The Impact of Fiscal Decentralization on Accountability, Economic Freedom, and Political and Civil Liberties in the Americas
Round 1
Reviewer 1 Report
This paper looks into how decentralisation contributes to the accountability and degree of freedom. A substantial literature review leads the authors to an econometric model that is tested on American states, leading to the conclusion that over
time, decentralisation contributes to more accountable government.
This is a well-written paper that looks into a relevant relationship that is essential for the understanding of how decentralised systems are set up, and how they are influenced by or contribute to democratic developments. The authors omit some essential parts
of the discussion on the enodgeneity of these variables, however, and so the reader wonders how such issues will be addressed. A major issue is indeed that maybe only in countries that have a certain degree of democracy (accountability, freedom), decentralisation
can be implemented. Hence, the causality runs the opposite way of the one discussed in this paper. I do not find the direction discussed in this paper implausible, as there is much to say for institutions creating their own dynamics, but it must be taken into
account.
The authors reveal these points only implicitly in the methodological part, where one starts to have many doubts on the econometric specification. In particular, they discuss some different methods, both a panel and a GMM estimation. Only the latter can really
account for the endogeneity problem. It is no surprise then to me that while in the panel estimation, all variables are significant, only the decentralisation indicator is in the latter. That’s still encouraging for the authors’ claim, yet there are few statistics
to asses the validity of this instrumental variabels approach. The J-test or a Stock Yogo test would have been useful. I’d drop the panel, and focus only on the GMM estimates.
Two other methodological points are not acceptable in the paper. The first one is the lack of clarity on the dataset. The dataset spans 1972-2015. There are 35 OAS members, so that potentially gives 33 times 35 data points, yet in the regressions we get at
most 158 observations. What implicit assumptions have been made on the timing of this dataset?
The second one concerns the interpretation of the non-linear terms in the regression model. It is vaguely mentioned in the methods section that this is capturing time dynamics in implementing decentralisation. While I agree with the dynamics, I do not see how
this non-linearity is related to time. It just reflects that higher values of decentralisation can lead to larger effects on the dependent. It is not a dynamic variable.
Finally, as I was reading, it looked to me the authors wanted to look into decentralisation and accountability at national level. The discussion seemed to suggest to me that decentralisation leads to different levels of accountability WITHIN a country. Arguably,
some regions have a tradition of more or less democracy at local level, leading to different degrees of accountability. The authors should be clearer that this is not their focus.
Author Response
Dear Reviewer:
Thank you for your insights. I’ve modified the document according to your recommendations.
Below you will find the responses, modifications, and clarifications to each one of your observations.
Please let me know if you need further modifications to the document.
Observation 1
The authors omit some essential parts of the discussion on the endogeneity of these variables, however, and so the reader wonders how such issues will be addressed. A major issue is indeed that maybe only in countries that have a certain degree of democracy (accountability, freedom), decentralisation can be implemented. Hence, the causality runs the opposite way of the one discussed in this paper. I do not find the direction discussed in this paper implausible, as there is much to say for institutions creating their own dynamics, but it must be taken into account.
Answer: This issue is addressed in section 2 (Review of the literature and theoretical foundations), lines 186-196, where it is stressed that even in countries with little or no democracy (i.e., Bolivia, China) decentralization can occur, hence highlighting that the direction of causality argued in the paper – from decentralization to political liberties and rights – is reasonable. Additionally, it is argued that from a methodological perspective, the issue of endogeneity underscored here can plausibly be addressed by GMM and Instrumental Variable estimations.
Observation 2
The authors reveal these points only implicitly in the methodological part, where one starts to have many doubts on the econometric specification. In particular, they discuss some different methods, both a panel and a GMM estimation. Only the latter can really account for the endogeneity problem. It is no surprise then to me that while in the panel estimation, all variables are significant, only the decentralisation indicator is in the latter. That’s still encouraging for the authors’ claim, yet there are few statistics to asses the validity of this instrumental variables approach. The J-test or a Stock Yogo test would have been useful. I’d drop the panel, and focus only on the GMM estimates.
Answer: All GLS specifications have been eliminated and tables 2-5 only report the instrumental variables (two-stage EGLS) and GMM estimations. In lines 327-336 and footnote 20 of section 3 (Data and methodology), it is argued that IV and GMM estimations are best equipped to deal with endogeneity issues, as well as cross-sectional and intrasectional correlation (in all IV and GMM regressions the Prais-Winsten Panel Corrected Standard Error Within estimator is used to correct for serial correlation).
Observation 3
Two other methodological points are not acceptable in the paper. The first one is the lack of clarity on the dataset. The dataset spans 1972-2015. There are 35 OAS members, so that potentially gives 33 times 35 data points, yet in the regressions we get at most 158 observations. What implicit assumptions have been made on the timing of this dataset?
Answer: In lines 258-269 and in footnote 8 of section 3 (Data and methodology) it is stated that of all American nations, data on fiscal decentralization is only available for 12 countries, and therefore the study focuses only on these nations. Additionally, even though the period of interest lasts from 1972 through 2015, the dates of available data for these 12 countries do not always coincide. The appendix of the article, inserted in line 586, shows that, for instance, revenue decentralization data for Bolivia is available for the period 1985-2014, but only for the period 2000-2014 for Costa Rica and 2005-2012 for Paraguay. The end result is an unbalanced panel data set for 12 American nations and a limited number of observations.
Observation 4
The second one concerns the interpretation of the non-linear terms in the regression model. It is vaguely mentioned in the methods section that this is capturing time dynamics in implementing decentralisation. While I agree with the dynamics, I do not see how this non-linearity is related to time. It just reflects that higher values of decentralisation can lead to larger effects on the dependent. It is not a dynamic variable.
Answer: In lines 318-324 of section 3 (Data and methodology) it is stated that the fiscal decentralization indicator (revenue- or expenditure-based) is deemed to have a non-linear impact on the dependent variable, indicating that the expected positive impact of the decentralization indicator may in fact be more nuanced. Additionally, in lines 334-336 of the same section, it is stated that generalized method of moments (GMM) was employed in order to analyze the dynamic nature of the relationship between decentralization and the dependent variables.
Observation 5
Finally, as I was reading, it looked to me the authors wanted to look into decentralisation and accountability at national level. The discussion seemed to suggest to me that decentralisation leads to different levels of accountability WITHIN a country. Arguably, some regions have a tradition of more or less democracy at local level, leading to different degrees of accountability. The authors should be clearer that this is not their focus.
Answer: In lines 31-36 of section 1 (Introduction) it is stated that the principal analysis is done at a national level (tables 2-5), and hence the results should not be taken to indicate that decentralization leads to different levels of accountability, economic freedom, or political and civil liberties within a country. It is also indicated that in the latter part of the paper (table 6) simulations are carried out to describe how different degrees of decentralization would impact the different dependent variables in each one of the countries analyzed in the study.
Reviewer 2 Report
This paper must be made tighter and more incisive. It must be motivated better, the introduction must more clearly present the thesis and argument, the theory section must be improved, the literature review must be organized by topic. In general, I would merge the first three sections while at the same time making them 30% shorter overall. The findings must be given some sort of explanation, however speculative. There is too much information as it stands that makes the paper less crisp than it should---it is too vague. I would recommend working on it much, much more.
Author Response
Dear Reviewer:
Thank you for your insights. I’ve modified the document according to your recommendations.
Below you will find the responses, modifications, and clarifications to each one of your observations.
Please let me know if you need further modifications to the document.
Observation
This paper must be made tighter and more incisive. It must be motivated better, the introduction must more clearly present the thesis and argument, the theory section must be improved, the literature review must be organized by topic. In general, I would merge the first three sections while at the same time making them 30% shorter overall. The findings must be given some sort of explanation, however speculative. There is too much information as it stands that makes the paper less crisp than it should---it is too vague. I would recommend working on it much, much more.
Answer: the paper has been adjusted, reorganized, and synthesized in a more coherent manner. Some of the changes made are these:
1. Section 1 presents more clearly the objective of the paper, namely that it aims to analyze the impact of fiscal decentralization on two indicators of economic freedom and two indicators of accountability and political and civil liberties in the Americas. Furthermore, it is stated that the principal analysis is done at the national level, and hence the results should not be taken to indicate that decentralization leads to different levels of accountability, economic freedom, or political and civil liberties within a country. It is also stated that in the latter part of the paper simulations are carried out to describe how different degrees of decentralization impact various indices of liberties, accountability, and economic freedom in each one of the countries analyzed in the study.
2. Sections 2 and 3 have been merged into one (Review of the literature and theoretical foundations). Additionally, the literature review has been condensed significantly and has also been organized according by topic.
3. All Generalized Least Square (GLS) specifications and corresponding estimations have been eliminated (tables 2-5) in order to focus only on the most important results (Two-stage EGLS and GMM estimates).
4. Section 5 (Conclusions and policy implications) has been shortened significantly in order to exclusively focus on the most relevant findings
The whole paper has been revised by two colleagues – an English professor and a Political Science professor – in order to correct any mistakes related to content, composition, and/or organizational formatting.
Round 2
Reviewer 1 Report
Changes are adequate.
Author Response
Dear Reviewer:
Thank you for your thoughtful insights and recommendations. I believe your suggestions have greatly improved my paper and for that I am truly thankful
Reviewer 2 Report
I would have spent much more time revising this paper.
Author Response
Dear Reviewer:
Thank you for your additional comments.
I have polished the paper further by making some format changes in section 2 (Review of the literature and theoretical foundations, lines 91, 186-187), section 4 (Results of the empirical estimation, footnote 22, lines 463, 510), and section 5 (Conclusions and policy implications, line 582).
With these additional changes I believe that the paper is now more incisive and - even though there is still a significant amount of informational data - easier to follow, just like you suggested. I'll be happy to make further changes if you feel that in any one of the particular sections of the paper specific changes should be made.
I thank you again for your time and effort in analyzing my paper and look forward to your response.
