Abstract
This current study is the first to apply Stochastic Data Envelopment Analysis (SDEA) approach to assess the efficiency of hotels. The determinants of hotel efficiency were also investigated employing the Tobit regression model approach. The SDEA results show that the SDEA efficiency measures are higher than the deterministic ones and the greater the stochastic variability of outputs, the closer the envelope moves successively to any given observation and the efficiency score approaches one. The optimal solution to SDEA involves the presence of some buffer (slack) of all outputs. In applications involving practice production situation, such buffer (slack) can be interpreted as the firms' need to hold inventory, excess capacity and organizational slack against stochastic uncertainty of market environment. Our results also indicate that resort hotels achieve better efficiency than those in metropolitan areas and operating as part of a chain is not the main determinant of the efficiency of international tourist hotels in Taiwan.