綠色社會責任 (GSR)，係民宿產業之重要外部性課題；且民宿總出租客房數，短期往往無法調整。有鑑於此，本文分別使用不可控制變數與傳統資料包絡分析法，推估日月潭國家風景區 21家民宿業者之短期與長期營運效率；並利用 Tobit迴歸模型，推估生態面積指數、企業社會責任、減碳措施指標、住宿價格與住房率，對效率五指標之衝擊效應，發現： (1) 總出租客房數視為固定時，效率邊界範圍將大於傳統 DEA，進而增加標竿業者數目 (效率等於 1之民宿業者)，並提高技術效率估計值；且發現效率五指標，分別有 25％~65％之效率改善空間。(2) 由於不同構面的 GSR投入對業者而言各有利弊，產業發展策略應由政府提供環保導向型之基礎建設，再引導業者降低住宿價格並提高住房率。

Green social responsibility (GSR) is an inevitable external issue for the development of the Bed and Breakfast (B&B) industry in Taiwan. As the total number of B&B rooms cannot be flexibly adjusted with changes in the operating conditions in the short term, this paper employed the data envelopment analysis with non-controllable variables and traditional Data Envelopment Analysis to estimate technical efficiency in both short-term and long-term stays for 21 B&B operators in the Sun Moon Lake Scenic Area. Finally, the Tobit regression model was applied to appraise how the biotope area factor, corporate social responsibility, carbon reduction index, room rate, and occupancy rate affect the five gauges of efficiency. Major findings are as follows: (1) As the total number of rooms is regarded as fixed, the efficiency frontier is greater than that estimated by traditional DEA, which increases both the number of benchmark operators whose efficiency equals one and the estimated value of technical efficiency. These five efficiency measures indicate that the potential for overall efficiency improvement is 25% to 65%. (2)Overall findings indicate that green social responsibility brings both benefit and cost for B&B operators. Hence, eco-orientated infrastructure investment by government, accompanied with a rise in the occupancy rate and a fall in the room price, may appear to be an indispensable strategy of sustainable management for Taiwan’s tourism industry.