台灣國際觀光旅館業效率與生產力之分析－由下而上分解法

Efficiency and Productivity Analysis of International Tourist Hotels in Taiwan: The Bottom-up Decomposition Approach

李文福(Wen-Fu Lee) ； 王媛慧(Yuan-Huei Wang)

《應用經濟論叢》 2012生產力與效率特刊 (2012/09) Pp. 39-89

https://doi.org/10.7086/TJAE.201209.0039

摘要

本文探討2002至2008年58家台灣國際觀光旅館之生產力變動，利用Balk（2001）由下而上的分解法將Malmquist TFP指數分解為技術變動、純技術效率變動、規模效率變動及產出組合變動等效果。DEA實證結果顯示，2002-2008年平均之VRS產出距離函數值為91.47％。第二階段迴歸分析總技術效率之影響因素，結果顯示：總體GDP影響為正、旅館規模影響為正、北中南之旅館效率較東部為高、政府經濟服務支出有助效率提升、市場集中度與效率呈現正相關，品牌連鎖關係並未能導致較佳之效率。整體而言，研究期間總要素生產力平均每年成長1.56％，驅動力主要來自技術進步（1.43％），次為純技術效率提升（0.2％）。至於規模效率與產出組合變動效果，整體指數相對穩定於1附近，但對部分廠商或部分年度而言變動仍顯著，值得關注。營運績效管理矩陣顯示，超過三成五的國際觀光旅館目前具有競爭力與長期發展潛力。

關鍵字

國際觀光旅館 ； Malmquist生產力指數 ； 技術進步 ； 產出組合 ； 截斷最小絕對離差估計

並列摘要

This study analyzes productivity changes of fifty-eight Taiwanese international tourist hotels with data from 2002 to 2008. Utilizing the bottom-up approach of Balk (2001), we decompose the Malmquist TFP indexes into changes in technology, pure technical efficiency, scale efficiency, and output mix. At the end, the operational performance management matrix is used to study hotels' existing competitiveness and potentials of long-run development.Several results are obtained. DEA analysis indicates that within the sample periods, on average, the pure technical efficiency was 91.47%. Regression analysis shows that GDP, local government economic service expenditures, and market concentration were positively related to overall technical efficiency, while the chain system hotels did not perform better than others. TFPC increased by 1.56% annually, which could mainly be attributed to the improvements in technical change (1.43%) and in pure technical efficiency change (0.2%). As for scale efficiency and output mix effect, the change indexes were stable around unity. The operational performance management matrix suggests that more than 35% of the hotels were competitive presently and with potentials of long-run development.

並列關鍵字

International Tourist Hotels ； Malmquist Productivity Index ； Technical Progress ； Output Mix ； Censored Least Absolute Deviations Estimation