Abstract
With a little help from a handful of scholars, the noise component of the composed error in a production model created the stochastic frontier analysis field. But after that glorious moment, it was confined to obscurity. We review what little research has been done on it. We present two cases where it torments us from the shadows, by sabotaging identification, and by distorting the sample skewness. We examine the relation between predicted noise and predicted inefficiency. For the Normal-Half Normal and the Normal-Exponential error specification, we provide its conditional expectation as predictor and we examine its distribution in relation to the marginal law. We also derive the conditional distribution of the noise and we compute confidence intervals and the probability of over-predicting it. Finally, we present a model where the noise, as the carrier of uncertainty, induces directly inefficiency. We conclude by showcasing our theoretical results through an empirical illustration.
Similar content being viewed by others
Notes
The article was accepted for publication already in January 1975.
More on this legendary birth can be found in the Peter Schmidt ET Interview, Sickles (2022).
It was Greene (1980) who showed formally that an error term with vanishing density at the extreme points of its support is a stronger and sufficient condition and restores the asymptotic properties of the MLE, even when the range of the dependent variable still depends on the unknown parameters.
In terms of poetic justice, this is how it punishes us for the neglect it suffered all these years.
This team’s review on the distributional forms used in SFA more generally (Stead et al. 2019) contains a little bit more on the noise component.
The same issue, coming from the other direction, arises in a cost stochastic frontier model.
In a personal communication, Christopher Frank Parmeter noted that the identification issue also vanishes if we fix a priori the probability value of the zero-quantile of the Asymmetric Laplace at any value, not specifically at 1/2, which is what we do for the symmetric Laplace. This is correct, but how are going to know this value? Symmetry, at least, can be argued from semi-philosophical premises, but once we allow for skewness, we have very many values to choose from. And we cannot rely on some auxiliary estimation that does not require distributional assumptions, because the estimated skewness will be the skewness of the composed error, not of the noise component.
We just remind the reader that empirical applications as well as simulations have indicated that we may face the “wrong skewness problem” in anywhere between 1/5 and 1/3 of the data samples we encounter.
The expression for \(E(v\mid \varepsilon )\) can of course be derived formally using integration and convolution. Such a derivation is included in a Technical Appendix, available from the author upon request, that contains also the other mathematical derivations of this work.
Cross-covariances would be zero even if the CEF errors were not the same variable, since the CEF error of a conditional expectation is uncorrelated with any function of the same conditioning variable, including any other conditional expectation based on that same variable.
It is straightforward to adjust these results for the cost SF model: we would find that the CEF errors are equal in magnitude but opposite in sign, and that the two predictors are positively correlated.
One can find related work in Waldman (1984).
Later we will also treat the Normal-Exponential specification.
See, e.g., Sickles and Zelenyuk (2019), pp. 373–374.
This sounds intuitive, and it also has a formal proof in Papadopoulos and Parmeter (2022).
For non-symmetric distributions, we would call the median a “neutral” predictor in the probabilistic sense, a predictor that claims “probabilistic” neutrality/balance with respect to over/under-prediction.
See, e.g., Jondrow et al. (1982), Theorem 1, p. 234.
Additional work on the distributions of the predictors can be found in Zeebari et al. (2021).
Consult Amsler and Schmidt (2021) for a full recent review on the use of Copulas in SFA.
In the Technical Appendix we derive the various formulas without the assumption of Normality of v, so they can be used with other zero-mean symmetric distributions like the Laplace.
The wage-cost data were taken from the Structural Business Statistics (SBS) data base of the European Union. The balance sheet and headcount data come from the ICAP data base, www.icap.gr. I thank professor Nikos Vettas, Director General of the Foundation for Economic and Industrial Research (IOBE, www.iobe.gr) for granting me access, and also IOBE researcher Alexandros Louka who prepared the data set. Any estimation and inference mistakes rest with the present author.
References
Aigner DJ, Lovell CAK, Schmidt P (1977) Formulation and estimation of stochastic frontier production functions. J Econom 6(1):21–37
Amsler C, Schmidt P (2021) A survey of the use of copulas in stochastic frontier models. In: Parmeter CF, Sickles R (eds) Advances in efficiency and productivity analysis. Springer, Cham, pp 125–138
Amsler C, Prokhorov A, Schmidt P (2021) A new family of copulas, with application to estimation of a production frontier system. J Prod Anal 55(1):1–14
Badunenko O, Henderson DJ (2021) Production analysis with asymmetric noise, MPRA Paper No. 110888, https://mpra.ub.uni-muenchen.de/110888
Bonanno G, De Giovanni D, Domma F (2017) The ‘wrong skewness’ problem: a re-specification of stochastic frontiers. J Prod Anal 47(1):49–64
El Mehdi R, Hafner CM (2014) Inference in stochastic frontier analysis with dependent error terms. Math Comput Simul 102:104–116
Fan Y (1994) Testing the goodness of fit of a parametric density function by kernel method. Economet Theor 10(2):316–356
Feller W (1957) An introduction to probability theory and its applications, 3rd edn. Wiley, New York
Greene WH (1980) Maximum likelihood estimation of econometric frontier functions. J Econom 13(1):27–56
Gupta AK, Nguyen N (2010) Stochastic frontier analysis with fat-tailed error models applied to WHO health data. Int J Innov Manag Inf Prod 1(1):43–48
Hadri K (1999) Estimation of a doubly heteroscedastic stochastic frontier cost function. J Bus Econ Stat 17(4):359–363
Horrace WC, Parmeter CF (2018) A Laplace stochastic frontier model. Economet Rev 37(3):260–280
Horrace WC, Schmidt P (1996) Confidence statements for efficiency estimates from stochastic frontier models. J Prod Anal 7:257–282
Jondrow J, Lovell CAK, Materov IS, Schmidt P (1982) On the estimation of technical efficiency in the stochastic frontier production function model. J Econom 19(2/3):233–238
Kumbhakar SC, Lovell CAK (2000) Stochastic Frontier analysis. Cambridge University Press, Cambridge
Makieła K, Mazur B (2022) Model uncertainty and efficiency measurement in stochastic frontier analysis with generalized errors. J Prod Anal 58:35–54
Meeusen W, van den Broeck J (1977) Efficiency estimation from Cobb-Douglas production functions with composed error. Int Econ Rev 18(2):435–444
Nguyen NB (2010) Estimation of technical efficiency in stochastic frontier analysis. PhD thesis, Bowling Green State University
Papadopoulos A (2015) The half-normal specification for the two-tier stochastic frontier model. J Prod Anal 43(2):225–230
Papadopoulos A, Parmeter CF (2020) The wrong skew problem in stochastic frontier analysis: a review and some new approaches. In: Presented at the virtual North American Productivity Workshop 2020
Papadopoulos A, Parmeter CF (2022) Quantile methods for stochastic frontier analysis. Found Trends Econom 12(1):1–120
Schmidt P (1976) On the statistical estimation of parametric frontier production functions. Rev Econ Stat 58(2):238–239
Sickles RC (2022) The ET Interview: Peter Schmidt. Econom Theory. https://doi.org/10.1017/S0266466622000299
Sickles RC, Zelenyuk V (2019) Measurement of productivity and efficiency. Cambridge University Press, Cambridge
Smith MD (2008) Stochastic frontier models with dependent error components. Economet J 11(1):172–192
Stead AD, Wheat P, Greene WH (2018) Estimating efficiency in the presence of extreme outliers: a Logistic-Half Normal stochastic frontier model with application to highway maintenance costs in England. In: Greene WH, Khalaf L, Makdissi P, Sickles RC, Veall M, Voia MC (eds) Productivity and inequality. Springer, Cham, pp 1–19
Stead AD, Wheat P, Greene WH (2019) Distributional forms in stochastic frontier analysis. In: ten Raa T, Greene WH (eds) The palgrave handbook of economic performance analysis. Springer, Cham, pp 225–274
Tancredi A (2002) Accounting for heavy tails in stochastic frontier models. Department of Statistical Sciences, University of Padua, Working Paper no. 2002.16
Tsionas MG (2020) Quantile stochastic frontiers. Eur J Oper Res 282(3):1177–1184
Waldman DM (1984) Properties of technical efficiency estimators in the stochastic frontier model. J Econom 25(3):353–364
Wang WS, Schmidt P (2009) On the distribution of estimated technical efficiency in stochastic frontier models. J Econom 148(1):36–45
Wang WS, Amsler C, Schmidt P (2011) Goodness of fit tests in stochastic frontier models. J Prod Anal 35(1):95–118
Wheat P, Stead AD, Greene WH (2019) Robust stochastic frontier analysis: a Student’s t-Half Normal model with application to highway maintenance costs in England. J Prod Anal 51(1):21–38
Zeebari Z, Månsson K, Sjölander P, Söderberg M (2021) Regularized conditional estimators of unit inefficiency in stochastic frontier analysis, with application to electricity distribution market. Ratio working paper no 345, The Ratio Institute
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
All authors declare that they have no conflicts of interest.
Ethical Approval
This article does not contain any studies with human participants or animals performed by any of the authors.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Below is the link to the electronic supplementary material.
Appendix: Identification in the asymmetric Laplace-Exponential model
Appendix: Identification in the asymmetric Laplace-Exponential model
A way to write the Asymmetric Laplace-Exponential composed density is
The correspondence with the two scale parameters from the alternative Laplace representation as a difference of two Exponentials is
A reason to use the \((\tau ,\, \sigma _v)\) parametrization is because \(\Pr (v\le 0) =\tau \), allowing for a likelihood-based estimation of a quantile probability. Using the above density, we present in Table 4 two cases of identification failure. In the left block, the true data generating process has a symmetric Laplace (case (iii) mentioned in the main text), while in the block to the right (which is the second example), the noise is Asymmetric Laplace (case (iv) mentioned in the main text). In each block, the first column with likelihood values is produced using the true \((\tau , \sigma _v,\sigma _u)\) parameter values, while the second corresponds to the parameter vector that characterize the alternative equivalent specification.
The \(\sigma _1,\,\sigma _2,\,\sigma _3\) are the true values holding in the sample, and \(\sigma _3\) relates to inefficiency. Based on these three values, for every value of \(\varepsilon \), we see that the Asymmetric Laplace-Exponential density \(g_{\varepsilon }(\varepsilon )\) takes the exact same value for two different combinations of values for the parameters \((\tau , \sigma _v, \sigma _u)\). So for any realized sample of values, the likelihood will have the same value under these two sets of estimates–but only one will be the true one (the left columns in each block of the Table). Moreover, these, identical as regards the value of the likelihood, parameter vectors represent totally different situations, e.g., in the case to the right, we may greatly overestimate the scale parameter (and so the mean) of the inefficiency (0.9 instead of the true 0.5).
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Papadopoulos, A. The noise error component in stochastic frontier analysis. Empir Econ 64, 2795–2829 (2023). https://doi.org/10.1007/s00181-022-02339-w
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00181-022-02339-w