A Dataset for Examining the Problem of the Use of Accounting Semi-Identity-Based Models in Econometrics

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article is interesting, but the conclusions from the conducted study are missing. The article should be expanded to two new chapters: 1) Detailed results from the conducted study, 2) Detailed conclusions from the conducted study.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Dataset for examining the Problem of the Use of Accounting Semi-Identity based Models in Econometrics

 

The data are available as open access files in EXCEL and STATA and R format. They are created by the author’s standard STATA or R code and appear to fulfil all the journal requirements for Data Descriptors. These data are generated for Monte Carlo simulations of a well-known econometric model to explain investment behaviour in firms. Consequently, this review is able to focus on the article submission itself.

The problem addressed is an important and interesting issue in the econometrics of finance since many models in this field use different components of published accounting data, which are often constrained to be related by identities. Hence the term accounting semi-identities. The paper correctly explains that that such accounting identity related variables when used to form the basis of an econometric model of firm behaviour give rise to estimating equations which suffer from omitted variable bias. This is a perennial problem in financial econometric research, but researchers are often unaware of the issue.

The paper suggests that Monte-Carlo simulations of models which are theoretically well designed, without being constrained to using only data from financial accounts that are defined by accounting semi-identities, can be used as a useful check on the plausibility of e.g., OLS results for models that use the accounting data. This is an interesting idea, but the article could benefit from a more detailed explanation of why this strategy could be helpful. The article sets out to test the FHP (1998) model that specifies that a firm’s investment decisions are determined by its cash flows in the same period, with a greater impact from the firm’s own cash flow as measured by the size of the OLS regression coefficient indicating the firm is more constrained in accessing external funding the implication being that firms with higher investments would have low sensitivity to their cash flow since they could be expected to have accessed external funding more easily. FHP tested this effect by splitting their sample into four subsamples in terms of investments and cash flows.  These variables are of course related by the firm’s accounting practices and result in an accounting semi-identity.

The paper therefore generates a Monte-Carlo simulation of an extended synthetic model in which Tonin’s Q is entered as an additional variable omitted in the FHP study. The aim is to test whether simulating the synthetic model generates the FHP same investment-cash flow propensity. However, the paper at this point is too compressed in its explanation of the logic and the comparisons. In particular Tables 1 and 2 need considerably more explanation in terms of the description of the subsamples in FHP and the nature of the Monte-Carlo results and findings. In essence, these findings are that that companies in the synthetic model with higher investments exhibit greater cash-flow sensitivity  contradicting  the predictions of the FHP (1988) model estimated from real world data.

This is an interesting finding and should be of interest to researchers in the field. However, the paper needs to bring out the deep generalization which the author seems to be offering: that Monte-Carlo simulations of a synthetic model on synthetic data can be used to validate the findings of accounting semi-identity models estimated from real world data. Is this what the author is claiming. If so, it needs some additional reasoning and explanation of why it could be a general strategy instead of simply a device that works in this particular case.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

I recommend to rewrite the paper according to scientific standards, which means to make seperate parts - introduction, methodology, results, discussion and conslucion. Mainly last two parts are missing. The paper needs strong dicsussion part reflecting to results of other studies in this field. 

Add information about situation in other countries and possibilities to replicate the study.

Provide discussion section which is missing and is a necessary part of scientific papers.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The revised version has been improved by responses to the reviewers' comments.