本文是研究美國職業軍人（服役滿二十年）與非職業軍人（服役滿二至六年）退役後轉業收入的 差異情形，研究資料採自1987年美國國防調查退役軍人資料檔（SOVIII)。而爲評量職業軍人與非職業 軍人退役後的收入情形，本研究係立基於人力資本的框架來建構纣數迴歸模式且運用Heckman’s兩階 段迴歸技術來修正因選擇性誤差導致使用一般最小平方法所可能產生的被扭曲的估計値。結果顯示： 男性的職業軍人與非職業軍人退役後在相同的服務年限水準上收入有明顯的差異存在，敎育程度對兩 者間在退役後的所得差異亦達顯著水準。本文冀望將此研究結果，提供予國軍做爲敎育訓練政策制訂 時的參考，俾有助吸引高素質人力投入軍中服務。

This paper analyzes differences in earnings between U.S. military retired veterans and non retired veterans with similar demographic characteristics. Using data from the 1987 National Survey of Veterans (SOVIII) Files, the analysis employs Heckman^ two stage regression technique to correct for the selectivity bias that distorts the estimations obtained using Ordinary Least Squares. A logearnings model is specified based on human capital theory. The intent of the model is to measure the effects of retired versus non - retired veteran status on the post-service earnings of male veterans. The findings reveal a statistically significant loss of post-service income incurred by male retired veterans. It was also determined, however, that increases in education had a statistically significant effect on earnings for both retired and non-retired veterans. Applying the results of this U.S. study to the Republic of China (R.O.C.), it is recommended that the R.O.C. Armed Forces conduct a study to determine whether a similar post-retirement earnings deficit applies.