Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

In 2005, WHA's policy resolution for WHO recommended that low- and middle-income countries to adopt social health insurance (SHI) as the health care financing strategy. SHI has indeed become a popular mode of financing health care in achieving the goal of universal coverage in the recent years. The value of insurance lies in its inherent design to avoid financial risk should adverse outcomes occur. When the exposure to risk is somewhat constrained by the availability of insurance benefits, one can expect to observe some changes in how the households allocate the resources freed up and become available to them as a result of health insurance. This study is hence set forth to study and evaluate the impact of health insurance on household consumption patterns and the empirical evidence can provide important policy references for countries considering social insurance approach.

As 57% of population were coverd by some form of social insurance programs with varying scope of coverage before National Health Insurance was introduced in 1995, Taiwan offers an unique laboratory to study the effect of health insurance in a natural experiment set-up. This study used data from 1993-2005 Taiwan's Survey of Family Income and Expenditure and adopted a difference-in-differences (DID) estimation model to assess the impact of health insurance on household consumption patterns. The household members insured by the government employee insurance were taken as the control group (as they are the least impacted by the introduction of NHI) and those without any form of social insurance coverage were treated as the treatment group. The major advantage of the DID model is to exclude the effect of natural growth over the years and can more accurately assess the effect of the policy of interest in a pre-post design with control. The effect of health insurance was evaluated by the changes in the proportion of the consumption expenditures devoted to medical items and non-medical items in the post-NHI periods (1996-2000 and 2001-2005 periods) compared to the pre-NHI period (1993-1994). In addition, a quantile regression model was employed to further examine the effect of insurance across households with different levels of spending shares. The empirical results have shown a reduction of approximately 27% (of the share) in the medical-item spending share and an average of 2% of increase (of the share) in the non-medical item spending share in the post-NHI periods, compared to the pre-NHI period. We also further examined the insurance effect by income quintile and found that the lower income group experienced a greater reduction in medical item spending share than their higher income counterpart. Among the non-medical consumption items, households in the lowest 20th income quintile devoted much of the increases to spending on educational, recreational and cultural activities, and rental expenses. The results from quantile regression indicated that NHI has the largest negative effect on medical item spending share at the top quantile; on the contrary, the largest positive effect on non-medical spending share was detected at the bottom quantile.