Revisiting the Spatial Autoregressive Exponential Model for Counts and Other Nonnegative Variables, with Application to the Knowledge Production Function
Round 1
Reviewer 1 Report
My impression from reading this paper is absolutely positive. The derivation of the proposed model is self-contained and intiutive (at least for the readership familiar with spatial econometrics at the handbook level; but I asssume this is the target group of the Journal). The literature review is focused on highly relevant papers, well-written, and the structure of the article including (i) (sketch) derivation, (ii) Monte Carlo investigation and (iii) empirical illustration - very convincing.
Notwithstanding the generally positive impression, I suggest to improve the paper in 3 aspects:
- Around line 155, the Authors state: "We propose the following two-step PPML procedure...", whereby on the next page, around line 166, the Authors' proposition is labelled as one-step, while - at the same time - the proposition of Lambert, Brown and Florax [12] is labelled as two step. This sounds like an inconsistency to be resolved.
- In the Monte Carlo study, the explanatory variable X1 is generated as i.i.d. across space. Bearing in mind the similar performance studies of spatial estimators, I'm curious to see a sensitivity check of the obtained results against the degree of internal spatial autocorrelation in X1 (since this may trigger some trade-off in bias between beta1 and rho).
- I suggest some language and editorial double-check. E.g. in the abstract, instead of writing "a specification of the model that is invertible", one could say "an invertible model specification"; or, the first sentence in Subsection 2.2 seems to be missing the subject. These are just examples.
My overall recommendation is positive, after these three minor improvements.
Author Response
We are very grateful for your comments that provided valuable insights to improve our paper. We detail the revision we made based on your comments in the attached file.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Dear Authors,
I proposed your paper to the minor revision. There are some shortcomings that, in my opinion, could be quickly improved. Please find my remarks enclosed in separated file.
Comments for author File: Comments.pdf
Author Response
We are very grateful for your comments that provided valuable insights to improve our paper. Please, see the explanation point-by-point of the details in the revisions in the manuscript and our responses to your comments.
Author Response File: Author Response.pdf
Reviewer 3 Report
Minor problems
The readers of the manuscript aren’t supposed to read all the papers referred to in text, the authors must include in their own manuscript all the information (a conclusion, a synthesis, state of the art) the readers need to hold in order to better understand and appreciate it: see [5], see [8], see [15] ?! the idea of [12] – what’s the idea of [12]? do I have to read first the paper of [5] / [8] / [15] / [12] ?! etc. All these situations must be reformulated.
The authors need to use the non-personal addressing instead of personal addressing (we ...).
The authors need to explain first PPML method before to apply it.
The authors must include an explanation for NUTS (Nomenclature of Territorial Units for Statistics), NUTS 1, NUTS 2, NUTS 3.
OLS regression is an international well-known standard that the readers must know about?
Major problems
The authors propose a new model for the spatial autoregressive exponential regression and try to prove its usefulness by analyzing an empirical example estimating a knowledge production function for the NUTS II European regions. Although it is introduced as a two-step procedure, finally only the first step is different, being rather an improvement of the procedure described by [12]. Still, the most results are in line with others before (e.g. [12]) and the new performances doesn’t look to significantly. Moreover, I am not convinced that a single example can support a new procedure. Anyway, the authors must use the results of the empirical application to prove the benefits of their new model, not as main conclusions of the study (lines 424-434). So, not all conclusions correspond to the main proposed purpose.
The 32-43 references were not used in manuscript?!
Author Response
We are very grateful for your comments that provided valuable insights to improve our paper. Please, see the explanation point-by-point of the details in the revisions in the manuscript and our responses to your comments.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
The authors need to pay more attention to the paragraph formatting.
