本研究在跨時動態架構下，推估我國婦女勞工的跨期替代彈性，這是國內首度嘗試這方面的研究。本文利用所謂「時間序列的橫斷面資料」（a time series of cross-sections, TSCS），建構準縱橫資料（pseudo panel data），以群組平均值（cohort mean）當作觀察值，可進行動態研究。本研究估計得到：我國已婚勞動供給的跨時替代彈性介於2.52與8.36間，高於美國和英國婦女的替代彈性。另外，本研究發現樣本選擇性偏誤問題，必須在迴歸模型中考慮，否則，迴歸係數估計值產生嚴重偏誤。

　　　　　This paper examines the intertemporal substitution elasticity of married female workers in Taiwan under an intertemporal optimization framework that allows for the presence of corner solution (nonworkers) and uncertainty. The demand for leisure equation is estimated using the so-called "time-series of cross sections" (TSCS) data. The estimation procedures follow the Heckman's (1976) two -stage method and Moffit's (1993) instrumental variable approach (IV) which differs substantially from the method proposed by Deaton (1985). The two -stage method is used to deal with the sample selection bias and IV to identify and consistently estimate the parameters in the leisure equation with TSCS data. The estimated intertemporal elasticities of substitution are between 2.52 and 8.36, which is higher than those obtained by other researchers using either panel or TSCS data from U.S. and U.K. Evidence is found that sample selection bias plays an important role and , therefore, must be considered in the regression models.