GEOSPATIAL AND INTERTEMPORAL IMPACT OF NATIONAL HEALTH INSURANCE SCHEME (NHIS) ON HOUSEHOLD HEALTHCARE EXPENDITURE: EVIDENCE FROM GHANA

GEOSPATIAL AND INTERTEMPORAL IMPACT OF NATIONAL HEALTH INSURANCE SCHEME (NHIS) ON HOUSEHOLD HEALTHCARE EXPENDITURE: EVIDENCE FROM GHANA | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 26 September 2025 V1 Latest version Share on GEOSPATIAL AND INTERTEMPORAL IMPACT OF NATIONAL HEALTH INSURANCE SCHEME (NHIS) ON HOUSEHOLD HEALTHCARE EXPENDITURE: EVIDENCE FROM GHANA Authors : Emmanuel Boadu 0009-0008-1197-9724 [email protected] , Anthony Abbam , and Anselm Komla Abotsi Authors Info & Affiliations https://doi.org/10.22541/au.175886290.09288122/v1 296 views 159 downloads Contents Abstract Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract This study examines the geospatial and intertemporal impact of Ghana’s National Health Insurance Scheme (NHIS) on household health expenditure using nationally representative data from the 2022 Annual Household Income and Expenditure Survey (AHIES). Employing propensity score matching (PSM) to control for selection bias, the analysis reveals that NHIS significantly reduces health spending across most of Ghana’s 16 regions, with Greater Accra, Ashanti, and Central showing the largest impacts. However, regional disparities persist, with Volta Region showing no statistically significant effect. Temporally, the scheme’s effectiveness also varies, with the largest cost reductions observed in the fourth quarter of 2022, suggesting seasonal sensitivity in healthcare access and NHIS performance. The findings underscore the need for region-specific interventions, improved service delivery, and stronger inter-ministerial coordination, particularly among the Ministry of Health, NHIA, and Ghana Health Service to ensure equitable and consistent financial protection for all households, especially the poor and vulnerable ones. GEOSPATIAL AND INTERTEMPORAL IMPACT OF NATIONAL HEALTH INSURANCE SCHEME (NHIS) ON HOUSEHOLD HEALTHCARE EXPENDITURE: EVIDENCE FROM GHANA Emmanuel Boadu 1* , [email protected] Anthony Abbam 1 , [email protected] Anselm Komla Abotsi 1 , [email protected] 1 Department of Economics Education University of Education, Winneba ABSTRACT This study examines the geospatial and intertemporal impact of Ghana’s National Health Insurance Scheme (NHIS) on household health expenditure using nationally representative data from the 2022 Annual Household Income and Expenditure Survey (AHIES). Employing propensity score matching (PSM) to control for selection bias, the analysis reveals that NHIS significantly reduces health spending across most of Ghana’s 16 regions, with Greater Accra, Ashanti, and Central showing the largest impacts. However, regional disparities persist, with Volta Region showing no statistically significant effect. Temporally, the scheme’s effectiveness also varies, with the largest cost reductions observed in the fourth quarter of 2022, suggesting seasonal sensitivity in healthcare access and NHIS performance. The findings underscore the need for region-specific interventions, improved service delivery, and stronger inter-ministerial coordination, particularly among the Ministry of Health, NHIA, and Ghana Health Service to ensure equitable and consistent financial protection for all households, especially the poor and vulnerable ones. Keywords: National Health Insurance Scheme (NHIS), Health Expenditure, Propensity Score Matching (PSM), Geospatial Analysis, Intertemporal Variation Highlights • Ghana’s National Health Insurance Scheme (NHIS) significantly cuts household health spending but its impact varies across regions and time. • NHIS’s financial protection is strong in urban centers but weak in rural regions. • The scheme’s impact peaks in Q4 2022, showing strong seasonal variation. • Achieving Universal Health Coverage requires spatially-targeted and seasonally-responsive policies. 1.0 INTRODUCTION Across the globe, millions of people are burdened by healthcare costs that consume a significant share of their income. Estimates suggest that around 208 million individuals spend more than 40% of their income (excluding food expenses) on healthcare, while over 800 million people allocate more than 10% of their total income to health-related expenses (Frimpong et al., 2021; Wagstaff et al., 2018). This means that after subtracting what these individuals spend on basic food needs (which are essential and non-negotiable), 40% or more of their remaining income goes to healthcare. This level of spending is often referred to as “catastrophic health expenditure” because it threatens a person’s ability to pay for other essential needs like education, housing, or transportation. This financial strain is particularly pronounced in sub-Saharan Africa, where out-of-pocket (OOP) payments make up roughly 36% of total health spending which is substantially higher than the global average of 22% (McIntyre et al., 2018). Numerous studies have found a strong link between high OOP spending and increased rates of catastrophic health expenditures, often pushing households deeper into poverty (Frimpong et al., 2021; van Doorslaer et al., 2007; Xu et al., 2018). In Lesotho, the impact is even more severe, with about one in five poor individuals pushed further into poverty because of direct health payments (Akinkugbe et al., 2012). In Ghana, for instance, between 3% and 5% of the population is driven below the poverty line due to these health-related costs (Aryeetey et al., 2016). Given the dire consequences of high out-of-pocket health expenditure, there is the need for countries in resource-constrained settings to explore the most sustainable healthcare financing model for providing financial risk protection for majority of the population in low and middle-income countries. It is against this backdrop that the World Health Assembly resolution 58.33 mandated member countries to prioritise attainment of Universal Health Coverage (UHC) by ensuring equality in access to healthcare services particularly when required with no financial barriers. Additionally, achievement of Universal Health Coverage (UHC) is outlined in Sustainable Development Goal 3 (SDG 3) particularly Target 3.8, which emphasises access to quality essential health services and financial risk protection. Achieving UHC requires that each person should be afforded leaving no one behind. It also demands that provision of healthcare services should be based on a person’s health needs implying that people with greater health needs should receive more services than those with less needs. The financial protection aspect of UHC indicates that peoples’ contributions towards healthcare financing should be based on their ability to pay (Witter et al, 2017). Notably, developing and implementing effective and sustainable healthcare financing scheme has ignited serious debates among decision-makers, health professionals and other stakeholders especially in resources-constrained environments (OECD, 2028; World Bank Group, 2019), where healthcare systems are severely under-funded (Palmer et al, 2004). During the implementation of Structural Adjustment Programmes in the 1980s most resource-poor countries including Ghana charged user fees at the point of healthcare delivery as a form of resource mobilisation for their health systems (WHO Alliance for Health Policy and Systems Research, 2004). Resultantly, the inability to pay for healthcare services at the point of delivery caused inequality in health care access. Meanwhile, providing the populace with universal healthcare services has been touted as a crucial intervention for acceleration socioeconomic development in resource-constrained settings (OECD, 2028; World Bank Group, 2019). It is in the light of this that the international community recognises Health Insurance as an important means of eliminating financial barriers to access and utilisation of healthcare services (WHO, 2005). In Ghana, as part of efforts toward UHC, the Government instituted the National Health Scheme (NHIS) in 2003 but implemented in 2004 to provide financial risk protection for all residence. The scheme is administered by the National Health Insurance Authority (NHIA) Alhassan et al (2016), covering 95% of disease conditions in Ghana. Membership to the NHIS scheme is renewable on yearly basis and according to the clients’ contributions, while card bearers access healthcare services from health facilities accredited by the NHIA. The nature of contribution to the scheme is such that members pay based on their income and clients received healthcare according to their health needs. Thus, health insurance goes a long way to subsidise the cost of healthcare for members and the rich in society pays for his/her household members including the vulnerable (Ministry of Health, 2004). Generally, the NHIS has helped to reduce out-of-pocket (OOP) expenditures and enhance financial protection, yet persistent disparities remain across regions and income groups (WHO & World Bank, 2023). Recent evaluations of scheme show that while NHIS enrollment has improved access to healthcare services, it has not uniformly translated into reduced financial burden, especially in rural and underserved regions (GSS, 2022; UN SDG Report, 2023). This mismatch between policy intent and observed outcomes underscores a pressing need to assess the geospatial and inter temporal dimensions of NHIS impact on health expenditure, to align national health financing strategies more effectively with SDG 3 and broader equity-focused reforms. The aim of this paper is to investigate the impact of health insurance on out-of-pocket health expenditure based on the propensity score matching analytical techniques. While several studies have assessed the NHIS’s general effectiveness (Alatinga & Fielmua, 2023; Aryeetey et al., 2016), no study has disaggregated its influence across time and geography which are key dimensions for understanding disparities in financial protection. As Ghana continues to grapple with health financing challenges, especially in the face of demographic and regional inequalities (Abrokwah et al., 2021; WHO, 2023), this study investigates how NHIS enrollment and coverage have influenced out-of-pocket health spending patterns across regions and over time. By leveraging recent household survey panel data and geospatial econometric methods, the study contributes timely insights into the scheme’s efficacy and sustainability amid shifting health policy priorities. We contribute to sparse literature on health insurance and healthcare expenditure in two significant ways. First, to our knowledge, no previous study has accessed the geospatial and intertemporal impact of health insurance on out-of-pocket payment for healthcare in resource constrained context. By leveraging recent household survey panel data, we have disaggregated the impact of health insurance across time and geography which are crucial dimensions for understanding disparities in financial protection especially among the vulnerable in society Second, the empirical design of the current paper is more amendable to causal conclusion. Considering the observational nature of the data used, the insured and the uninsured are presumed to be equivalent at baseline, making it difficult to determine whether linear adjustment (e.g. using multivariate regression method) would produce biased estimates of the treatment effect on health services and health outcomes. We circumvent these problems by deploying the propensity score matching techniques, where the insured and uninsured subjects were matched into observable baseline characteristics thereby allowing these characteristics to be eliminated as potential confounders of the estimated impact of national health insurance on out-of- pocket spending on healthcare. The remaining sections of the paper are organised as follows. Section 2 presents the theoretical framework underpinning the study. Section 3 discusses methodological and analytical strategies employed for the study. The results and findings of the impact of health insurance on out-of-pocket expenditure are presented in section 4. The paper concludes with a summary of key findings and recommendations. 2.0 THEORETICAL UNDERPINNING: WHY DO PEOPLE DEMAND HEALTH INSURANCE? To explore this issue, it is essential to examine how risk and uncertainty influence individuals’ demand for health insurance. Traditional utility theory operates under the assumption that outcomes are known with certainty. However, real-life situations often involve uncertainty, particularly regarding health, where illness occurs unpredictably. This unpredictability challenges the standard utility model. The von Neumann-Morgenstern utility framework addresses this by modeling preferences under conditions of risk, suggesting that a rational individual will opt for the scenario that offers the highest expected utility. This model helps explain how people derive satisfaction (utility) from wealth, especially when faced with different possible outcomes (Nketiah-Amponsah & Sarkodie, 2014). Under a basic expected utility framework, an individual’s utility U depends on their disposable income Y, and there is a probability \(p\) that the individual may fall ill and incur a medical cost L. Let: \(q=coverage\ \left(\text{amount\ an\ insurence\ company\ pays\ incase\ of\ loss}\right)\) \begin{equation} \ \Pi=\ premium\ per\ cedis\ \ (₵)\ \nonumber \\ \end{equation} In this senario, Income level of the individual in the good state and bad state after taking insurance is given as : \(Y_{g}=Y-\Pi q\) and\(Y_{b}=Y-L+q-\Pi q\) (1) Given \(\Pi\), the individual chooses \(q\) so as to maximize his/her expected utility. \({}_{q}^{\text{Max}}\{\ \)pU(\(Y-L+q-\text{Πq})\ \)+ (1- p)U(\(Y-\text{Πq})\}\) (2) Where (1– p) is the probability of not falling sick, p is the probability of getting sick. \(\text{pU}^{/}(Y-L+q-\Pi q)\ (1-\Pi)+{(1-p)U}^{/}(Y-\Pi q)\ (-\Pi)=0\)(3) \(\text{pU}^{/}(Y-L+q-\Pi q)\ (1-\Pi)={(1-p)U}^{/}(Y-\Pi q)\ (\Pi)\) \(\frac{U^{/}(Y-L+q-\Pi q)\ }{U^{/}(Y-\Pi q)\ }=\frac{(1-p)(\Pi)}{p(1-\Pi)}\text{\ \ \ }\)(4) Suppose the insurance market is competitive ( a real case scenario), then it enjoys a zero expected Profit : E(Profit) = \(p[\text{Πq}-q]\)+\((1-p)[\text{Πq}]\) = 0 (5) \begin{equation} p\Pi q-pq+\Pi q-p\Pi q=0\nonumber \\ \end{equation} \(\Pi q=pq\) or \(\Pi=p\) (6) By substituting (5) back into (4): \(\frac{U^{/}(Y-L+q-\Pi q)\ }{U^{/}(Y-\Pi q)\ }=\frac{(1-\Pi)(\Pi)}{\Pi(1-\Pi)}=1\ \ \ \)thus \(U^{/}(Y-L+q-\Pi q)=U^{/}(Y-\Pi q)\ \) (7) This clearly shows \(Y-L+q-\Pi q=Y-\Pi q\) Hence,\(q^{*}=L\) (Full insurance) (8) From (8), the maximum amount that an individual would want to be insured equals the total loss incurred by him/her. 3.0 METHODS 3.1 DATA In this study secondary data analysis was conducted using the Annual Household Income and Expenditure Survey conducted by the Ghana Statistical Service. The Annual Household Income and Expenditure Survey conducted from the first quarter of the year 2022 to the third quarter of 2023 is Ghana’s first high-frequency, nationally representative household panel survey. Its primary aim is to collect both quarterly and annual information on household consumption and a broad array of demographic, economic, and welfare-related indicators. These include data on employment, food security, poverty (in both monetary and multidimensional terms), and health, all of which are crucial for evidence-based policymaking and research. The survey contributes to the generation of major macroeconomic indicators such as regional and quarterly GDP, unemployment and underemployment rates, inequality measures, and poverty statistics. The survey sample was derived from the 2021 Population and Housing Census, comprising 10,800 households across 600 Enumeration Areas (EAs), almost evenly split between urban (50.67%) and rural (49.33%) zones. Within each selected EA, eighteen households were randomly chosen to form a comprehensive dataset capable of producing reliable, region-specific expenditure estimates for GDP calculations. 3.3 ESTIMATION TECHNIQUES AND EMPIRICAL MODELING As clearly discussed by Chakraborty & Mukherjee (2020), causal impact evaluations typically rely on randomised experiments, where if the randomization is done effectively, the treatment and control groups are equivalent in every respect, except for their exposure to the intervention. This means any differences in outcomes can be directly attributed to the treatment. While randomized controlled trials (RCTs) are often considered the gold standard for establishing causality, they are frequently impractical or unethical in many real-world social and behavioral research settings. In such cases, researchers turn to non-experimental or observational methods. However, the reliance on non-experimental or observational approaches is vulnerable to selection bias, where differences in group characteristics may distort the estimated treatment effect. To address this challenge, various econometric techniques such as multivariate regression, instrumental variables, quasi-experimental designs, and propensity score matching (PSM) are employed to reduce bias. For the purposes of this study, the PSM approach is adopted to account for potential selection effects. The propensity score, denoted as {P(X)} represents the likelihood that an individual will participate in a program, given a specific set of observed characteristics X. Unlike randomized experiments, where treatment and control groups are equivalent in both observed and unobserved traits, propensity score matching (PSM) ensures similarity only in the variables that are measured. Matching units across many covariates can be analytically complex and burdensome due to computational limitations and data constraints. To simplify this, Rosenbaum and Rubin (1985) introduced the concept of the propensity score as a balancing score, which condenses multiple covariates into a single measure. This one-dimensional score enables more manageable matching when many variables are involved. Essentially, PSM involves matching individuals from the treatment and control groups who have similar propensity scores, with the possibility of excluding units that do not have comparable counterparts (Rubin, 2001). In applying PSM, this study equally adopts the three-step analytic process described by Guo and Fraser (2014). The steps are as follows: Step 1: Selection of background covariates In evaluating the impact of Ghana’s National Health Insurance Scheme (NHIS) on total household health expenditure, it is crucial to match households on a set of socio-demographic variables that are known to influence both healthcare utilization and insurance enrollment. Age is a key factor, as older individuals typically have greater health needs and are more likely to spend on medical care and enroll in NHIS, which could otherwise introduce bias in estimating the program’s effect (Kusi et al., 2015). Sex is also relevant, given gender differences in health-seeking behavior and service utilization, women for instance, tend to use more reproductive and maternal services, which are covered under NHIS (Abrokwah et al., 2014). Location of residence is another critical variable, as urban households generally have better access to healthcare facilities and information about NHIS, leading to higher enrollment and potentially different spending patterns compared to rural households (Dalaba et al., 2014). Education significantly affects health awareness, understanding of insurance benefits, and decision-making capacity, influencing both enrollment in NHIS and the way health services are utilised and paid for (Nguyen et al., 2011). Employment status plays a dual role by influencing ability to pay for premiums and access to formal sector health insurance schemes (Aryeetey et al., 2016). Finally, marital status often correlates with household size and support systems, which can affect health expenditure patterns and insurance uptake. Married individuals for instance, may be more likely to be insured, thus influencing total household spending (Aryeetey et al., 2016). Matching on these covariates helps ensure that treated and control households are comparable in key characteristics, thereby improving the credibility of causal inferences drawn using propensity score matching. Step 2: Calculation of propensity scores The propensity scores are calculated for each household using the psmatch2 command in Stata 17 to utilize a binary probit regression. Denoting the binary treatment condition as \(D_{i}\), this study define\(D_{i}=1\) if a household is in the treatment condition (insured) and\(D_{i}=0\) if a household is in the control condition (not insured). The vector of the conditioning variables is denoted as Xi and the vector of the regression parameters is denoted as \(\beta_{i}\). The binary probit regression depicts the conditional probability of receiving the treatment as follows: \(P(D_{i}/X_{i})=F(I_{i})=\int_{-\mathbf{\infty}}^{X_{i}\beta_{i}}e^{\frac{-z^{2}}{2}}\text{dz\ }\)where Z is the standard normal variable, i.e., Z ∼ N(0,\(\sigma^{2}\)), F is the standard normal CDF and \(I_{i}=X_{i}\beta_{i}\). Step 3: Matching After the propensity score estimation, the PSM is implemented. The study constructs matched pairs between the two types of households, i.e., the insured households (treatment group) and the uninsured households (control group) based on the maximum closeness of their propensity scores. Before matching, Kernel matching was employed to satisfy certain assumptions. Kernel matching enhances the validity of Propensity Score Matching (PSM) by addressing key assumptions more effectively than traditional matching methods. First, it strengthens the common support assumption by using a weighted average of all control units within a defined bandwidth, rather than discarding unmatched observations, thus minimizing bias due to poor overlap (Heckman et al., 1997; Smith & Todd, 2005). Second, it reduces model dependence and misspecification bias, as its non-parametric approach does not rely on exact matches or arbitrary caliper restrictions, making it more robust when the propensity score model is imperfectly specified (Zhao, 2004; Imbens & Wooldridge, 2009). Additionally, kernel matching improves covariate balance by smoothing differences across the propensity score distribution, which helps approximate the conditional independence assumption by ensuring treated and control groups are more comparable (Rosenbaum & Rubin, 1983; Becker & Ichino, 2002). By retaining more observations and efficiently weighting controls, kernel matching also enhances precision and efficiency, particularly in cases where the treatment group is small or the propensity score distribution is sparse (Heckman et al., 1998; Leuven & Sianesi, 2003). Thus, kernel matching offers a more flexible and reliable approach to meeting PSM’s core assumptions compared to exact or nearest-neighbor matching. 4.0 RESULTS AND DISCUSSION 4.1 Descriptive Statistics Table 1 provides the descriptive statistics of the 135,696 households included in the study. A majority of the respondents (63.88%) were not covered by the National Health Insurance Scheme (NHIS), while only 36.12% were insured. The gender distribution shows a slightly higher proportion of females (54.75%) compared to males (45.25%). In terms of residence, the sample is relatively balanced, with 52.26% living in rural areas and 47.74% in urban settings. Educational attainment is skewed toward lower levels, with 59.24% having basic education and only 4.85% reaching tertiary education. Employment status is nearly evenly split, with 51.09% employed and 48.91% unemployed. Regarding marital status, a notable 73.07% of respondents were single, while 26.93% were married. The average health expenditure among households was GHS 21.20, but the standard deviation of 184.99 suggests substantial variability, likely driven by a small proportion of households with extremely high medical costs (up to GHS 45,750). The average age of respondents was approximately 24.86 years, ranging from 15 to 115 years, indicating a youthful sample with some older adults included. These statistics reflect significant socioeconomic and demographic diversity, justifying the need for careful matching in subsequent impact analysis to ensure comparability between insured and uninsured groups. Table 1: Descriptive statistics of study variables Variable Category Frequency Percentage NHIS coverage Insured 49,019 36.12 Not insured 86,677 63.88 Sex Male 61,408 45.25 Female 74,288 54.75 Location Rural 70,908 52.26 Urban 64,788 47.74 Education level No education 29,858 22.00 Basic 80,386 59.24 Secondary 18,871 13.91 Tertiary 6,581 4.85 Employment status Employed 69,324 51.09 Unemployed 66,372 48.91 Marital status Married 36,539 26.93 Single 99,157 73.07 Obs Mean Std Dev. Min. Max. Health expenditure 135,696 21.20396 184.9859 0 45750 Age 135,696 24.86058 20.04672 15 115 4.2 Geospatial Distribution of Total Health Expenditure in Ghanaian Households The geospatial distribution of total health expenditure among Ghanaian households, based on the 2023 Annual Housing Income and Expenditure Survey (sample size: 135,696), reveals stark regional disparities that underscore critical gaps in healthcare equity and financing. The Ashanti Region leads with the highest total health expenditure at GHS 413,251.7, followed by Greater Accra (GHS 293,211.8) and Eastern Region (GHS 248,994.2). These three regions alone account for over one-third of national household health spending, pointing to concentrated financial flows toward more urbanized, populous, and economically active regions. Several factors may account for these patterns. First, population density and service availability are key drivers. Ashanti and Greater Accra, being highly urbanized, have larger populations and more healthcare facilities, both public and private which naturally result in higher overall expenditures (GSS, 2022). Additionally, these regions have a broader income base, enabling more households to afford and seek healthcare services. However, this does not necessarily imply better financial protection. On the contrary, high absolute spending may reflect significant out-of-pocket (OOP) burdens in the absence of effective pooling mechanisms, particularly for specialized or private care (Frimpong et al., 2021). Conversely, the Upper West (GHS 95,462.6), Ahafo (GHS 94,116.0), and Savannah (GHS 101,088.1) regions report the lowest total health expenditures, which raises concerns about access and utilization rather than Figure 1: Regional distribution of health expenditure efficiency. These areas are historically underserved, with fewer health facilities, limited transport infrastructure, and lower household incomes, factors that suppress health spending not by choice, but by constraint (Alatinga & Fielmua, 2023). These disparities reinforce the urgent need to strengthen the National Health Insurance Scheme (NHIS) as a mechanism for both financial risk protection and equitable access. While NHIS has improved utilization in some regions, its reach and impact remain uneven, particularly in the northern and newly created regions (Aryeetey et al., 2016). By integrating geospatial expenditure data into policy planning, Ghana can better target NHIS subsidies, expand coverage in low-spending areas, and reduce OOP burdens where they are most concentrated. 4.3 Geospatial Impact of NHIS on Health care Expenditure in Ghana The impact of the NHIS on household health expenditure shows significant regional variation across Ghana, revealing how local socioeconomic and infrastructural characteristics shape the effectiveness of the scheme. In heavily urbanized regions such as Greater Accra and Ashanti, NHIS coverage is associated with substantial reductions in healthcare costs. Using Kernel matching, the ATT estimates as shown in table 2 were -18.53 GHS and -15.33 GHS, respectively, while the Nearest Neighbor estimates in table 5 at appendix produced very similar magnitudes of -17.67 GHS and -13.96 GHS. The consistency across methods strengthens the evidence that the scheme is particularly effective in reducing out-of-pocket spending in regions with better access to accredited healthcare providers, higher awareness of NHIS benefits, and relatively efficient administrative oversight. Figure 2a : Geospatial impact of NHIS on health care expenditure in Ghana In newer or less urbanized regions such as Western North and Ahafo, the Kernel matching results (-19.41 GHS and -11.19 GHS) are echoed by the Nearest Neighbor estimates (-16.90 GHS and -10.66 GHS), both confirming strong reductions in health expenditures. This reflects the relatively low baseline access in these areas, meaning NHIS membership sharply reduces the cost burden for those who do seek care. Similarly, in Central and Eastern regions, the Kernel-based reductions of -9.66 and -6.28 GHS align closely with Nearest Neighbor estimates of -7.31 and -6.36 GHS, respectively. These findings reinforce the insurance’s role in mitigating financial risk in areas that mix rural and urban health service dynamics. Figure 2b : Geospatial impact of NHIS on health care expenditure in Ghana The northern and more economically disadvantaged regions such as the Northern, Savannah, Upper East, and Upper West also present a consistent pattern across both matching methods. Kernel estimates ranged from -5.27 to -6.75 GHS, while Nearest Neighbor estimates showed reductions between -4.72 and -5.98 GHS. In all cases, the effects remain statistically significant, underscoring that NHIS delivers tangible financial protection in areas of high vulnerability despite persistent gaps in medication and diagnostic coverage. This convergence between the two methods enhances confidence that NHIS is indeed serving its protective role where poverty and limited infrastructure are most pressing. In Bono, Bono East, and Oti, both Kernel (-3.77 to -7.33 GHS) and Nearest Neighbor (-3.20 to -6.65 GHS) estimates indicate moderate but significant cost reductions. The close correspondence of results again supports the interpretation that these semi-urban regions benefit meaningfully from NHIS, though challenges such as uneven provider accreditation and reimbursement bottlenecks may temper the full potential of the scheme. The Volta Region remains an exception. While Kernel matching showed a statistically insignificant reduction of -4.99 GHS, the Nearest Neighbor estimate of -7.01 GHS is also negative but statistically insignificant at conventional levels. This consistency across methods strengthens the interpretation that the NHIS effect is muted in Volta, likely due to lower enrollment, weak provider capacity, and persistent out-of-pocket payments. This highlights the need for targeted interventions to improve facility resourcing, service delivery, and benefit awareness in the region. Interestingly, the convergence between Kernel and Nearest Neighbor results confirms the robustness of the findings: NHIS significantly reduces household health expenditures across most regions of Ghana, though the magnitude and significance vary with local socioeconomic and infrastructural contexts. These results highlight the importance of region-specific strategies in health financing policy, ensuring that NHIS can fully deliver on its promise of financial risk protection. Table 2: Geospatial Impact of NHIS on Health care expenditure in Ghana REGION Treated Controls Difference S.E. Obs. WESTERN Unmatched 15.8366 19.3715 -3.5348 2.2053 7,726 ATT 15.8366 21.2419 -5.4052** 2.3502 CENTRAL Unmatched 17.8386 25.1475 -7.3089** 2.5852 8,355 ATT 17.8386 27.4996 -9.6610*** 2.4101 GREATER ACCRA Unmatched 23.7095 41.3836 -17.6740*** 6.3915 8,372 ATT 23.7095 42.2444 -18.5349*** 5.2611 VOLTA Unmatched 21.7828 28.7940 -7.0112** 3.4952 8,411 ATT 21.7828 26.7791 - 4.9962 3.1270 EASTERN Unmatched 20.125 26.4852 -6.3602** 2.7582 10,138 ATT 20.125 26.4063 -6.2813** 2.3610 ASHANTI Unmatched 21.6240 35.5851 -13.9613*** 3.9295 13,501 ATT 21.6240 36.9492 -15.3255*** 3.3363 WESTERN NORTH Unmatched 19.7746 36.6765 -16.9018 13.7913 7,054 ATT 19.7746 39.1802 -19.4055* 11.7174 AHAFO Unmatched 8.7488 19.4107 -10.6618*** 2.4446 6,112 ATT 8.7488 19.9383 -11.1894*** 2.1985 BONO Unmatched 13.1167 19.7701 -6.6534*** 2.5546 8,023 ATT 13.1167 20.4516 -7.3349*** 2.0012 BONO EAST Unmatched 13.5238 16.7242 -3.2004** 1.5481 8,575 ATT 13.5238 17.2998 -3.7758*** 1.5289 Table 2 con’t: Geospatial Impact of NHIS on Health care expenditure in Ghana OTI Unmatched 16.9186 22.6179 -5.6993** 2.2573 6,708 ATT 16.9186 23.8710 -6.9523*** 2.1777 NORTHERN Unmatched 10.9020 16.8769 -5.9749*** 1.9541 9,975 ATT 10.9020 17.6563 -6.7543*** 1.9932 SAVANNAH Unmatched 14.1452 18.8608 -4.7156** 2.1025 6,092 ATT 14.1452 19.1515 -5.0063** 2.1754 NORTH EAST Unmatched 11.1126 13.5853 -2.4727 1.5121 5,857 ATT 11.1126 13.6526 -2.5400* 1.4320 UPPER EAST Unmatched 16.4154 21.9594 -5.5439** 2.1077 9,747 ATT 16.4154 22.7830 -6.3675*** 1.9216 UPPER WEST Unmatched 9.0020 13.9441 -4.9421** 2.5139 7,713 ATT 9.0020 14.2733 -5.2713** 2.1089 AGGTREGATE Unmatched 16.0960 24.0928 -7.9970*** 1.0451 135,696 ATT 16.0960 25.7058 -9.6100*** 0.8984 4.4 Intertemporal Impact of NHIS on Health Care Expenditure in Ghana Table 3 presents the intertemporal (quarterly) impact of the National Health Insurance Scheme (NHIS) on healthcare expenditure in Ghana in 2022. The Kernel results indicated in table 3show consistent reductions across all quarters, with the strongest effect in the fourth quarter (-41.73 GHS). The Nearest Neighbor estimates from Table 6 in appendix confirm this overall pattern, though with slightly different magnitudes: Q1 (-6.37 GHS), Q2 (-11.84 GHS), Q3 (-4.50 GHS), and Q4 (-41.18 GHS). The similarity across methods strengthens confidence that the NHIS reduces household healthcare spending throughout the year, while also suggesting that seasonal and structural dynamics shape the intensity of the effect. Figure 3: Intertemporal impact of NHIS on health care expenditure in Ghana In the first quarter (Q1), Kernel matching estimated an ATT of -6.94 GHS, closely mirrored by the Nearest Neighbor estimate of -6.37 GHS. This convergence indicates that insured households consistently benefit from lower healthcare costs at the start of the year. In Q2, Kernel (-13.18 GHS) and Nearest Neighbor (-11.84 GHS) results both show a stronger effect, likely reflecting seasonal illness peaks during the rainy season, when malaria, respiratory infections, and water-borne diseases increase healthcare utilization. The consistency between the two methods reinforces the interpretation that NHIS shields households from sharp seasonal shocks in healthcare costs. In Q3, the Kernel estimate of -4.86 GHS is in line with the Nearest Neighbor estimate of -4.50 GHS. Both results suggest a slight moderation in NHIS’s impact, perhaps due to seasonal stabilization in disease prevalence or temporary challenges in health system readiness, such as drug shortages or delayed reimbursements that reduce scheme effectiveness. Despite the smaller reduction, the fact that both methods find significant impacts indicates that NHIS continues to provide consistent financial protection across quarters. The fourth quarter (Q4) shows the most dramatic reductions under both approaches: Kernel (-41.73 GHS) and Nearest Neighbor (-41.18 GHS). This striking convergence highlights the robustness of the finding. The unusually large impact may stem from seasonal surges in healthcare demand, end-of-year policy initiatives, or fiscal-year effects in NHIS reimbursements that improve benefit delivery. It may also reflect deferred care from earlier quarters, leading to a spike in utilization during the final months of the year. Either way, the near-identical estimates across methods underscore the strength of NHIS’s protective role during periods of heightened financial risk. Taken together, the consistency between Kernel and Nearest Neighbor results across all four quarters demonstrates the robustness of the intertemporal findings. The scheme reliably lowers healthcare spending for insured households, with seasonal fluctuations amplifying its effect at particular times of the year. The nationwide significance of these reductions reinforces the conclusion that NHIS is not only regionally impactful but also temporally reliable in offering financial risk protection. Table 3: Intertemporal Impact of NHIS on Health care expenditure in Ghana QUARTER Treated Controls Difference S.E. Obs. 2022Q1 Unmatched 13.50487 19.8748 -6.3699*** 1.1445 42,926 ATT 13.50487 20.4485 -6.9436*** 1.0577 2022Q2 Unmatched 21.7153 33.5513 -11.8361*** 2.7927 42,129 ATT 21.7153 34.8905 -13.1753*** 2.2741 2022Q3 Unmatched 10.8338 15.3378 - 4.5040*** 0.8859 41,500 ATT 10.8338 15.6928 - 4.8590*** 0.8103 2022Q4 Unmatched 58.7506 99.9266 -41.1760*** 12.6894 3,826 ATT 58.7506 100.4835 - 41.7329*** 11.1925 5.0 CONCLUSIONS AND POLICY RECOMMENDATIONS This study presents compelling evidence on the geospatial and intertemporal disparities in the financial protection offered by Ghana’s National Health Insurance Scheme (NHIS). Geospatially, the analysis reveals that NHIS significantly reduces healthcare costs in nearly all of Ghana’s 16 administrative regions, though the magnitude of impact varies substantially. Urbanised and relatively well-resourced regions such as Greater Accra, Ashanti, and Central experience the largest financial relief from NHIS coverage. These reductions are likely to be linked to higher awareness, better access to accredited facilities, and more efficient health delivery service. In contrast, some rural or economically disadvantaged regions such as Volta, where the effect is statistically insignificant show weaker impacts, suggesting gaps in coverage effectiveness, service availability, or benefit awareness. The spatial disparities underscore the need for region-specific interventions, including improved health system capacity and outreach, to ensure equitable protection across the country. Intertemporally, the study finds that NHIS consistently reduced household healthcare spending across all four quarters of 2022, with the most substantial impact observed in the fourth quarter. This quarterly variation is likely influenced by seasonal disease trends, healthcare-seeking behavior, and policy or administrative factors such as enrollment drives or provider reimbursements. The steep drop in fourth quarter expenditures may reflect intensified healthcare utilization due to end-of-year illness surges or improved NHIS responsiveness during that period. Overall, the findings suggest that NHIS is not only effective in lowering health-related financial burdens but that its impact is sensitive to both geographic and temporal contexts. To maximise its effectiveness, NHIS reforms must be both spatially targeted and seasonally responsive, ensuring that healthcare protection is equitable, timely, and consistent across various regions in Ghana. As the country strives to achieve Universal Health Coverage (UHC), the government should strengthen the NHIS to achieve its mandate of ensuring equitable healthcare access and utilisation across the various regions and in all seasons. Further, proper regulations should be made to control user charges and other by-way expenses at the point of service utilisation and improve the availability of health services. Efforts should be made to institute effective surveillance and resource allocation systems to better anticipate seasonal disease trends and incidence, particularly in the second and fourth quarters of the year, when health expenditure burdens tend to spike. Additionally, proper and more targeted health insurance literacy campaigns, focusing on rural and peri-urban populations with historically low enrollment rates, should be implemented to bridge knowledge gaps about NHIS benefits, especially among the poor and underserved in society. Finally, concrete steps should be taken to explore mechanisms to expand the NHIS benefit package to cover high-cost services such as diagnostics and chronic illness care, while ensuring predictable and reliable funding flows to health facilities to prevent seasonal lapses in service quality. While the NHIS remains a critical pillar of Ghana’s health financing architecture, its current performance is uneven, with significant gaps across both geography and time. The next phase of reform must focus on equity, responsiveness, and institutional coordination, anchored by evidence and led by a coalition of ministries and agencies. REFERENCES 1. Abrokwah, S. O., Moser, C. M., & Norton, E. C. (2014). The effect of social health insurance on prenatal care: The case of Ghana. International Journal of Health Care Finance and Economics , 14(4), 385–406. 2. Abrokwah, S. O., Moser, C. M., & Norton, E. C. (2021). The effect of social health insurance on health care utilization and out-of-pocket payments: Evidence from Ghana. Health Economics , 30(4), 697–714. 3. Abuosi, A., Adzei, F., Anarfi, J., Badasu, D., Atobrah, D., & Yawson, A. (2015). Investigating parents/caregivers financial burden of care for children with non-communicable diseases in Ghana. BMC Pediatrics , 15(1). 4. Alatinga, K. A., & Fielmua, N. (2023). Health insurance and equity in access to healthcare in Ghana: A regional perspective. Health Policy and Planning , 38(2), 123–134. 5. Alatinga, K. A., & Fielmua, N. (2023). Health insurance and equity in access to healthcare in Ghana: A regional perspective. Health Policy and Planning , 38(2), 123–134. 6. Aryeetey, G. C., Jehu-Appiah, C., Spaan, E., Agyepong, I., & Baltussen, R. (2016). Costs, equity, efficiency and feasibility of identifying the poor in Ghana’s NHIS . Health Policy , 120(7), 914–922. 7. Aryeetey, G. C., Jehu-Appiah, C., Spaan, E., Agyepong, I., Baltussen, R. (2016). Costs, equity, efficiency and feasibility of identifying the poor in Ghana’s NHIS . Health Policy , 120(7), 914–922. 8. Aryeetey, G. C., Westeneng, J., Spaan, E., Jehu-Appiah, C., Agyepong, I., & Baltussen, R. (2016). Can health insurance protect against out-of-pocket and catastrophic expenditures and also support poverty reduction? Evidence from Ghana’s National Health Insurance Scheme. International Journal for Equity in Health, 15(1), 116. 9. Becker, S. O., & Ichino, A. (2002). Estimation of average treatment effects based on propensity scores. The Stata Journal , 2(4), 358-377. 10. Chakraborty, S., & Mukherjee, V. (2020). Revisiting the Economic Costs of Arsenicosis: A PSM Approach. Ecology, Economy and Society-the INSEE Journal, 3(2), 33-58. 11. Chankova, S., Sulzbach, S., & Diop, F. (2008). Impact of mutual health organizations: Evidence from West Africa. Health Policy and Planning , 23(4), 264–276. 12. Dalaba, M. A., Akweongo, P., Aborigo, R., Awine, T., Azongo, D., & Awoonor-Williams, J. K., et al. (2014). Does the national health insurance scheme in Ghana reduce household cost of treating malaria in the Kassena-Nankana districts? Global Health Action, 7(1), 23848. 13. Frimpong, A. O., Amporfu, E., & Arthur, E. (2021). Effect of the Ghana National Health Insurance Scheme on exit time from catastrophic healthcare expenditure. African Development Review , 33(3), 492-505. 14. Frimpong, J. A., Helleringer, S., & Asuming, P. O. (2021). Catastrophic health expenditures and the role of health insurance in Ghana. Health Economics Review , 11(1), 1–13. 15. Ghana Statistical Service (GSS). (2022). Ghana Living Standards Survey Round 8 (GLSS 8): Main Report. Accra: GSS. 16. Ghana Statistical Service (GSS). (2022). Ghana Living Standards Survey Round 8 (GLSS 8) : Main Report. Accra: GSS. 17. Guo, S., & Fraser, M. W. (2014). Propensity score analysis: Statistical methods and applications (Vol. 11). SAGE publications. 18. Heckman, J. J., Ichimura, H., & Todd, P. E. (1997). Matching as an econometric evaluation estimator: Evidence from evaluating a job training program. Review of Economic Studies , 64 (4), 605-654. 19. Heckman, J. J., Ichimura, H., & Todd, P. E. (1998). Matching as an econometric evaluation estimator. Review of Economic Studies , 65(2), 261-294. 20. Imbens, G. W., & Wooldridge, J. M. (2009). Recent developments in the econometrics of program evaluation. Journal of Economic Literature , 47(1), 5-86. 21. Kusi, A., Hansen, K. S., Asante, F. A., & Enemark, U. (2015). Does the National Health Insurance Scheme provide financial protection to households in Ghana? BMC Health Services Research , 15(1), 331. 22. Leuven, E., & Sianesi, B. (2003). PSMATCH2: Stata module to perform full Mahalanobis and propensity score matching. Statistical Software Components S432001, Boston College Department of Economics 23. Nguyen, H. T., Rajkotia, Y., & Wang, H. (2011). The financial protection effect of Ghana National Health Insurance Scheme: Evidence from a study in two rural districts. International Journal for Equity in Health, 10(4). 24. Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41-55. 25. Rosenbaum, P. R., & Rubin, D. B. (1985). Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. The American Statistician , 39(1), 33-38. 26. Rubin, D. B. (2001). Using propensity scores to help design observational studies: application to the tobacco litigation . Health Services and Outcomes Research Methodology , 2, 169-188. 27. Smith, J. A., & Todd, P. E. (2005). Does matching overcome LaLonde’s critique of nonexperimental estimators? Journal of Econometrics, 125(1-2), 305-353. 28. United Nations. (2023). The Sustainable Development Goals Report 2023 . New York: UN. 29. World Health Organization & World Bank. (2023). Tracking Universal Health Coverage: 2023 Global Monitoring Report . Geneva: WHO. 30. World Health Organization. (2023). Tracking universal health coverage: 2023 global monitoring report . WHO and World Bank. 31. Zhao, Z. (2004). Using matching to estimate treatment effects: Data requirements, matching metrics, and Monte Carlo evidence. Review of Economics and Statistics , 86(1), 91-107. Funding The authors have no support or funding to report. Availability of Data Data for this study were sourced from the Ghana Statistical Service and available here https://www.statsghana.gov.gh Ethics approval and consent to participate Ethics approval of this study is not required since the data are secondary and are in the public domain. Competing Interests The authors declare that they have no competing interest Acknowledgements The authors gratefully acknowledge the Ghana Statistical Service for providing the data from the Annual Household Income and Expenditure Survey (AHIES) which made this study possible. We also thank the editors and anonymous reviewers for their constructive feedback. APPENDIX 1 Table 4: Measurement and description of study variables Variable Description Measurement NHIS coverage 1 = Insured Categorical 0 = Not insured Sex 1 = Male Categorical 0 = Female Location 1 = Urban Categorical 0 = Rural Education level 0 = No education Categorical 1 = Basic 2 = Secondary 3 = Tertiary Employment status 1 = Employed Categorical 0 = Unemployed Marital status 1 = Married Categorical 0 = Single Health expenditure Total health expenditure (GH₵) Continuous Age Age in years Continuous APPENDIX 2 Table 5: Geospatial Impact of NHIS on Health care expenditure in Ghana [Nearest Neighbor Marching] REGION ATT Std. Err. T- statistic WESTERN Nearest Neighbor -3.535** 1.800 -1.9639 CENTRAL Nearest Neighbor -7.309*** 2.311 -3.163 GREATER ACCRA Nearest Neighbor -17.674*** 5.152 -3.431 VOLTA Nearest Neighbor -7.011 4.860 -1.443 EASTERN Nearest Neighbor -6.360** 2.317 -2.745 ASHANTI Nearest Neighbor -13.961*** 3.250 -4.295 WESTERN NORTH Nearest Neighbor -16.902** 7.623 -2.217 AHAFO Nearest Neighbor -10.662*** 2.142 -4.977 BONO Nearest Neighbor -6.653*** 1.972 -3.374 BONO EAST Nearest Neighbor -3.200** 1.491 -2.146 Table 5 con’t: Geospatial Impact of NHIS on Health care expenditure in Ghana OTI Nearest Neighbor -5.699** 2.119 -2.689 NORTHERN Nearest Neighbor -5.975*** 1.942 -3.076 SAVANNAH Nearest Neighbor -4.716** 2.102 -2.243 NORTH EAST Nearest Neighbor -2.473* 1.368 -1.807 UPPER EAST Nearest Neighbor -5.544*** 1.802 -3.077 UPPER WEST Nearest Neighbor -4.942** 1.966 -2.513 AGGTREGATE Nearest Neighbor -9.650*** 1.028 -9.384 APPENDIX 3 Table 6: Intertemporal Impact of NHIS on Health care expenditure in Ghana [Nearest Neighbor Marching] QUARTER ATT Std. Err. T- statistic 2022Q1 Nearest Neighbor -6.370*** 1.029 -6.189 2022Q2 Nearest Neighbor -11.836*** 2.197 -5.387 2022Q3 Nearest Neighbor -4.504*** 0.792 -5.687 2022Q4 Nearest Neighbor -41.176*** 10.893 -3.780 Information & Authors Information Version history V1 Version 1 26 September 2025 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords geospatial analysis health expenditure intertemporal variation national health insurance scheme (nhis) propensity score matching (psm) Authors Affiliations Emmanuel Boadu 0009-0008-1197-9724 [email protected] University of Education Winneba View all articles by this author Anthony Abbam University of Education Winneba View all articles by this author Anselm Komla Abotsi University of Education Winneba View all articles by this author Metrics & Citations Metrics Article Usage 296 views 159 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Emmanuel Boadu, Anthony Abbam, Anselm Komla Abotsi. GEOSPATIAL AND INTERTEMPORAL IMPACT OF NATIONAL HEALTH INSURANCE SCHEME (NHIS) ON HOUSEHOLD HEALTHCARE EXPENDITURE: EVIDENCE FROM GHANA. Authorea . 26 September 2025. DOI: https://doi.org/10.22541/au.175886290.09288122/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.175886290.09288122/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9fed1ccbbf8dad07',t:'MTc3OTI5ODY3OQ=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();