本研究探用Gonzdlez et al.(2005)提出的縱横平滑移轉迴歸模型，並考慮內生性偏誤問題，以 1995-2011年的155個國家為實證樣本，探討健康支出的所得彈性，實證模型以人均實質所得和預期壽命為 解釋變數，預期壽命為門檻變數，人均健康支出為內生變數，探討人均健康支出與人均實質所得的非線 性關係及健康支出的所得彈性。實證結果顯示，存在兩個門檻值（3.830與4.303)，表示在此兩個點存在結 構性轉變，將此縱横資料分成低、中、高三個狀態。健康支出與實質所得在預期壽命的三個狀態均為正 向顯著關係，意謂在低預期壽命狀態下(低於46歲），實質所得愈高，健康支出急速增加；當預期壽命介於 46~74歲中間狀態時，健康支出增加速度趨緩；但是到了高生命預期狀態時（超過74歲），健康支出增加速 度又逐渐加快。健康支出對人均實質所得的彈性在低預期壽命與高預期壽命兩種狀態時為「奢侈品」。此 實證模型進一步檢測從低生命預期到高生命預期轉換的平滑度。

This study applies a Panel Smooth Transition Regression (PSTR) model which developed by Gonzalez et al. (2005) and takes into account the potential endogeneity biases in order to establish country-specific and time-specific elasticities of healthcare expenditure with respect to real income for 155 selected countries for the period 1995-2011. This study sets real income per capital and life expectancy as the explanatory variable, life expectancy as the transition variable, and healthcare expenditure per capital as dependent variable to investigate the nonlinear relationship between healthcare expenditure per capital and real income per capital. Our empirical results indicate that there exist two threshold values of 3.830 and 4.303, which shows that there are structural changes at these two points. Both thresholds divide our panel into three regimes. There is a significantly positive relationship between healthcare expenditure and real income since the estimated coefficients of LGDP are significantly positive (2.319, 0.613 and 1.677) in the three regimes, suggesting that as real income rises, healthcare expenditure rapidly increases in low life expectancy countries, then slowly increases in the middle stage when life expectancy between approximately 46 and 74, and last gradually increases in the high regime after the values of life expectancy exceed approximately 74 old. The healthcare expenditure is elastic with to real income for the low and high regime, suggesting that they are a luxury good. This technique further estimates the smoothness of the transition from a low life expectancy to a high life expectancy regime.