The Effect of Health Expenditure and Immunisation on Under-Five Mortality Rate in Selected African Countries.

Background: As a way of tackling child mortality, many countries in the world depend on their respective health-care system. But governments of most countries in Africa are yet to provide robust funding of their health-care systems as many people still depend on the out-of-pocket payment to receive health services. Against this backdrop, this study used annual panel data to assess the effect of health-care expenditure and immunisation on the under-ve mortality rate in 30 selected African countries for the period 2000-2017. Methods: Multiple regression technique was adopted for the data analysis and the robust xed regression estimator was preferred to the random effects as determined by Hausman test. Results: The ndings indicated that domestic government general health expenditure had a signicant negative effect on the under-ve mortality rate. However, the effect of domestic private health expenditure on under-ve mortality was not signicant while external health expenditure had a signicant negative effect on under-ve mortality rate. The impact of diphtheria immunisation on under-ve mortality was signicant. Conclusions: Except domestic private health expenditure, government and external forms of health expenditure coupled with diphtheria immunisation were signicant factors for the reduction of the under-ve mortality in the selected countries.

under-ve mortality by 2030 [24] is likely to be reached by a few Sub-Saharan African countries if other variables remain constant.
Medically, certain vaccine-preventable diseases such as measles, pertussis, diphtheria, polio, tuberculosis and tetanus have been linked to child mortality [23]. The problem of child mortality is remotely and complexly related to certain living conditions such as access to clean drinking water and sanitation facilities [36]. Interestingly, as captured in both the MDG4 and Sustainable Development Goals (SDGs), under-ve mortality, which is the death of a child from day one to ve years, has generated a great deal of interest and concern [32]. The phenomenon of child mortality transcends medical causes to include certain socio-economic and demographic variables in a broader context. As several studies [35,18,5] have suggested, a multi-dimensional approach is needed to tackle this menace.
In different but related studies, a mother's age was considered as a non-factor for the uneven distribution of mortality under the age of ve [35,32]. Some scholars have further noted that house settlement patterns and rural-urban differentials were risk factors for child mortality [3,12,15]. Health care systems, health care nancing, poverty and access to maternal health services are remotely and complexly identi ed as correlates of child mortality [31,33,39,1]. For example, as a complex phenomenon, child mortality has been largely associated with maternal health status because children under-ve years of age have always been considered to be under the custody of their mothers except in certain exceptional circumstances. Consequently, the conditions of mothers in uence the survival chances of the children [30].
Child death has some cross-cutting impacts on culture, economics, government and families as a signi cant public health problem for the population. For example, based on the infant mortality trend study of 2,976 million child deaths as of 2013, it was estimated that the African continent would suffer the effect of child deaths amounting to around 6 percent of its non-health Gross Domestic Product (GDP) [20]. Rising infant mortality still has the ability in the near future to decrease the productive in uence of the working community. Moreover, infant death leaves behind some devastating physical, psychological and emotional impacts on family members. Child death deprives ageing parents of future care in Africa, where the care of elderly persons is one of the major traditional responsibilities of family members.
As a way of tackling child mortality, many countries in the world depend on their respective health-care system. But governments of most countries in Africa are yet to provide robust funding of their health-care systems as many people still depend on the out-of-pocket payment to receive health services. Similarly, child immunisation in Africa is still being confronted by myriad of challenges ranging from religious belief, negative politicising, geographical inaccessibility, to poor or lack of physical infrastructural facilities. In the light of the foregoing, this study assessed effects of health care expenditure and immunisation on the under-ve mortality rate in selected African countries for the period of 17 years (2000-2017). In measurable terms, the study is bound by the following speci c objectives, namely to: assess the effect of Domestic Government General Expenditure on Health (DGGEH) on Under-Five Mortality Rate (UFMR) in selected African countries; assess the effect of Domestic Private Health Expenditure (DPHE) on under-ve mortality rate; examine the impact of External Health Expenditure (EHE) on UFMR in the selected countries; examine the impact of Diphtheria Immunisation (DI) on UFMR in the selected African countries.
Domestic government general expenditure on health is used as one of the proxies for health expenditure in this study. It can be referred to both capital and recurrent health expenditure of the government, including revenue as domestic grants, transfers, health subsidies, organisational health nancing programmes, voluntary health insurance bene ciaries and national health insurance scheme [21]. Some documented studies have linked health expenditure to health outcomes (child mortality). Some of these studies used data from time series in the health economics literature, while others used panel data sets. A study found a signi cant long-term association between government spending on health and child mortality in several developing countries [22]. Similarly, several different but related study results found that government spending on health was signi cantly associated with decreasing child mortality [4,8,6,11,21]. In contrast, a few studies have found that public health spending in high-income countries has not had a signi cant impact on child mortality [10,2].
Another form of expenditure is Domestic Private Health Expenditure (DPHE). This encompasses household out-of pocket, funds from corporations and non-governmental organisations. A group of authors has noted that DPHE had a signi cant in uence on under-ve mortality [22]. That is, private expenditure on health reduced child mortality signi cantly. Also, total health expenditure in 47 African countries has been found to have a signi cant impact on under-ve mortality [4]. However, another study discovered that the association between private health spending and child mortality was negligible one [21]. As part of the expenditure component, External Health Expenditure (EHE) comprises all nancial in ows and resources that come into the national health system from outside the country [21]. More importantly, the issue is the effects of such in ows and resources on the health outcome like child mortality. Against this background, it was opined that external spending on health in the form of aid had a signi cant impact on the mortality of children under ve [25]. Another research has also rea rmed that external health investment is strongly correlated with child and neonatal mortality [21]. In the same way, a study showed that when external health resources are improved, they contribute to a reduction in infant mortality [8]. A research affecting 10 selected African countries equally found that external health spending in the form of aids and immunisation was an effective factor in reducing mortality for children under ve [2].
One of the main ways of addressing childhood morbidity worldwide has been the regular and periodic immunisation of children under ve. Some of the diseases that can be treated by the Expanded Immunisation Programme (EPI) include measles, pertussis, diphtheria, polio, tuberculosis, and tetanus (Matthews and Diamond, 1999). However, the interest of this study is partly to assess the effect of diphtheria immunisation (% of children ages 12-23 months) on the under-ve mortality rate in selected African countries. In addition, a couple of studies have stated that immunisation is an important factor in reducing the mortality of children [26,7]. Similarly, it has been revealed that immunisation raises the probability of infant survival and was viewed as one of the most effective ways to stem the tide of child mortality [37]. The accessible literature reviewed in this study has indicated that there is still a need for further research relating private health spending, external health expenditure and immunisation to underve mortality, as the current body of literature on the topic still leaves some gaps. Besides, the time frame covered in the available literature needs to be updated as the current study lls in the gaps in terms of time frame (2000-2017) and combined variables (health care expenditure and immunisation).
With regard to the theoretical perspective, Grossman's health production and investment model is being adopted for this research. According to Becker, health in this case is conceptualised as both a consumable item and an investment resource [29]. The model opined that an individual chooses their level of health and by extension, their life span. It further postulated that time was a variable factor that contributes to declining health status in both objective and subjective terms. However, through certain social determinants such as diet, exercise, medical care and life-style decisions, it could be replenished and counter-balanced. On the other hand, ageing, smoking, extreme opioid use and exposure to other toxic compounds may make the health capital of an individual depreciate. In this context, the level of health an individual has is a function of the amount of resources allocated to its investment. Also, an individual is viewed as a "manufacturer" as well as a "user" of health and that the demand for health is not for immediate consumption but for the production of other goods. So, one's level of health is not attributed to certain external social factors but endogenously in uenced. In other words, Grossman's main argument for health investment is that health care services are not requested for their own purposes, but for the impact on health itself. Ageing in uences the state of health and requires constant health care services. However, the impact of ageing on wellbeing is mitigated by educational level and a balanced style of living [29].
The model can be sustained in this study when health expenditure (domestic government general, domestic private, external health expenditures and diphtheria immunisation) is viewed as an investment tailored towards improving maternal and child health status (reduction in under-ve mortality). In this case, expenditure on and utilisation of maternal and child health services is considered as derived demand capable of impacting on child mortality rate. Within the theoretical purview of the household health production function of Grossman, both increase and decrease in child mortality is predicated on the amount of resources allocated to health. It is taken for granted that more increase in health and education is likely to translate into reduction in under-ve mortality in the selected African countries.
Moreover, it can be deduced that the age-long uneven distribution of child mortality across and within various regions of the world is neither accidental nor incidental but it can be related to varying investment in health capital among other factors. "Other things being equal", the insightful understanding of Grossman's model is that, more investment in health capital leads to desirable child health status and by extension declining child mortality. Even though Grossman's model is subject to some methodological and empirical aws, however, the theory has continued to stand the test of time with regard to investment in health capital.

Methods
The aim of the study is to assess the effect of health expenditure and immunisation on under-ve mortality rate in selected African countries. The ex post facto research design was considered as appropriate for the study since it involved examining the statistical effects of the independent or explanatory variables on the dependent or outcome variable over past time. The independent variable in this study is health expenditure, which is proxied by Domestic Government General Expenditure on Health (DGGEH), Domestic Private Health Expenditure (DPHE) and External Health Expenditure (EHE). Immunisation as an independent variable is also represented by Diphtheria Immunisation (DI). On the other hand, Under-Five Mortality Rate (UFMR) serves as the dependent variable.
Data Description and Sources The study utilised the annual panel data that comprised 30 African countries for the period of 17 years involved. This is to enable the authors to determine the impacts of DGGEH, DPHE, EHE and DI on UFMR in the selected countries. The analysis of data was facilitated by the application of STATA 12 software version and the study hypotheses were extrapolated from the results and tested at the pre-determined 5% level of signi cance. The analysis was patterned after the econometric styles used in the literature [  Since the correlational relationships between the independent variables have not exceeded the 85% "rule-of-thumb" suggested in the literature [17,16], it could be deduced that multicollinearity was not an issue or a problem in this study. However, the authors subjected the deduction to the test of the Variance In ation Factor (VIF) for further validation in order to avoid any doubt of in ated results or spurious regression outcome.  Table 3, the Variance In ation Factor (VIF) values of the independent variables were less than 10; which implies a total absence of multicollinearity in the analysis of the independent variable components. This is also in tandem with the recommendation in the existing literature [28]. At this juncture, the problem of heteroskedasticity in the model has been detected as revealed in Table 4 above. The Breusch Pagan/Cook-Weisberg test shows the chi2 (1) value of 4.41 and the corresponding probability value of 0.035 which is statistically signi cant at 1% alpha level (p-value < 0.05).
Consequently, the problem of heteroskedasticity in the model has outed the homoskedastic principle of the general linear regression model. It was on this basis that the authors considered it imperative to correct the pitfall by adopting the robust xed effect regression model after conducting the Hausman speci cation test.  Table 5 revealed that UFMR, DGGEH, DPHE, EHE and DI did not follow the normal pattern of distribution around their central means as shown in the respective probability values of less than 0.05 level of signi cance. Again, an assumption of the general linear regression model of normal distribution of data has been violated at this point. The possible approach to be adopted was the robust xed effect estimator to correct this statistical anomaly. Since the study involved the panel data, the authors opted for the Hausman speci cation test to determine whether the xed effects or the random effects estimator was appropriate. The result of the Hausman test suggested the robust xed effect model being appropriate for this study as indicated by in the chi2 value of 1954.27 and the corresponding signi cant Prob. Chi2 of 0.000. Besides, the problem of the non-normal distribution of the data equally warranted the application of the robust xed effect model.  Table 7 above re ects the overall coe cient of determination of 0.169. This suggested that the independent variables, namely: DGGEH, DPHE, EHE and DI exerted an estimate of 16% joint effects on the changes in the dependent variable (UFMR) for the period. The probability value of the F-Statistic is signi cant at 1% which implies the tness of the study model and its suitability for reliable decision making.

Test of Hypotheses
Ho1: Domestic government general expenditure on health has no signi cant impact on underve mortality rate in selected African countries.
Ho2: Domestic private health expenditure has no signi cant effect on under-ve mortality rate in the selected countries.
Ho3: External health expenditure has no signi cant effect on under-ve mortality rate in the selected countries.
Ho4: Diphtheria immunisation has no signi cant effect on under-ve mortality rate in the selected African countries.
It was noted in Table 7 that the domestic government general expenditure on health had a signi cant negative effect on the under-ve mortality rate in the 30 selected African countries as shown in the probability value less than 5% (0.046) and the corresponding coe cient of -.128. On the strength of this result, the null hypothesis which stated that domestic government general expenditure on health has no signi cant impact on under-ve mortality rate is hereby rejected. Also, the impact of domestic private health expenditure on under-ve mortality was not signi cant as re ected in the probability value of 0.470 (47%) which is above the a priori assumption of 5% level of signi cance and the coe cient of 0.002 respectively. In consequence, the null hypothesis which stated that domestic private health expenditure has no signi cant effect on under-ve mortality rate is accepted.
Furthermore, the result in Table 7 shows that external health expenditure exerted a signi cant negative effect on under-ve mortality rate. This was underscored by the P-value of 0.000 and the coe cient of -0.0117. Since the P-value is signi cant at 1%, the null hypothesis three (Ho3) is rejected. Similarly, diphtheria immunisation has a signi cant negative effect on under-ve mortality as indicated by the Pvalue of 0.000 and the corresponding coe cient of -0.534. On this premise, the null hypothesis which stated that diphtheria immunisation has no signi cant effect on under-ve mortality rate in the selected African countries is hereby rejected.

Discussion
The study revealed that domestic government general expenditure on health had a signi cant negative effect on the under-ve mortality rate in the 30 selected African countries. This shows that a 1 percent unit increase in government health spending will reduce under-ve mortality rate by -0.128 if other variables are held constant. The nding is consistent with some previous studies [ 22,4,8,6,11,21].
However, it is at variance with a study involving some high income countries where no signi cant relationship was found between public health spending and child mortality [10].
Contrary to the a priori expectation, this study found that the impact of domestic private health expenditure on under-ve mortality was not signi cant. Similarly, when other variables remain unchanged, a unit increase in domestic private health expenditure will lead to an increase of 0.002 in under-ve mortality rate per time. While the nding is in agreement with a couple of studies conducted in the previous years [4,22], it is however, in contrast to some other studies [26,2]. Moreover, the study found that external health expenditure had a signi cant negative effect on under-ve mortality rate for the period under review. Also, an increase of 1% in external health expenditure will produce a corresponding negative impact on under-ve mortality rate in the 30 African countries within the determined time frame. The above nding is in consonance with some studies in the literature [25,8,4,21]. Moreover, immunisation in the form of diphtheria was found to have a signi cant negative effect on under-ve mortality. In other words, holding domestic government general health expenditure, domestic private health expenditure and external health expenditure constant, 1% increase in immunisation will lead to -0.534 signi cant decline in the under-ve mortality. The above nding is anchored on some previous studies [26,7,4] that established a negative signi cant relationship between immunisation of children and child mortality.

Research Limitations:
This study assessed the effects of health care expenditure and immunisation on the under-ve mortality in selected African countries for the period of 17 years (2000-2017). However, it did not differentiate the effects of health care expenditure and immunisation on the under-ve mortality of individual countries. Moreover, this study did not cover other socio-cultural, economic and political factors that underscore the under-ve mortality in the affected countries. Notwithstanding, the study provides a basis for regional comparison vis a vis other parts of the world.

Conclusions
To tackle the problem of the under-ve mortality in African countries, a multi-dimensional approach is required as this study indicated it. Apparently, government general health expenditure, external health-care expenditure and diphtheria immunisation have been noted as signi cant factors for the reduction of the under-ve mortality in the selected countries. On the other hand, domestic private health expenditure did not have any signi cant contributory impact on the under-ve mortality rate. This may not be unconnected with the harsh socio-economic realities prevalent in most African countries that constrain access to antenatal, prenatal and post-natal health care services. So, many mothers and their under-ve children rely more on health services being offered by government.
On the basis of the study ndings above, it is suggested that governments of the various African countries should step up domestic health expenditure in the area of maternal and child health. Besides, the issue of diversion of funds meant for the health sector by some corrupt individuals and their syndicates must be checkmated by the various governments if the intended health outcome must be maximised. The countries should create an enabling environment comprising good roads, electricity, pipeborne water, security and other health-related infrastructure that may attract more grants for maternal and child health from international external sources.
Also, the governments of African region should pool their resources together and start the production of vaccines locally to address the issue of certain childhood killer diseases like diphtheria, measles, tuberculosis and many other preventable health a ictions, instead of relying on the foreign-based vaccines for immunisation. This is so because a time may come that the donor agencies supporting the immunisation of the under-ve African children may not be able to sustain it due to some unforeseeable circumstances and vagaries surrounding international politics and economy. African governments should try and make access to maternal and child health-care services free and compulsory for mothers and their under-ve children. In effect, it may help to solve the problem of access and further reduce the challenge of the under-ve mortality in the region.