Effects of Health Factors on GDP Growth: Empirical Evidence from Saudi Arabia
Round 1
Reviewer 1 Report
I would like to thank the authors for this research that aims to examine the direct and indirect links between health factors (infant mortality rate, maternal mortality ratio, road traffic mortality rate, and healthcare expenditure) and sustainable GDP growth in Saudi Arabia, which has been identified as one of the countries most vulnerable to changes in health factors.
The paper is well written and well structured.
The research subject is timely, innovative, and highly interesting. It also fits the aim and scope of the journal.
The research is well designed and follows a sound scientific research method.
Results and recommendations are clear and could have an impact among the community of researchers.
I have a very minor comments that do not affect the excellent quality of the paper.
It would be better to divide section 4 Results and Discussion to two separate sections. So, keep the section 4 as results and add section 5 as discussion (starting line 545). This will facilitate the reading of the paper.
Lines 570 to 573 : check the structure of sentences as some ideas are not very clear or easy to understand.
Author Response
Please find the attached response.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Well executed and timely study.
Author Response
Please find the attached response.
Author Response File: Author Response.docx
Reviewer 3 Report
Nowadays, one of the most important research fields in macroeconomics is the issue of identifying, analyzing, and assessing the social and humanistic effects of economic development and the role of social factors in shaping economic processes. The negative phenomena of economic development of the last half-century include, among others, a significant increase in morbidity and mortality. Therefore, the topic undertaken by the authors is part of the research trend of selected factors of economic growth. After getting acquainted with the content of the study, I suggest the authors, improve the study, and supplement the content in the subsections. Below are detailed comments on the article:
1. I suggest supplementing the abstract with the purpose of the article.
2. I suggest supplementing the introduction with research questions.
3. In the results and discussion chapter, I suggest removing the information on the characteristics of the L397-L403 tests because they have been characterized in the materials and methods chapter.
4. I propose, under the research carried out in the chapter results and discussion, to inform about the rejection or confirmation of the four hypotheses in the literature review chapter.
5. The conclusions should answer the question: What did the authors find new during the research? Did the results confirm the hypotheses?
Author Response
Please find the attached response.
Author Response File: Author Response.docx
Reviewer 4 Report
Comments to the Author
Review of “Effects of Health Factors on GDP Growth: An Empirical Evidence from Saudi Arabia”
Manuscript ID: sustainability-2365661
Summary and recommendation
The paper examines the effect of four key measures of health performance on economic growth in Saudi Arabia from 1990-2019. Using standard time-series analysis (co-integration and VECM models) it fins that infant mortality, road accident mortality and healthcare expenditures have negative effects on growth, while maternal mortality has an insignificant effect on economic growth. The study also finds that infant mortality rates and traffic mortality rates affect health expenditure (but not vice-versa).
The paper is generally well-written and the econometric analysis is competently done in a narrow statistical sense (though some steps in the analysis need to be clarified). The motivation and theoretical framework of the analysis is, however, weaker. The lack of a strong theoretical argument that health care measures are primary drivers of GDP growth makes the estimating equations less compelling, and any assertions of causality do not seem warranted. In the absence of theoretical foundations, the study is much less interesting since, in the end, all it does is provide a story of how some variables behave in similar ways.
Major issues
1. I find the literature review and hypothesis development to be unclear. There are three very distinct processes analysed in the literature: GDP, GDP growth, and GDP per capita. While these are all related, there is no clear exposition of the presumed theoretical and causal links between these outcomes and health measures, some of which are clearly are bi-directional. Explicitly recognizing the linkages between these different processes may reconcile what are reported in the review as contradictory results. For example, Acemoglu and Johnson’s finding on GDP per capita is identified as contradictory, when in fact it could be perfectly consistent with the other studies looking at GDP and GDP growth. Having GDP grow while GDP per capita falls can be completely consistent if population increases faster than GDP. This is a simple theoretical relationship that is not directly stated, and yet could allow for a more realistic and structured analysis of the potential causal relationships. The relationships are complex in terms of whether population rises faster than GDP, whether there is a change in the dependency ratio, and whether a decline in maternal mortality and infant mortality is sufficient to lead to meaningful changes in human capital investment. These different causal routes will behave differently in different countries depending on their population structures and growth rates, and their underlying economic structures and growth rates. I think clarity in the theory would help the discussion. In fact, Saudi Arabia’s demographic structure could be presented in the context of these previously identified connections.
A related problem is that the paper states that it is trying to explain growth (a dynamic change process) with levels of health indicators. Attempting to explain growth using levels is a little problematic. Looking at figure 2a it is highly volatile and exhibits no real trend…it is clearly stationary (and the tests confirm this). The health variables (figures 3a-5a) are all exhibiting trend (and in IMR and probably MMR there is a trend even to the first difference) which follows a standard evolution for these variables, i.e. gradual improvement, but with improvements declining as the level reaches levels of better performance. In the end, the fact that everything is first differenced (in effect) sort of acknowledges this problem. By contrast, figure 2b is uninteresting, it is the first difference of a variable that is already based on a first difference. The underlying analysis actually seems to be about cointegrating health levels with GDP levels, all of which exhibit trend and often non-stationarity. Finally, it is unclear whether the econometric analysis uses the first difference of growth as the dependent variable or just the growth itself.
The second element of the theory that is not always discussed comparatively for the studies is the problem of timing; some useful references are made to specific time horizons or a “long run”, but not consistently enough to determine whether the study results are as inconsistent or not. I would be rather surprised to see much of a short run causal connection running to economic growth. The time frame for the analysis also complicates the causality analysis. I have a much easier time buying the argument that higher GDP per capita allows for higher investment in health care in the future, and thus better health outcomes. I think that relationship is sort of a stylized fact. Disentangling the reverse causality will be much trickier in my view, though I could be persuaded. That suggests that perhaps there should be some review of the literature examining the reverse causal argument.
The discussion of RTM and GDP growth is also not really compelling. In general GDP growth and increases in GDP per capita will both generally be associated with more kilometres driven and thus a higher rate of traffic accidents. I would be surprised to find that the causality is reverse, though perhaps in extreme cases that could be the case. I would want to see more details of the studies, including whether they controlled for road quality (which could lead to both higher fatalities and lower GDP).
In general, these are not necessarily fatal problems. Greater theoretical clarity would be desirable to make some of the presumed causal links more clear. Even if the results of previous studies are not as contradictory as suggested in the current text, there are clearly multiple potential causal links that are contingent on the demographic and socio-economic conditions, which alone is sufficient to justify specific country studies such as this one on Saudi Arabia.
2. The second, and perhaps more serious, problem that I have is with the estimating models and the lack of reasonable theory in them, though I suppose this is common in modern time series analysis. VECM does not provide a theory of what explains the values of the main dependent variable, it just says that there are other variables that have similar behaviour. For example, I don’t understand why we would try to understand, let alone explain, the GDP growth rate in Saudi Arabia without including oil prices. Instead, four related health variables are all thrown in as the cointegrating vectors; I don’t really know of any growth models that rely that extensively or exclusively on health measures. I also suspect that some of these health measures are collinear, making the individual equations potentially complicated by multicollinearity. The only conclusions that seem to be possible here are non-causal, so the statements on page 13 suggesting that changes in the health variables lead to changes in GDP growth are not really justified, it is only that there is a statistically significant, possibly non-causal, co-movement between these variables. Similarly, later on page 13, the sentence “This study confirms that there is a strong long-run equilibrium causality between GDP growth and health factors” does not seem supportable by the evidence, only that there is a long-term correlation between the variables.
The Granger causality tests also make it clear that very few of the variables Granger cause another one, which also throws into doubt the causality of the relationships. In fact, the one exception in the Granger tests is of health care expenditures ,which appear to be Granger caused by traffic fatalities and infant mortality rates. This causal argument actually makes some sense, since higher demand on health care services may indeed generate subsequent increases in expenditure. Of course, there is also a likely reverse causality where prior higher expenditures improve heath care capacity and reduces mortality rates.
In fact, given that the trajectory of some of the health variables (figure 4a and 5a in particular) I expect that there would be a strong cointegrating equilibrium between the Saudi Arabian growth rate and the infant and maternal mortality rates of a lot of other countries over this time period. The series may be cointegrated, but it doesn’t mean they are in any real way actually connected, let alone causally linked.
Given the absence of a compelling theoretical causal relationship, and the absence of an empirical one (in the Granger sense), I have no idea why you would want to bother trying to tell a causal story. If I was told that nearly 75% of the variation in the Saudi Arabian growth rate can be explained by these four variables (and with no reference to the oil market) then I would be very sceptical…I don’t think there is a causal story to really tell here. The only exception I can see to this lack of a compelling underlying structural model is health care expenditures which, as a component of aggregate expenditure and hence of GDP, has a clear theoretical connection to GDP over short time periods.
I could be persuaded that the model has some value as a potential predictive model, which is what a lot of modern time series seems to be about. The problem with this approach to prediction, however, is that if there is a fundamental causal relationship that is being left out of the model then a shock to the missing factor could lead to completely incorrect predictions.
3. The technical work seems fine. My one technical concern is that the stationarity tests presented in table 4 are not totally clear. I am not really convinced that IMR (and probably not MMR) are I(0), though I could agree they are stationary with trend or drift. Were the stationarity tests done with trend or drift adjustments? More details are required.
Minor Issues:
1. In tale 7 HCE is written as HCF.
2. I also find the adjustment speed to be terribly fast, though perhaps it could be true. How does it compare to the results of other similar studies?
Final Assessment:
I do not find that the underlying theoretical discussion is clear enough in this draft. As a result, I have a difficult time accepting the story as anything other than correlational. I recommend resubmission after major revisions. Specifically, to be to be publishable I think I would need to see the author(s) do the following:
- Set up the theoretical relationships between health and GDP in a clearer fashion.
- Resolve and clarify the issue of whether the analysis is in levels or changes in levels with GDP growth being the latter.
- Clarify whether the dependent variable of growth is actually growth, or the first difference of growth.
- Justify the absence of more compelling explanatory variables, such as oil revenues or prices.
- Remove any references to “effects” or “causality”, these are not justifiable even in the Granger sense (with two exceptions).
- Reduce or remove the text attempting to explain the results as causal in some way.
Author Response
Dear Reviewer,
We are delighted to receive your valuable comments and suggestion to improve our paper. We tried our level best to address your valuable comments and suggestions. Please find the attached responses.
Author Response File: Author Response.docx
Round 2
Reviewer 4 Report
The authors have revised the paper sufficiently to address my main concerns. I would encourage them to reflect on the comments of their referees carefully to consider how they might incorporate these more fully in future work.
I detected a few minor errors that should be fixed in the editing process, but otherwise the paper was clearly written.
