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

Abstract

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 this study draw from forty-five Sub-Saharan African (SSA) countries covering the period 1960 to 2022. The analysis follows a country-specific comparative manner using the Autoregressive Distributed Lag (ARDL) model as the major estimation technique. The results indicate that poor health outcomes are not due to inadequate budgetary allocations alone. Specifically, this 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.

1. Introduction

Africa trails the rest of the world on human development indicators, including life expectancy, infant mortality, undernourishment, the incidence of HIV/AIDS, malaria, and tuberculosis Kirigia and Barry (2008), Mallaye and Yogo (2012) and Thoumi and Bandara (2019). According to Songwe (2019), and Novignon and Lawanson (2017), health outcomes in Africa still lag other regions in the world, pointing to large inequalities in access to health care and high health-related out-of-pocket expenses. Further, Songwe (2019) stated that Africa has approximately 15 percent of the global population but, unfortunately, carries one-fourth of the worldwide disease burden. Tandon and Cashin (2010) estimated that Sub-Saharan Africa (SSA) and Asia account for over 50% of the global disease burden. In SSA, children die from diseases such as malaria, diarrhea, and tuberculosis that could have been treated at a low cost or avoided altogether with basic hygienic practices (Anyanwu and Erhijakpor 2009; Filmer et al. 2000). According to the World Health Organization, (WHO 2016), 36 percent of children with fever in SSA do not receive any medical attention. Further, in 2018, the WHO (2016) stated that malaria affected 228 million people globally and killed an estimated 405,000 mostly in SSA. This unsavory statistic is made severe by the COVID-19 pandemic and the low vaccination figure for the entire SSA relative to other regions of the world (World Bank 2020).

These indicators continue to worsen despite considerable improvements in funding for the health sector in the region declarations made in Abuja in the year 2001; Gambia Banjul in the year 2006; Quagodougou in Burkinafaso in the year 2009; Tunis in Tunisia in the year 2012; Angola, Luanda in the year 2014; Addis Ababa, Ethiopia in the year 2015 and most recently in 2019 (Ogbuoji et al. 2019; WHO 2016; Anyanwu and Erhijakpor 2009). In fact, since 2000, Africa’s government health spending has grown by more than 4 percent each year, with the highest period of growth being from 2000 to 2015. This is comparable to global spending on health between 2000 and 2016, which grew in real standings at 4 percent per annum (UNECA 2018). Total health expenditure equaled USD 9.7 billion in 2000 and increased to USD 68.7 billion in 2010 (Avila et al. 2013). The number of countries spending more than USD 44 per capita per year doubled from 15 to 31 in 2000 to 2015 (UNECA 2018). Similarly, twenty-nine countries have increased total health spending per capita between 2000 and 2015. These countries include Mauritius, South Africa, Algeria, Botswana, Namibia, Botswana, Rwanda, Zambia, Madagascar and Togo, Angola, Gabon, Botswana, Equatorial Guinea, Lesotho, Seychelles, São Tomé and Principe, and Swaziland.

Theorists have argued that for countries to make good progress in health outcomes, they must make some level of benchmark expenditure in the sector (Musgrove 1996). According to the Keynes (1936) hypothesis government needs to intervene in an economy by spending more on government projects and social programs. The theory is based on the principles that increases in government on infrastructure and social programs contribute to creating a more favorable environment for the private sector to invest in; thus, it is better at creating jobs and supporting economic activity. Two other key theories largely support the Keynes (1936) assertion in this study. The benefit theory of government revenue and spending popularized by Erik Lindahl as presented in the works of Samuelson (1954) and Musgrave and Musgrave (1973), which is a normative assessment of the fact that public expenditure such as health should show beneficial evidence on the people who pay taxes into public treasury. In this context, this theory presents a justification for public health spending and the necessity of health outcomes arising therefrom. Secondly, the positivity of the benefit theory can be linked to institutional theory. Suffice it to say that benefits from public spending can only have a pass-through effect on the people if strong institutions exist. This benefit (health outcome) accruing from public spending can only be guaranteed by strong institutions with voice and accountability, rule of law, government effectiveness and devoid of corruption (see Debelle et al. 2024). The SDG Goal 3 of UNICEF also points to promotion of healthy lives and well-being for all as a sign of the existence of institutional support for health for all targets. In consistency with this study, the goal addresses major health priorities such as reproductive, maternal, newborn, child and adolescent health; communicable and noncommunicable diseases; universal health coverage; and access for all to safe, effective, quality and affordable medicines and vaccines. UNICEF’s role in collaboration with the government supports the advocacy of institutional theories as a driver of not just health spending but also health outcomes.

With this backdrop, it may be pertinent to ask if these increased spendings are enough to bring about improved population health in SSA countries. Piatti-Fünfkirchen et al. (2018), and Jowett et al. (2016) each suggest that government spending on health should be in the region of 5 percent of GDP, and the Abuja Declaration of 2001 recommended that governments allocate 15 percent of its spendings to the health sector. However, the WHO (2016) revealed that ten years after the commencement of the Abuja Declaration, there was only one country (Tanzania) that had achieved the target and that eleven countries had actually cut the share of government expenditures devoted to healthcare. Kirigia and Barry (2008) and Preker and Langenbrunner (2005) had identified a lack of adequate and sustained levels of resources as the biggest constraint to achieving health outcomes in Africa.

Conversely, Savedoff (2007) noted that no specific level of health spending should be applied across all countries and that factors such as the nature of the health challenges, policy objectives in the health sector, fiscal capacity, and competing demands on public resources need to be considered. Similarly, the WHO (2016) held that even at a comparative level of spending, health results vary considerably across African countries due to major system inefficiencies. In some cases, health outcomes are possible without major increases in spending.

The arguments on spending level bring to the forefront the issues of efficiency, governance, and corruption in health spending. It is evident that funding alone does not determine performance (see Jowett et al. 2016; Tandon and Cashin 2010). Notably, Thoumi and Bandara (2019) held that higher health budgetary allocations will not necessarily translate into an improvement of its outcomes unless specific measures are implemented to correct the underlying inefficiency in spending.

The primary purpose of this paper is to reexamine the evidence concerning the effect of government effectiveness of health expenditures on health outcomes in forty-five African economies using current data, but with a more appropriate testing procedure. The number of studies investigating the impact of government expenditure on health outcomes is quite limited in the specific context of SSA countries and in its mediating role in the provision of health services. Even more limited is the literature which includes government effectiveness in measuring the health spending and health outcomes nexus. This study further extends these previous studies in several additional dimensions.

First, in contrast with previous studies based on pooled data or panel estimation methods, it is well established, as pointed out in Arize et al. (2010), that any panel or pooled data study also suffers from aggregation bias. This stems from the concern that what is valid for one cross-sectional unit may not be accurate for another one.1 Work by Ucal et al. (2016) point out that individual country effects may differ substantially or even cancel each other out to some degree and not be captured by fixed or random effects. They suggest that individual country studies are preferable. The current research follows their suggestion to avoid omitted variable misspecification. We examine whether a long-run relationship exists among health spending, life expectancy, and health outcomes (like maternal mortality rate and infant mortality), with SSA as the geography of interest.

Second, the econometric techniques used in the previous studies of health outcomes, life expectancy, and health spending are generally outdated, lacking new developments in econometrics. For example, flexible cointegration does not typically require a knowledge of the order of integration of variables or specifically handle time series that does not need the classification of variables into I(0), I(1) or partially integrated. We employ the Autoregressive Distributed Lag Model (ARDL) and the Bounds testing approach to cointegration and error-correction modelling proposed by (Pesaran et al. 2001) to compute and test the short- and long-run estimates. The ARDL cointegration method accounts for endogeneity, as well as short and long-run estimates separately in a single-equation model. Therefore, provided the long-run equilibrium exists, we determine the sign, magnitude, and statistical significance of the long-run effects of the cointegrating variables. The importance of the long-run estimates of the relationship between health outcomes and government expenditure would permit the identification of a benchmark against which one can identify the health policy stance adopted by governments.

Third, we investigate the short-run dynamics of the estimating function for each SSA country’s short-run relationship between health spending and health outcomes like maternal mortality rate, infant mortality, and life expectancy. In a policy sense, the short-run adjustment of health outcomes to changes in government effectiveness and other variables is also frequently necessary. How quickly health outcomes respond to changes in their determinants is essential for understanding future effects that may occur because of changes in health care policy and for interpreting recent events (Arize et al. 2015). Our results suggest that there is inter-country variation in the adjustment speed to the last period’s disequilibrium, and that overlooking the cointegrated nature of the variables would have introduced misspecification in the underlying dynamic structure.

Based upon the above empirical and policy inconclusiveness, the focus of this study is to inquire how health expenditures in SSA have impacted health outcomes. Previous studies have presumed (explicitly or implicitly) that the relationship is stable. This is unlikely to be the case. Therefore, we tested for cointegration and investigated whether such a relationship was structurally stable over the sample period. significant evidence of such a relationship must be ascertained by testing for statistical cointegration and, following, an investigation into the stability of the cointegrating relationship. The analysis presented here confirms the importance of government effectiveness in achieving a stable health system.

It is worth noting in this study that government effectiveness in this analysis serves as a perception of government commitment to policies, as well as the quality of civil and public services (Chang et al. 2018; Montes et al. 2019). As emphasized by the World Bank, this governance indicator is vital in assessing the methods of government policy implementation and their effectiveness (World Bank 2020). It is important to note, however, that there has been considerable conflict in the literature as to the true impact of measured government effectiveness.

Some researchers have noted that high scores in government effectiveness may be perceived as favoring countries that are likely to prioritize efficiency over factors such as inclusivity (Andrews and Van de Walle 2013). In certain contexts, this potentially promotes streamlined policy-making over greater public input, de-valuing democratic processes. Of additional concern by some is the potential influence of business interests on governments scored highly for effectiveness, prioritizing economic goals over social equity (Merry 2011). Regions with highly ranked government effectiveness may frame public policy towards enticing foreign investment, possibly disregarding public welfare concerns to some extent. However, recent studies suggest that countries demonstrating highly effective governance do so from targeting public needs, such as health and education, as part of social and economic policy implementations (Floerkemeier et al. 2021).

Health outcomes may also be influenced indirectly by government effectiveness through progress and development of infrastructure and regulatory environments that aid health care delivery (Ssozi and Amlani 2015). This, in turn, should help to improve health outcomes in the country. As illustrated in recent research, infrastructure development has a significant and positive impact on government effectiveness (Kyriacou et al. 2019). This further suggests that enhanced government effectiveness, and the scores reflecting such, affect public health through multiple pathways.

The contributions of this study relate specifically to the Sustainable Development Goals, particularly SDG 8 and 3, which seeks to achieve sustainable, inclusive economic growth, full and productive employment, and support a healthy life for all. Full and productive employment can be achieved if there is a healthy population built on government effective implement of health policies that heighten good health outcomes.

No prior study has considered this topical area on country-specific comparative basis for this number of SSA countries. Africa provides a good regional focus for this study, characterized by low-income levels, large disease burdens, poor health outcomes, and apparent government ineffectiveness. These weak economic and social circumstances can aid the evaluation of, and policy-making towards, government effectiveness in the responsiveness of health outcomes to health expenditure (See, Rana et al. 2018; Baldacci et al. 2003).

2. Literature Review

2.1. Review of the Related Literature

Public expenditures allocate financial resources into sectors decided by the government in accordance with national objectives and priorities for public policy. Thus, the magnitude and direction of such spending may influence the pattern and levels of services available to the citizens. Since the Second World War, there has been continual and substantial increase in the level of government expenditures both in absolute terms and also in relation to the national income. Obviously, it is expected that public spending will positively influence outputs and bring efficient utilization, making welfare maximization a central concept in economic and social activities. According to Pigou (1928), “Expenditure should be so distributed between battleships and poor relief in such wise that the last shilling devoted to each of them yields the same real return”. Thus, serious attention has been focused on the effective utilization of government resources and the importance of avoiding wasteful expenditure. As researched, public spending is expected to produce beneficial results for the population regarding education, health and other social services (See, Anyanwu and Erhijakpor 2009; Barroy et al. 2016).

Theories that explain and justify public expenditure to derive optimal efficiency include Wagner’s (1883), Peacock and Wiseman’s (1967), and the Keynesian propositions, Keynes (1936). Conversely, the classical economists have a critical reservation on the idea of public intervention via spending. Wagner’s law (Wagner 1883) advocates a connection between the growth of an economy and the instantaneous growth of the public spending, He supports the “law of increasing state activity”. However, the law maintains that public expenditure is an endogenous variable, a consequence rather than a cause of economic growth (Wagner 1883). In other words, as the economy develops over time, the spending activities of the government increase. Overall, it generally believes that public expenditure is needed for three main reasons: social, administrative and protective, and welfare activities. Governments will then provide welfare services like education, public health, food, increased subsidy, natural disaster aid, and environmental protection programs.

Peacock and Wiseman (1967), despite some slight divergence about the coveted public spending, still agree on the necessity of public expenditure. They hypothesized that government spending responds to changes in society, that is “displacement effect”. For the classical economists, though, Smith (1776) is of the view that government spending causes more harm than good to an economy and advocates that the private sector should be allowed most of the activities. According to this thought, government spending is dangerous because it stifles private investment.

The Keynesian theory, on the other hand, believes in government’s intervention for correcting inefficiencies and imbalances. It broadly states that public spending is needed to help economies. The role of government should involve increased spending in order to optimize social and economic welfare under this. The theory by Keynes (1936) comprises of models on how public spending in the economy strongly influences output. Further it envisages multiplier effects resulting from public intervention; this suggests an astronomical output. Keynes labels public expenditure as an exogenous variable for growth instead of an endogenous phenomenon. This theory is not without its shortcomings, however, as it leans toward a centrally planned economy where government finds it easy to understand the priorities of the economy. This study will serve as a further test on the relevance of the Keynes proposition especially in developing countries as Sub-Saharan Africa countries fall predominantly into this classification.

2.2. Empirical Review

Related studies on health spending and health outcomes in SSA include works by Micah et al. (2019), Novignon and Lawanson (2017), Odhiambo et al. (2015), Makuta and O’Hare (2015), Novignon et al. (2012), and Mallaye and Yogo (2012). Others are Anyanwu and Erhijakpor (2009). Some of the above studies focused mainly on the impact of health spending on health outcomes with conclusions of positive impact. Some of them used the infant and child mortality rates and life expectancy for measuring health outcomes. Maternal mortality and neonatal rates, though, were rarely employed.

Specifically, Anyanwu and Erhijakpor covered the entirety of African countries with a time span of 5 years (1999–2004). Using robust ordinary least square with variables such as per capita total health expenditure and per capita public health expenditure on infant and under-five mortality, they found that a 10 percent increase in health expenditures reduced infant and under-five mortality by 21 and 25 percent, respectively. This finding is consistent with Novignon et al. (2012) and suggests significant improvements in population health may be found still from increased spending, avoiding the flat-of-the-curve concern faced by many developed nations.

Makuta and O’Hare (2015) factored in the effect of quality of governance and spending on two outcomes (under 5 mortality and life expectancy). Covering 1996 to 2011, the result aligned with the other findings where public spending positively and significantly impacted health outcomes. With respect to under-five mortality, the coefficient of elasticity is between 0.33 and 0.60. The quality of governance was found to have a reducing impact on mortality rate at between −0.17 and −0.19 in most of the investigated countries. This position is canvassed by such other studies as Bein et al. (2017), and Sango-Coke and Bein (2018), Mallaye and Yogo (2012), Micah et al. (2019) and Odhiambo et al. (2015).

Studies outside Africa on the impact of health expenditures on health outcomes have been conducted by (Rana et al. 2018; Nixon and Ulmann 2006; Rahman et al. 2018; Baldacci et al. 2003). Specifically, Bokhari et al. (2007) found that the elasticity estimates of the outcomes were high in magnitude, further supported by Kim and Lane (2013). However, Issa and Quattara (2005) discovered that though health expenditure improves health outcomes, health expenditure per capita is more significant than aggregate public health expenditure. This was supported by Filmer and Pritchett (1999), whose finding revealed that impact of public spending on health is quite small with coefficients that were extremely small and insignificant. Notably, however, Rana et al. (2018) posited that this does not hold for maternal mortality. Similarly, Nixon and Ulmann (2006) determined health expenditures improved health outcomes but with very low impact on life expectancy. Rana et al. (2018) when they found that the impact of public health expenditures on health outcomes was stronger in low-income countries than in higher income countries.

In developed countries, there are a good number of works on the role of quality of governance on health outcome. Virtually all works reviewed in this paper on the role of governance in facilitating the efficacy of public spending showed convergence of findings that countries with good quality governance significantly improved health outcomes. All the works reviewed used OLS as the technique for data analysis while focusing on infant, under-five, and maternal mortality, and life expectancy (Kim and Wang 2019).

In contrast, however, there are some findings that public health spending does not significantly improve health outcomes. For instance, Kamiya (2010), using GMM on data across 141 developing countries, found that spending did not lead to mortality reduction. This is also corroborated in the findings of Politi et al. (1995) who concluded that health spending had no statistically significant effect on health outcomes.

In sum, the lack of consensus in the previous studies emphasizes the need for better knowledge and understanding of this relationship. It is essential to not only investigate the health expenditure and health outcome linkage, but also to do so in the light of government effectiveness in managing health spending in a disease prevalent region like SSA.

3. Materials and Methods

This study focuses on establishing a relationship between health expenditure (HEXP) and health outcomes (HO) expressed thus:

$${HO}_t=a_t+{\mathrm{H}\mathrm{E}\mathrm{X}\mathrm{P}}_t+GEFF+{\epsilon }_t$$

(1)

where ${HO}_t$ represents the natural logarithm of health outcomes (infant mortality—INFMORT, life expectancy—LEXP and maternal and adult mortality rate—MRAD); ${\mathrm{H}\mathrm{E}\mathrm{X}\mathrm{P}}_t$ represents health expenditure disaggregated into total health expenditure scaled by GDP (THEGDP) and public health expenditure per capita (HEPC) and $GEFF$ is government effectiveness2. The variables are described in

Table 1. Description of the Model Variables.

S/No	Variable Name /Description	Notation	Source of Data	Role
1	Infant Mortality The infant mortality rate is the number of infants dying before reaching one year of age, per 1000 live births in a given year.	INFMORT	United Nations Population Division. World Population Prospects: 2022 Revision referred from World Development Indicator	Dependent Variable
2	Maternal and Adult Mortality The maternal and adult mortality rate is the probability of dying between the ages of 15 and 60—that is, the probability of a 15-year-old dying before reaching age 60, if subject to age-specific mortality rates of the specified year between those ages.	MAdMR	United Nations Population Division. World Population Prospects: 2022 Revision referred from World Development Indicator	Dependent Variable
3	Life Expectancy (Total) Life expectancy at birth indicates the number of years a newborn infant would live if prevailing patterns of mortality at the time of its birth were to stay the same throughout its life.	LEXPTO	United Nations Population Division. World Population Prospects: 2022 Revision referred from World Development Indicator	Dependent Variable
4	Total Health Expenditure to Gross Domestic Product Level of current health expenditure expressed as a percentage of GDP. Estimates of current health expenditures include health care goods and services consumed during each year. This indicator does not include capital health expenditures such as buildings, machinery, IT and stocks of vaccines for emergency or outbreaks.	THEGDP	World Health Organization Global Health Expenditure database retrieved through World Development Indicator (WDI)	Independent Variable
5	Health Expenditure Per Capital Current expenditures on health per capita in current US dollars. Estimates of current health expenditures include health care goods and services consumed during each year.	HEPC	World Health Organization Global Health Expenditure database retrieved through World Development Indicator (WDI)	Independent Variable
6	Government Effectiveness Government effectiveness captures perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government’s commitment to such policies. Estimate gives the country’s score on the aggregate indicator, in units of a standard normal distribution, i.e., ranging from approximately −2.5 to 2.5.	GEFF	Detailed documentation of the World Governance Indicator, retrieved through World Development Indicator (WDI)	Independent Variable

Source: Authors’ summary.

and the model is presented in full functional and estimable forms as:

For the response variables that are centered on mortality (adult, maternal and infant), the a priori expectation is negative; and for life expectancy and government effectiveness, the a priori expectation is positive.

$${\mathsf{\Delta }HO}_t={\pi }_p+\sum _{i=1}^k{\delta }_{ip}{\mathsf{\Delta }HO}_{t-i}+\sum _{i=1}^k{\tau }_{ip}{\mathsf{\Delta }THEPGDP}_{t-i}+\sum _{i=1}^k{\theta }_{ip}{\mathsf{\Delta }LHEPC}_{t-i}+\sum _{i=1}^{k3}{\sigma }_{ip}{\mathsf{\Delta }LGEFF}_{t-i}+{\varpi }_{1p}{HO}_{t-1}+{\varpi }_{2p}{LTHEGDP}_{t-1}+{\varpi }_{3p}L{HEPC}_{t-1}+{\varpi }_{4p}{LGEFF}_{t-1}+{\xi }_{1t}$$

(2)

Three equations are estimated with the three response variables (infant mortality—INFMORT, life expectancy—LEXP and maternal and adult mortality rate—MAMR) using the same set of explanatory variables as contained in Equation (2).

Our estimation follows a country-specific comparative approach using the Autoregressive Distributed Lag cointegration model by (Pesaran et al. 2001) to simultaneously evaluate long- and short-run effects. This is in addition to the bound testing approach, which is a generalized Dickey–Fuller type regression that tests the significance of lagged levels of variables in a conditionally unrestricted error correction model (UECM). This is a preferred estimation method because of its comparative advantages.3 The first step in the estimation involves the determination of cointegration using the bound testing approach following the FPSS critical values provided by (Pesaran et al. 1997) and Pesaran et al. (2001). The decision rule is of the form contained in

Table 2. Summary ARDL Decision Rule.

State	Decision	Conclusions
FPSS > I(1)	HO is rejected	Cointegration
FPSS < I(0) and I(1)	HO cannot be rejected	No cointegration
FPSS within I(1) and I(0)	Inconclusive	Inconclusive

ARDL = Autoregressive Distributed Lag model; FPSS = F-stat of (Pesaran et al. 2001).

.

However, the test statistic is sensitive to the number of lags used, hence, using fewer lags may not capture every important information from the model (Bahmani-Oskooee and Bohl 2000). As indicated by Stock and Watson (2012), too many lags overfits the model. Secondly, the long-run coefficients that relate to the exogenous variables with fixed lags as follows are determined:

$$\widehat{\psi }=\frac{\widehat{\delta }(\widehat{p},{\widehat{q}}_1,{\widehat{q}}_2,\dots ,{\widehat{q}}_k)}{1-{\widehat{\varphi }}_1-{\widehat{\varphi }}_2-\dots -{\widehat{\varphi }}_p}$$

(3)

where $\widehat{\delta }(\widehat{p},{\widehat{q}}_1,{\widehat{q}}_2,\dots ,{\widehat{q}}_k)$ are the ordinary least square (OLS) estimates for the Autoregressive Distributed Lag model (ARDL) framework (Arize 2017).

The unrestricted error correction regressions are estimated to further confirm cointegration among the variables. In addition, a selection of tests is carried out to confirm validity of our estimates.

4. Results

First, we present the summary statistics in pooled form for all the investigated series. The distributional characteristics are shown majorly by the mean, standard deviation, and the relative standard deviation (RSD). This is shown in

Table 3. Summary of Panel Descriptive Statistics.

Variable Name	Mean	Median	Std. Dev.	Skewness	Kurtosis	Jarque–Bera	RSD
GEFF	−0.39	0.00	0.60	−0.91	3.13	331.78	−1.53
HEPC	38.18	0.00	99.69	4.17	22.39	44,539.16	2.61
INFMORT	79.56	74.40	40.86	0.23	2.69	30.18	0.51
LEXPTO	52.56	53.28	11.32	−2.11	11.59	9160.64	0.22
MATMORT	210.38	0.00	347.87	1.81	6.46	2511.29	1.65
THEGDP	2.03	0.00	2.96	1.36	4.56	984.41	1.46

below:

With due attention on the coefficient of variation, it can be said that the health outcome proxies like infant mortality and life expectancy are not highly dispersed as their RSD fall below one (1). All the regressors have RSD above unity which shows the volatility of government expenditure and effectiveness.

The linear association of the variables is shown by the panel correlation matrix reported in

Table 4. Panel Correlation Matrix.

Correlation
t-Statistic
Probability	GEFF	HEPC	INFMORT	LEXPTO	MATMORT	THEGDP
GEFF	1.000000
	-----
	-----
HEPC	0.080953	1.000000
	3.977252	-----
	0.0001	-----
INFMORT	0.113309	−0.343612	1.000000
	5.584645	−17.91745	-----
	0.0000	0.0000	-----
LEXPTO	0.001489	0.300848	−0.291878	1.000000
	0.072897	15.44800	−14.94383	-----
	0.9419	0.0000	0.0000	-----
MATMORT	−0.553341	0.082174	−0.092303	0.099220	1.000000
	−32.53089	4.037655	−4.539381	4.882843	-----
	0.0000	0.0001	0.0000	0.0000	-----
THEGDP	−0.364185	0.498798	−0.338484	0.300746	0.604553	1.000000
	−19.14893	28.18199	−17.61513	15.44223	37.16527	-----
	0.0000	0.0000	0.0000	0.0000	0.0000	-----

.

A negative correlation is found between maternal mortality and government effectiveness as well as total health expenditure. This implies that government effectiveness varies inversely with maternal mortality and health spending. Life expectancy positively correlates with government effectiveness and government expenditure. Life expectancy rises side by side with government effectiveness and health expenditure per head.

Without reporting every result, it is worthy of note that none of our time series variables are integrated of order two (I(2)). The remaining results are in

Table A1. Autoregressive Distributed Lag Model Estimates for the 45 Studied African Countries.

Countries/Models	Coefficients Estimates			Diagnostic Tests
	THEGDP	HEPC	GEFF	${\mathsf{\lambda }\left(\mathbf{t}-\mathbf{t}\mathbf{e}\mathbf{s}\mathbf{t}\right)}^{\mathbf{a}}$	${\mathbf{F}}^{\mathbf{b}}$	${\mathbf{L}\mathbf{M}}^{\mathbf{c}}$	${\mathbf{R}\mathbf{E}\mathbf{S}\mathbf{E}\mathbf{T}}^{\mathbf{d}}$	${\mathbf{H}\mathbf{E}\mathbf{T}}^{\mathbf{e}}$	${\mathbf{S}\mathbf{T}\mathbf{A}\mathbf{B}}^{\mathbf{f}}$	${\overline{\mathbf{R}}}^2$
ANGOLA:
Model 1	−3.55 (2.96) **	−0.34 (30.92) **	−34.71 (6.19) **	−0.12 (22.12) **	69.88	1.88	0.84	1.59 (0.35)	Stable	0.99
Model 2	−52.62 (4.18) *	−52.23 (8.31)	−90.75 (2.64)	−0.02 (17.04) **	29.05	25.09	3.02	0.65 (0.75)	Stable	0.99
Model 3	3.26 (3.45) **	2.04 (3.50) **	0.06 (0.05)	−0.01 (6.55) **	8.71	1.10	1.66	0.28 (0.94)	Stable	0.99
BENIN:
Model 1	10.45 (2.25) *	−49.95 (32.81) **	−0.30 (0.02)	−0.72 (13.93) **	21.56	3.89	0.17	1.84	Stable	0.98
Model 2	109.01 (3.46) **	−72.11 (14.21) **	−25.23 (1.01)	0.08 (6.60) **	5.44	5.45	0.55	3.32 (0.18)	Stable	0.98
Model 3	−9.86 (7.18) **	6.39 (33.71) **	−1.76 (2.11)	0.13 (2.9)	107.55	2.47	8.73	2.17 (0.29)	Stable	1.00
BOTSWANA
Model 1	−0.20 (0.58)	−0.43 (3.13) **	0.37 (2.18) **	−0.18 (6.63) **	8.23	2.19	1.34	1.34	Stable	0.95
Model 2	0.39 (2.29) **	−0.76 (9.92) **	−0.38 (1.56)	−0.26 (6.35) **	7.05	4.81	0.78	0.18	Stable	0.95
Model 3	0.02 (1.05)	0.18 (19.41) **	0.08 (2.39) **	−0.13 (32.31) **	189.78	1.01	1.63	1.18	Not Stable	0.99
BURKINA FASO
Model 1	0.23 (7.76) **	−0.32 (19.71) **	0.68 (4.93) **	0.30 (9.21) **	16.31	0.79	2.37	0.78	Stable	0.87
Model 2	−0.15 (0.47)	−0.97 (17.41) **	1.67 (3.58) **	−0.07 (37.03) **	249.26	2.16	6.83	1.24	Stable	0.99
Model 3	−0.21 (7.70) **	0.21 (17.02) **	0.06 (1.50)	0.19 (11.24) **	23.71	2.59	90.76	1.18	Stable	0.93
BURUNDI
Model 1	−0.13 (1.74)	−0.09 (1.84) *	−0.11 (1.74)	−0.10 (10.36) **	20.65	4.06	0.56	1.46	Stable	0.88
Model 2	−0.98 (3.21) **	0.50 (1.42)	0.63 (1.42)	−0.07 (5.79) **	6.10	2.23	151.10	2.87	Stable	0.82
Model 3	−0.09 (1.18)	0.19 (2.12) *	−0.05 (0.50)	−0.07 (9.36) **	15.33	0.35	2.91	0.50	Stable	0.86
CAMEROUN
Model 1	0.86 (12.12) **	−0.02 (4.96) **	−1.45 (3.67) **	−0.03 (29.82) **	123.14	0.16	0.77	0.17	Stable	0.91
Model 2	0.06 (0.01) **	−0.01 (6.25) **	0.06 (5.00)	−0.22 (15.96) **	36.38	1.98	1.09	0.83	Stable	0.94
Model 3	−0.06 (4.97) **	0.01 (3.75) **	−0.04 (1.21)	−0.11 (22.72)	68.85	0.99	0.32	0.78	Stable	0.97
CABO VERDE
Model 1	−14.94 (2.69) **		−9.25 (1.64)	−0.23 (20.96) **	55.92	0.88		0.51	Stable	0.96
Model 2	7.97 (0.64)	0.28 (3.44) **	12.33 (2.37) **	0.01 (31.82) **	194.77	0.34	0.10	1.87	Stable	0.91
Model 3	−0.12 (4.60) **	−0.28 (36.16) **	−0.18 (8.50) **	0.56 (31.83) **	101.32	1.32	1.49	1.04	Stable	0.98
CENTRAL AFRICAN REPUBLIC
Model 1	0.86 (12.12) **	−0.02 (4.96) **	−1.45 (3.67) **	−0.03 (29.82) **	123.14	0.16	0.77	0.17	Stable	0.91
Model 2	0.06 (0.01) **	−0.01 (6.25) **	0.06 (5.00)	−0.22 (15.96) **	36.38	1.98	1.09	0.83	Stable	0.94
Model 3	−0.06 (4.97) **	0.01 (3.75) **	−0.04 ( (1.21)	−0.11 (22.72)	68.85	0.99	0.32	0.78	Stable	0.97
CHAD
Model 1	−4.18 (1.27)	4.34 (2.50) *	−1.01 (0.80)	0.002 (17.83) **	62.45	0.65	1.59	3.92	Stable	0.67
Model 2	−1.22 (1.70)	1.79 (5.23) **	−1.45 (1.70) **	0.01 (37.81) **	280.81	0.78	2.98	0.30	Stable	0.95
Model 3	−2.33 (3.56) **	2.29 (7.16) **	−1.22 (4.69) **	−0.002 (42.36) **	352.44	0.06	0.02	0.97	Stable	0.94
COMOROS
Model 1	−4.18 (1.27)	4.34 (2.50) *	−1.01 (0.80)	0.002 (17.83) **	62.45	0.65	1.59	3.92	Stable	0.67
Model 2	−1.22 (1.70)	1.79 (5.23) **	−1.45 (1.70) **	0.01 (37.81) **	280.81	0.78	2.98	0.30	Stable	0.95
Model 3	−2.33 (3.56) **	2.29 (7.16) **	−1.22 (4.69) **	−0.002 (42.36) **	352.44	0.06	0.02	0.97	Stable	0.94
CONGO
Model 1	93.81 (2.37) **	−112.82 (7.11) **	27.00 (0.31)	−0.27 (5.88) **	6.66	0.04	0.23	0.85	Stable	0.73
Model 2	51.48 (3.92) **	−53.26 (11.48) **	−9.00 (0.30)	−0.35 (6.49) **	7.65	0.38	0.20	0.75	Stable	0.84
Model 3	−10.15 (2.47) **	10.29 (5.41) **	0.97 (0.15)	−0.18 (8.86) **	14.28	0.15	0.24	0.54	Stable	0.90
COTE D’IVORIE
Model 1	18.99 (5.91) **	−1.67 (16.45) **	24.95 (4.51) **	−0.62 (29.50) **	108.76	2.48	1.52	25.20	Stable	1.00
Model 2	13.66 (2.11) **	−10.86 (2.70) **	0.29 (0.24)	−0.25 (9.02) **	16.73	4.22	0.25	2.03	Stable	0.94
Model 3	−2.24 (7.93) **	0.13 (45.73) **	−1.43 (3.47) **	−0.11 (73.12) **	668.26	27.00	7.10	5.29	Stable	1.00
EQUITORIAL GUINEA
Model 1	−1.87 (2.54) **	0.58 (2.38) **	0.84 (0.56)	−0.01 (6.00) **	7.82	1.97	1.43	1.31	STABLE	0.57
Model 2	210.6 (16.55) **	−5.00 (0.85) **	−15.34 (13.45) **	−0.13 (3.73) **	11.29	0.69	0.54	0.92	STABLE	0.99
Model 3	0.02 (0.37)	−0.21 (2.08)	−0.05 (1.62)	−0.05 (2.33) **	11.29	2.59	24.92	0.92	STABLE	0.08
ERITREA
Model 1	440.86 (2.38) **	−40.86 (6.06) **	110.83 (2.20) **	−0.57 (14.79) **	51.89	2.69	1.14	1.05	STABLE	0.94
Model 2	−924.2.84) **	−433.98 (4.54) **	−1215.03 (4.11) **	−0.92 (7.73) **	14.18	0.93	0.001	0.36	STABLE	0.50
Model 3	−11.53 (5.71) **	4.63 (2.47) **	1.98 (0.51)	−0.86 (12.61) **	35.67	0.31	75.43	1.51	STABLE	0.97
ETHIOPIA
Model 1	43.73 (5.56) **	−4.93 (7.61) **	−63.92 (4.82) **	−0.77 (16.71) **	51.47	0.76	1.41	0.83	STABLE	0.99
Model 2	25.71 (5.70) **	−2.21 (6.43) **	−44.13 (5.09) **	−0.46 (11.76) **	25.37	1.72	1.58	1.34	STABLE	0.99
Model 3	−3.72 (6.88) **	0.37 (7.28) **	4.72 (4.97) **	−0.87 (21.19) **	82.77	1.11	1.33	0.66	STABLE	0.99
GABON
Model 1	22.31 (3.88) **	0.03 (2.12) **	35.14 (4.84) **	−0.15 (4.03)	2.95	0.75	1.32	0.76	STABLE	0.93
Model 2	15.35 (5.79) **	0.01 (2.10) **	17.17 (5.60) **	−0.39 (4.27) **	3.31	0.54	1.19	1.64	STABLE	0.97
Model 3	2.94 (3.95) **	0.00 (0.40)	47.06 (15.52) **	−0.25 (4.97) **	5.78	0.20	1.67	1.12	-	-
GAMBIA
Model 1	−3.74 (3.15) **	−0.58 (12.80) **	12.53 (1.89) *	−0.83 (40.12) **	586.75	0.65	141.15	2.30	STABLE	0.98
Model 2	−38.60 (11.59) **	0.18 (5.96) **	62.56 (4.86) **	−0.49 (4.31) **	4.35	1.11	9.41	1.47	STABLE	0.91
Model 3	0.12 (1.67)	0.05 (12.89) **	−0.78 (1.35)	−0.94 (44.53) **	574.46	1.27		2.72	STABLE	0.98
GHANA
Model 1	7.38 (2.89) **	−1.39 (2.89) **	−5.73 (5.24) **	−0.90 (4.05) **	3.38	0.13	6.56	0.97	STABLE	0.80
Model 2	3.36 (9.32) **	−0.07 (0.49)	−2.36 (1.88) *	−0.68 (4.23) **	4.20	0.52	0.24	1.44	STABLE	0.57
Model 3	3.35 (2.27) **	−0.09 (0.14)	−4.17 (2.53) **	0.66 (4.62)	15.19	0.32	1.65	0.99	STABLE	0.74
GUINEA
Model 1	−0.09 (2.31) **	−0.05 (2.03) **	0.12 (5.04) **	−0.01 (2.13) **	1.07	2.92	13.36	2.17	STABLE	0.79
Model 2	−1.58 (5.14) **	0.74 (3.75) **	0.60 (3.92) **	−0.85 (11.54) **	24.77	0.49	0.04	5.87	STABLE	0.85
Model 3	0.27 (5.23) **	−0.09 (2.51) **	−0.11 (5.40) **	−0.86 (14.12) **	37.12	2.03	2.89	2.53	STABLE	0.93
GUINEA BISSAU
Model 1	0.09 (3.42) **	−0.04 (6.35) **	0.01 (0.72)	−0.79 (4.92)) **	21.74	0.92	0.00	1.24	STABLE	0.90
Model 2	−0.03 (0.16)	−0.47 (13.64) **	0.03 (0.59)	−0.92 (29.22) **	248.14	2.01	1.34	3.14	STABLE	0.95
Model 3	−0.33 (0.43)	1.28 (2.96) **	−0.30 (1.07)	−0.04 (10.74) **	27.25	3.78	9.74	0.28	STABLE	0.80
KENYA
Model 1	0.20 (2.78) **	−0.36 (24.08) **	0.05 (1.23)	−0.97 (64.52) **	769.73	0.14	3.89	1.39	STABLE	0.99
Model 2	−21.31 (5.82) **	3.32 (5.53) **	19.49 (5.09) **	−0.81 (6.63) **	7.82	0.72	0.94	0.79	STABLE	0.75
Model 3	−0.13 (5.43) **	0.15 (28.47) **	0.03 (2.31) **	−0.92 (8.17) **	21.95	1.20	0.00	2.06	STABLE	0.99
LESOTHO
Model 1	−3.57 (5.05) **	2.63 (5.21) **	4.94 (5.38) **	−0.77 (3.24) **	7.08	3.24	8.24	1.01	STABLE	0.91
Model 2	−0.48 (5.56) **	−0.01 (0.33)	−0.24 (2.55) **	−0.93 (11.56) **	43.74	0.31	10.11	2.48	STABLE	0.92
Model 3	2.31 (4.94) **	−1.64 (4.91) **	−3.05 (5.05) **	−0.80 (3.11) **	6.19	3.82	15.92	0.88	STABLE	0.98
LIBERIA
Model 1	−0.04 (3.43) **	−0.01 (6.93) **	0.19 (1.97) **	−0.98 (20.01) **	73.39	1.88	18.74	3.51	STABLE	0.96
Model 2	0.44 (7.53) **	−0.61 (8.89) **	−0.49 (7.53) **	−0.98 (15.08) **	53.37	0.30	142.69	1.79	STABLE	0.99
Model 3	0.47 (6.33) **	0.16 (25.83) **	2.14 (3.43) **	−0.84 (6.99) **	8.94	2.02	4.21	1.15	STABLE	1.00
MAURITANIA
Model 1	−0.55 (0.51)	1.26 (4.40) **	1.27 (5.71) **	−0.66 (6.04) **	8.46	0.93	0.04	0.34	STABLE	0.66
Model 2	−0.41 (2.32)	−0.25 (3.73) **	−0.18 (3.91) **	−0.87 (11.43) **	24.31	1.04	75.50	4.92	STABLE	0.94
Model 3	−0.04 (2.33) *	0.05 (7.88) **	0.01 (3.46) **	−0.90 (18.00) **	60.31	0.31	0.35	1.22	STABLE	0.97
MADAGASCAR
Model 1	0.12 (1.75)	−0.17 (6.79) **	0.12 (4.37) **	0.89 (20.25) **	76.24	0.08	11.26	10.20	STABLE	0.90
Model 2	0.99 (2.06) **	−0.75 (9.22) **	0.84 (2.98) *	−0.97 (19.98) **	74.22	0.36	0.93	10.08	STABLE	0.89
Model 3	1.55 (7.00) **	0.50 (3.94) **	−0.55 (2.54) *	−0.24 (4.65) **	4.88	0.22	5.40	2.27	STABLE	0.94
MALAWI
Model 1	0.05 (1.43)	−0.14 (5.26) **	−0.85 (2.94) *	−0.18 (21.05) **	55.37	7.69	0.18	0.58	STABLE	0.99
Model 2	−1.44 (1.73)	0.97 (2.41) *	−1.40 (5.77) **	−0.32 (8.59) **	621.18	0.16	7.73	0.93	STABLE	0.91
Model 3	−0.14 (4.98) **	0.14 (8.34) **	0.02 (0.76)	−0.32 (25.26) **	81.24	1.79	37.21	0.39	STABLE	0.95
MALI
Model 1	0.06 (3.91) **	−0.01 (24.56) **	−0.10 (5.67) **	−0.82 (42.35) **	514.61	0.32	1.50	2.24	STABLE	0.98
Model 2	0.18 (6.73) **	−0.02 (43.56) **	−0.21 (3.45) **	−0.29 (2.45) **	22.34	0.33	1.03	0.62	STABLE	0.99
Model 3	−0.03 (6.46) **	0.01 (33.18) **	0.09 (4.68) **	−0.91 (76.22) **	1321.58	0.18	0.05	1.16	STABLE	1.00
MAURITIUS
Model 1	−0.02 (2.62) **	−0.00 (4.46) **	−0.12 (3.93) **	−0.78 (10.74) **	49.03	1.58	38.53	11.32	STABLE	0.97
Model 2	−0.47 (1.25)	1.11 (3.34) **	−2.00 (2.37) **	−0.14 (2.48) **	1.45	2.08	5.90	2.83	STABLE	0.88
Model 3	−0.00 (0.22)	6.45 (5.15) **	0.02 (3.07) **	−0.69 (22.59) **	237.89	0.46	1.09	15.01	STABLE	0.96
MOZAMBIQUE
Model 1	0.01 (2.83) **	−0.13 (10.45) **	−0.26 (7.17) **	−0.20 (2.74) **	35.46	0.32	19.74	3.05	STABLE	0.76
Model 2	0.27 (3.73) **	−0.61 (24.87) **	0.30 (3.73) **	−0.31 (2.38) **	19.10	0.88	0.85	0.95	STABLE	0.98
Model 3	−0.04 (2.35) **	0.10 (21.61) **	−0.05 (4.86) **	−0.35 (2.80) **	21.03	0.41	0.10	0.86	STABLE	0.97
NAMIBIA
Model 1	−0.18 (1.01)	−0.23 (6.12) **	0.33 (2.40) **	−0.19 (7.06) **	16.09	0.52	0.77	1.45	STABLE	0.98
Model 2	0.19 (0.59)	−0.36 (3.85) **	−0.85 (2.96) **	−0.59 (2.76) **	5.47	0.21	6.66	0.86	STABLE	0.99
Model 3	0.16 (2.10) **	0.09 (3.26) **	0.15 (3.26) **	−0.85 (10.95) **	22.32	0.41	0.27	3.33	STABLE	0.97
NIGER
Model 1	0.36 (12.04) **	−0.14 (8.05) **	0.00 (0.06)	−0.56 (20.66) **	370.54	0.33	98.09	1.84	STABLE	0.96
Model 2	0.15 (10.31) **	−0.04 (12.16) **	−0.41 (4.98) **	−0.40 (8.51) **	103.59	0.13	4.62	3.25	STABLE	0.98
Model 3	−0.04 (10.39) **	0.01 (9.87) **	0.13 (5.77) **	−0.44 (21.94) **	501.03	0.43	18.16	6.37	STABLE	0.98
NIGERIA
Model 1	−0.00 (0.38)	−0.06 (8.73) **	0.11 (2.99) **	−0.71 (5.48) **	13.42	0.02	12.84	0.31	STABLE	0.95
Model 2	−0.35 (2.10) **	−0.27 (6.63) **	0.49 (2.41) **	−0.38 (2.83) **	11.72	0.57	12.91	3.01	STABLE	0.92
Model 3	0.01 (1.37)	0.08 (8.01) **	0.09 (1.37)	0.76 (5.74) **	13.96	0.13	6.44	2.26	STABLE	0.94
RWANDA
Model 1	−0.26 (4.51) **	−0.25 (8.48) **	−0.04 (3.77) **	−0.93 (6.70) **	8.19	0.04	41.07	1.85	STABLE	0.99
Model 2	−0.53 (2.66) **	−0.66 (4.13) **	−0.56 (3.88) **	−0.77 (11.09) **	22.50	0.38	7.54	1.05	STABLE	0.99
Model 3	0.29 (3.93) **	0.12 (2.10) **	0.12 (4.53) **	−0.99 (6.82) **	8.18	1.14	16.46	1.10	STABLE	0.99
SAOTOME
Model 1	0.01 (5.03) **	−0.09 (8.51) **	0.16 (3.52) **	−0.93 (27.62) **	140.87	0.26	3.28	2.48	STABLE	0.97
Model 2	−0.04 (4.02) **	−0.45 (7.90) **	−0.39 (2.93) **	−0.92 (26.01) **	124.92	0.03	7.81	4.13	STABLE	0.96
Model 3	−0.00 (5.27) **	0.04 (8.40) **	−0.24 (8.66) **	−0.92 (27.26) **	137.19	0.12	7.15	3.28	STABLE	0.97
SENEGAL
Model 1	0.35 (6.43) **	−0.17 (6.55) **	0.09 (3.03) **	−0.93 (26.89) **	133.74	0.07	10.69	1.22	STABLE	0.97
Model 2	1.19 (4.84) **	−0.84 (6.31) **	0.45 (3.03) **	−0.93 (25.78) **	122.88	0.37	4.60	0.92	STABLE	0.96
Model 3	−0.18 (5.02) **	0.12 (88.58) **	−0.06 (3.06) **	−0.93 (25.29) **	121.12	0.34	6.67	0.90	STABLE	0.96
SEYCHELLES
Model 1	0.12 (5.87) **	−0.09 (5.30) **	−0.06 (3.58) **	−0.78 (10.34) **	44.96	0.14	0.61	4.95	STABLE	0.87
Model 2	0.20 (4.85) **	−0.00 (6.86) **	−0.53 (6.95) **	−0.97 (24.03) **	106.57	0.35	14.33	0.62	STABLE	0.93
Model 3	−0.94 (7.68) **	0.00 (3.69) **	−0.62 (0.88) **	−0.99 (15.87) **	46.77	1.68	0.34	2.90	STABLE	0.85
SIERRA LEONE
Model 1	0.24 (3.48) **	−0.00 (7.72) **	−0.25 (5.09) **	−0.81 (18.72) **	122.31	0.02	212.38	1.82	STABLE	0.95
Model 2	0.30 (2.88) **	−0.01 (11.13) **	−0.20 (2.91) **	−0.74 (22.60) **	200.42	0.38	40.04	2.37	STABLE	0.97
Model 3	0.19 (2.84) **	−0.00 (9.53) **	0.28 (4.63) **	−0.87 (20.10) **	103.56	0.75	156.92	2.61	STABLE	0.95
SOUTH AFRICA
Model 1	−1.28 (5.29) **	−0.16 (3.31) **	−0.42 (3.86) **	−0.78 (9.76) **	17.62	0.74	80.25	1.47	STABLE	0.89
Model 2	−2.07 (12.10) **	−0.31 (7.39) **	0.18 (2.35) **	−0.82 (27.78) **	223.09	1. 15	11.52	1.61	STABLE	0.95
Model 3	1.29 (5.40) **	0.16 (2.25) **	0.55 (3.08) **	−0.23 (5.87) **	8.14	1.36	7.14	1.09	STABLE	0.87
SOUTH SUDAN
Model 1	260.67 (4.64) **	−12.71 (2.34) **	-	−0.16 (8.28) **	22.10	1.16	331.07	2.49	STABLE	0.54
Model 2	3.16 (6.32) **	−0.33 (11.17) **	34.51 (8.32) **	−1.14 (36.14) **	242.62	1.09	2.06	3.20	STABLE	0.98
Model 3	−282.82 (1.70) *	59.03 (1.73) *	-	−0.004 (6.38) **	12.96	193.37	5.76	2.26	STABLE	0.39
SUDAN
Model 1	−0.03 (3.15) **	−0.05 (8.37) **	0.24 (12.03) **	−0.94 (30.60) **	221.71	0.06	28.99	0.81	STABLE	0.99
Model 2	0.04 (2.47) **	−0.14 (13.65) **	0.42 (10.25) **	−0.96 (23.44) **	102.35	0.60	4.76	1.40	STABLE	0.99
Model 3	−0.003 (0.82)	0.03 (12.55) **	−0.10 (10.05) **	−0.96 (22.03) **	90.37	0.66	7.19	1.54	STABLE	0.99
TANZANIA
Model 1	0.03 (2.80) **	−0.02 (13.22) **	−0.27 (3.24) **	−1.19 (11.02) **	22.51	0.28	130.96	2.23	STABLE	0.97
Model 2	0.05 (2.59) **	−0.03 (14.99) **	−0.67 (4.25) **	−1.12 (12.23) **	27.84	0.97	160.83	1.73	STABLE	0.97
Model 3	−0.01 (2.22) **	0.01 (13.67) **	0.18 (3.92) **	−1.12 (11.76) **	25.74	0.64	137.55	2.26	STABLE	0.96
TOGO
Model 1	1.26 (3.31) **	0.60 (4.73) **	−1.60 (4.55) **	−0.57 (4.87) **	5.50	1.88	23.52	1.01	STABLE	0.49
Model 2	0.88 (2.15) *	0.56 (3.75) **	−1.48 (2.85) **	−0.61 (7.29) **	12.52	0.64	2.04	0.96	STABLE	0.52
Model 3	−0.07 (0.68)	−0.05 (2.02) *	−0.02 (0.68)	−0.96 (8.20)	12.55	0. 59	1.14	1.85	STABLE	0.53
UGANDA
Model 1	−0.47 (2.53) *	0.08 (1.36)	0.82 (3.07) *	−1.33 (5.33) **	5.11	0.34	2.90	1.54	STABLE	0.87
Model 2	−8.10 (3.15) **	21.67 (2.02) **	148.98 (3.95) **	−1.89 (6.98) **	8.74	0.89	3.89	1.92	STABLE	0.86
Model 3	−0.07 (1.11)	−0.002 (2.12) *	−0.02 (0.68)	−0.96 (8.26) **	12.73	0.66	0.74	1.75	STABLE	0.54
ZAMBIA
Model 1	0.05 (16.47) **	−0.28 (10.90) **	−0.15 (3.37) **	−1.35 (9.71) **	16.82	1.66	1.01	11.76	STABLE	0.99
Model 2	0.05 (12.10) **	−0.55 (16.59) **	−0.21 (2.70) **	−1.14 (22.46) **	91.90	0.10	0.17	0.88	STABLE	0.99
Model 3	−0.03 (15.59) **	0.18 (17.23) **	0.11 (4.69) **	−1.08 (30.14) **	165.94	0.83	1.15	0.91	STABLE	0.99
ZIMBABWE
Model 1	0.03 (0.29)	−0.33 (8.39) **	0.03 (0.44)	−1.13 (14.45) **	38.96	0.16	99.35	3.25	STABLE	0.78
Model 2	0.02 (0.45)	−0.21 (8.42) **	0.02 (0.45) **	1.14 (14.50) **	39.20	0.23	145.22	3.08	STABLE	0.78
Model 3	−0.01 (0.16)	0.19 (7.56) **	−0.02 (0.47)	−1.13 (13.19) **	32.44	0.20	83.54	4.73	-	0.75

Model 1: MRAdMa = f(THEGDP, HEPC, GEFF) Model 2: INFMORT = f(THEGDP, HEPC, GEFF) Model 3: LEXPTO = f(THEGDP, HEPC, GEFF). Notes: The numbers in parentheses beside the estimated coefficients are the absolute t-statistic. An * and ** indicate significance at the 10% and 5% levels, respectively. ^a. The upper boundary critical value of the F-test for cointegration when there is one exogenous variable is 4.78 (5.73) at the 10% (5%) level of significance. These come from (Pesaran et al. 2001), (Table CI, Case III, p. 300). ^b. The upper bound critical value of the t-test is −2.91 (−3.22) at the 10% (5%) level when k = 1. These come from (Pesaran et al. 2001), (Table CII, Case III, p. 303). ^c. LM is Lagrange Multiplier test of residual serial correlation. It is distributed as x−2 with four degrees of freedom (fourth order). Its critical value at 5% (10%) significance level is 9.48 (7.78). ^d. RESET is Ramsey’s test for misspecification. It is distributed as x−2 with two degrees of freedom. The critical value is 5.99 at the 5% level and 2.70 at the 10% level. ^e. White’s Heteroskedasticity Test. ^f. CS/CS2 refers to CUSUM and CUSUMQ tests for stability, respectively.

in Appendix A, while a summary of the results is shown in a cluster of the form reported in

Table 5. Summary of the Estimation Results (Full Result in Table A1 in Appendix A).

	HEPC	THEGDP vs. MRAD	GEFF vs. MRAD	HEPC vs. INFMORT	THEGDP vs. INFMORT	GEFF vs. INFMORT	HEPC vs. LEXP	THEGDP vs. LEXP	GEFF vs. LEXP
Number of Countries *	39	40	34	42	36	34	43	35	28
% of Compliance with a prori expectation	87	89	76	93	80	76	96	78	62

* A denominator of 45 representing the total number of investigated countries was used in calculating the percentage compliance.

.

4.1. Health Expenditure versus Infant, Maternal and Adult Mortality, and Life Expectancy

Looking at the country-specific result and using Angola as an example, Materanl mortality reduces as government spending increases. A unit change in total health spending, health spending per head, respectively, reduces maternal mortality by 3.55 units and 0.34 units. This is also the same for infant mortality but by 52.62 poimnts and 52.23 points, respectively. Life expectancy is found to be a positive function of total health spending, and health spending per head. While total health spending is significant at a coefficient of 3.26 points, health expenditure per head stands at 2.04 points.

On a general note, the findings show health expenditure as an influencer of infant mortality, adult and maternal mortality, and life expectancy. In 39 out of the 45 countries (representing approximately 87%) and 42 out of 45 (representing 93%) investigated, health expenditure per capita (HEPC) was observed to exert a reducing effect on infant mortality and adult/maternal mortality. On the other hand, we find that in 40 out of the 45 countries (89%) and 36 out of 45 (80%), total health expenditure as a ratio of the GDP had a reducing impact on maternal and adult mortality and infant mortality, respectively. In the life expectancy model, we demonstrate that in 43 countries and 35 countries out of 45, respectively (representing 96% and 78%, respectively), both health expenditure per capita (HEPC) and total health expenditures scaled by the gross domestic product (THEGDP) exerted a positive and significant impact on life expectancy.

4.2. Health Outcome versus Government Effectiveness

The country-specific result using Nigeria as an example across the three models shows, Materanl mortality reduces as government effectiveness increases. A unit change in government effectiveness reduces maternal mortality by 0.11 units and infant mortality by 0.49 units. The inpact on life expectancy is found to be insignificant.

On a general note, in 34 out of 45 (76%) of the investigated countries, it was found that government effectiveness (GEFF) had a reduced impact on the maternal and adult mortality and infant mortality rates, respectively. Conversely, we find life expectancy to be a positive and significant function of government effectiveness in 28 out of the 45 studied SSA countries. This means that in 62% of the countries of interest, life expectancy increases as the government allocates more funds to the health sector and becomes more effective in its processes and spending. By implication, in 38% of the studied countries, we found no significant relationship between life expectancy and government effectiveness. It can be deduced that such effectiveness is nonexistent or too minor to produce significant impact.

4.3. Long-Run Dynamics of Health Expenditure, Health Outcome and Government Effectiveness

The bound test illustrates a cointegrating relationship for all of the models for every country except Eritrea (maternal, adult and infant mortality models) and Mauritius (infant mortality model). Regarding short-run results, which are presented in the complete results in Appendix A, an indicator of interest is the coefficient of the lagged error-correction term. A significant negative coefficient is expected if health outcomes are adjusted towards the steady state following shocks and distortions arising from health expenditure and government effectiveness. We first find that in the model with maternal and adult mortality as the dependent variable, the error correction in 39 out of the 45 studied countries (87%) carried the correct sign. This shows that maternal and adult mortality adjusts appreciably to the combined shocks and dynamics of health expenditure and government effectiveness. Second, in the infant mortality model as the outcome variable, the error correction term in 78% of the studied countries (35 out of 45) were found to be negatively significant. This observation indicates an inevitable return to long-run equilibrium by health outcome to deviations triggered by the explanatory variables. Third, the highest number of rightly signed error correction terms are present in the model with life expectancy as the regressor. In 42 out of 45 countries (approximately 93% of the studied SSA countries), health outcomes were found to adjust to the shocks and dynamics of health expenditure and government effectiveness (see the speeds of adjustment for the countries in the Appendix A). All the rightly signed error correction terms were less than unity (1), removing every suspicion of unpredictability and oscillatory explosions. Also, a relatively quick adjustment was observed in most of the countries in almost all the models, suggesting that the predictive content of the health outcomes, health expenditures, and government effectiveness interaction has not waned over time.

Several diagnostic tests were employed in this study. The adjusted R² values imply a good fit for each of the evaluated models. We used the Lagrange multiplier (LM) tests for the autoregressive residual process to erase the suspicion of autocorrelation. The functional form of the models and absence of specification errors/biases were examined by employing Ramsey’s RESET test using the square of the fitted values. In contrast, the homoscedasticity of the model residuals was confirmed by the White processes of heteroscedasticity test (details are reported in the Appendix A). Furthermore, we utilize the cumulative sum of recursive residuals (CUSUM and CUSUMSQ) to test for the parameter stability of the overall models. The results of all the post-estimation tests prove that the estimated coefficients of all models are stable and conform to the assumptions underlying our estimation technique.

5. Conclusions

While various econometric techniques have been used to study the interaction between health expenditure and health outcomes with due consideration for government effectiveness, we provide new insights into the health outcomes and health expenditures nexus by bringing to fore the role government effectiveness plays in developing countries. In most of the studied countries, we obtained results favorable to the hypothesis that health expenditure reduces mortality rate and increases life expectancy. It was illustrated that poor health outcomes are not just due to inadequate budgetary allocations to the sector but are also symptomatic of the inefficiency associated with the management of public resources in SSA (Akinlo and Sulola 2019).

These findings carry significant implications for decision making. By way of policy, it is important to advocate for pro-poor spending, especially towards primary and cottage health spending. An observation by the WHO (2016) makes this even more significant as it is typical that lower priority is given to primary care. Many African governments spend less than 40% of their health service expenditures on primary care. It, therefore, means that much of the money that is spent goes to secondary care (curative care). Public health expenditures largely benefit the poor who are not just in the majority, but also cannot afford private health care. The poor also are exposed to poor standards of living and face greater mortality. Pro-poor spending will take to a higher level the responsiveness of health outcomes to public health expenditures. Second, improved efficiency in public expenditure management and strong institutional frameworks will have the proclivity to act as safety net against corruption and its negative payoffs. This will address leakages and increase the beneficial impact of health and public expenditures in general.

This work does not claim to be exhaustive but is believed to have added to the conversation on the health outcome and health expenditure nexus with due consideration for government effectiveness. Like studies emanating from most developing countries, this study has been constrained by data availability and reliability and may also have been affected by cross-sectional biases. These constraints were addressed by relying on the integrity of the data repository and using estimation methods that accounted for biases and endogeneity problems.

Generally, every society desires improved health and well-being driven by informed and implemented public health spending that reduces adverse health outcomes and improves positive results. This need is of extraordinary dimension in SSA countries generally characterized by high disease prevalence and mortality rates, as well as abysmal life expectancies in the face of mismanagement of public resources and commonwealth. Having found the degree that health expenditures drive health outcomes, especially in the face of enhanced government effectiveness, it is imperative for managers of public resources in SSA to increase health-related budgetary allocations to the acceptable threshold (see WHO) effectively and efficiently.