How Effective Is Fiscal Decentralization for Inequality Reduction in Developing Countries?
Round 1
Reviewer 1 Report
The manuscript is clear, relevant for the field and presented in a well-structured manner, the gap in knowledge is identified, the results are clear. One minor suggestion, that expressions like "has a negative effect on inequality" from statistical point of view mean that dependent variable decreases when independent variable increases (i.e inequality decreases). However, from economic point of view the expression "the negative effect on inequality" can be associated with increase in inequality. I would recommend to avoid positive or negative effect and change to increase or decrease in inequality.
Author Response
We agreed with these comments and suggestions and thus, to avoid confusion, we already replaced some of the expressions of “negative” and “positive” in the text.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper tackles with an interesting research question: the relationship between fiscal decentralization and inequality. However, it adds a twist to the analysis by focusing on institutional quality and military spending.
However, I have three concerns:
- The manuscript ignores the rich literature on conflict and decentralization:
The multi-disciplinary literatures on the relationship between internal and external conflicts and decentralization are rich with theoretical formulations and empirical testing of relationships. However, both the theoretical formulations and empirical findings remain inconclusive. In terms of internal conflicts, while one group of researchers theorizes that decentralization often intensifies internal conflicts, another group suggests that decentralization can mitigate conflicts. Please see the relevant literature:
Brancati D. 2006. Decentralization: fueling the fire or dampening the flames of ethnic conflict and secessionism? International Organization 60: 651–685.
Brancati D. 2008. Peace by Design: Managing Intrastate Conflict through Decentralization. Oxford University Press: New York
Bunce V, Watts S. 2005. Managing diversity and sustaining democracy: ethnofederal versus unitary states in the postcommunist world. In Sustainable Peace: Power and Democracy After Civil Wars. Cornell University Press: Ithaca; 133–58.
Cornell SE. 2002. Autonomy as a source of conflict: Caucasian conflicts in theoretical perspective. World Politics 54(2): 245–76.
Lake DA, Rothchild D. 2005. Territorial decentralization and civil war settlements. In Sustainable Peace Power and Democracy after Civil Wars, Roeder PG, Rothchild D (eds). Cornell University Press: Ithaca.
Schou, Arild and Marit Haug. 2005. “Decentralisation in conflict and post-conflict situations.” Norwegian Institute for Urban and Regional Research Working Paper No 139.
Snyder J. 2000. From Voting to Violence: Democratization and Nationalist Conflict. Norton: New York.
Edwards, B. and Yilmaz, S. 2016. “Decentralization as a Post-Conflict Stabilization Tool: the Case od Sierra Leone” Public Administration and Development 36: 347-358.
On the other hand, the literature on external conflicts predicts that the increase in conflicts increases the propensity for internal cohesion or centralization, often referred to as the in-group/outgroup hypothesis. This hypothesis has been argued and extensively tested by:
Stein, A. 1976. “Conflict and cohesion: A review of literature.” Journal of Conflict Resolution, 20(1): 143–172.
Lewis, W. H. 1961. “Feuding and social change in Morocco.” Journal of Conflict Resolution, 5(1): 43–54.
Markides, K. C. and S. F. Cohn. 1982. “External conflict/internal cohesion: A reevaluation of an old theory.” American Sociological Review, 47(1): 88–98.
Gibler, D. M. 2010. “Outside-in: The effects of external threat on state centralization.” The Journal of Conflict Resolution Vol. 54, No. 4 (August 2010), pp. 519-542
Tosun, M. S. and Yilmaz, S. 2010. “Centralization, Decentralization and Conflict in the Middle East and North Africa” Middle East Development Journal Vol. 2, No. 1 (2010) 1–14.
- The use of GFS for fiscal decentralization variable
The GFS provides a breakdown of expenditures by function and economic type, it does not identify the degree of local expenditure autonomy, therefore it fails to represent the degree of fiscal decentralization. In GFS, local expenditures that are mandated by the central government or are spent on behalf of central government appear as subnational expenditure. On the revenue side, GFS does not distinguish the sources of tax and non-tax revenues, intergovernmental transfers, and other grants. Hence, there is no information on whether revenues are collected through shared taxes, piggybacked taxes, or locally determined own source revenues. In addition, GFS does not disclose what proportion of intergovernmental transfers is conditional as opposed to general purpose, and whether transfers are distributed according to an objective criterion or a discretionary measure.
- The names of 56 countries selected for the analysis are not included in the manuscript
There is no information about the selected countries. In general, cross-country analysis include countries that have different governance and intergovernmental systems.
Author Response
Response 1: We added some explanation in page 4.
Response 2: Yes, we agreed on that. Although the GFS has consistent definitions across countries over time, it ignores the degree of control exerted by central governments over local revenues and expenditures. In addition, it aggregates all subnational governments into a single group, overlooking the number of subnational governments within the country, their types of intergovernmental transfer, and their differences of revenue and expenditure (see Ebel & Yilmaz, 2002; Stegarescu, 2005). However, despite these weaknesses of GFS, Letelier (2005) argues that there is no evidence of systematic measurement error across countries.
Response 3: We already put the list of developing countries in the Appendix.
Author Response File: 
Author Response.pdf
Reviewer 3 Report
On the whole, I would recommend some minor changes.
The introduction could explain a little about how the study was carried out and what the main results are.
It is necessary to include a discussion section, in which the results obtained are contrasted with those obtained by other researchers in previous studies.
In the conclusions, it is necessary to comment on whether there are any limitations to the study.
Author Response
Response 1: We already revised the manuscript according to your suggestion. Please see the improvement on the introduction and conclusion, as well as the discussion.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
I think there are two issues to be resolved; one minor and one major. The minor point is about the GFS data. I think the paper should discuss the limitations of GFS dataset. The major point is about the selection of countries which includes high centralized unitary countries (Afghanistan, Bangladesh, Egypt) and highly decentralized federal countries (Argentina, Brazil, Ethiopia, India, Malaysia, Nigeria, Pakistan). I am not sure how the authors calculated subnational revenues and expenditures in these two very different categories of countries. If subnational in Argentina (or in any federal country) means everything below the federal government, then you can't take deconcentrated local governments in Egypt (Egypt doesn't have devolved local governments) as subnational governments and do the analysis in one dataset. It is like comparing apples with oranges. The least the authors can do is to run the regressions separately for unitary and federal countries.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 3
Reviewer 2 Report
On the issue of GFS, I am not sure if you can move forward if you admit that there are problems with measurement. As you cited, Stegarescu (2005) eloquently shows the shortcomings of the GFS dataset. Letelier (2005) makes the argument that there is no evidence of systematic measurement error, without providing any evidence himself. I think the evidence is provided in Ebel&Yilmaz (2003). He choses to ignore it conveniently.
In terms of your regression results, it seems you are admitting that you have problems with your RE and FE estimations with the following sentence: “Overall, we count on the results from GMM settings, instead of RE and FE estimations, since the former pass the autocorrelation and validity tests.” Then the question is: why are you still reporting them in the tables?
In your reply you admit that there are shortcoming of GFS data and in order “[t]o compensate the limitation on GFS, [you] incorporated RAI dataset…” Again, the same question: if you accept the shortcomings of GFS data, why are you still reporting Rev and Exp in the tables?
On the issue of RAI, Appendix B includes both federal and unitary countries, therefore, I would like to have more information about how the index measures regional autonomy in federal and unitary: I am worried about aggregation problem, as federal countries usually have three tiers of government versus unitary countries’ two tiers.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 4
Reviewer 2 Report
I am satisfied with the changes.
