Threshold effect of governance quality in the relationship between public health expenditure and life expectancy at birth in the West African Economic and Monetary Union

Abstract

Background

The population in the West African Economic and Monetary Union (WAEMU) suffers poor health, as manifested in low life expectancy at birth. Public health expenditure (PHE) has long been viewed as having the potential to improve health. However, the relationship between PHE and health outcomes is inconclusive. In view of the above, poor governance could be a factor inhibiting the effect of public health expenditure on certain health indicators. This article analyzes the threshold effect of governance quality on the relationship between public health spending and life expectancy at birth in WAEMU.

Methods

The data used come from the Worldwide Governance Indicators (WGI) database and the World Bank’s World Development Indicators and cover the period 1996–2018. Threshold-effect analysis for Hansen (1999) was performed to assess the relation between governance quality (using the six (06) indicators and index of governance quality for robustness) and public health expenditure with life expectancy at birth.

Result

The results show that governance quality is a transition variable at which public health expenditure affects life expectancy. Indeed, the effect of public spending on life expectancy at birth varies according to the thresholds reached by the different levels of governance on a 0–1 scale.

Conclusion

Public health expenditure affects life expectancy if the quality of governance is high. Improving governance quality must be a prerequisite for any allocation of public resources in the health sector in WAEMU.

Introduction

According to the World Health Organization [1], Public health expenditure includes current and capital expenditure from budgets, external borrowing and grants (including grants from international agencies and nongovernmental organizations or mandatory health insurance funds). The provision of health services requires the availability of adequate infrastructure, diagnostic medical equipment, drugs and well-trained medical personnel [2]. Through these different channels, public expenditure on health enables an improvement in life expectancy at birth.

Public expenditure on health enables the development and maintenance of health infrastructure, the provision of essential medical supplies and equipment, and the recruitment and training of health professionals [3]. The literature has shown that investments in public health infrastructure and improved access to health services play an important role in improving the health and well-being of populations [4, 5]. Health infrastructure can directly protect health (such as public sanitation systems) or support other activities that protect and improve health [6]. The fact that health expenditures are in a form that will strengthen the health system contributes positively to the strengthening of the institutional infrastructure for health and the improvement of the health conditions of individuals [7]. Human resources for health are necessary for the delivery of health services; only by ensuring a sufficient, equitably distributed, adequately supported and well-performing health workforce can health goals set by national governments or the international community be achieved [8]. The availability of health workforce will allow the population to be more assured of accessing necessary health services, maintain the status of public health in promotion, prevention, care and rehabilitation and generally improve the health status of the community [9–11]. Huang et al. [12] show that medical personnel have a positive effect on life expectancy at birth. Bayraktar et al. [13] showed that number of hospital beds, number of doctors have statistically significant and positive effects on the ratio of Covid-19 recovered/cases. In addition, increasing the health care budget expands the accessibility of health services while reducing their cost, which leads to a decrease in the probability of death [13].

Although studies have demonstrated the importance of public health expenditure in improving life expectancy at birth, some authors have found a negative or non-significant effect [14–17]. Public health expenditure is a potential determinant of health indicators, but does not in itself guarantee better health indicators [18]. Efficiency in the use of available resources is necessary to achieve the desired improvements [19]. The quality of governance is considered an important determinant of the efficiency of public health expenditure [20, 21]. “Governance is the manner in which power is exercised in the management of a country’s economic and social resources for development” [22]. It includes voice and accountability, political stability and absence of violence, government effectiveness, quality of government regulation, rule of law, and control of corruption [23].

The quality of governance can influence the effect of public health expenditure in improving life expectancy at birth through different channels. First, quality health services are likely to be provided when health worker productivity is high, medicines are available, in-kind payments are reduced and there is no diversion of funds from the system [24]. He adds that public resources allocated to health can be used inefficiently through absenteeism of doctors or nurses in health centers, especially public ones, initialization of equipment due to lack of necessary parts. Friedman [25] and Hussmann [26] find that higher absenteeism rate of qualified doctors and medical staff, demand for informal payments or bribes in exchange for free hospital care, beds and medicines are the main causes of decline in efficiency of health expenditure to an alarming level. Corruption in the health sector is not only manifested in the form of embezzlement, but also in the form of absenteeism of health workers, purchase of substandard equipment due to embezzlement, theft of medicines and other devices meant for public consumption for better health, demand for gratuities, and many other related attitudes [24].

Second, government effectiveness, absence of corruption, transparency, accountability and the rule of law are essential elements that can influence the impact of public health expenditure. Indeed, when public officials are not transparent and accountable in managing health budgets and are not sanctioned in accordance with civil service rules, absenteeism, bribery, theft of medicines and consequently low quality of services provided can occur. Corruption, through the leakage of public resources, limits the achievement of social objectives, desired economic output and other political goals [27]. The rule of law ensures access to effective medicines, regulates the production and importation of unhealthy foods and medicines [28]. When this indicator is respected, it guarantees effective delivery of public services such as health, as arbitrary actions such as corruption and violence are banned [27]. An effective rule of law enables all citizens, communities and institutions to have easy access to justice and the legal system and ensures accountability of all stakeholders (healthcare providers, communities, governments and hospital authorities) [29]. A less corrupt government and regulatory system allows the population and relevant stakeholders to have a voice on irregularities that contribute to maximum and efficient use of resources in the health sector [30].

In the other, promote better institutions contribute to improving the quality of medical infrastructure and equipment, promote better spatial distribution of health facilities, and subsequently improve the quality of health services and health outcomes [31, 32]. Voice and civil accountability, which is an indicator of the political and individual rights enjoyed by citizens, can have an effect on the quality of public health expenditure. It can improve the efficiency of public health expenditure through better monitoring of the implementation of health programs or projects as well as the utilization of funds allocated to the sector.

Datta et al. [33] also argue that greater accountability at regular intervals to citizens motivates political parties to allocate more public budgets and resources to health care. Regarding political instability, it can lead to the destruction of social infrastructure, especially hospitals, as well as a lack of resources for the health sector, which impacts access to health services and therefore the health status of the population. Furthermore, in a situation of political instability, we may see the movement of health personnel in unstable areas to more stable areas, which will reduce the number of health service providers in these struggling areas.

The Sustainable Development Goals (SDGs), which reflect the aspirations of people around the world, call on countries to build strong institutions.¹Aron [34] argues that the majority of African countries suffer from a problem of low institutional quality.

Then, what is the Threshold effect of the quality of governance in the relationship between public health expenditure and life expectancy at birth in the WAEMU?

The aim of this research is to identify the threshold effect of the quality of governance in the relationship between public health expenditure and life expectancy at birth in the WAEMU. The contribution of this research to the existing literature on public spending on health, the quality of governance and life expectancy at birth is four orders.

First, research that links public health expenditure to health outcomes and governance to health outcomes only captures part of the picture [35]. In fact, public spending, governance and development outcomes are linked, and it is this impact that we seek to evaluate empirically in our research.

Secondly, this research uses a methodological approach based on Threshold panel models to identify the non-linear effect of public spending on health on life expectancy at birth via the quality of governance. Threshold panel models are the best suited models to model nonlinear relationships. Indeed, Wang et al. [36] argue that other standard methods for estimating nonlinear relationships such as adding quadratic terms, indicator variables, interaction terms in regression or exogenously determining thresholds would raise multicollinearity issues and may lead to biased estimates. One of the advantages of threshold panel modeling introduced by Hansen [37] is that the threshold is endogenously determined. To our knowledge, this analysis has not been the subject of an empirical study in the literature.

Thirdly, in addition to determining individual thresholds for the various governance indicators, a robustness analysis is carried out by constructing a composite index. Existing studies that have analyzed the role of governance, focused on one [38, 39] in or two [27, 35] and or all of the six governance indicators [18, 21, 40, 41]. The use of individual indicators could limit the conclusions drawn about governance as a whole and is not very informative for robust policy consideration, because governance is multidimensional.

We specifically focus on WAEMU includes eight countries located in West Africa (Benin, Burkina Faso, Côte d’Ivoire, Guinea-Bisseau, Mali, Niger, Senegal and Togo) for two related reasons. Firstly, in the West African Economic and Monetary Union (WAEMU) countries, the performance of the health sector, in terms of improving the health status of the population remains weak. The population in this area suffers poor health as manifested in low life expectance at birth. Indeed, according to World Bank [42], in 2018, apart from Niger and Senegal (with 62 and 68 years respectively), the other countries of the Union had a life expectancy lower than that of low-income countries (62 years). Life expectancy in the WAEMU 2018 was approximately 60 years, compared with 63, 73, 74 and 72 respectively in Africa Eastern and Southern, Middle East & North Africa, Latin America & Caribbean and the world. Likewise, six countries (Niger and Senegal excluded) had life expectancies lower than Sub Saharan Africa (SSA) (about 61 years). Secondly, the countries of the West African Economic and Monetary Union (WAEMU) are characterized by poor institutional quality. During 1996–2018, the zone’s scores for control of corruption, quality of regulation, and political stability were all negative, with respective averages of −0.58, −0.46, and − 0.40 [43]. In 2019, most WAEMU countries also recorded negative scores on all six² governance indicators. For example, corruption control³ in Benin, Burkina Faso, Côte d’Ivoire, Mali, Niger, Senegal, and Togo was − 0.32, −0.19, −0.53, −0.70, −0.55, 0.05, and − 0.75, respectively. Coupled with these low levels of governance quality, WAEMU countries also ranked last in terms of life expectancy at birth according to the United Nations Development Programme rankings [44].

The remainder of this paper is divided into two sections. The first section discusses some empirical evidence on the effect of governance quality on population health. The final section addresses methodological issues, the presentation and analysis of the main results.

Literature review

The new institutional economics is the theoretical approach that integrates governance into the improvement of people’s health [45], one of the main authors of this theoretical, defines institutions as the “rules of the game” or norms that constrain and shape interactions between human beings in a society. They define society’s framework of incentives and sanctions, provide information about which actions are encouraged and discouraged, and the likely benefits and/or costs associated with such actions. Thus, if the institutional environment is such that economic agents can obtain the returns from an investment in physical capital, human capital and new ideas, they are more inclined to make these investments [46]. Acemoglu et al. [47] have shown that the development of certain areas currently in Africa or elsewhere can be explained by the nature of the institutions set up by the colonizers in these countries during the colonization phase.

North [45] said that “Third World countries are poor because the institutional constraints define a set of payoffs to political/economic activity, which do not encourage productive activity”. As a result, countries experience a decrease in public expenditures that could have improved health specifically and other key socio-economic conditions due to weaknesses in institutions that result in increased incentives to use public funds for less productive activities.

Kowalewska et al. [48] have shown that indicators referring to the measurement of institutions such as government efficiency, political stability, economic freedom or the number of days needed to set up a business play an important role in the pursuit of development (both economic and human).

Empirical studies have examined the effect of governance in the relationship between public health expenditure and health outcomes including life expectancy at birth. A first set of studies is that which uses time-series data in their analysis [39, 40, 49, 50]. Indeed, Osakede [40] analyzed the effect of public health expenditure on population health by considering the six indicators of governance (political stability, control of corruption, regulatory quality, citizen voice and accountability, rule of law, government effectiveness) in Nigeria over the period 1980–2017. They found that the interaction between public health expenditure and governance effectiveness as well as control of corruption improved health by leading to a decrease in maternal mortality, while public health expenditure in interaction with the rule of law increased maternal mortality. The interaction between public health expenditure and regulatory quality improved life expectancy, while the interaction between political stability and public health expenditure led to a decrease in life expectancy and a deterioration in maternal and child health. This suggests that the existence of effective policies regulating the quality of goods, which leads to better health status. The rule of law also ensures access to effective medicines, regulates the production and importation of unhealthy foods and medicines.

Mhango and Chirwa [39] for the case of Malawi, found that the lower the level of corruption, the more effective public health expenditure is in improving health. On the other hand, public health expenditure deteriorates health outcomes in cases of poor governance through inefficiency in service delivery in most public places. They conclude that in the public health system, corruption negatively affects the availability of resources and decreases the quality of services offered. It also prevents populations from having equal access to health care and increases the cost of services provided. Yaqub et al. [30] had found a similar result in Nigeria.

The second group of studies uses panel data. Among these studies, some used a single indicator to measure governance [32, 38, 51]. In contrast, others measured governance with multiple indicators [27, 35, 52, 53].

Farag et al. [38] examined the role of health expenditure on health outcomes in 133 low- and middle-income countries by considering the quality of governance (captured by governance effectiveness). They find that public health expenditure improves health indicators and the return on public investment in health care depends on the level of government effectiveness achieved by the country. Public health spending may have a greater impact in countries that spend their money on the “right” health interventions. For example, poor targeting of investment in health infrastructure and other inputs may reduce the impact of public health spending on improving population health.

Dhrifi [51] in 93 developed and developing countries with the generalized method of moments (GMM) finds that institutional quality (composite index constructed with the six governance indicators) plays an important and significant role in mediating the relationship between public health expenditure and infant mortality rate. Bousmah et al. [32] find that the effect of health expenditure on health outcomes (measured by life expectancy and infant mortality) depends on the efficiency of institutions in allocating health resources. They find that the effect of health expenditure on life expectancy is positive when institutional quality is high (Freedom House = 1), but it decreases when institutional conditions deteriorate, until it becomes negative when the quality of institutions is very poor (Freedom House = 7). They conclude that importance is attached to the “voice of the people”, that is, the ability of individuals to fully express their real opinions, needs and preferences in order to make more effective decisions on resource allocation. Odhiambo et al. [54] find the same results with the Corruption Perception Index (CPI) on 41 sub-Saharan African countries.

Regarding panel data studies that looked at multiple indicators to capture governance, Murshed and Ahmed [53] find that poor governance as measured by political stability, voice and accountability, government effectiveness, control of corruption, and rule of law reduces the effectiveness of health spending in terms of desired health outcomes. Public investments (health infrastructure and medical equipment, etc.) aimed at improving the existing level of health status in a country may not be effective in achieving the desired outcomes in the absence of good governance. Hu and Mendoza [27] for the period 1960–2005 find that the interaction between health spending and the quality of governance (quality of bureaucracy and control of corruption) on health indicators is insignificant and has an unexpected sign. Furthermore, in sub-Saharan African countries, Makuta and O’Hare [18] find that the impact of public health expenditure on health indicators depends on the quality of governance. Specifically, government effectiveness, rule of law, control of corruption, and accountability mediate the impact of public health expenditure on population health. They show that in countries with better governance quality (i.e. above the average score), life expectancy relative to public health expenditure is about 6 years, while in countries with poor governance it is about 3 years. This observation is likely explained by the fact that improved governance leads to better resource allocation, better targeting, and more efficient use of available resources.

As for Kim and Wang [21], they show that the quality of public administration measured by five indicators of governance (except political stability) plays an important role in moderating the relationships between public expenditure and public health. These authors recommend developing policies that reflect citizens’ opinions on the side of government system contributions, improving transparency in the policy-making process, and providing effective and efficient service in policy implementation. Rajkumar and Swaroop [35] had reached the same result in 91 countries subjected by using the corruption index and the quality of bureaucracy as an indicator of governance. They find that a 1% increase in public health expenditure leads to a decrease of 0.32%, 0.20%, and 0% in infant mortality rates in countries where governance (corruption index) is good, average, or bad, respectively. For them, managing public resources to promote development requires well-trained and competent staff, working within an institutional framework with an incentive system that reduces fraud and promotes cost efficiency.

Finally, Langnel and Buracom [52] in 32 sub-Saharan African (SSA) countries over the period 2000 to 2015 find that the interaction term between government effectiveness and health expenditure is significant on child mortality. This suggests that improving government effectiveness is likely to enhance the influence of health expenditure on health outcomes in sub-Saharan Africa. Therefore, countries that have an effective and efficient civil service, deliver quality public services without manipulation by political elites, and are able to implement growth-oriented policies have the potential to improve health outcomes, including child mortality, in sub-Saharan Africa. In contrast, for other governance indicators, the interaction terms are negative but not significant.

Empirical evidence suggests that governance quality may be a channel through which public spending on health care can influence population health indicators. However, most of these studies have not addressed the threshold effects of governance quality on public health spending. As a result, most of them have conducted their analysis by introducing indicators of governance quality either simultaneously or separately. However, indicators of governance quality are collinear, so conducting an analysis in the form of a composite index is a contribution.

This article aims to fill this empirical gap through a threshold effect analysis. A panel threshold regression modeling analysis is conducted using Hansen [37] transition model. The panel threshold regression (PTR) method is used for data analysis.

Methods

Model structure

In this section, we discuss the multi-regime model originally introduced by Hansen [37]. This model allows not only the determination of the number of regimes but also the introduction of thresholds in a static panel.

The general mathematical formulation of Hansen [37] model is as follows:

$$\:{y}_{it}={\alpha\:}_{i}+{\beta\:}_{1}^{{\prime\:}}{x}_{it}+{\beta\:}_{2}^{{\prime\:}}{x}_{it}I({q}_{it},\:\gamma\:)+{\epsilon\:}_{it}$$

(1)

$\:{y}_{it}$ represents the dependent variable, $\:{\alpha\:}_{i}$ represents the individual fixed effects, $\:{\beta\:}_{1}^{{\prime\:}}$ and $\:{\beta\:}_{2}^{{\prime\:}}$ are vectors of parameters associated with the explanatory variables of the mode, $\:{x}_{it}$ is a vector of the explanatory variables without the lagged part of the explained variable, $\:I({q}_{it},\:\gamma\:)$ represents the transition function associated with a transition variable, $\:{q}_{it}$ is a vector of threshold parameters $\:\gamma\:$, and $\:{\epsilon\:}_{it}$ is the error term.

The formulation of Hansen’s [37] two-regime model is as follows:undefined

$$\begin{aligned} {y}_{it} & ={\alpha\:}_{i}+{\beta\:}_{1}^{{\prime\:}}{x}_{it}I\left({q}_{it}\le\:\:\gamma\:\right)\\ &\quad +{\beta\:}_{2}^{{\prime\:}}{x}_{it}I\left({q}_{it}>\:\gamma\:\right)\\ &\quad +{\theta\:}^{{\prime\:}}{z}_{it}+{\epsilon\:}_{it} \end{aligned}$$

(2)

\:\gamma\:\right)\\ &\quad +{\theta\:}^{{\prime\:}}{z}_{it}+{\epsilon\:}_{it} \end{aligned}]]>

$\:{z}_{it}$, a set of variables whose associated coefficients do not undergo a regime change.

Inspired by Hansen’s model [37], our model can be presented as follows:

$$\begin{aligned} {lnLE}_{it}&={\alpha\:}_{i}+{\beta\:}_{1}^{{\prime\:}}{lnPHE}_{it}I\left({GOQ}_{it}\le\:\:\gamma\:\right)\\&\quad +{\beta\:}_{2}^{{\prime\:}}{lnPHE}_{it}I\left({GOQ}_{it}>\:\gamma\:\right)\\&\quad +{\theta\:}^{{\prime\:}}ln{z}_{it}\end{aligned}$$

(3)

undefined\:\gamma\:\right)\\&\quad +{\theta\:}^{{\prime\:}}ln{z}_{it}\end{aligned}]]>

$\:{LE}_{it}$, life expectancy at birth; $\:{PHE}_{it}$, the public health expenditure variable; $\:{\alpha\:}_{i}$, fixed effect; $\:{GOQ}_{it}$, indicator of governance quality; $\:{z}_{it}$, matrix of control variables; $\:{\beta\:}_{1}^{{\prime\:}}$ et $\:{\beta\:}_{2}^{{\prime\:}}$ are coefficients that capture the elasticity of public health spending on health in two regimes, $\:{\epsilon\:}_{it}$, is the error term.

The transition function represented by $\:I(.)$ takes the value 1 when the arguments it contains are valid and 0 otherwise.

Data

The data used are from the Worldwide Governance Indicators (WGI) [43] database and the World Bank’s World Development Indicators [42] and cover the period 1996–2018.

Institutions refer to the set of formal and informal rules (value systems and beliefs, representations, social norms) that govern human interactions [55].

Formal governance quality encompasses a set of variables classified by Kaufman and the World Bank. The term governance is so multifaceted that several indicators have been developed to try and capture its different dimensions. In this paper, we employ the World Governance Indicators (WGI) by the World Bank. A total of six indicators have been were used to take into account the quality of institutions (especially political ones) or governance.

Citizen Voice and Accountability, an indicator that measures the political and individual rights of citizens, guarantees freedom of information, civil liberties, and the general participation of citizens in political decision-making [56]. In health policy, when citizens enjoy this freedom, it enables the provision of adequate social services through investments in health care [57]. Moreover, because governments can be held accountable for their actions, they are more likely to respond to the needs and demands of the population for improved healthcare provision [58].

Political stability and the absence of violence, which measures the likelihood of violent regime or government change, and threats to public order (including terrorism) play a role in health. Wars and conflict lead to the destruction of economic potential, resulting in population displacement, increased poverty, and poor healthcare provision [59]. Political instability is the result of ineffective governance [60]. Regulatory quality, on the other hand, is a measure of the barriers that prevent the market from functioning properly. Government effectiveness refers to the competence of public bureaucrats and the quality of public services. These two governance indicators focus on the outcome of public action and ensure that people have access to quality public goods and health services [27].

Control of corruption, an indicator of governance quality that captures the use of prerogatives by public authorities for personal purposes, in this case embezzlement, leads to inefficient use of public resources and can reduce investment in social sectors such as health and education. Rule of law, a measure of respect for the law by the judiciary and police. When respected, this indicator guarantees the efficient delivery of public services such as healthcare, as arbitrary actions such as corruption and violence are outlawed [27].

For all these different variables, we standardized⁴ to obtain data between 0 and 1.

Public spending on health care (as a % of GDP) is used as the variable of interest. This variable accounts for medical and sanitation expenditures. Following [19], private health expenditures, including household health expenditures (out-of-pocket expenditures), also known as out-of-pocket payments, private insurance, charitable donations, and payments for direct services by private firms, are used. Education in general is an important determinant of household health, particularly in relation to prenatal and postnatal care in developing countries. Baldacci et al. [61] and Schultz [62] have shown that women’s educational attainment plays a critical role in the health status of infants and children, as well as the population as a whole. An additional year of education per woman reduces infant and maternal mortality rates.

Finally, Chou et al. [63] show that female education has a positive effect on reducing infant mortality. However, due to data availability, this variable is approximated by elementary school enrollment rates. Gross domestic product is used to measure income. According to Hojman [64] and Banerjee et al. [65] there is a direct and positive relationship between wealth and survival, depending on the ability of households to supply goods and the demand for medical products. Economic theory adds that if health is considered a necessary good, then, all else being equal, an increase in income will lead to an increase in demand for medical care. Gupta and Baghel [66] stated that the health status of the population improves as per capita income increases. In addition, economic growth increases technological progress, including medical technology development and technological advances in health, which also positively affect health outcomes [67–69]. Economic growth leads to increased wages and a higher quality of life, which improves health [70].

Access to drinking water, as measured by the proportion of the population with access to a source of drinking water relative to the population with access to water, is integrated into this model. Clean, accessible water is essential to human health, reducing mortality and morbidity. A lack of clean drinking water contributes to the outbreak of diseases such as diarrhea, cholera, and viral hepatitis A. Targets 6.1⁵ and 6.3⁶ of the Millennium Development Goals state that safe water, sanitation, and hygiene are necessary to reduce maternal mortality rates, Arthur and Oaikhenan [71] have highlighted the importance of access to drinking water in improving population health.

Similar to Arthur and Oaikhenan [71], the proportion of the population with access to adequate sanitation services is used. Improved sanitation can reduce diarrheal disease and can significantly reduce the adverse health effects of other diseases. Many international organizations use hygienic sanitation facilities to measure progress in the fight against poverty and disease, and target 6.2⁷ of the MDGs focuses on sanitation services.

Estimation strategy

Two successive tests are performed: (i) the test for determining the number of regimes and (ii) the test for significance of the slope coefficients. After this test, we create a confidence interval on the institutional quality thresholds. Finally, the calculated threshold is used to estimate our model using the OLS method.

Inference

Before adopting and estimating the above non-linear model, a linearity test must be performed to ensure a non-linear relationship, determine the number of regimes and confidence intervals for the threshold(s), and finally test the significance of the threshold(s).

The linearity test consists of verifying that the relationship between the variable of interest and the dependent variable can be represented according to a regime-switching model and proving that the effect of the threshold(s) is significant. For us, this involves testing the null hypothesis of equality of coefficients against the alternative hypothesis of equality of coefficients:

$\:{H}_{0}\::{\beta\:}_{1}={\beta\:}_{2}$ versus $\:{H}_{1}\::{\beta\:}_{1}\ne\:{\beta\:}_{2}$. The test is based on the following model:

$$\:{y}_{it}={\alpha\:}_{i}+{\beta\:}_{1}^{{\prime\:}}{x}_{it}I\left({q}_{it}\le\:\:\gamma\:\right)+{\beta\:}_{2}^{{\prime\:}}{x}_{it}I\left({q}_{it}>\:\gamma\:\right)+{\theta\:}^{{\prime\:}}{z}_{it}$$

(4)

undefined\:\gamma\:\right)+{\theta\:}^{{\prime\:}}{z}_{it}]]>

$$\begin{aligned} & \left\{\begin{array}{c}H_0\::{\beta\:}_1={\beta\:}_{2\:}\\\:H_1\::{\beta\:}_1\neq\:{\beta\:}_2\end{array}\right.<=> \\ & \left\{\begin{array}{c}{y}_{it}={\alpha\:}_{i}+{\beta\:}_{1}^{{\prime\:}}{x}_{it}\\\:{y}_{it}={\alpha\:}_{i}+{\beta\:}_{1}^{{\prime\:}}{x}_{it}I\left({q}_{it}\le\:\:\gamma\:\right)+{\beta\:}_{2}^{{\prime\:}}{x}_{it}I\left({q}_{it}>\:\gamma\:\right)+{\theta\:}^{{\prime\:}}{z}_{it}\end{array}\right. \end{aligned}$$

(5)

undefined\right. \end{aligned}]]> \\ & \left\{

Validation of the null hypothesis $\:{H}_{0}$ implies that Eq. 5 is a linear model. On the other hand, failure to reject the alternative hypothesis $\:{H}_{1}$ means that Eq. 1 is a nonlinear model.

The decision is made on the basis of the following likelihood ratio or Fisher statistic:

$\:{LR}_{1}ou\:F=\frac{{S}_{0}-{S}_{1}\left({\widehat{\gamma\:}}_{1}\right)}{{\widehat{\sigma\:}}^{2}}$ where $\:{\widehat{\sigma\:}}^{2}=\frac{1}{N(T-1)}{S}_{1}\left({\widehat{\gamma\:}}_{1}\right)$ with$\:{\:S}_{0}$ the sum of squared residuals from the linear model and $\:{S}_{1}({\widehat{\gamma\:}}_{1}$) is the sum of squared residuals from the threshold model. The bootstrap method on the critical values of the F statistic is used to test the significance of the threshold effect.

If a threshold effect is present, the number of regimes is determined. The procedure is to test the presence of k regimes against k-1 regimes. In practice, we can test for the presence of 3 thresholds. In this case, if we cannot distinguish any regimes, we test for the presence of two thresholds. If there is no regime in this case either, we test for a single threshold. Thus, if we find the presence of a threshold r, we estimate the nonlinear model with k regimes (k = r + 1).

Table 1 presents the linearity test between public spending on health and population health (life expectancy at birth). Analysis of the table reveals that there is a non-linear relationship between public spending on health and population health in the WAEMU in terms of the LR-tests and the associated probability. Thresholds of 0.4499 for control of corruption, 0.3864 for government effectiveness and quality of regulation, 0.4206 for political stability and absence of violence, rule of law and citizen voice, and accountability at which regime change occurs are 0.4206 (in an interval between 0 and 1) for life expectancy. The confidence interval calculated on the basis of the distribution simulated by the methodology of Hansen [37] indicates that at a first species risk of 5%, the values are between 0.3952 and 0.4886 (control of corruption), 0. 358 and 0.389 (government effectiveness and regulatory quality), 0.385 and 0.424 (political stability and absence of violence), 0.388 and 0.424 (rule of law) and 0.3815 and 0.424 (citizen voice and accountability).

The three-regime test on life expectancy at birth rejects the hypothesis of two thresholds (probability greater than 5%), as the second threshold appears insignificant. There is a single threshold, i.e. a double regime, in the relationship between public spending on health and life expectancy at birth in relation to governance indicators.

Table 1 Linearity test and determination of the number of regimes with the six indicators of governance

Discussion

This analysis is done using the six indicators of governance of Kaufmann et al. [72]. These are control of corruption, government effectiveness, political stability and absence of violence, quality of regulation, rule of law and citizen voice and civil responsibility. Table 2 summarizes the effect of public health expenditure on life expectancy at birth under the condition of a certain threshold of these six (06) indicators.

Table 2 Estimation of the threshold effect of governance quality with the six governance indicators in the link between public health expenditure and life expectancy at birth

The results present in the Table 2 show that public health expenditure has a positive and non-significant effect on life expectancy at birth in the low levels of the quality of governance indicators (value below the threshold); i.e. 0.4499 for control of corruption and 0.4206 for political stability, rule of law and citizen voice and accountability. Once these thresholds are exceeded, the effect becomes positive and significant respectively for all of these different variables.

Regarding the control of corruption, its low level can lead to inefficient use of public resources. Indeed, in a situation of low level of control of corruption, there can be misappropriation of resources, theft of health equipment, resale of medical devices and fraud. High corruption thus leads to a reduction in the quality of care and increases the prices of medical and health equipment, which reduces access to care. Thus, in a situation of low level of control of corruption, there are obstacles to medical care for the poor through the increase in health care costs due to the demand for bribes. Such a situation leads to inefficiency in converting public health expenditure into key health outcomes and improving life expectancy at birth. This result corroborates that found by Rajkumar and Swaroop [35] who found that an increase in public spending in a situation of poor governance has no effect on health.

On the other hand, the more the level of corruption control tends towards one (01), the more we see a reduction in cases of corruption, thus leading to a more efficient use of public resources and more trust of the populations in health centers, because any act of corruption is perceived and sanctioned by the authorities. We can observe an increase in the health offer, but also in its quality and accessibility through the elimination of certain costs linked, among other things, to bribes, to the resale of medical devices or products and therefore an improvement in health indicators. This result corroborates those found by Odhiambo et al. [54] and Bouchard et al. [73] respectively in sub-Saharan Africa and Uganda. Indeed, according to Odhiambo et al. [54], when the level of perception of corruption is low, increasing public health spending helps improve the health of populations.

Political stability can influence health indicators through its effect on economic potential. Furthermore, conflict and violence that characterize an unstable political environment predispose individuals to poor health conditions, increase poverty, reduce access to health care and ultimately lead to worsening health status. Indeed, in a situation of low level of political stability (political instability) characterized by the wars and conflicts, there is a destruction of social infrastructure including hospitals, economic potential with its corollary of population displacement, increase in poverty, and also poor provision of health care.

In WAEMU countries, the index for political stability and non-violence rose from − 0.156 in 2000 to −0.509 in 2010 and − 1.072 in 2020. This trend indicates a worsening of political instability, reflected on the one hand in the many attempted (or actual) illegal overthrow of a government in some countries in the region (Burkina Faso, Mali, Guinea-Bissau), and on the other hand in electoral or insurrectionary violence and loss of life (e.g. Côte d’Ivoire in 2002 and 2010–2011, Togo in 2005, Burkina Faso in 2014, Senegal in 2012). This instability is also marked by recurring terrorist attacks in some countries in the region (Niger, Mali, Burkina Faso) and terrorist incursions in others (Togo, Benin, Côte d’Ivoire).

In short, populations are faced with budgetary constraints on the one hand, and on the other hand there is less expenditure by the authorities in the health sector due to the lack of resources for adequate healthcare, which explains the non-significant effect of public health expenditure below the threshold. Likewise, politically instability political instability (as in the case of terrorism) leads to populations moving to other areas, leading to an overload on their health system which means resources allocated to the health sector are no longer adequate to meet the increased demand. On the other hand, its improvement, which is a sign of stability, generates more resources for individuals and the State and increases life expectancy at birth through more expenditure in health infrastructure and more medical personnel. A higher value of political stability indicates less political risk in a country and this situation can increase the efficiency of public health expenditure. This result corroborates that found by Osakede [40].

For indicators such as citizen voice and civil responsibility and the rule of law, below the threshold determined, a positive and non-significant effect is observed. On the other hand, when this threshold is crossed, public health expenditure positively and significantly influences life expectancy at birth. An increase in public health expenditure of 1% leads to an increase in life expectancy at birth of 0.145% and 0.135% respectively for citizen voice and civil responsibility and the lower rule of law.

A higher of threshold of voice and civil responsibility shows that people are in a situation to choose their own government. In this case the government is relatively more alert in its spending, which leads to higher efficiency of public spending in all the sector and in the health specifically. This result could also be explained by the fact that a low level of citizen voice and civil responsibility translates into very low accountability and less transparency in the management of public resources. On the other hand, improving transparency and accountability in the use of resources is imperative to improve the health of populations. Indeed, quality health services are likely to be put in place when the misappropriation of health system funds is reduced in a situation of transparency in the management of public resources.

In the case of the rule of law, its low level generates detrimental effects on users of health services. This result is explained by the fact that, in many African countries, legislation, policies and practices are contrary to human rights obligations and hostile to public health objectives [40]. Also, when the rule of law is made in a country, certain practices can be observed at the level of the health system. These practices can involve discrimination of users based on their socio-economic status. Thus, non-compliance with the rule of law leads to individual behavior by health actors that can ultimately harm the health of users. Non-compliance with the rule of law would be synonymous with non-compliance with legal channels by the judicial system and the police. In such a situation, citizens are generally less likely to resort to the judicial system to denounce cases of violence and corruption in the health sector, which inhibits the effect of public health spending on the health status of the population. Low progress in the rule of law can constitute a structural obstacle to improving health in the country [74].

On the other hand, in a situation where the rule of law reaches a certain level, citizens have confidence in judicial institutions on the one hand and health professionals practice less behaviors contrary to ethics and medical deontology, which guarantees access to health services. Indeed, abnormal practices are sanctioned by the judicial or disciplinary system in a state of law. In addition, the rule of law also guarantees access to effective medicines, regulates the production and importation of unhealthy foods and medicines [28]. In such a situation, public health expenditure injected into the health sector helps improve the health of populations. This result is consistent with that found by Bousmah et al. [32] and Osakede [40].

For the indicators of government effectiveness and regulatory quality, when they are below the estimated threshold (0.3864), public health expenditure negatively and significantly influences life expectancy at birth. Indeed, an increase in public health expenditure of 1% leads to a reduction in life expectancy at birth of 0.130% and 0.125% respectively for government effectiveness and regulatory quality. However, the effect becomes insignificant once the threshold is exceeded. Indeed, in a situation of low government effectiveness, there is absenteeism of health sector workers in public services, which can lead to a lack of credibility of users of health services.

Also, government ineffectiveness can result in a situation of lack of training of actors on ethics, the professional conduct of professionals in the sector, which can result in a negative effect of public health expenditure on life expectancy at birth. However, its effectiveness could translate into an improvement in the quality of medical personnel and consequently the provision of good quality health services. There is also a better allocation of resources and better targeting in public health policies in a situation of government effectiveness. All this helps to improve the performance of public spending in the health sector. This result corroborates those found by Murshed and Ahmed [53] and Bousmah et al. [32] who show that the quality of governance can reduce the effectiveness of public spending on health when it is of poor quality.

Good regulatory quality guarantees access to public goods and health services for the population. On the other hand, poor regulatory quality would reduce accessibility to health centers, because health services will be of poor quality, which negatively influences the health status of the population. Regulation in the health sector consists of protecting the interests of consumers through price control of pharmaceutical products. The absence of control could lead to an increase in the price of products, but also the provision of poor-quality products, which negatively influences the health of the population. Thus, when the quality of regulation improves, we may witness a greater use of health services and consequently an increase in their life expectancy. Thus, the existence of effective policies regulating the quality of the sector inevitably induces a better state of health. Our result corroborates that found by Bousmah et al. [32] who find that public health spending can have a negative effect on population health when the quality of governance (measured by civil liberties) does not reach a certain level.

Moreover, households expenditure more in their health when the quality of governance is good. This could be related to greater trust in health services, which improves health outcomes. This result confirms the findings of [32, 40], who show that corruption and nepotism in the health sector are factors that negatively affect the quality of health services. They also add that private expenditure in health care depends on other factors, such as political stability. Thus, incentives to spend in one’s health are weakened in a war situation.

Private health expenditures have a negative effect on life expectancy at birth. Private health expenditures are a financial barrier to accessing health care facilities. This can lead to a deterioration in the health status of the population as doctor visits are postponed and life expectancy at birth decreases. These practices are common in developing countries where the public health system is inadequate because health insurance is virtually nonexistent and households are forced to pay directly for health care. In WAEMU countries 95% of private health expenditures are paid directly by households because they do not benefit from social insurance. When their health deteriorates, they either postpone consultation or resort to traditional medicine, which has a negative impact on the health of the poorest.

Access to sanitation has a positive and significant effect on life expectancy at birth. In the WAEMU region, 50.7% of the population has access to latrines. Lack of sanitation is a cause of certain diseases (diarrhea, for example, or malaria), and further expansion of these facilities can reduce these diseases and increase life expectancy at birth. This result is consistent with the findings of Kiross et al. [75].

Drinking water has a positive and significant influence on life expectancy at birth. When the population has access to a drinking water source, it reduces certain epidemics, such as cholera, diarrhea and hepatitis A. WAEMU countries have made efforts to improve access to drinking water. In fact, the percentage of the population with access to drinking water in the WAEMU is 80%. This result is consistent with that found by Arthur and Oaikhenan [71] for sub-Saharan African countries. Gross domestic product has a positive and significant effect on life expectancy at birth. An increase in per capita income leads to an increase in life expectancy at birth. On the one hand, this could be explained by an increase in the supply of health care services by the public sector, but on the other hand, it could also be explained by a demand for healthcare services by households as a result of an improvement in their living conditions.

Robustness analysis with the index of governance quality

This robustness analysis is conducted using the governance index. Before moving on to the actual estimate, we first proceed with preliminary tests for the construction of the index as well as the linearity test in the relationship between public health expenditure and life expectancy at birth.

Building a composite index with principal component analysis (PCA)

Principal component analysis (PCA) is a method for compressing information when data are redundant. Governance indicators are highly correlated, and their use in the same equation can lead to multicollinearity. There are three main conditions in the literature for testing the need for PCA: the multicollinearity test, Bartlett’s sphericity, and the Kaiser‒Meyer‒Olkin index (KMO).

Multicollinearity test between governance quality variables

This test is used to check for the presence of correlation between variables. Tabachnick and Fidell [76] recommended using this test when correlation coefficients between explanatory variables are greater than 0.3. Hair et al. [77] used a rule of thumb to categorize the degree of correlation according to the associated coefficient. Coefficients of 0.3, 0.4, and 0.5 (all absolute values) are considered minimum, significant, and practically significant correlations, respectively. Table 3 shows that most correlation coefficients are greater than 0.3 and significant at the 5% level.

Table 3 Multicollinearity results for governance quality indicators

Bartlett’s sphericity test

This test proposes a global measure and aims to determine the extent to which the correlation matrix (observed matrix) is significantly different from the unit matrix (theoretical matrix). Under the Ho hypothesis, the absolute value of the determinant is equal to one, whereas under perfect collinearity, it is equal to zero. When the determinant is less than 0.00001, it is generally assumed that the data are highly redundant and provide only one type of information. When the determinant is close to one (01), the variables are nearly orthogonal, and principal component analysis (PCA) is not needed. The test results give a low determinant (0.045). It is closer to zero than to one, and this correlation matrix deviates from the unit matrix at the 1% threshold (P value = 0.000).

The Kaiser‒Meyer‒Olkin (KMO) index

The KMO index, which ranges from 0 to 1, is used to assess the relevance of the use of PCA. If it is close to zero (0), the partial correlations are equal to the raw correlations. On the other hand, if it is close to 1, the variables are correlated, and the compression provides an excellent summary of the information. According to authors such as Tabachnick and Fidell [76] and Hair et al. [77], an index of 0.5 indicates that factor analysis is appropriate because the variables are correlated. The Table 4 confirms this with an index of 0.75.

Table 4 Results of the KMO index and Bartlett sphericity test

Inference

The analysis of the linearity test in Table 5 shows that there is a nonlinear relationship between public health expenditure and life expectancy in the WAEMU with respect to the LR tests and the associated probability. We also found that the threshold for governance quality of regime change is 0.4206 (within an interval between 0 and 1). The confidence interval calculated based on the distribution simulated by the method of Hansen [37] shows that this value for institutional quality ranges from 0.388 to 0.4206 for a risk of the first type of 5%.

The three-regime test for life expectancy at birth rejects the two-threshold hypothesis (probability greater than 5%) because the second threshold appears to be insignificant. The relationship between public health expenditures and life expectancy at birth in the context of governance quality has a single threshold, i.e., a double regime.

Table 5 Linearity test and determination of the number of regimes

Results

Table 6 summarizes the effect of public health spending on life expectancy at birth conditional on a given level of governance quality index. The results show that public health spending has a positive and non-significant effect on life expectancy at birth when the level of institutional quality is low (value below the threshold). Once the threshold is exceeded, the effect becomes positive and significant. These results confirm those obtained with the six indicators of governance.

Table 6 Estimation of the threshold effect of governance quality in the relationship between public health expenditure and life expectancy at birth

Conclusion

The aim of this article was to identify the effect of public health expenditure on population health in WAEMU, considering governance quality. A threshold effect model was used, following the theoretical model of Hansen [37]. The results show that governance quality is a transition variable through which public spending affects life expectancy at birth across a transition threshold.

Our results reveal that the quality of governance measured by the six (06) indicators is a key factor in the transition from public health expenditure to life expectancy at birth. The results of the six indicators are confirmed by the composite index of governance quality. Improving governance quality must be a prerequisite for any allocation of public resources in basic social sectors in general and in the health sector in particular in the WAEMU. Our results are in line with those found by Bousmah et al. [32], Osakede [40] and Langnel and Buracom [52]. It is therefore imperative that governments in the WAEMU improve the quality of governance in order to improve health outcomes.

This analysis shows that the effect of public spending on health is not guaranteed unless governance quality reaches a certain level. These results indicate that in the WAEMU countries, the improvement of governance quality is crucial to maximize the benefit from public health expenditure in the health sector. This can be done through strengthening regulation, control and the fight against corruption in the health sector on the one hand and establishing the strengthening of the prevention and risk management system for poor governance in the sector through internal controls and inspections.

Also, the establishment of a mechanism for reporting cases of corruption through the establishment of a secure telephone line dedicated to the fight against corruption for citizens and professionals in the sector is necessary. In addition, it is necessary to strengthen education and awareness-raising campaigns aimed at the public on the dangers of corruption and the means of combating it in order to create a culture of rejection of the phenomenon in society. Ensuring fair working conditions and competitive salaries for civil servants on the one hand, and depoliticizing public administration on the other could also be a solution.

Furthermore, it would be important to strengthen the orders of health professionals in training their members to respect transparency, ethics and professional conduct in the exercise of their function. Transparency and accountability can enhance efficiency of resources allocated to health care by reducing corruption and improving the operation of health systems. It is necessary to apply effective sanctions in the event of corrupt practices and mismanagement of financial resources. The regulatory and legislative framework also needs to be strengthened to provide a framework for the involvement of public health professionals in private health structures.

Also, it is important to put establish a mechanism for capitalizing and valuing experiences in the health sector in order to improve the efficiency of the sector. In addition, strengthening accountability and rendering of accounts and pursuing or establishing administrative sanctions for individual violations will improve the citizen voice and civil liability. It would be important to strengthen medical, paramedical and pharmaceutical inspection and organize unannounced visits to health establishments and carry out the evaluation of health services by users in order to have a well-regulated sector. It is essential to ensure that the law is applied fairly and impartially for all citizens, and to strengthen the capacity of judicial institutions to enable them to play their role effectively in order to consolidate the rule of law.

To ensure political stability and the fight against violence and terrorism, states should set up a sub-regional army to pool their efforts in the fight against terrorism in WAEMU. In most cases, the illegal overthrow of a government, popular uprisings, insurrections and conflicts are the hallmarks of political instability in WAEMU countries. These manifestations may be the result of the countries’ structural conditions, which include high unemployment rates, income inequalities, the absence of normal power transition mechanisms and ethnic problems. And so, improving the economic through measures aimed at reducing inequalities, creating jobs and providing equitable opportunities for all. In addition, the rules governing the mandate process in the region need to be strengthened (and respected).

Access to hygiene and sanitation services contributes to improving institutional quality in the WAEMU. It is therefore essential that the various governments ensure access to drinking water and sanitation for the entire population through joint action by the various ministries. Indeed, in sub-Saharan Africa, drinking water supply and sanitation are mediocre. WAEMU countries must make more efforts to ensure access to water and sanitation for all and to ensure sustainable management of water resources in order to improve the well-being of their populations and consequently increase life expectancy at birth. This means increasing the number of water supply and sanitation facilities and improving consultation between the various players in the sub-sector to ensure a balanced distribution of these facilities. Finally, improving economic growth by increasing investment in physical capital and creating a stable environment, among other things.

This research has limitations in that it focuses on a single indicator of health capital (life expectancy at birth). First, futures research may take into account other indicators of health capital such as the infant, maternal and child mortality rates. Second, our research used the estimation by a static threshold panel model, namely the abrupt transition threshold panel (PTR) model of Hansen [37] on a subset of African countries. Future research will be able to use the dynamic threshold panel model developed by Kremer et al. [78] on a larger sample (developing countries) who is a novel panel threshold technique.