Abstract
This paper makes two important contributions to the literature on prediction intervals for firm specific inefficiency estimates in cross sectional SFA models. Firstly, the existing intervals in the literature do not correspond to the minimum width intervals and in this paper we discuss how to compute such intervals and how they either include or exclude zero as a lower bound depending on where the probability mass of the distribution of \( u_{i} |\varepsilon_{i} \) resides. This has useful implications for practitioners and policy makers, with greatest reductions in interval width for the most efficient firms. Secondly, we propose an ‘asymptotic’ approach to incorporating parameter uncertainty into prediction intervals for firm specific inefficiency (given that in practice model parameters have to be estimated) as an alternative to the ‘bagging’ procedure suggested in Simar and Wilson (Econom Rev 29(1):62–98, 2010). The approach is computationally much simpler than the bagging approach.
Similar content being viewed by others
Notes
Notable exceptions are point estimates of firm inefficiency from the class of time invariant panel models (Pitt and Lee 1981) and the class of deterministically time varying models (Battese and Coelli 1992 and Cuesta 2000) which yield consistent estimates as \( T \to \infty \). For the purpose of this paper, however, we restrict our attention to cross sectional models (or equivalently pooled panel models).
As discussed in Simar and Wilson (2010, footnote 9), but also alluded to in Coelli et al. (2005) and Greene (2008), Horrace and Schmidt (1996) incorrectly used the terminology ‘confidence intervals’ when in fact they are prediction intervals for the random variable u i (and not a parameter), using the information available in the realized composite error term. Importantly, a prediction interval does not collapse in width as \( N \to \infty \), which is clearly the case here.
Bera and Sharma (1999) suggest that the computed \( E\left[ {u_{i} |\varepsilon_{i} } \right]/\left( {var\left[ {u_{i} |\varepsilon_{i} } \right]} \right)^{0.5} \) should be compared to critical values derived from one sided percentiles of the conditional distribution. But this is not hypothesis testing for u i and not even hypothesis testing for \( E\left[ {u_{i} |\varepsilon_{i} } \right] \) given \( var\left( {E\left[ {u_{i} |\varepsilon_{i} } \right]} \right) \ne var\left( {u_{i} |\varepsilon_{i} } \right) \).
An issue that arises in comparing the 'bagging' and asymptotic approaches is the possibility that the point estimate of the variance of the inefficiency term is zero - this is the 'wrong skewness' problem. In this instance, the asymptotic approach produces a zero width interval, by construction, but the bagging approach may still produce a nonzero width interval (Simar and Wilson 2010). This issue has attracted some attention in recent discussion of stochastic frontier modelling. We note the possibility, however, we have not attempted to confront this substantive issue in this paper (our analysis assumes a nonzero estimate of the variance). We leave this question for further research, by others as well as ourselves.
Given efficiency is often expressed as a percentage it is worth clarifying that the percentage reductions given above are percentages of the interval width rather than absolute percentage points.
References
Aigner DJ, Lovell CAK, Schmidt P (1977) Formulation and estimation of stochastic frontier production function models. J Econom 6(1):21–37
Alvarez A, Amsler C, Orea L, Schmidt P (2006) Interpreting and testing the scaling property in models where inefficiency depends on firm characteristics. J Prod Anal 25:201–212
Amsler C, Leonard M, Schmidt P (2010) Estimation and inference in parametric deterministic frontier models, working paper
Battese GE, Coelli TJ (1992) Frontier production functions and the efficiencies of Indian Farms Using Panel data from ICRISAT’s village level studies. J Quant Econ 5:327–348
Bera AK, Sharma SC (1999) Estimating production uncertainty in stochastic frontier production frontier models. J Prod Anal 12:187–210
Coelli T, Rao DSP, O’Donnell CJ, Battese GE (2005) An introduction to efficiency and productivity analysis, 2nd edn. New York, Springer
Cuesta RA (2000) A production model with firm-specific temporal variation in technical inefficiency: with application to Spanish dairy farms. J Prod Anal 13(2):139–152
Econometric Software Inc. (2010) LIMDEP, user’s manual. http://www.limdep.com, Plainview, NY
Farrell MJ (1957) The measurement of productive efficiency. J R Stat Soc Ser A Gen 120(3):253–290
Flores-Lagunes A, Horrace WC, Schnier KE (2007) Identifying technically efficient fishing vessels: a non-empty, minimal subset approach. J Appl Econom 22:729–745
Greene WH (2008) The econometric approach to efficiency analysis. In: Fried HO, Lovell CAK, Schmidt SS (eds) The measurement of productive efficiency growth, 2nd edn. Oxford University Press, New York
Greene WH (2011) Econometric analysis, 7th edn. Prentice Hall, New York
Hjalmarsson L, Kumbhakar SC, Heshmati A (1996) DEA, DFA and SFA: a comparison. J Prod Anal 7:303–327
Horrace WC (2005) On ranking and selection from independent truncated normal distributions. J Econom 126:335–354
Horrace WC, Schmidt P (1996) Confidence statements for efficiency estimates from stochastic frontier models. J Prod Anal 7(2/3):257–282
Horrace WC, Schmidt P (2000) Multiple comparisons with the best, with economic applications. J Appl Econom 15:1–26
Jondrow J, Lovell CAK, Materov IS, Schmidt P (1982) On estimation of technical inefficiency in the stochastic frontier production function model. J Econom 19:233–238
Kim Y, Schmidt P (2008) Marginal comparisons with the best and the efficiency measurement problem. J Bus Econ Stat 26(2):253–260
Krinsky I, Robb A (1986) On approximating the statistical properties of elasticities. Rev Econ Stat 68(4):715–719
Kumbhakar SC, Löthgren M (1998) A Monte Carlo analysis of technical inefficiency predictors. Working paper series in economics and finance, no. 229, Stockholm School of Economics
Meeusen W, van Den Broeck J (1977) Efficiency estimation from Cobb-Douglas production functions with composed error. Intern Econ Rev 18(2):435–444
Pitt M, Lee L (1981) The measurement and sources of technical inefficiency in Indonesian weaving industry. J Dev Econ 9:43–64
R Development Core Team (2010) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0. http://www.R-project.org
Simar L, Wilson PW (2010) Inferences from cross-sectional, stochastic frontier models. Econom Rev 29(1):62–98
Smith A, Wheat P (2012) Evaluating alternative policy responses to franchise failure: Evidence from the passenger rail sector in Britain. J Transp Econ Policy 46(1):25–43
Taube R (1988) Möglichkeiten der Effizienzmess ung von öffentlichen Verwaltungen. Duncker & Humbolt GmbH, Berlin
Train KE (2009) Discrete choice methods with simulation, 2nd edn. Cambridge University Press, New York
Waldman DM (1984) Properties of technical efficiency estimators in the stochastic frontier model. J Econom 25:353–364
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix 1: Derivation of minimum width predictive intervals for the truncated normal distributions
Consider \( \left( {u_{i} |\varepsilon_{i} } \right)\sim N^{ + } \left( {\mu_{i*} ,\sigma_{*}^{2} } \right) \)
There are two solutions to the Langrangean problem. Either:
Otherwise
\( U^{*} \) for the case in (12) is given by Horrace and Schmidt (1996) and reproduced in Eq. (3) as
Now consider (11).
Define, X ~ N \( (\mu_{i*} ,\sigma_{*}^{2} ) \)
Then
Given the symmetry of \( f\left( X \right) \), for \( f(L^{*} ) = f(U^{*} ) \),
Yielding
Intuitively, L* and U* in (11) are the boundaries of the central interval of the untruncated normal distribution with mean \( \mu_{i*} \) and variance \( \sigma_{*}^{2} \), since the normal distribution is symmetric. However they do not correspond to the usual \( \frac{\alpha }{2} \) and \( \left( {1 - \frac{\alpha }{2}} \right) \) percentiles of the normal distribution since the actual distribution is truncated and thus a correction is necessary for the untruncated distribution to integrate to unity.
Appendix 2: Model output for the empirical example
Table 3 gives the output for the preferred model in Smith and Wheat (2012) reestimated for a normal-half normal pooled model. See Smith and Wheat (2012) for more details on the model formulation and interpretation. We consider that the model parameter estimates are broadly in line with those from the Smith and Wheat model, which was a panel data model, but here we analyse the data as a pooled model. Importantly our conclusions regarding constant economies of scale are the same as that found in Smith and Wheat, although we no longer find economies of train density at the sample mean (see Smith and Wheat (2012) for details of computation in this context). The average point efficiency scores \( \left( {exp\left( { - E\left[ {u_{i} |\varepsilon_{i} } \right]} \right)} \right) \) are 0.90 for the panel model and 0.91 for the pooled model, although the correlation between the scores is only 0.6 which is not surprising given the added structure imposed to efficiency variation in the panel model. Overall, while a pooled model is not our preferred model for modelling TOC costs, we consider that it is a reasonably credible alternative for the illustrative purpose of this paper.
Rights and permissions
About this article
Cite this article
Wheat, P., Greene, W. & Smith, A. Understanding prediction intervals for firm specific inefficiency scores from parametric stochastic frontier models. J Prod Anal 42, 55–65 (2014). https://doi.org/10.1007/s11123-013-0346-y
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11123-013-0346-y