Nonparametric efficiency analysis: a multivariate conditional quantile approach

This paper proposes a unifying probabilistic framework for efficiency and productivity analysis in a complete multivariate setup (multiple inputs and multiple outputs). Properties of the Farrell’s efficiency scores are derived in terms of the characteristics of the probability distribution of the data generating process. This allows to introduce a notion of -quantile efficiency scores related to a non-standard conditional -quantile frontier. Nonparametric estimators are then naturally introduced providing estimators of the production frontier more robust to outliers and/or extreme values than the traditional envelopment estimators (FDH/DEA). The paper defines a new concept of efficiency measurement, analyzes its properties and proposes a nonparametric estimator with all its asymptotic properties. Numerical examples (simulated data and mutual funds data) illustrate its practical use. This extends and generalizes previous works of Cazals, Florens and Simar (2002), Aragon, Daouia and Thomas-Agnan (2002) and Daraio and Simar (2003).

• metadata
• relations

Document type	Document de travail (Working Paper)
Access type	Accès libre
Publication date	2004
Language	Anglais
Number of pages	36 pages
Collection	STAT Discussion Papers - 0419
Affiliations	Université de Toulouse - GREMAQ UCL - EUEN/STAT - Institut de statistique
Keywords	Efficiency ; Frontier estimation ; Robust nonparametric estimators ; Conditional quantiles
Links	http://sites.uclouvain.be/IAP-Stat-Phase-V-VI/ISBApub/ISBAdp.html#2004 http://hdl.handle.net/2078.1/122924[Handle]

See also
	[boreal:23569]	Nonparametric efficiency analysis: a multivariate conditional quantile approach