LTCI is a form of health insurance that compensates for the costs of care services required by individuals due to old age, illness, or disability. Originating in the United States in the 1970s, LTCI later expanded to European countries such as France, Germany, the United Kingdom, Ireland, and South Africa. In 2000, Japan formally integrated LTC coverage as a public service product into its national social security system. Although LTCI has achieved considerable global uptake, it remains in its early stages in China.

In 2016, the Ministry of Human Resources and Social Security issued the Guiding Opinions on Carrying Out the Pilot LTCI System, which outlined fundamental policies, administrative guidelines, and supportive measures for LTCI. Fifteen cities—including Chengde, Changchun, and Qiqihar—were designated as pilot sites. Under this policy, LTCI covers participants of basic medical insurance, with differentiated coverage levels and payment rates according to care needs and service modalities and reimburses a proportion of incurred care costs.

In 2020, the National Healthcare Security Administration, in collaboration with the Ministry of Finance, introduced the Guiding Opinions on the Expansion of the LTCI System Pilot Program, advocating the development of a more comprehensive LTCI framework. By the end of 2023, China’s LTCI pilot program had expanded to 49 regions, covering 183.31 million people, including 1.34 million who were already receiving LTCI benefits. The pilot involved 8,080 designated service organizations and 302,800 nursing service personnel, reflecting the rapid advancement of LTCI in China.¹

To estimate the effect of LTCI policy implementation on labor-force participation of adult children, we exploit LTCI pilots reforms during the sample period as a quasi-natural experiment. Specifically, we employ a DID approach to compare changes in labor-force participation in newly implemented LTCI regions versus non-LTCI regions, before and after the policy’s introduction. Given our use of panel data and binary explanatory variables, we specify the baseline model as an xtprobit regression (see Equation 1):

Among them, trea t ikt is the LTCI pilot dummy variable, which was set to 1 if in a LTCI pilot region and eligible for the type of health insurance coverage, and 0 otherwise; pos t ikt is the LTCI pilot period dummy variable, which was set to 1 for the period when LTCI was carried out and for each period thereafter, and 0 otherwise; pos t ikt · trea t ikt is the independent variable, a dummy variable for the implementation of LTCI, which is 1 for the period when LTCI was implemented in the individual’s region and each subsequent period, and 0 otherwise. If its coefficient α 1 is less than 0, LTCI reduced the likelihood of offspring’s labor force participation. If it is greater than 0, LTCI raised the likelihood of offspring’s labor force participation. X ijt is an individual- and household-level control variable, with α j is its estimated coefficient; α 0 is a constant term, λ t is a time fixed effect, γ k is an city fixed effect, λ t · γ k is the cross-multiplication term of the time fixed effect and the city fixed effect, and ε ikt is a randomized disturbance term.

We draw on individual-level data from the 2011, 2013, 2015, and 2018 waves of the China Health and Retirement Longitudinal Study (CHARLS). CHARLS samples Chinese residents aged 45 and above (and their spouses) across approximately 10,000 households in 150 cities, totaling around 17,000 respondents. In addition to collecting standard demographic and socioeconomic information, the survey gathers extensive details on health status, health service utilization, health insurance, income and consumption, financial support, and caregiving. Notably, CHARLS includes questions on each household’s children, including their employment status, thereby enabling this study’s focus on labor-force participation of adult childrenlabor-force.

In defining the sample, adult children were restricted to those aged 16 to 60. This criterion aligns with China’s labor regulations, which prohibit the employment of minors under 16 (Labor Law of the People’s Republic of China) and typically recognize 60 as the retirement age. Observations missing key variables, including age, gender, work status, and income, were excluded. After constructing an unbalanced panel, the final analytic sample contains 38,847 valid person-period observations, with adult children as the primary unit of analysis. Information on LTCI coverage was collected from official policy documents of pilot cities. China has introduced LTCI in phases, beginning with Qingdao’s Opinions on Establishing a Long-Term Medical Care Insurance System (Trial) in July 2012, followed by national guidance in 2016 and again in 2020. Given data constraints and the fact that the second round of national pilot sites was launched in 2020, this study focuses on the first set of pilot regions. Two pilot sites—Nantong City in Jiangsu Province and Shihezi City in the Xinjiang Production and Construction Corps—were excluded due to unavailability of relevant data. Ultimately, 13 pilot cities (e.g., Chengde City in Hebei Province and Qingdao City in Shandong Province) included in CHARLS are classified as the treatment group, while all remaining cities serve as the control group.

In order to align our sample with the study’s focus on individuals who may require long-term care, we first restrict the parent sample to those aged 60 or older—consistent with the standard LTCI coverage criteria for older adults with disabilities. Second, we remove outliers on key variables, as well as observations with missing essential data. This results in a final pooled sample of 38,800 person-wave observations, of which 3,270 are classified as covered by LTCI (i.e., they reside in a pilot region and are in a period after the pilot’s introduction). We further divide the dataset into LTCI-covered and LTCI-uncovered subsamples and compare mean differences.

Table 1 presents the descriptive statistics for the total sample and by LTCI coverage. Overall, 86.65% of adult children in the sample report being employed. In the LTCI-covered subsample, the labor-force participation rate is slightly higher than in uncovered regions, providing an initial indication that LTCI may positively influence labor-force participation of adult children labor-force. The remaining variables capture individual and parental characteristics (e.g., gender, age, health status), which serve as controls in our subsequent econometric analyses.

Table 2 presents the estimated effects of LTCI on labor-force participation of adult children labor-force. Column (1) excludes control variables, whereas column (2) introduces time and city fixed effects, and column (3) adds the full set of individual- and parent-level controls. In all three specifications, the interaction term, the interaction term Post×Treat carries a positive and statistically significant coefficient, indicating that LTCI policy implementation is associated with higher labor-force participation among adult children. Notably, the magnitude of this coefficient remains relatively stable across models, suggesting that the core finding is robust to the inclusion of various controls.

The results in column (3) further show that, among child-level covariates, being male (Cgender) is negatively associated with labor-force participation, while factors such as age (Cage) and marital status (Cmarry) exhibit positive associations. Household registration in a non-rural region (Chukou) also increases the likelihood of employment. Regarding parent-level variables, parental gender (Pgender), age-squared (Page2), and marital status (Pmarry) have significant effects.

Overall, these findings suggest that LTCI coverage plays a key role in facilitating labor-force participation of adult children, above and beyond the influence of individual and household characteristics.

This section presents two robustness checks to validate the DID results. First, we examine whether the parallel trends assumption hold by investigating whether labor-force participation of adult children labor-force in treatment and control regions followed similar trajectories before LTCI implementation. Second, we conduct a placebo test by randomly assigning “treatment” status to cities to ensure that the observed effects are indeed attributable to LTCI.

LTCI influences labor-force participation of adult children labor-force, we perform subgroup analyses by gender, age, cohabitation, and skill level. Tables 3, 4 present the corresponding regression results.

The implementation of LTCI provides new care options for older adults, particularly in light of its service payment approach in many pilot cities (14 out of 21 cities by 2018). Under this model, professional care from institutions, community workers, or specialized home services can effectively substitute for informal family-based care. To examine this substitution effect, we use the CHARLS question “Who helps you with dressing, bathing, eating, etc.?” and code assistance from employed caregivers, volunteers, nursing home personnel, or community workers as formal care (value = 1). Care offered by spouses, children, or other family members is coded as informal care (value = 0).

As shown in column (1) of Table 5, the LTCI pilot policy significantly increases the likelihood of using formal care. This finding aligns with prior evidence that subsidized LTC services can reduce reliance on informal care within the family, allowing caregivers to reallocate time to alternative activities, such as employment (2, 8, 11, 32).

Although the proportion of households that actually use LTCI within the sample is relatively small, the anticipated effects of LTCI can still meaningfully influence labor-force participation among adult children. Because children may adjust their current behavior (e.g., adjust labor-force participation earlier) based on expectations of future LTCI coverage. Any positive impact on their labor outcomes is more plausibly linked to the expectation that LTCI will reduce future caregiving costs or uncertainties. In other words, if adult children foresee that their caregiving burden could be lessened when their parents need care, they may be more willing to engage in or remain in the workforce.

To assess this, we compare the effect of LTCI on labor-force participation of adult children labor-force versus its effect on parents’ labor-force participation. As shown in Table 5, column (2) and (3), the LTCI coefficient is statistically insignificant for parents’ employment but positive and significant for adult children’s employment. This discrepancy indicates that LTCI exerts its strongest effect through anticipation, rather than immediate utilization. In other words, even though parents may not yet be fully utilizing LTC benefits, adult children respond to the expectation of reduced caregiving obligations, thus increasing their labor-force participation.