Discussion Paper

No. 2017-70 | September 28, 2017
Normalized CES supply-side system approach: how to replicate Klump, McAdam, and Willman (Review of Economics and Statistics, 2007)
(Published in Special Issue The practice of replication)


This paper lays out a replication plan for the influential paper by Klump et al. (Factor Substitution and Factor-augmenting Technical Progress in the United States: a Normalized Supply-side System Approach, Review of Economics and Statistics, 2007) on using a normalized CES supply-side system approach to estimate the value of the elasticity of substitution between factors of production and identify the growth patterns for biased technical progress. The authors begin with a general discussion of basic principles on carrying out a replication study. Further, they outline key steps to follow to replicate the chosen paper and establish criteria that can be used to determine if the replication confirms or disconfirms the original findings. This paper contributes to the increased interest in improving replications in economics research.

JEL Classification:

O41, O47, E24


  • Downloads: 70


Cite As

Gerald Eric Daniels Jr. and Venoo Kakar (2017). Normalized CES supply-side system approach: how to replicate Klump, McAdam, and Willman (Review of Economics and Statistics, 2007). Economics Discussion Papers, No 2017-70, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2017-70

Comments and Questions

Anonymous - Referee Report 1
October 24, 2017 - 09:15

Report on
Normalized CES supply-side system approach: How to replicate
Klump, McAdam, and Willman (2007)?

In this paper the authors try to illustrate how a particular paper (KMW, 2007) could be replicated and the benefit of doing so. They discuss the importance of the paper itself in the ...[more]

... growth theory field and the mechanics of the required replication steps. In addition they touch on the general benefits of replication exercises for the profession.

Like most researchers I strongly support the increasing emphasis on replication in economics. Moreover I know the KMW paper well and how it has been received. The current paper is well-written and well-crafted and would be of interest to a number of scholars. Consequently I have no hesitation in recommendation acceptance subject to fairly minor revision points:

•The paper talks a lot about replication methods in general and explains the mechanics of KMW. I think however – largely for readers’ sakes – that it’s important to point out that the model had a life beyond this 2007 publication. I think therefore the KMW 2012 normalization survey paper should also be cited, as should the Leon-Ledesma, McAdam, Willman AER paper (2010) since they contain additional information on the framework and additional results on its robustness and characteristics.

•I think it’s also worth (if trivial) to point out that the KMW study cannot 100% be replicated unless you have exactly the same data that they used. The authors write:
" … When replicating this study, we believe that a verification approach should be used. The data should be remeasured using similar methods to verify and rectify any potential measurement errors or coding errors in the original study. Further, the sample period should be extended to the most recent data available as the result should not be sensitive to a larger sample period."
It is impossible to create original data set, because new vintages of NIPA data (also history) differ. Typically differences across vintages are small but especially regarding capital stock data differences may be larger. Hence, it is clear that it is impossible to do exact replications but this shouldn’t be a big issue if results are reasonably close to original. However, if not, there may be difficulties to clarify whether differences are real or reflect some unnoticed mistakes in recreating the data that is used in replication. Regarding the use of an expanded data horizon, this raises the question is the study any longer a replication, but an extension which may or may not produce as “good” results as KMW found. In short the authors should maybe tone that down a little bit, or nuance that message a little more.

•Equations 3 - 4 presentationally look awful. Could the authors somehow try to improve them, getting them all on one line for example?

•The Kmenta approximation presented actually is the Kmenta approximation with additionally non neutral technical progress, so it should rightly be attributed to the KMW, 07 paper as an extension of the original Kmenta approximation.

Venoo Kakar - Reply to referee report 1
November 08, 2017 - 14:16

see attached file

Anonymous - Referee Report 2
November 06, 2017 - 08:08

see attached file

Venoo Kakar - Response to referee 2's comments
November 14, 2017 - 06:32

see attached file