts.
- Example quantitative statement preserved from source:
  - On average, the grip strength of a 70-year-old individual in 2022 was approximately equivalent to that of a 60-year-old in 2006.
- Healthy aging is evident across multiple physical health indicators (measured and self-reported).
- Comparable healthy aging trends found in the sample of Asian peer economies, although results for Korea are generally somewhat stronger.

### Main Findings: Effects of Health on Labor Supply
- Better old-age health is associated with:
  - Increased likelihood of participating in the labor force.
  - Postponed retirement.
- 2SLS estimates using chronic diseases as an instrument indicate exogenous improvements in health lead to increases in labor supply.
- The estimated impact from the IV approach is substantially larger than implied by OLS estimates, highlighting endogeneity concerns.
- Results are robust to alternative instruments and to controlling for year fixed effects.

### Policy-Relevant Implications (implied by findings)
- Healthy aging can partly mitigate adverse economic effects of population aging by increasing older individuals’ labor market attachment.
- Policies that promote healthy lifestyles and reduce chronic disease incidence may yield labor supply gains among older populations.
- Given the stronger results in Korea, country-specific health and labor policies can matter for harnessing the labor supply benefits of healthy aging.

### Relation to Literature and Ongoing Work
- Connects to literature on healthy aging across cohorts (Abeliansky and Strulik 2019; Abeliansky, Erel, and Strulik 2020; Old and Scott 2023) and on health effects on labor force and growth (Kotschy and Bloom, 2023).
- Distinguishes itself by using multiple health measures for Korea and Asian peers and by using chronic disease incidence as an instrument for health.
- Related concurrent work expands the sample to 41 economies and a broader set of health and labor outcomes (Gruss et al., 2025; IMF 2025).

*Source: IMF Working Paper — Can Healthy Aging Boost Labor Supply? Evidence from Korea (content and figures/tables referenced within the document).*

### 2.3 Labor force status

### 2.3 Labor force status

### Labor force status patterns and differences
- Two distinct labor market inactivity measures are analyzed: (i) whether the individual participates in the labor market, and (ii) whether the individual is retired.  
- Korea: tendency for individuals to transition from activity into retirement as they age, notably from 60 years old onward, with only a small fraction moving into non-retirement inactivity.  
- Japan: individuals are more likely to shift into non-retirement inactivity, especially before 65.  
- Gender gaps in labor force participation are particularly pronounced in Korea, even at relatively younger ages (50-59), contributing to a lower aggregate labor force participation rate.

### Empirical approach to link health and labor supply
- Two-step approach to study health → labor market outcomes among older individuals:
  - Correlations: OLS regressions of each labor market indicator (LMI_i,t) on each health indicator (H_i,t), controlling for a vector Z_i,t of socioeconomic characteristics (age, gender, education, log household wealth) and a time trend (or time fixed effects as robustness):  
    LMI_i,t = β0 + β1 H_i,t + θ Z_i,t + β2 t + ε_i,t
  - Causal effects: 2SLS using exogenous health shocks proxied by the incidence of chronic diseases as instruments for health. Lifestyle risk factors (smoking, poor nutrition, physical inactivity, excessive alcohol use) are controlled for in regressions.
- Health indicator cohort-trend analysis: regress health H_i,t on year of birth (YOB_i) controlling for X_i,t (age, gender, education, log household wealth). Standard errors clustered at year-of-birth level. Pooled OLS and Mundlak (1978) regressions are estimated; pooled sample of Asian peers additionally controls for country fixed effects.
- Mundlak regression specification: augments random effects with within-individual averages of time-varying explanatory variables (X̄_i) to mitigate bias from individual-specific effects correlated with regressors:
  H_i,t = β0 + β1 YOB_i + δ X_i,t + X̄_i + u_i + e_i,t

### Evidence of healthy aging in Korea vs. peers
- Broad finding: later-born cohorts in Korea exhibit better physical health than earlier-born cohorts after controlling for age and socioeconomic characteristics.
- OLS estimates for Korea show significant improvements across all health indicators considered; the most pronounced gains are observed for grip strength (a measured metric). Positive and statistically significant trends are also found for:
  - principal component summary of self-reported health indicators,
  - overall health status,
  - ADLs,
  - IADLs.
- Mundlak estimates are qualitatively similar and suggest somewhat stronger healthy aging trends compared to OLS.
- Pooled sample of five Asian peer countries (China, India, Japan, Malaysia, Thailand), excluding Korea:
  - No significant evidence of healthy aging for the grip strength metric on average.
  - Self-reported indicators show significant improvements over time for the principal component, overall health status, and IADLs.
  - Note: limited time coverage of health data in peer countries (often spanning only a few survey waves) may lead to imprecise trend estimates.

### Magnitude of cohort health gains (age-equivalent effects)
- Age-equivalent gains are calculated using the coefficient on year of birth multiplied by the number of years in the sample (17 years) and divided by (coefficient on age minus coefficient on year of birth).  
- Key age-equivalent findings for Korea:
  - By 2022, the average 70-year-old in Korea had grip strength comparable to that of a 60-year-old in 2006.
  - Self-reported health for a 70-year-old in 2022 was equivalent to that of someone aged 59-64 in 2006—depending on regression specification.
  - Among subcomponents, ADLs and IADLs showed the largest gains, with age-equivalent improvements of 8–13 years.
  - The smallest gains are for self-reported overall health status, where a 70-year-old in 2022 is comparable to someone aged 64-68 in 2006.
  - Overall characterization: “the 70s are the new 60s” in Korea, with roughly a decade of improvement in age-equivalent health over the sample period (17 years).

### Socioeconomic determinants and heterogeneity
- Cross-sectional determinants observed in regressions:
  - Health deteriorates with age.
  - Health is generally better among individuals with higher education and greater household wealth.
  - Gender differences: men tend to have higher grip strength and overall self-rated health; women in Korea report better outcomes in ADLs (contrasting with the regional average where men tend to fare better).
- Heterogeneity of cohort trends in Korea (interaction results in Appendix Table 6):
  - Within-country health disparities have persisted.
  - Health gaps have persisted across household wealth quintiles (given equal trends).
  - Health gaps have widened by gender (stronger healthy aging trends for males relative to females).
  - Health gaps have narrowed across education groups (stronger healthy aging trends for individuals with lower levels of education).

### Cross-country health heterogeneity
- Even after controlling for age, gender, education, and wealth, significant cross-country variation in health remains:
  - Self-rated overall health status is notably higher in Thailand, Japan, and Malaysia (relative to China benchmark).
  - Grip strength is notably lower in India and Malaysia.
  - Cross-country differences in ADLs and IADLs are smaller but meaningful; most countries score slightly higher than China—except India.
- Results are robust across OLS and Mundlak specifications.

*Source: IMF Working Paper chapter section "2.3 Labor force status" (figures and regression results as reported in the source).*

### 5. Effect of health on labor market outcomes

### 5. Effect of health on labor market outcomes

### Data and identification
- Data source: KLoSA (Korean Longitudinal Study of Aging) and authors’ calculations.
- Health measures considered: grip strength, self-reported principal component (PC), overall health status, ADLs, IADLs.
- Controls: age, gender, education, household wealth, lifestyle factors (ever smoked, over/underweight, infrequent moderate physical activity), and survey year.
- Identification strategy: OLS for correlations; 2SLS with incidence of chronic diseases as instrument to estimate causal effects. First-stage F-statistics exceed Stock and Yogo (2005) critical value of 10, ruling out weak instruments.

### Main empirical findings (OLS vs. 2SLS, Korea)
- OLS estimates (coefficients on health measures; dependent variables: Labor force participation (dummy) and Retired (dummy)):
  - Labor force participation (dummy): Grip strength 0.056*** (0.007); Self-reported PC 0.080*** (0.007); Overall health status 0.072*** (0.006); ADLs 0.059*** (0.005); IADLs 0.073*** (0.006).
  - Retired (dummy): Grip strength -0.042*** (0.006); Self-reported PC -0.068*** (0.007); Overall health status -0.065*** (0.005); ADLs -0.048*** (0.006); IADLs -0.062*** (0.007).
- 2SLS (IV) second-stage estimates (coefficients on health measures; Korea):
  - Labor force participation (dummy): Grip strength 1.013*** (0.144); Self-reported PC 0.287*** (0.037); Overall health status 0.196*** (0.020); ADLs 0.493*** (0.091); IADLs 0.559*** (0.095).
  - Retired (dummy): Grip strength -0.938*** (0.124); Self-reported PC -0.266*** (0.033); Overall health status -0.181*** (0.021); ADLs -0.456*** (0.077); IADLs -0.518*** (0.083).
- IV first-stage coefficients (chronic disease → health measures):
  - Grip strength -0.326*** (0.050)
  - Self-reported PC -1.334*** (0.094)
  - Overall health status -2.021*** (0.052)
  - ADLs -0.751*** (0.108)
  - IADLs -0.666*** (0.089)
- Weak IV F-statistics (first-stage): 4320, 2149, 6496, 485, 6 (as listed in Table 1).

### Effects of a 1-decade improvement in health (rescaled 2SLS estimates; Korea)
- Effect on labor force participation (point estimates, standard errors in parentheses):
  - Grip strength 0.191*** (0.027)
  - Self-reported PC 0.025*** (0.003)
  - Overall health status 0.007*** (0.001)
  - ADLs 0.036*** (0.007)
  - IADLs 0.048*** (0.008)
- Effect on retirement:
  - Grip strength -0.177*** (0.027)
  - Self-reported PC -0.024*** (0.003)
  - Overall health status -0.006*** (0.001)
  - ADLs -0.033*** (0.007)
  - IADLs -0.045*** (0.008)
- Interpretation from text: a 10-year gain in grip strength increases the probability of labor force participation by about 19 percentage points and decreases the probability of retirement by about 18 percentage points. Improvements in self-reported health over one decade increase labor force participation by about 2.5 percentage points.

### Heterogeneity (2SLS by subgroup, labor force participation)
- Age groups (coefficients for Grip strength; Self-reported PC; Overall health; ADLs; IADLs):
  - Ages 50-59: 1.172***; 0.391***; 0.194***; 1.082***; 1.110***
  - Ages 60-69: 0.892***; 0.348***; 0.210***; 0.675***; 0.751***
  - Ages 70-79: 0.891***; 0.221***; 0.205***; 0.307***; 0.346***
  - Ages 80-89: 0.517***; 0.115***; 0.183***; 0.125***; 0.160***
- Gender:
  - Female: 0.584***; 0.248***; 0.155***; 0.449***; 0.529***
  - Male: 1.301***; 0.348***; 0.263***; 0.561***; 0.623***
- Education:
  - < Upper secondary: 0.926***; 0.301***; 0.218***; 0.503***; 0.540***
  - Upper secondary: 1.580***; 0.340***; 0.215***; 0.623***; 0.737***
  - Tertiary: 0.210; 0.107; 0.072; 0.164; 0.254
- Wealth:
  - Quintile 1 (lowest): 2.190**; 0.367***; 0.295***; 0.602***; 0.610***
  - Quintile 5 (highest): 0.800**; 0.219***; 0.122***; 0.437**; 0.600**
- Summary points from heterogeneity:
  - Larger effects for relatively younger older groups (50s and 60s).
  - Effects for men almost twice as large as for women.
  - Stronger effects for lower-education and lower-wealth individuals.

### Robustness checks and alternative instruments
- Results robust to:
  - Restricting instrument to subset of chronic diseases plausibly exogenous to work (Korea: cataracts, cancer, stroke).
  - Weighting diseases by severity (weights from regression of principal component of self-reported health on diseases).
  - Using a dummy for top 10 chronic diseases by severity (list includes Parkinson’s, Alzheimer’s, psychological disorders, stroke, kidney disease, lung disease, osteoporosis, arthritis, asthma, urinary incontinence).
  - Instrument definitions: total number of chronic diseases; dummy for any of the 17 chronic diseases.
  - Controlling for year fixed effects.
- Health instrumented by chronic diseases has no significant impact on probability of being unemployed, conditional on labor market participation (Appendix Table 9).

### Comparative evidence (Asia-5 excluding Korea)
- OLS: grip strength and self-rated overall health status not statistically significant in comparator group.
- 2SLS (instrument: incidence of chronic disease) finds significant effects for all health indicators across China, India, Japan, Malaysia, Thailand.
- Magnitude: causal effects are substantially larger in Korea than in comparator countries (nearly twice as large for most indicators), except self-rated overall health where effects are similar.
- Selected IV second-stage coefficients, Asia-5 (Excl. Korea):
  - Labor force participation (dummy): Grip strength 0.260* (0.151); Self-reported PC 0.101** (0.041); Overall health 0.108*** (0.037); ADLs 0.142** (0.059); IADLs 0.149** (0.064).
  - Retired (dummy): Grip strength -0.237** (0.105); Self-reported PC -0.105*** (0.032); Overall health -0.101*** (0.028); ADLs -0.140*** (0.045); IADLs -0.165*** (0.047).
  - IV first-stage chronic disease coefficients: -0.971***; -2.619***; -2.098***; -1.993***; -1.622***.
  - Weak IV F-statistics: 26.872; 56.711; 6.886; 6.617; 7.73.

### Aggregate contribution to Korea’s labor force participation (2006–20)
- Healthy aging measured by grip strength more than accounted for the total rise in elderly labor force participation (Figure 10).
- Healthy aging measured by self-reported PC contributed about 40 percent of the total rise in older workers’ labor force participation.
- Over the sample period 2006-20, healthy aging implied an increase in the labor supply of older individuals in Korea by around 24 percentage points (text reference).

### Mechanisms and interpretation
- IV estimates are substantially larger than OLS estimates, indicating endogeneity bias in OLS and supporting a causal interpretation of health → higher labor supply and lower retirement probability.
- Stronger effects for objective health measures (e.g., grip strength) in Korea suggest objectively measured health is a particularly strong determinant of labor supply.
- The pattern of diminishing effects at older ages (weaker for 70s and 80s) suggests constraints from skill obsolescence, pension incentives, and age discrimination.
- Positive health effects reflect stronger voluntary attachment to the labor market by healthier older individuals, not primarily driven by country-level policy changes or firm decisions.

### Policy implications and recommendations
- Health promotion and prevention policies that reduce incidence of chronic disease can help expand effective labor supply of older individuals. Potential measures include:
  - Expanding preventative healthcare (e.g., immunization and health screenings).
  - Expanding mental health services.
  - Targeting healthy behaviors: substance-abuse prevention, tobacco and junk food taxes, smoke-free regulations, promoting physical exercise.
- Pair health policies with labor-market measures to extend working lives:
  - Incentives to postpone retirement, phased retirement, reduced early-retirement benefits.
  - Lifelong reskilling and training.
  - Age-friendly workplaces, flexible work arrangements.
  - Combating age discrimination.
- Consider demand-side reforms to address structural labor market rigidities (e.g., seniority-based wage and promotion systems) that create incentives for early retirement and discourage hiring older workers.

*Source: IMF Working Paper — “Can Healthy Aging Boost Labor Supply? Evidence from Korea”, Section 5 (Effect of health on labor market outcomes).*

### Appendix Table 1. Data Sources and Availability: Household Surveys by Country

### Appendix Table 1. Data Sources and Availability: Household Surveys by Country

### Survey coverage by country and measured indicators
- China: The China Health and Retirement Longitudinal Study (CHARLS) — 4 waves (2011, 2013, 2015, 2018)
  - Measured: Grip strength, Overall health status, ADLs, IADLs (X X X X)
- India: Longitudinal Aging Study in India (LASI) — 1 wave (2017-19, 2021)
  - Measured: Grip strength, Overall health status, ADLs, IADLs (X X X X)
- Japan: Japanese Study of Aging and Retirement (JSTAR) — 3 waves (2007, 2009, 2011)
  - Measured: Overall health status, ADLs, IADLs (X X X)
- Korea: Korean Longitudinal Study of Aging and Retirement (KLoSA) — 8 waves (2006, 2008, 2010, 2012, 2014, 2016, 2018, 2020)
  - Measured: Grip strength, Overall health status, ADLs, IADLs (X X X X)
- Malaysia: Malaysia Aging and Retirement Survey (MARS) — 1 wave (2018-19)
  - Measured: Grip strength, Overall health status, ADLs (X X X)
- Thailand: Health Aging and Retirement in Thailand (HART) — 4 waves (2015, 2017, 2020, 2022)
  - Measured: Overall health status, ADLs (X X)

### OLS and Mundlak regression evidence on health indicators (Korea and Asia-5 excluding Korea)
- OLS regressions, Korea (Source: KLoSA)
  - Sample sizes and fit:
    - Observations: Grip strength 33,149; other indicators 37,307
    - R-squared: Grip strength 0.539; Self-reported PC 0.096; Overall health 0.179; ADLs 0.028; IADLs 0.056
  - Key coefficient signs and significance (Year of birth):
    - Grip strength: 0.019*** (0.002)
    - Self-reported PC: 0.009*** (0.002)
    - Overall health: 0.003** (0.002)
    - ADLs: 0.007*** (0.002)
    - IADLs: 0.009*** (0.002)
  - Controls included: individuals’ age, gender, education, and wealth
  - Age effects (negative and significant):
    - Age on Grip strength: -0.012*** (0.001)
    - Age on Self-reported PC: -0.015*** (0.002)
    - Age on Overall health: -0.027*** (0.001)
    - Age on ADLs: -0.007*** (0.001)
    - Age on IADLs: -0.007*** (0.002)
- OLS regressions, Asia-5 (Excl. Korea) (Source: CHARLS, HART, JSTAR, LASI, MARS)
  - Observations and fit:
    - Observations vary: Grip strength 67,253; Self-reported PC 78,169; Overall health 84,669; ADLs 99,505; IADLs 93,863
    - R-squared: Grip strength 0.493; Self-reported PC 0.135; Overall health 0.337; ADLs 0.046; IADLs 0.106
  - Year-of-birth coefficients (self-reported measures, with controls including country identifier):
    - Grip strength: -0.018 (0.014)
    - Self-reported PC: 0.055*** (0.015)
    - Overall health: 0.081*** (0.010)
    - ADLs: 0.013 (0.013)
    - IADLs: 0.031*** (0.008)
  - Country identifiers show distinct intercept differences (examples):
    - Country identifier = 2, India: Grip strength -0.762*** (0.111); IADLs -0.338*** (0.047)
    - Country identifier = 3, Japan: Grip strength 0.841*** (0.121); Self-reported PC 1.451*** (0.080)
    - Country identifier = 5, Malaysia: Grip strength -0.763*** (0.115); Self-reported PC 0.633*** (0.096)
    - Country identifier = 6, Thailand: Grip strength 1.500*** (0.101); Self-reported PC 0.403*** (0.092)

- Mundlak regressions, Korea (Source: KLoSA)
  - Observations: Grip strength 33,149; other indicators 37,307
  - R-squared: Grip strength 0.539; Self-reported PC 0.109; Overall health 0.183; ADLs 0.040; IADLs 0.067
  - Year of birth coefficients (controlling for mean age and mean wealth):
    - Grip strength: 0.026*** (0.004)
    - Self-reported PC: 0.051*** (0.010)
    - Overall health: 0.016** (0.006)
    - ADLs: 0.046*** (0.009)
    - IADLs: 0.044*** (0.007)
  - Mean age and mean wealth enter positively and are often significant:
    - Mean age on Self-reported PC: 0.053*** (0.011)
    - Mean wealth on Overall health: 0.080*** (0.008)

- Mundlak regressions, Asia-5 (Excl. Korea)
  - Observations: same as OLS Asia-5
  - R-squared comparable to OLS: Grip strength 0.495; Self-reported PC 0.136; Overall health 0.338; ADLs 0.048; IADLs 0.106
  - Year of birth coefficients:
    - Grip strength: 0.001 (0.015)
    - Self-reported PC: 0.047** (0.018)
    - Overall health: 0.084*** (0.012)
    - ADLs: -0.000 (0.013)
    - IADLs: 0.031** (0.015)

### Heterogeneity in healthy aging trends (Korea)
- Estimates from OLS regressions of health indicators on year of birth interacted with socioeconomic groupings (controls: age, gender, education, wealth)
- By age groups (Year of birth x Ages 50-59; 60-69; 70-79; 80-89):
  - Coefficients identical across reported age groups for each indicator:
    - Grip strength: 0.018*** 
    - Self-reported PC: 0.008*** 
    - Overall health: 0.004** 
    - ADLs: 0.006*** 
    - IADLs: 0.008***
- By gender:
  - Year of birth x Female: Grip strength 0.013***; Self-reported PC 0.007***; Overall health 0.006***; ADLs 0.004**; IADLs 0.006***
  - Year of birth x Male: Grip strength 0.024***; Self-reported PC 0.010***; Overall health 0.001; ADLs 0.008***; IADLs 0.011***
- By education:
  - Lower: Grip strength 0.021***; Self-reported PC 0.011***; Overall health 0.003*; ADLs 0.009***; IADLs 0.011***
  - Upper secondary: Grip strength 0.018***; Self-reported PC 0.007***; Overall health 0.004**; ADLs 0.005***; IADLs 0.007***
  - Tertiary: Grip strength 0.014***; Self-reported PC 0.004**; Overall health 0.001; ADLs 0.002; IADLs 0.005**
- By wealth quintiles (Quintile 1 through Quintile 5):
  - Year of birth x Quintile 1 to Quintile 5: Grip strength 0.020***; Self-reported PC 0.009***; Overall health 0.004**; ADLs 0.006***; IADLs 0.008***

### Effect of better health on labor market outcomes — 2SLS evidence (Korea)
- Instruments considered:
  - Incidence of chronic diseases unrelated to work
  - Incidence of chronic diseases weighted by their severity
  - Dummy for the top 10 most severe chronic diseases
- 2nd stage estimates (IV) — Effects on labor force participation (dummy) and retirement (dummy)
  - IV: Chronic diseases unrelated to work — Labor force participation coefficients:
    - Grip strength: 0.738*** 
    - Self-reported PC: 0.282*** 
    - Overall health: 0.290*** 
    - ADLs: 0.367*** 
    - IADLs: 0.451***
    - Retirement (negative effects):
      - Grip strength: -0.712*** 
      - Self-reported PC: -0.273*** 
      - Overall health: -0.281*** 
      - ADLs: -0.356*** 
      - IADLs: -0.437***
  - IV: Severity-weighted chronic diseases — Labor force participation:
    - Grip strength: 1.182*** 
    - Self-reported PC: 0.250*** 
    - Overall health: 0.222*** 
    - ADLs: 0.401*** 
    - IADLs: 0.344***
    - Retirement:
      - Grip strength: -1.055*** 
      - Self-reported PC: -0.223*** 
      - Overall health: -0.199*** 
      - ADLs: -0.358*** 
      - IADLs: -0.307***
  - IV: Dummy — top 10 most severe chronic diseases — Labor force participation:
    - Grip strength: 0.806*** 
    - Self-reported PC: 0.231*** 
    - Overall health: 0.190*** 
    - ADLs: 0.345*** 
    - IADLs: 0.406***
    - Retirement:
      - Grip strength: -0.671*** 
      - Self-reported PC: -0.192*** 
      - Overall health: -0.158*** 
      - ADLs: -0.287*** 
      - IADLs: -0.337***
- 1st stage coefficients (example signs and significance):
  - Chronic diseases unrelated to work (first-stage):
    - Grip strength: -0.309*** 
    - Self-reported PC: -1.001*** 
    - Overall health: -0.998*** 
    - ADLs: -0.757*** 
    - IADLs: -0.619***
  - Severity-weighted chronic diseases:
    - Grip strength: -1.042*** 
    - Self-reported PC: -5.716*** 
    - Overall health: -6.677*** 
    - ADLs: -3.468*** 
    - IADLs: -4.055***
  - Dummy — top 10 most severe chronic diseases:
    - Grip strength: -0.066*** 
    - Self-reported PC: -0.299*** 
    - Overall health: -0.385*** 
    - ADLs: -0.193*** 
    - IADLs: -0.165***
- Instrument strength (Weak IV F-statistics):
  - Chronic diseases unrelated to work: 55.27; 283.7; 411.9; 109.3; 136.2 (by indicator)
  - Severity-weighted chronic diseases: 59.23; 181.1; 1462; 59.14; 69.86
  - Top 10 most severe chronic diseases: 34.97; 121.7; 528.6; 47.02; 47.35
- Controls used in 2SLS specifications:
  - Socio-economic controls: Yes
  - Lifestyle controls: Yes
  - Survey year controlled for in many specifications; some specifications include Year FE

### Robustness: Year fixed effects and normalization to 1-decade improvements (Korea)
- 2SLS with Year FE — 2nd stage effects on labor force participation and retirement (coefficients with standard errors in parentheses)
  - Effect on labor force participation:
    - Grip strength: 1.054*** (0.151)
    - Self-reported PC: 0.291*** (0.037)
    - Overall health: 0.199*** (0.020)
    - ADLs: 0.499*** (0.091)
    - IADLs: 0.566*** (0.095)
  - Effect on retirement:
    - Grip strength: -0.950*** (0.125)
    - Self-reported PC: -0.262*** (0.033)
    - Overall health: -0.179*** (0.021)
    - ADLs: -0.450*** (0.077)
    - IADLs: -0.512*** (0.083)
- Normalization: Effect of 1-decade health improvements (with Year FE) — effects on labor force participation and retirement
  - Effect on labor force participation (1-decade improvement):
    - Grip strength: 0.199*** (0.028)
    - Self-reported PC: 0.026*** (0.003)
    - Overall health: 0.007*** (0.001)
    - ADLs: 0.036*** (0.007)
    - IADLs: 0.049*** (0.008)
  - Effect on retirement (1-decade improvement):
    - Grip strength: -0.179*** (0.024)
    - Self-reported PC: -0.023*** (0.003)
    - Overall health: -0.006*** (0.001)
    - ADLs: -0.033*** (0.006)
    - IADLs: -0.044*** (0.007)

### Effect of health on unemployment probability (Korea)
- 2SLS estimates of unemployment (dummy) on health indicators (controls: socio-economic and lifestyle)
  - Reported IV: 2nd stage coefficients (not significant as presented)
    - Grip strength: -1.233 (3.288)
    - Self-reported PC: -0.044 (0.029)
    - Overall health: -0.016 (0.010)
    - ADLs: -0.481 (0.384)
    - IADLs: -0.336 (0.302)
  - Socio-economic controls: Yes
  - Lifestyle controls: Yes

*Source: Appendix Table 1 and associated appendix tables and figures from the provided IMF Working Paper content.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025260-source-pdf.pdf_