Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Report on “Public health expenditure and sustainable outcomes in 45 Sub-Saharan African countries: Does government effectiveness matter?”

 

This paper examines how public health expenditures and government effectiveness affect health outcomes in 45 Sub-Saharan African countries. This is an important question but not easy to answer since there are two-way relationships between health outcomes and public health expenditures, and there are many “third factors” that are correlated with these three variables. I have the following comments on the paper.

1. Introduction

The authors have not laid out the framework for why they decided to examine how public health expenditure and government effectiveness affects health outcomes, and how those questions stand in the health economics theories. It is important to understand it from the theoretical point of view because of three reasons. First, many factors that affect health outcomes are correlated with public health expenditure and government effectiveness, and that will cause an estimation bias when estimating the effects of public health expenditure and government effectiveness on health outcomes. For example, higher income and education are two of the most important factors that improve health outcomes. If a government experiences a better economic growth period and decides to invest more in public education and public health at the same time. Not taking into account higher spending in education or policies that increase the population’s education will lead to a significant overestimation of the effects of public health spending. Second, health outcomes and public health spending have a two-way relationship. Public health spending can lead to better health outcomes, but better health outcomes can also lower public health spending. And third, public health spending is not total medical spending. The total medical spending is the sum of public health spending and private health spending, Usually, when the public health spending (including medical spending covered by the government) increases, the private health spending decreases, and vice versa. In the introduction, you should discuss those channels and how your methods navigate the challenges raised by them.

2. Literature Review

The authors provide a detailed summary of the literature in the text, and the current literature review accounts for over 3 pages of a less than 10-page article. This is not a meta-analysis, so I suggest that Table 1 be omitted or put in the appendix.

 3. Materials and Methods

First, the current version does not describe the data and data sources nor explain how the dependent variables and independent variables are constructed and what their units are. I suggest that the authors add the Data section to the manuscript. Besides the above content, there should be a summary statistics table, which presents the mean of infant mortality, adult mortality, life expectancy, GDP, government effectiveness, and other variables. I am interested in knowing how the government effectiveness index is constructed and what areas of effectiveness are taken into account.

Second, the three models presented are basically one model with three different dependent variables and their lags. The authors can shorten this section and make it less confusing by using one model, but state that the dependent variables (and their lags) are infant mortality, MRAD, and LEXP.

Third, I have several questions and comments on the specification of the models.

Take Model 1 as an example. The difference of infant mortalities between time t and time (t-1) is the dependent var. So you already demeaned the infant mortality in period (t-1). You included another infant mortality at time (t-1) in the right-hand side. So you demeaned it twice. That should not happen in the model. There's a direct correlation between delta INFMORT and INFMORT at t-1. If INFMORT at (t-1) is big, delta INFMORT at t-1 is small and vice versa. You wouldn’t want to include a variable that is mechanically correlated with the dependent var like that.

In multiple years, it's hard to know how much of the correlation estimated reflects the true effect of healthcare spending on health because health outcomes also affect healthcare spending. Eg. better health outcomes will lead to less healthcare expenses; also, better outcomes lead to higher productivity and higher GDP, which in turns make the government have more money to invest in healthcare. What are the mechanisms that the authors think apply here?

You have GDP, public health expenditure, and effectiveness at (t-1)) in the model. But you also have changes in GDP, health expenditure, and effectiveness for the periods from (t-1) to (t-k) in Model 1, to (t-k1) in Model 2, and (t-k2) in Model 3. What are the values of k, k1 and k2?

4. Results

The main results should be presented in the main text.

The correlation estimated by the models cannot be interpreted as a causal impact. The current models cannot separate the effects of other factors, such as education, on health outcomes.

I don’t see the authors’ interpretation of the results in the main text. What are the magnitude of the correlations between public health education and government effectiveness in terms of percentage and the outcomes? For example, I want to see “x percent change in public health education is correlated with y percent change in infant mortality”, or “one standard deviation in government effectiveness is associated with z percent change in infant mortality.”

With regard to the appendix table titled “Autoregressive Distributed Lag Model Estimates for the Studied 45 African Countries”, what are the units of infant mortality here? What does the coefficient of -3.55 mean? Is that -3.55 percent or per thousand? How are the coefficients compared to findings in the previous literature? Explain how they could be different. Please discuss that in the Result section.

 

 

 

 

Comments on the Quality of English Language

The English presentation needs moderate editing. 

Author Response

Reviewer One's Comments

1. Introduction - Theoretical Framework Clarification:
   - Expand the introduction to discuss the theoretical frameworks or hypotheses regarding the relationship between health expenditure, government effectiveness, and health outcomes. Address how the study fills the gaps in current knowledge and navigates potential biases the reviewer mentions.

Response

A section has been added to the introduction to show the study's theoretical framework and the linkage between health expenditure, health outcomes, and government effectiveness. Equally highlighted in the first section is the value added by this study and the gaps the study fills in the literature.

2. 
   Literature Review - Streamlining and Focus:
   - Streamline the literature review to focus on directly relevant studies and theories, reducing the length and possibly moving the detailed table to an appendix. Highlight the gaps in existing literature that the study aims to fill, specifically concerning the SSA context.

Response

The literature section has been retouched and resized by moving the former Table 1 to the appendix section.

3. Materials and Methods - Data Description and Model Specification:
   - Include a new section detailing the data sources, variables used (with units), and how these variables were constructed. Provide summary statistics for all main variables. Clarify the construction of the government effectiveness index, detailing what aspects of effectiveness it covers. Simplify the explanation of the econometric models used, correct the mentioned issues with variable specifications, and explain the choice of lagged variables.

Response

The models and equations have been arranged based on the reviewer's suggestions. Table 1 has been added to describe the model variables, showing names, notation, source, role, and expected signs. Summary statistics have also been included in Table 3 in Section 4 in response to this reviewer's comment.

 

4. 
   Results - Clarity and Interpretation:
   - Ensure that the main results are summarized in the text while acknowledging the limitations of the Analysis. Discuss how the results compare with existing literature and the implications of the findings in practical terms, using clear, non-technical language where possible.

Response

Thanks to the reviewer for pointing out that the outcome variables have been used as part of the regressors in the lag form. This makes the model an Autoregressive Distributed lag (ARDL) model and enhances its capacity to handle endogeneity problems and such diagnostic problems as autocorrelation. Therefore, within the ARDL framework, right-hand side variables are not correlated with the error term (zero cross-correlation among lagged errors), yielding estimated parameters that are consistent and efficient (see Pesaran, Shin, and Smith, 2001, pp 299-308). Also, our results are robust to omitted- variable bias and are unaffected by whether a variable is stationary or nonstationary.

K,  K1 and K2 are lag terms. Demeaning of variables through lagging of the dependent and independent variables ensures that residual autocorrelation will not affect any of the parameters. The ARDL Bound Testing Approach ensures consistent and efficient estimates of the level variables (long-run) and first differences (short-run) . Pesaran, Shin, and Smith (2001) show that appropriately modifying the lag orders of the ARDL model is adequate to simultaneously correct for residual autocorrelation and the problem of endogenous regressors, thus giving the ARDL an advantage over other estimation techniques. Inders (1993) has argued that including dynamics in an ARDL model helps correct the endogeneity bias, ensuring that efficient estimates are obtained with valid t-statistics. In our case, we started with two lags and found that residual noncorrelation is achieved by lag 1.

We followed a cluster analysis of the results, as can be shown in Table 5; however, to address the issue of the magnitude of the correlations between public health education and government effectiveness in terms of percentage and outcomes, we discussed the specific results of some countries as reference points for the rest.

We acknowledge the nice and kind contributions of the reviewer, and we believe that this revised version is a good piece. The above shows that great care and dexterity has been shown in addressing the points raised by the reviewer. Undoubtedly, the paper now has greater clarity because of the reviewer’s helpful comments. For this, the author(s) wish to thank the reviewer.

Pesaran, M. H., Shin, Y., & Smith, R. J. (2001). Bounds Testing Approaches to the Analysis of level relationships. Journal of Applied Econometrics,16(3), 289–326.

Arize, Augustine, C. (2017) “ A convenient Method for the Estimation of ARDL parameters and test Statistics: USA Trade Balance and Real Effective Exchange Rate Relation” International Review of Economics and Finance 50, 75-84.

Reviewer 2 Report

Comments and Suggestions for Authors

Journal: ECONOMIES

Manuscript ID: economies-2917733

Type: Research article

Title: Public Health Expenditure and Sustainable Health Outcomes  in 45 Sub-Saharan African Countries: Does Government Effectiveness Matter?

 

Synthesis: This paper examines the interaction between health expenditure and health outcomes with due consideration for government effectiveness across developing African economies. The rich data for the study draws from forty-five Sub-Saharan African Countries (SSA) covering the period     1960 to 2022. The analysis follows a country-specific comparative manner using the Autoregressive Distributed Lag Model (ARDL) as the major estimation technique. The results indicate that poor health outcomes are not due to inadequate budgetary allocations alone. Specifically, the study found a cointegrating relationship and strong adjustment of health outcomes deriving from the shocks and dynamics of not just health expenditures, but also government effectiveness. It is therefore recommended that strong institutions and safety nets be created to guard against corruption and leakages that derail the beneficial impact of public health spending. Also, government expenditures should be focused more on cottage and primary health dimensions to better mitigate adverse health conditions in SSA countries.

The study is relevant and publishable in the journal Economics. Ssome minor issues require further exploration. Therefore, my decision is 'Minor revision'. Below are my comments.

 

Major comments

INTRODUCTION: It is not clear what the study contributes to the existing research context: why are the study results relevant in the healthcare domain? It would be appropriate to allow the reader to grasp the added value of the study from the outset.

What is the 'lack' in the literature that justifies the study? Furthermore, what need justifies the analysis of the forty-five Sub-Saharan African Countries? Why and how does the case add to the context of existing studies?

 

CONCLUSIONS: Provide a description of the study's limitations in the 'Conclusions' section.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

The quality of English language in this paper is acceptable. While there are areas for improvement, such as occasional grammatical errors or word choice issues, overall communication remains understandable. With further attention to detail and practice, the authors can enhance clarity and coherence in their expression.

Author Response

Reviewer Two's Comments

1. Introduction - Contribution to Existing Research:
   - Clearly state the novel contributions of the study to the field and its relevance to healthcare policy and outcomes, especially within the SSA context.
   Response

We thank the reviewer  for helpful comments. A section has been added to the introduction to show the study's theoretical framework and the linkage between health expenditure, health outcomes, and government effectiveness. Equally highlighted in the first section is the value added by this study and the gaps the study fills in the literature.

2. Conclusions - Study Limitations:
   - Explicitly discuss the limitations of the study in the conclusions section, including potential biases, data quality issues, and the generalizability of the findings.

Response

The concluding part of the manuscript has been revised which now includes the limitations of the study.

It is our view that the contributions of the reviewer assisted this revised version and the author’ are unreservedly grateful for this.

Reviewer 3 Report

Comments and Suggestions for Authors

This fascinating article explores the effects of governmental expenditures and government effectiveness on outcomes. The paper uses a sophisticated econometric modeling approach that allows for modeling the process in each country rather than aggregating and examining how the relationships between variable is shaped over time. I think this important work....and also challenging for readers like myself without strong econometic background. While I acknowledge these gaps, there are elements of the paper that I think can be improved.

1) The data source is not explained well...and there is no effort to assess the relative quality of data from each country. It seems to me that the quality (reliability and validity) of the health data and government effectiveness would be strongly correlated and that this may introduced bias. Without more info on data sources and specific efforts to ensure comparability across countries, it is difficult to assess how much the findings are biased by this factor. 

2) The World Bank government effectiveness indicator has some history of use but there is little specific data on how the underlying measures were dveloped in each country and the reliability of the index over time. More competent and internationally-focused governments may also collect these data differently and this is particularly a potential factor over time. Moreover the concept of government effectiveness can not be seen as value neutral: the values that different governing groups put on health and well-being of the full populace compared to other goals (such as enriching a governing junta's members/making war against internal opposition) may or may not be reflected in the effectiveness. It seems to me that factors such as stability of the government, commitments to poverty reduction, commitments to minority inclusion, gender equity etc. may not be reflected in the GE measure. I think the measure may get close to assessing what the World Bank consider effectiveness and clearly this is a controversial concept. Most readers would want to know what it is about Tanzania, RSA and Nigeria that yeild different score.

3) There is a general need to spend more time explaining constructs used in the analysis such as on line 313 "THRGDP"....this should have been introduced and discussed in the methods. You also use a phase around "shocks" that is not explained----is this a reference to revolutions and other changes in government or an unnecessary reference to Keynesian views of quick changes in funding levels as policy choices?.

4) This is a complex analysis based on time series data from 45 country. There needs to be some discussion of the impacts of the multiple measures and tests on reliability. If power analysis is not possible in the ARDL model, this should be noted.

5) A major point of the authors is that the ADLR analysis is preferred because of how it handles outliers and independent variables relationships with dependent variables over time. I think it would be extremely useful to offer a simplistic model---how do population health outcomes per country correlate with expenditures and GE aggregating across several recent years, for example, in an OLS model--and show what the new model adds to our understanding.

6) I think the exciting opportunities in publishing in this journal include the broad and diverse audience. Throughout the paper, the authors must reach beyond the secluded world of World Bank concepts and analyses to show a broader health economics and policy audience understand your primary conclusion that the quality of governance as well as health expenditures shape health outcomes. I agree with this idea because I believe the social determinants of health have powerful impacts beyond those of healthcare expenditures to shape health outcomes.

 

Comments on the Quality of English Language

i found some typos....the language is great...it just needs some dumbing down for people who are not following new methods in econometrics.

Author Response

Reviewer Three's Comments

1. Data Source and Quality Assessment:
   Provide a thorough explanation of the data sources used, including an assessment of the data quality and comparability across countries. Address the issue of potential bias due to data quality variability among countries and its impact on the findings.

 

Response

The reviewer's comment is found agreeable, and it is essential to acquiesce that working with developing economies challenges data reliability and availability. In this study, we had to depend on data from the World Development Indicator, a data repository of the World Bank that offers a reasonable answer to the data reliability question. We also added Table 1 to give a detailed description of the datasets used in this study.

 

2. Government Effectiveness Indicator:
   - Offer a detailed discussion on the construction of the government effectiveness indicator, its limitations, and how it may affect the study's conclusions. Consider alternative or additional measures that might capture government commitment to health and welfare more comprehensively.

Response

This has been done (See response to Reviewer One)

 

3. Clarification of Constructs and Methods:
   - Clarify all constructs used in the analysis, ensuring terms like "THRGDP" and "shocks" are introduced and explained in the methods section. Discuss the impact of using multiple measures and tests on the reliability of your findings and the rationale behind choosing the ARDL model over simpler models.

Response

It is important to note that we used the ARDL procedure as an OLS procedure rearranged to capture short and long-run effects. In addition, it allows estimates to be efficient and consistent, which makes it a superior regression estimator compared to regular OLS. The ARDL estimates are robust to small-sample bias. The regular OLS produces estimates that are only consistent in large samples. Though both can measure the elasticities (response) of the regressand to changes in the regressors (i.e., variables are in the logarithmic format), the ARDL is conceptually and practically superior to the OLS.

 

4. Broadening the Appeal of the Paper:
   - Simplify the explanation of econometric methods and results to make them accessible to a broader audience, including those not specialized in econometrics or health economics. Highlight the implications of the findings for broader discussions on health economics, policy, and the social determinants of health.

Response

Section 5 of the study has been rewritten to further elucidate the findings and make it more relatable to people with less econometric knowledge. Also, the study's social, policy and research implications have been further strengthened.

We found the reviewer’s comments thoughtful, detailed, and extremely useful. We have acknowledged the limitations of our work in the text. We are grateful to the reviewer for his/her recommendations. Thank you because it will help us in our further studies.

 

 

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

This is an interesting and informative study with an enormous scope and important findings. There are several places that still need a bit more explication to facilitate reader comprehension.

1) The short discussion of government effectiveness in the literature review needs further explication. The measure reflects a certain view of government and there is considerable debate about whether it values less democratic push and more accomodation to for-profit/international corporations. And there is a discussion of several possible links between government effectiveness and positive health outcomes. Even if you quite reasonably select not to focus on the several indirect pathways from government effectivness and ideology on health, you should still acknowledge these debates.

2) The paper contains several equations. It is customary to offer definitions for each of the terms. This would make the data clearer.

3) It would be helpful to add text to clarify the data: are all measures available for all years for all countries? how are countries that changed names or changed borders and subdivisions addressed?  It would seem that all data come from several publications by one international agency. You do not explicitly acknowledge that this is a limitation.

4) More about the analysis would be helpful: how were the lag terms computed and how many lags are included,,,,maybe this would be clear with more on the notation.

5) In table 1, I do not understand the contribution of the last column to the right....it would be better to drop this column and then say in text for which variables there are expectations of sign.

6) Table 5 needs much more explanation. The text in later sections comes to explain this more....Perhaps the overall comments on Table 5 are out of place, and instead you could point to in each of the subsequent analyses. Am I correct in understanding that when you say in "x" countries, variable y reduces infacr mortality, you are counting the number of countries with significant relationships.

7) Using your method, I understand it, there are tests for multiple dependent variables for multiple independent variables, for multiple countries with each country compared to itself. Is there any concern about needing to correct p values for multiple tests?

8) It seems from my reading of the tables and appendix, that there are several countries for which the GEFF variable is not related to the key outcomes---Benin, Burundi, Congo and others. Is there anything you can say about what is unique about these countries and how this influences these outcome.

9) It would be great to provide a quick guide to the columns in the Appendix table of results....such as 1 sentence explaining each column and how to interpret.

10) This is really a question that reflects coming from a different discipline.....Does this analysis provide any clarity about the interactions of government efficiency and expenditures on health?....Are there examples of relatively stable low investment/high efficient or high investment/low efficient  countries that can be identified in these analyses? Or examples of countries that made more dramatic changes in efficiency or expenditures at certain points and how did this effect outcomes. There is a broad literature on age/period/cohort effects. The people who experienced major changes in government efficiency or investments at key point in their lives may be informative.

Comments on the Quality of English Language

the editors did a great job on the text.

the formatting of the tables, particularly the appendix table, needs some more work

 

Author Response

Response 1. We have added a section that explains how we measure the effectiveness of government (see p.5).This is in addition to the footnote previously added (see p 10). 2. There are now three equations, and Table 1 describes the model variables. The inclusion of the data source compensates for data limitations. Despite its limitations, the World Development Indicator is one of the most credible data sources from the World Bank. As a result, our datasets can be relied upon to be reliable. 3. According to the reviewer's recommendation, the sign column in Table 1 has been removed, and a short note is included at the foot of the table explaining the a priori expectation. 4. The results of the study were discussed in more detail. In our study, which involves joint estimation, there is no need for a common p-value. It has been found unnecessary in our field. 5. The tables and appendices have also been addressed, and if our study is accepted, any other formatting problems can be resolved at the copyediting stage. In sum, thank you for the comments. They are excellent. We greatly appreciate the effort this reviewer put into them. It has resulted in a much-improved version. As is evident from the text, we have responded to the reviewer's comments. We have gone much further than the reviewer asked us to address their concerns.

Author Response File: Author Response.docx