固定效應隨機前沿模型估計方法的性能評估

Performance Evaluation of Estimation Methods for Fixed­EffectStochastic Frontier Models

賴建州(Chien-Chou Lai)

指導教授 ： 陳宜廷

共同指導教授 ： 王泓仁(Hung-Jen Wang)

國立臺灣大學/社會科學院/經濟學研究所/碩士(2021年)

https://doi.org/10.6342/NTU202103941

摘要

在本論文中，我們回顧了三種可用於估計Greene (2005a, b)真實固定效果隨機前沿模型的方法，包括Chen et al. (2014)基於封閉偏斜常態假設的最大概似法，Chen etal. (2015)動差法和Belotti與Ilardi (2018)的最大模擬概似法。我們通過蒙地卡羅模擬比較這三種估計方法的小樣本表現，並使用真實資料評估這些方法的實用性。模擬結果顯示動差法和基於封閉偏斜常態假設的最大概似法和最大模擬概似法之間的特色差異。實證案例表明，這三個估計方法在實證資料中的表現可能不同。

關鍵字

隨機前沿模型 ； 固定效果 ； 追蹤資料

並列摘要

In this thesis, we review three estimation methods, including the closed­skew­normal (CSN)­basedmaximumlikelihood (ML)methodofChenetal. (2014), themethod­of­moments (MM) estimator of Chen et al. (2015) and the maximum simulated like­lihood (MSL) method of Belotti and Ilardi (2018) for the true fixed­effect stochasticfrontier model of Greene (2005a, b). We also compare the finite­sample performanceof these three estimation methods by Monte Carlo experiments and assess the empiri­cal performance of these methods using real data. The simulation shows that the MMestimator performs comparably to the CSN­based estimator and the MSL estimator.The empirical example shows that these three estimators might perform differently inreal data.

並列關鍵字

Stochastic frontier ； fixed-effect ； panel data

參考文獻

References

Aigner, D., Lovell, C. and Schmidt, P., 1977. Formulation and estimation of stochastic frontierproduction function models.Journal of Econometrics,6, 21-­37.

Azzalini, A., 1985. A class of distributions which includes the normal ones.Scandinavian Journalof Statistics,12, 171-178.

Battese, G.E. and Coelli, T.J., 1992. Frontier production functions, technical efficiency and paneldata: With application to paddy farmers in India.Journal of Productivity Analysis,3, 153-­169.

Belotti, F. and Ilardi, G., 2018. Consistent inference in fixed­effects stochastic frontier models.Journal of Econometrics,202, 161-177.