Top

Published in:

2015 | OriginalPaper | Chapter

3. An Introduction to CNLS and StoNED Methods for Efficiency Analysis: Economic Insights and Computational Aspects

Authors : Andrew L. Johnson, Timo Kuosmanen

Published in: Benchmarking for Performance Evaluation

Publisher: Springer India

Activate our intelligent search to find suitable subject content or patents.

search-config

AI-assisted search

Off

Abstract

This chapter describes the economic insights of the unifying framework known as Stochastic semi-Nonparametric Envelopment of Data (StoNED), which combines the virtues of the widely used neoclassic production models, Data Envelopment Analysis (DEA), and Stochastic Frontier Analysis (SFA). Like DEA, StoNED is able to estimate an axiomatic production function relaxing the functional form specification required in most implementations of SFA. However, StoNED is also consistent with the econometric models of noise, providing a distinct advantage over standard DEA models. Further, StoNED allows for the possibility that systematic inefficiency is negligible consistent with neoclassical theory, thus providing a unifying framework. StoNED is implemented by estimating a conditional mean using convex nonparametric least squares (CNLS) followed by using standard SFA techniques to estimate the average efficiency and decompose the residual. Detailed descriptions of General Algebraic Modeling System (GAMS) and matrix laboratory (MATLAB) code will aid readers in implementing the StoNED estimator.

Dont have a licence yet? Then find out more about our products and how to get one now:

Springer Professional "Wirtschaft+Technik"

Online-Abonnement

Mit Springer Professional "Wirtschaft+Technik" erhalten Sie Zugriff auf:

über 102.000 Bücher
über 537 Zeitschriften

aus folgenden Fachgebieten:

Automobil + Motoren
Bauwesen + Immobilien
Business IT + Informatik
Elektrotechnik + Elektronik
Energie + Nachhaltigkeit
Finance + Banking
Management + Führung
Marketing + Vertrieb
Maschinenbau + Werkstoffe
Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

inform now

Springer Professional "Wirtschaft"

Online-Abonnement

Mit Springer Professional "Wirtschaft" erhalten Sie Zugriff auf:

über 67.000 Bücher
über 340 Zeitschriften

aus folgenden Fachgebieten:

Bauwesen + Immobilien
Business IT + Informatik
Finance + Banking
Management + Führung
Marketing + Vertrieb
Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

inform now

previous chapter Data Envelopment Analysis for Performance Evaluation: A Child’s Guide

next chapter Dynamic Efficiency Measurement

Available only for authorised users

We have found CVX, an additional toolbox that must be downloaded separately, for MATLAB performs well. Also our experience is, CPlex, Minos, XA are solvers for GAMS that perform well. However, because the computational optimization algorithms differ between software, often slight differences in the results exist for both QP and NLP problems.

For extensions to the general multi-input multi-output setting, see Kuosmanen et al. (2014).

See Sect. 2.3.2 of the Chapter by Ray and Chen in this book for a more detailed description of the assumptions regarding the production possibility set.

Modeling heteroskedastic inefficiency and noise is discussed in Kuosmanen et al. (2014), Sect. 8.

Our discussion centers on estimators based on ordinary least squares. The attempts of Banker and Maindiratta (1992) to combine axiomatic estimation with standard models of noise in a maximum likelihood framework should also be recognized. However, to the best of our knowledge no applications of this maximum likelihood approach exist do to computational challenges.

We follow the terminology of Chen (2007), who provides the following intuitive definition: “An econometric model is termed ‘parametric’ if all of its parameters are in finite dimensional parameter spaces; a model is ‘nonparametric’ if all of its parameters are in infinite-dimensional parameter spaces; a model is ‘semiparametric’ if its parameters of interests are in finite-dimensional spaces but its nuisance parameters are in infinite-dimensional spaces; a model is ‘semi-nonparametric’ if it contains both finite-dimensional and infinite-dimensional unknown parameters of interests” Chen (2007, p 5552, footnote 1).

The Finnish Energy Market Authority measures CAPEX as the replacement value of the capital stock owned by the distributor depreciated by a constant depreciation rate. Thus, CAPEX is directly proportional to the total capital stock.

The only distinction between parameters and variables in GAMS is variables are determined as the results of an optimization problem, whereas parameters are assigned values via calculations or assign statements.

When entering data, be sure to use good practices regarding significant figures. If you include data with many significant figures, this will increase computational time significantly.

Note the path should be adjusted to point to the location where the data file is saved.

The parallel literature of isotonic regression (Ayer et al. 1955; Brunk 1955; Barlow et al. 1972) considers estimation of monotonic increasing or decreasing curves without imposing concavity or convexity. Keshvari and Kuosmanen (2013) introduced isotonic regression to efficiency analysis.

From this point forward, we will refer to convex regression, recognizing that concave regression can be achieved through reversing an inequality, discussed in Sect. 3.3.2.

In a power curve or s-shape single-input production function, the inflection point in the input value at which the second derivative changes sign or in other words where the production function changes from being a convex function to a concave function.

www.cvxr.com.

Note in our notation, \({\varvec{\upbeta}}_{i}^{\prime } {\mathbf{x}}_{i} = \beta_{i1} x_{i1} + \beta_{i2} x_{i2} + \cdots + \beta_{im} x_{im} .\) Further, this formulation is intended to show the relationship to other mathematical models, i.e., classic OLS regression and the Afriat inequalities. For computational purposes, the problem may be reformulated to reduce the number of variables and/or constraints as discussed in Sect. 3.4.1.

For those familiar with DEA, the parameters \(\alpha_{i}\) and \({\varvec{\upbeta}}_{i}\) are analogous to \(u_{0}\) and \({\mathbf{u}}\) in the multiplier formulation of DEA.

The linear program used to calculate the lower bound function \(\hat{f}_{\hbox{min} }^{{\text{CNLS}}}\) is equivalent to the DEA estimator under the assumption of variables returns to scale and replacing the observed output levels with the estimated output level \(\hat{f}^{{\text{CNLS}}} ({\mathbf{x}}_{i} )\) coming from (3.5).

Our experiments with GAMS were performed on a personal computer with an Intel Core i7 CPU 1.60 GHz and 8-GB RAM. The optimization problems were solved in GAMS 23.3 using the CPLEX 12.0 Quadratically Constrained Program (QCP) solver. Our experiments with MATLAB were performed on a laptop computer with an Intel Core i5 CPU 2.50 GHz and 4-GB RAM.

From this point forward, we refer to only Eq. (3.5), but issues regarding (3.5) apply equally to (3.7) below.

Approximation algorithms are also possible strategies, but we focus on calculating the exact solution to the CNLS formulation.

Lee et al. found that if there were more than 100 observations, the group strategy for adding constraints was always preferred to other methods tested and that the sweet spot strategy’s threshold value could be adjusted based on the number of observations and the dimensionality of the data. In the experiments of Lee et al., they generate input data uniformly and do not correlate the inputs. However, when input variables are correlated CNLS becomes easier to solve. Thus, in observed data where the inputs are typically highly correlated, the computational improvement will allow problems even larger than 1,000 observations to be solved.

Setting \(\delta_{i}\) to zero implies that the set V is empty and (3.8a, b, c) is solved with only the (3.8a) and (3.8c) constraints. V still grows via the addition of violated constraints in the algorithm.

Some preliminary test indicates that XA is very effective for solving CNLS problems.

For a more extensive summary, see Kuosmanen et al. (2014).

Also called “the no free lunch” axiom, it states that the production of positive output is impossible without the use of at least one input.

They limit their computational time to 5 h and use a GAMS/CPlex implementation.

Of course, CNLS (3.5) and (3.12) differ from DEA in that these methods account for noise; Sect. 3.6.2 describes the equivalence of CNLS and DEA under the deterministic assumption.

The DEA literature defines nonincreasing returns to scale and nondecreasing returns to scale production functions. Within CNLS, similar production functions can be estimated by imposing restrictions on the coefficients \(\alpha_{i}\)

Nonincreasing returns to scale (NIRS): impose \(\alpha_{i} \ge 0\,\forall \,i\)
Nondecreasing returns to scale (NDRS): impose \(\alpha_{i} \le 0\,\forall \,i\)

Here, we construct a vector call it r, such that \(r_{i} = \ln y_{i} - \ln (\hat{\phi }_{i} )\), then r is regressed on z without an intercept term.

We do not advocate this solution to limited data. We see deterministic estimators as useful when the deterministic assumption is likely to hold.

In (3.18), since all of the \(\varepsilon_{i}^{\text{CNLS - }}\) are nonpositive, squaring the objective is simply a monotonic transformation, and thus, it is not necessary.

However, Stochastic semi-Nonparametric Envelopment of Data (StoNED) can be used.

The average value, \(\mu\), is typically a function of the parameters of the distribution of u. For example, if u is distributed half-normally, then \(E(u) = \sqrt {2/2\pi } \sigma_{u}\) where \(\sigma_{u}\) is the pretruncated standard deviation of u. More discussion related to this point is provided in Sect. 3.7.2.

Equations 3.24, 3.25, and 3.26 are shown as separate equations for ease of reading.

Afriat, S.N. 1967. The construction of a utility function from expenditure data. International Economic Review 8: 67–77.CrossRef

Afriat, S.N. 1972. Efficiency estimation of production functions. International Economic Review 13(3): 568–598.CrossRef

Aigner, D., and S. Chu. 1968. On estimating the industry production function. American Economic Review 58: 826–839.

Aigner, D., C.A.K. Lovell, and P. Schmidt. 1977. Formulation and estimation of stochastic frontier production function models. Journal of Econometrics 6: 21–37.CrossRef

Almanidis, P., and R.C. Sickles. 2011. The skewness issue in stochastic frontier models: Fact or fiction? In Exploring research frontiers in contemporary statistics and econometrics, ed. I. van Keilegom, and P.W. Wilson. Berlin: Springer.

Alminidis, P., Qian, J., and Sickles, R. 2009. Stochastic Frontiers with bounded inefficiency. MIMEO, Rice University.

Ayer, M., H.D. Brunk, G.M. Ewing, W.T. Reid, and E. Silverman. 1955. An empirical distribution function for sampling with incomplete information. Annals of Mathematical Statistics 26: 641–647.CrossRef

Banker, R.D., and A. Maindiratta. 1992. Maximum likelihood estimation of monotone increasing and concave production frontiers. Journal of Productivity Analysis 3:401–416.

Banker, R.D. 1993. Maximum likelihood, consistency and data envelopment analysis: A statistical foundation. Management Science 39: 1265–1273.CrossRef

Banker, R.D., and R. Morey. 1986. Efficiency analysis for exogenously fixed inputs and outputs. Operations Research 34(4): 513–521.CrossRef

Banker, R.D., A. Charnes, and W.W. Cooper. 1984. Some models for estimating technical and scale inefficiencies in data envelopment analysis. Management Science 30(9): 1078–1092.CrossRef

Barlow, R.E., D.J. Bartholomew, J.M. Bremner, and H.D. Brunk. 1972. Statistical inference under order restrictions; the theory and application of isotonic regression. New York: Wiley.

Battese, G.E., and T.J. Coelli. 1995. A model for technical inefficiency effects in a stochastic frontier production function for panel data. Empirical Economics 20(2): 325–332.CrossRef

Bogetoft, P., and L. Otto. 2011. Benchmarking with DEA. Springer, New York: SFA and R.

Brunk, H.D. 1955. Maximum likelihood estimates of monotone parameters. Annals of Mathematical Statistics 26: 607–616.CrossRef

Carree, M.A. 2002. Technological inefficiency and the skewness of the error component in stochastic frontier analysis. Economics Letters 77: 101–107.CrossRef

Charnes, A., W.W. Cooper, and E. Rhodes. 1978. Measuring the efficiency of decision making units. European Journal of Operational Research 2: 429–444.CrossRef

Chen, X. (2007). Large sample sieve estimation of semi-nonparametric models. In: Handbook of Econometrics, eds. J. Heckman and E. Leamer , vol. 6. North Holland.

Cobb, C.W., and P.H. Douglas. 1928. A theory of production. American Economic Review 18: 139–165.

Coelli, T. 1995. Estimators and hypothesis tests for a stochastic frontier function: A Monte Carlo analysis. Journal of Productivity Analysis 6: 247–268.CrossRef

D’Agostino, R., and Pearson, E.S. (1973). Tests for departure from normality, empirical results for the distributions of b2 and \(\sqrt {b_{1} }\). Biometrika 60(3):613–622.

Diewert, W.E., and T.J. Wales. 1987. Flexible functional forms and global curvature conditions. Econometrica 55(1): 43–68.CrossRef

Du, P., C.F. Parmeter, and J.S. Racine. 2013. Nonparametric kernel regression with multiple predictors and multiple shape constraints. Statistica Sinica 23(3): 1347–1371.

Fan, Y., Q. Li, and A. Weersink. 1996. Semiparametric estimation of stochastic production frontier models. Journal of Business and Economic Statistics 14: 460–468.

Farrell, M.J. 1957. The measurement of productive efficiency. Journal of the Royal Statistical Society Series A120: 253–281.CrossRef

Gabrielsen, A. (1975). On estimating efficient production functions. Working paper no. A-85, Chr. Michelsen Institute, Department of Humanities and Social Sciences, Bergen, Norway.

Greene, W.H. 1980. Maximum likelihood estimation of econometric frontier functions. Journal of Econometrics 13: 26–57.

Greene, W.H. 2008. The econometric approach to efficiency analysis. In The Measurement of productive efficiency and productivity growth, ed. H.O. Fried, C.A.K. Lovell, and S.S. Schmidt, 92–250. New York: Oxford University Press Inc.CrossRef

Groeneboom, P., G. Jongbloed, and J.A. Wellner. 2001a. A canonical process for estimation of convex functions: the “invelope” of integrated Brownian motion +t^4. Annals of Statistics 29(6): 1653–1698.CrossRef

Groeneboom, P., G. Jongbloed, and J.A. Wellner. 2001b. Estimation of a convex function: Characterizations and asymptotic theory. Annals of Statistics 29(6): 1620–1652.CrossRef

Hackman, S.T. 2008. Production economics: Integrating the microeconomic and engineering perspectives. Heidelberg: Springer.

Hannah, L.A., and D.B. Dunson. 2013. Multivariate convex regression with adaptive partitioning. Journal of Machine Learning Research 14: 3207–3240.

Hanson, D.L., and G. Pledger. 1976. Consistency in concave regression. Annals of Statistics 4(6): 1038–1050.CrossRef

Hildreth, C. 1954. Point estimates of ordinates of concave functions. Journal of the American Statistical Association 49: 598–619.CrossRef

Johnson, A.L., and T. Kuosmanen. 2011. One-stage estimation of the effects of operational conditions and practices on productive performance: Asymptotically normal and efficient, root-n consistent StoNEZD method. Journal of Productivity Analysis 36(2): 219–230.CrossRef

Jondrow, J., Lovell, C.A.K., Materov, I.S., and Schmidt, P. (1982). On estimation of technical inefficiency in the stochastic frontier production function model. Journal of Econometrics 19:233–238.

Keshvari, A. (2014). An enhanced Fourier-Motzkin method for data envelopment analysis, Working paper.

Keshvari, A., and T. Kuosmanen. 2013. Stochastic non-convex envelopment of data: Applying isotonic regression to frontier estimation. European Journal of Operational Research 231: 481–491.CrossRef

Kneip, A., and L. Simar. 1996. A general framework for frontier estimation with panel data. Journal of Productivity Analysis 7: 187–212.CrossRef

Kumbhakar, S.C., and C.A.K. Lovell. 2000. Stochastic frontier analysis. New York: Cambridge University Press.CrossRef

Kumbhakar, S.C., S. Ghosh, and J.T. McGuckin. 1991. A generalized production frontier approach for estimating determinants of inefficiency in U.S. dairy farms. Journal of Business and Economic Statistics 9(3): 279–286.

Kumbhakar, C.S., B.U. Park, L. Simar, and E.G. Tsionas. 2007. Nonparametric stochastic frontiers: A local likelihood approach. Journal of Econometrics 137: 1–27.CrossRef

Kuosmanen, T. 2008. Representation theorem for convex nonparametric least squares. Econometrics Journal 11: 308–325.CrossRef

Kuosmanen, T. 2012. Stochastic semi-nonparametric frontier estimation of electricity distribution networks: Application of the StoNED method in the Finnish regulatory model. Energy Economics 34: 2189–2199.CrossRef

Kuosmanen, T., and M. Fosgerau. 2009. Neoclassical versus frontier production models? Testing for the skewness of regression residuals. Scandinavian Journal of Economics 111(2): 351–367.CrossRef

Kuosmanen, T., and A.L. Johnson. 2010. Data envelopment analysis as nonparametric least-squares regression. Operations Research 58: 149–160.CrossRef

Kuosmanen, T., and M. Kortelainen. 2012. Stochastic non-smooth envelopment of data: Semi-parametric frontier estimation subject to shape constraints. Journal of Productivity Analysis 38(1): 11–28.CrossRef

Kuosmanen, T., Johnson, A.L., and Saastamoinen, A. 2014. Stochastic nonparametric approach to efficiency analysis: A unified framework. In: Handbook on Data Envelopment Analysis, eds. J. Zhu, vol. II. Berlin: Springer.

Lee, C.-Y., A.L. Johnson, E. Moreno-Centeno, and T. Kuosmanen. 2013. A more efficient algorithm for convex nonparametric least squares. European Journal of Operational Research 227(2): 391–400.CrossRef

Meeusen, W., and J. Vandenbroeck. 1977. Efficiency estimation from cobb-douglas production functions with composed error. International Economic Review 18(2): 435–445.CrossRef

Ondrich, J., and J. Ruggiero. 2001. Efficiency measurement in the stochastic frontier model. European Journal of Operational Research 129: 434–442.CrossRef

Olesen, O.B., and J. Ruggiero. 2014. Maintaining the regular ultra passum law in data envelopment analysis. European Journal of Operational Research 235: 798–809.CrossRef

Pitt, M.M., and L.F. Lee. 1981. Measurement and sources of technical inefficiency in the indonesian weaving industry. Journal of Development Economics 9: 43–64.CrossRef

Ray, S.C. 1988. Data envelopment analysis nondiscretionary inputs and efficiency: An alternative interpretation. Socio-Economic Planning Science 22: 167–176.CrossRef

Ray, S.C. 1991. Resource-use efficiency in public schools: A study of Connecticut data. Management Science 37: 1620–1628.CrossRef

Ray, S. 2004. Data envelopment analysis: theory and techniques for economics and operations research. Cambridge: Cambridge University Press.

Reifschneider, D., and R. Stevenson. 1991. Systematic departures from the frontier: A framework for the analysis of firm inefficiency. International Economic Review 32: 715–723.CrossRef

Robinson, M.P. 1988. Root-n-consistent semiparametric regression. Econometrica 56: 931–954.CrossRef

Schmidt, P. 1985. Frontier production functions. Econometric Reviews 4(2): 289–328.CrossRef

Schmidt, P., and T. Lin. 1984. Simple tests of alternative specifications in stochastic frontier models. Journal of Econometrics 24: 349–361.CrossRef

Simar, L., and P.W. Wilson. 1998. Sensitivity analysis of efficiency scores: How to bootstrap in nonparametric frontier models. Management Science 44(1): 49–61.CrossRef

Simar, L., and P.W. Wilson. 2000. A general methodology for bootstrapping in non-parametric frontier models. Journal of Applied Statistics 27(6): 779–802.CrossRef

Simar, L., and P.W. Wilson. 2010. Inferences from cross-sectional, stochastic frontier models. Econometric Reviews 29(1): 62–98.CrossRef

Stevenson, R.E. 1980. Likelihood functions for generalized stochastic frontier estimation. Journal of Econometrics 13(1): 57–66.CrossRef

Timmer, C.P. 1971. Using a probabilistic frontier production function to measure technical efficiency. Journal of Political Economy 79: 767–794.CrossRef

Varian, H.R. 1984. The nonparametric approach to production analysis. Econometrica 52: 579–598.CrossRef

Wang, H., and P. Schmidt. 2002. One step and two step estimation of the effects of exogenous variables on technical efficiency levels. Journal of Productivity Analysis 18: 129–144.CrossRef

Winsten, C.B. 1957. Discussion on Mr. Farrell’s paper. Journal of the Royal Statistical Society Series A 120(3): 282–284.

Title: An Introduction to CNLS and StoNED Methods for Efficiency Analysis: Economic Insights and Computational Aspects
Authors: Andrew L. Johnson
Timo Kuosmanen
Publisher: Springer India
Book: Benchmarking for Performance Evaluation
Print ISBN: 978-81-322-2252-1

Electronic ISBN: 978-81-322-2253-8

Copyright Year: 2015
DOI: https://doi.org/10.1007/978-81-322-2253-8_3

Springer Professional

Abstract

Please log in to get access to your license.

Dont have a licence yet? Then find out more about our products and how to get one now:

Springer Professional "Wirtschaft+Technik"

Springer Professional "Wirtschaft"

Premium Partner