One-Stage DEA introduced

In the follow-up study of the StoNEZD paper in the Journal of Productivity Analysis, Andrew Johnson and Timo Kuosmanen examine the properties of Two-Stage DEA (2-DEA) and develop a new One-Stage DEA (1-DEA) approach. The 1-DEA paper is forthcoming in European Journal of Operational Research, and it is available online since 24 Jan 2012. The permanent link to the 1-DEA paper in EJOR is the following:

http://dx.doi.org/10.1016/j.ejor.2012.01.023

1-DEA can be seen as a truncated version of the StoNEZD approach, which facilitates the joint estimation of the axiomatic, DEA-style piecewise linear frontier and a linear regression function involving z-variables that influence productivity (but are not inputs as such). In addition to the development of 1-DEA, the authors show that 2-DEA is consistent even if z-variables are correlated with inputs, and examine the sources of bias in 2-DEA estimation. Extensive Monte Carlo simulations show that the 1-DEA method systematically outperforms 2-DEA in precision, even when the truncation point is misspecified.

Energy Economics publishes StoNED application

In a forthcoming Energy Economics article, Maethee Mekaroonreung and Andrew Johnson apply StoNED to estimate the shadow prices of SO2 and NOx emissions from U.S. coal-fired power plants. To model undersirable outputs, the authors employ the axiom of weak disposability, adapting the weakly disposable technology by Kuosmanen (2005) to the stochastic framework. Applying the StoNED method with a composite disturbance term, the authors obtain the average shadow prices estimates of SO2 are between 201 and 343 $/ton and average shadow prices of NOx are between 409 and 1,352 $/ton. These shadow price estimates were in reasonable ranges compared to the EPA’s allowance prices, which suggests that using the proposed method to estimate the
shadow prices of other pollutants will help regulators and the power industry establish references for marginal abatement costs. The authors recommend the use of weak disposability StoNED method over weak disposability DEA, which is likely to overestimate shadow prices due to extreme observations.

New redesigned StoNED website launched

The style of the StoNED website has been modernized, in connection to the development of the new websites for the NOMEPRE network and the EWEPA”13 conference.  Our efforts to update the content of the StoNED homepage will continue, and this might cause some technical problems in the coming months. We hope the redesigned website will serve our readership better, and we apologize for any inconvenience due to the development and maintenance efforts.

Helsinki hosts the EWEPA conferences in 2013 and 2015

The scientific committee of the European Workshop on Efficiency and Productivity Analysis EWEPA has accepted the bid by TIMO KUOSMANEN and Antti Saastamoinen to organize the next EWEPA conferences in 2013 and 2015 in Helsinki, Finland. The next EWEPA"13 conference will take place in June 2013 at the AALTO UNIVERSITY SCHOOL OF ECONOMICS (former Helsinki School of Economics). The authoritative EWEPA Conference and its companion North American Productivity Workshop (NAPW) bring together academics and practitioners in the field of efficiency and productivity. Hosted for the first time in Finland, the EWEPA"13 will combine state-of-the-art research, innovative use of information technology, and attractive social events. Further details about the EWEPA"13 will be provided at the EWEPA website.

StoNEZD paper published online

Article by ANDREW JOHNSON and TIMO KUOSMANEN that extends the StoNED method to deal with z-variables that characterize operating environment of firms (referred to as StoNEZD) has been published online at the website of JOURNAL OF PRODUCTIVITY ANALYSIS. The permanent DOI reference of the article is: 10.1007/S11123-011-0231-5.

Journal of Productivity Analysis publishes the z-variable extension StoNEZD

Understanding the effects of operational conditions and practices on productive efficiency can provide valuable economic and managerial insights. Recent article by ANDREW JOHNSON and TIMO KUOSMANEN extends the StoNED method to deal with z-variables that characterize operating environment of firms. The extended method is referred to as "stochastic semi-nonparametric envelopment of z-variables data (StoNEZD)." The StoNEZD article has been accepted for publication in the JOURNAL OF PRODUCTIVITY ANALYSIS. It is expected to appear in 2012, Volume 36 Issue 2.

The proposed StoNEZD method has several advantages for estimating the effects of contextual variables. The article by Johnson and Kuosmanen shows that the StoNEZD estimator is statistically consistent under more general assumptions than those required by the conventional two-stage semiparametric DEA approaches. In particular, the StoNEZD estimator remains consistent even when the noise term is unbounded and the contextual variables are correlated with inputs. Importantly, Johnson and Kuosmanen show that the StoNEZD estimator for the contextual variables is asymptotically efficient, asymptotically normally distributed, and converges at the standard parametric rate. These are the key properties to ensure that the conventional methods of statistical inference (such as t-tests for statistical significance and the confidence intervals) are valid for asymptotic inference, despite the nonparametric specification of the frontier and the presence of an asymmetric, non-normal disturbance term.

The article also reports evidence from Monte Carlo simulations. The results show that the StoNEZD estimator yields systematically more precise estimates than the conventional two-stage DEA method, whether noise is present or not. The Monte Carlo evidence provides strong support for using the one-stage StoNEZD method for joint estimation of the frontier and the contextual variables.

The article by Johnson and Kuosmanen was accepted to the special symposium issue of JOURNAL OF PRODUCTIVITY ANALYSIS consisting of the keynotes of the North American Productivity Workshop (NAPW) in Houston, 2-5 June 2010. The symposium issue is edited by Knox Lovell and Robin Sickles. The keynote speakers include William Cooper, Erwin Diewert, Dale Jorgenson, Peter Schmidt, Paul Wilson, William Greene, and Timo Kuosmanen.

StoNED article ranks #1 in the most downloaded from JPA list

Since its online publication in 6 Dec 2010, the STONED ARTICLE by TIMO KUOSMANEN and MIKA KORTELAINEN was downloaded 146 times in the first month and 451 times in the first 3 months. The download statistics of the JOURNAL OF PRODUCTIVITY ANALYSIS are available on the journal website. In the latest "most downloaded article of the journal" rankings, the StoNED paper ranks #1 on both 30 days and 90 days list, and appears consistently on the top-5 list of the past 7 days.

The online article is available with open access, free of charge, from the following permanent link: HTTP://DX.DOI.ORG/10.1007/S11123-010-0201-3.

Finnish Energy Market Authority adopts StoNED

The Finnish Energy Market Authority (ENERGIAMARKKINAVIRASTO EMV) has decided to apply StoNED as the frontier estimation method within the regulatory model of the electricity distribution companies in Finland. The StoNED method is adopted for the 3rd regulation period in years 2012-2015. EMV uses the StoNED method for estimating the efficient cost frontier for the electricity distribution sector. Based on the efficient cost level, EMV sets firm-specific efficiency improvement targets and defines the acceptable annual cost levels for each firm within the regulation period.

The guidelines of the regulatory model to be implemented in the 3rd regulation period were published in 14 January 2011 (available in Finnish HERE). The energy companies and other interested parties have an opportunity to provide feedback by 11 March 2011. Based on this feedback, EMV may conduct further investigations to further improve and finalize the regulatory model. EMV will specify the final model, the efficiency imrovement targets, and the acceptable cost levels for the year 2012 during the second half of 2011.

The first StoNED paper published online

6 December 2010

The first StoNED article by TIMO KUOSMANEN and MIKA KORTELAINEN that introduces the new method has been published online at the website of JOURNAL OF PRODUCTIVITY ANALYSIS. The permanent DOI reference of the article is: 10.1007/S11123-010-0201-3.

The article is published in the Springer's Open Choice program, which means the article is available with open access, free of charge, anytime, and anywhere in the world.

Journal of Productivity Analysis accepts the first StoNED paper

Article by TIMO KUOSMANEN and MIKA KORTELAINEN that introduces the StoNED method has been accepted for publication in the JOURNAL OF PRODUCTIVITY ANALYSIS. It is expected to appear in 2011, Volume 35 Issue 2.

JOURNAL OF PRODUCTIVITY ANALYSIS is the leading specialized field journal in the area of productivity and efficiency analysis. The journal’s broad scope spans economics, management sciences, operations research, and business and public administration. A partial list of topics includes productivity theory, organizational design, index number theory, and related foundations of productivity analysis. The journal also publishes research on computational methods employed in productivity analysis and empirical research based on data at all levels of aggregation.

The original title of the paper was changed in order to address the critique from the anonymous reviewers. Editor Paul Wilson writes “your estimation idea is not fully non-parametric, since specific distributional assumptions are required. Consequently, the method is semi-parametric. This is fine, but call it what it is.” As a response, Kuosmanen and Kortelainen coined the new name “Stochastic Non-smooth Envelopment of Data”, which preserves the original acronym StoNED. Their main priority was to introduce StoNED as the new brand-name in the business of frontier estimation. The exact meaning of the "N" in the acronym can be left for the reader: one can use non-smooth, nonparametric or semi-nonparametric, whichever term is most appropriate, preferred or useful.

Regarding the classification of the StoNED method, the term semi-nonparametric seems the most appropriate in most cases. Quoting Chen (2007): “Large Sample Sieve Estimation of Semi-Nonparametric Models”, Handbook of Econometrics, Volume 6B Chapter 76

“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.”

The acronym StoNED has also caused some resentment since the publication of the original working papers. According to Kuosmanen and Kortelainen, the readership of their work is mature enough to take the acronym as it is. They emphasize that the acronym does not convey any reference to the slang term for cannabis intoxication or any other inappropriate hidden connotations (see WIKIPEDIA for the disambiguation of the term STONED).