Explaining Efficiency in Nonparametric Frontier Models: Recent Developments in Statistical Inference
The explanation of efficiency differentials is an essential step in any frontier analysis study that aims to measure and compare the performance of decision making units. The conditional efficiency measures that have been introduced in recent years (Daraio and Simar, J. Prod. Anal. 24:93–121, 2005) represent an attractive alternative to two-step approaches, to handle external environmental factors, avoiding additional assumptions such as the separability between the input-output space and the space of external factors. Although affected by the curse of dimensionality, nonparametric estimation of conditional measures of efficiency eliminates any potential specification issue associated with parametric approaches. The nonparametric approach requires, however, estimation of a nonstandard conditional distribution function which involves smoothing procedures, and therefore the estimation of a bandwidth parameter. Recently, Bădin et al. (Eur. J. Oper. Res. 201(2):633–640, 2010) proposed a data driven procedure for selecting the optimal bandwidth based on a general result obtained by Hall et al. (J. Am. Stat. Assoc. 99(486):1015–1026, 2004) for estimating conditional probability densities. The method employs least squares cross-validation (LSCV) to determine the optimal bandwidth with respect to a weighted integrated squared error (WISE) criterion.This paper revisits some of the recent advances in the literature on handling external factors in the nonparametric frontier framework. Following the Bădin et al. (Eur. J. Oper. Res. 201(2):633–640, 2010) approach, we provide a detailed description of optimal bandwidth selection in nonparametric conditional efficiency estimation, when mixed continuous and discrete external factors are available. We further propose an heterogeneous bootstrap which allows improving the detection of the impact of the external factors on the production process, by computing pointwise confidence intervals on the ratios of conditional to unconditional efficiency measures.We illustrate these extensions through some simulated data and an empirical application using the sample of U.S. mutual funds previously analyzed in Daraio and Simar (J. Prod. Anal. 24:93–121, 2005; Eur. J. Oper. Res. 175(1):516–542, 2006; Advanced Robust and Nonparametric Methods in Efficiency Analysis: Methodology and Applications, Springer, New York, 2007a).
KeywordsMutual Fund Efficiency Score Nonparametric Estimator Optimal Bandwidth Conditional Measure
Previous discussions with Léopold Simar as well as his valuable comments and suggestions contributed a lot in improving the quality of the present paper and are gratefully acknowledged. The usual disclaimers apply.
- Cooper, W.W., Seiford, L.M., & Tone, K. (2000). Data envelopment analysis: A comprehensive text with models, applications, references and DEA-solver software. Boston: Kluwer Academic Publishers.Google Scholar
- Daraio, C., Simar, L., & Wilson, P. (2010). Testing whether two-stage estimation is meaningful in nonparametric models of production, Discussion Paper #1030, Institut de Statistique, Université Catholique de Louvain, Louvain-la-Neuve, Belgium.Google Scholar
- Deprins, D., Simar, L., & Tulkens, H. (1984). Measuring labor-efficiency in post offices. In M. Marchand, P. Pestieau, & H. Tulkens (eds.) The Performance of public enterprises – Concepts and Measurement (pp. 243–267). Amsterdam: North-Holland.Google Scholar
- Fan, J., & Gijbels, I. (1996). Local polinomial modelling and its applications. London: Chapman and Hall.Google Scholar
- Farrell, M.J. (1957). The measurement of the Productive Efficiency. Journal of the Royal Statistical Society, Series A, CXX, Part 3, 253–290.Google Scholar
- Färe, R., Grosskopf, S., & Lovell, C.A.K. (1994). Production frontiers. Cambridge: Cambridge University Press.Google Scholar
- Jeong, S.O., Park, B.U., & Simar, L. (2008). Nonparametric conditional efficiency measures: asymptotic properties. Annals of Operations Research, doi: 10.1007/s10479-008-0359-5.Google Scholar
- Pagan, A., & Ullah, A. (1999). Nonparametric Econometrics. Cambridge: Cambridge University Press.Google Scholar
- Simar, L., & Wilson, P.W. (2008). Statistical inference in nonparametric frontier models: recent developments and perspectives. In Harold O. Fried, C.A. Knox Lovell, & Shelton S. Schmidt (Eds.), The Measurement of Productive Efficiency, 2nd edn. Oxford: Oxford University Press.Google Scholar
- Simar, L., & Wilson, P.W. (2009). Inference by Subsampling in Nonparametric Frontier Models, Discussion Paper #0933, Institut de Statistique, Université Catholique de Louvain, Louvain- la-Neuve, Belgium.Google Scholar
- Simar, L., & Wilson, P.W. (2010). Two-Stage DEA: Caveat Emptor, Discussion Paper #10xx, Institut de Statistique, Université Catholique de Louvain, Louvain-la-Neuve, Belgium.Google Scholar