بیمه سلامت و تامین نیروی کار زن در تایوان

We examine whether the availability of subsidized health insurance to the non-working population in Taiwan affected the labor force participation of married women. Our empirical identification exploits the fact that such insurance was first made available to wives of government employees, before being made universally available in Taiwan in 1995. We find that the availability of insurance for non-workers was associated with a 4 percentage point decline in labor force participation among married women, with larger declines among women from low income households. Countries considering universal health insurance as in Taiwan should anticipate similar declines in labor force participation.

This paper provides an empirical examination of how the expansion of subsidized health insurance to the non-working population in Taiwan affected the labor supply of married women. We estimate reduced form labor force participation models for married women using a series of cross-sectional surveys, the Survey of Family Income and Expenditure, collected annually in Taiwan from 1969 to the present. Our empirical work identifies the effect of this insurance expansion on labor force participation by exploiting the fact that this insurance was only made available to wives of government employees beginning in 1982, before being made universally available in 1995. We find that the availability of subsidized health insurance for non-working women was associated with a roughly 4 percentage point decline in labor force participation among married women, with larger declines among women whose husbands have low earnings potential (as measured by their educational attainment). Both of these results are best explained by a simple static model of labor supply with incomplete health insurance markets, in which health insurance for non-working women is valued above its expected financial value due to a combination of risk aversion, highly variable health care expenses, and the lack of a market for private health insurance in Taiwan. There is a large empirical literature, almost exclusively based on the US experience, that investigates the impact of health insurance on individual labor supply behavior (see Currie and Madrian, 1998 and Gruber, 1998 for recent surveys). The work most closely related to this paper falls into four categories. (1) Spouse insurance coverage and female labor force participation, (2) medicare and retirement, (3) employer provided health insurance and “job-lock” and (4) medicaid and the labor supply of single mothers. The common thread connecting our results to the recent papers in these four categories is that the link between employment and health insurance is found to have significant impacts on labor supply behavior which are larger than one would anticipate based on the expected dollar value of this insurance alone. However, if there is market failure in the private insurance market (say, due to adverse selection), then one would expect such large effects of health insurance on labor supply: risk averse individuals who cannot obtain insurance otherwise will value employer-provided health insurance above its expected dollar value. Thus, our paper adds to the evidence that employer-provided health insurance is an important determinant of labor supply behavior in the presence of incomplete insurance markets. Section 2 provides some background on the key features of health insurance in Taiwan that provide the basis for our empirical work. Until recently, health insurance in Taiwan has only been available to working adults with no coverage of spouses or dependents. In 1982, the health insurance plan for government employees was amended to provide coverage of spouses, allowing government employees’ wives to withdraw from the labor market and still maintain health insurance coverage. In 1995, National Health Insurance was implemented in Taiwan and provided coverage to spouses of all workers. These changes provide the basis for our empirical work, as we compare the labor force participation of government employees’ wives to that of other married women in the years just before and after these changes in coverage. Section 3 develops a simple static labor supply model as a guide for interpreting the empirical results. This model suggests that the impact of these insurance expansions should be largest on the labor force participation of women whose husbands have low earnings, which we proxy for in the empirical work with husband’s education. Such women would face low levels of consumption if they chose not to work and as a result are more concerned about both the level and the uncertainty of health expenses in the absence of insurance. The model also implies that the impact of these insurance expansions on labor force participation should be considerably larger than one would anticipate based on the expected value of the health insurance subsidy being provided (which is on the order of 1–3% of a women’s earnings). The remaining sections of the paper present the data, empirical strategy and estimation results. All of our analysis relies on the Survey of Family Income and Expenditure, a cross-section survey of approximately 15,000 households collected annually in Taiwan since 1969. We use the surveys from 1979 to 1985 to analyze the 1982 change in insurance coverage and the surveys from 1992 to 1997 to analyze the 1995 change. In both analyses, we estimate reduced form probit models of labor force participation among married women on the pooled data. We identify the impact of the change in insurance coverage by comparing the change in labor force participation among the affected women to the change among women whose insurance options were unaffected. We find that labor force participation of married women declines by about 4 percentage points after subsidized health insurance is made available to them through their spouses or through National Health Insurance. For wives of less-educated husbands, the effects are larger, up to twice as large depending on the specification. The magnitudes of both of these estimates are large relative to the expected value of the insurance subsidy being given to women, but are consistent with our model in which risk-averse women are reluctant to leave the labor force if they cannot purchase health insurance. Moreover, the magnitudes are large from a macro-perspective and may help to explain recent concern in Taiwan over a decline in labor force participation rates. Countries considering insurance expansions similar to Taiwan’s want to consider these potential effects on their labor supply.

In 1996, the Taiwanese government announced that the labor force participation rate was at its lowest point in the last 10 years. Experts and scholars pointed to a variety of potential factors, ranging from the economic recession to the uncertain economic environment. In this paper, we point to the implementation of National Health Insurance as another important factor causing the recent decline in labor force participation in Taiwan. Overall, we find strong evidence that the availability of subsidized health insurance for non-working women in Taiwan has reduced labor force participation among married women by approximately 4 percentage points. The results based on the availability of National Health Insurance in 1995 reinforce the findings based on the expansion of government employee insurance to spouses in 1982. In both cases, making subsidized health insurance available to women if they chose not to work was associated with a significant decline in the fraction of women working. Moreover, this decline occurred primarily among wives of less educated men as predicted by a standard model of labor force participation. Are these estimates large? One way of answering this question is to ask whether they are consistent with other evidence on female labor force participation. As discussed in Section 3, for reasonable parameter values a simple model of labor supply suggests that the impact of these health insurance expansions should have been comparable to the impact of a 2–6% fall in wages. Thus, our estimates are consistent with a wage elasticity of non-participation ranging from just under 1 to just over 2. Recent estimates from women in the US (Eissa and Liebman, 1996) and Canada (Card and Robins, 1996) yield elasticities of non-participation of around 1, suggesting that our estimates are large but are not unreasonable. Not surprisingly, our results suggest that insurance is a powerful work incentive. When social insurance programs offer subsidized health insurance to non-working adults, as was the case in Taiwan, this provides significant work disincentives for individuals who could not otherwise purchase health insurance. Recent health care reform proposals in the US and other countries have, to various degrees, proposed expanding subsidized health insurance coverage among non-working adults. The Taiwan experience suggests that these countries should anticipate a decline in labor force participation if such proposals are implemented. Conversely, our results imply that creation of government insurance programs to cover the working poor (such as the recent medicaid expansions) may have much larger impacts than equivalent financial incentives (such as the earned income tax credit) because of the value individuals place on insurance. For policy makers looking to increase work incentives, health insurance may provide a very cost-effective tool.