Financial Literacy and Health Insurance Uptake: Evidence From the Households in Tanzania

Health insurance is a key component of global healthcare systems, providing broad financial protection and access to healthcare services. In developing countries, however, health insurance coverage remains low, and many individuals and households face significant barriers to enrolment. For instance, in many African countries, including Tanzania, the health insurance enrolment rate is reported to be below 10 (C​h​e​m​o​u​n​i​,​ ​2​0​1​8). Many countries have begun introducing health insurance for people experiencing poverty because it allows households to prepay for healthcare, reducing the share of catastrophic health expenditures they must pay out of pocket (G​e​n​g​ ​e​t​ ​a​l​.​,​ ​2​0​1​8). Low health insurance penetration in these regions often stems from limited awareness, affordability issues, and cultural attitudes toward insurance (B​a​y​k​e​d​ ​e​t​ ​a​l​.​,​ ​2​0​2​3). Low awareness and limited understanding of health insurance among individuals threaten uptake in developing countries (S​e​k​h​r​i​ ​&​ ​S​a​v​e​d​o​f​f​,​ ​2​0​0​5).

The Tanzanian health system has to confront significant challenges in achieving universal health coverage (UHC). Many households there rely on out-of-pocket payments, leading to financial hardship and inequity in accessing healthcare services (B​i​n​y​a​r​u​k​a​ ​e​t​ ​a​l​.​,​ ​2​0​2​1; F​i​n​a​n​c​i​a​l​ ​S​e​c​t​o​r​ ​D​e​e​p​e​n​i​n​g​ ​T​a​n​z​a​n​i​a​,​ ​2​0​2​3​b). The Community Health Fund (CHF) and the National Health Insurance Fund (NHIF) together play a pivotal role in providing affordable health insurance. However, their coverage is limited, particularly among the informal sector and rural populations (D​u​r​i​z​z​o​ ​e​t​ ​a​l​.​,​ ​2​0​2​2). Financial literacy is essential for shaping household decisions about health insurance by improving understanding of insurance products, strengthening risk management, and fostering all-inclusive financial planning. For instance, individuals with higher financial literacy are more likely to comprehend the benefits and terms of insurance policies, resulting in increased uptake (L​u​s​a​r​d​i​ ​&​ ​M​i​t​c​h​e​l​l​,​ ​2​0​1​4).

From a governance and risk-management perspective, health insurance is a formal household risk-pooling mechanism that converts uncertain and potentially catastrophic medical expenditure into a more predictable premium. This function is especially significant wherever out-of-pocket payment could compel households to borrow, sell assets, reduce essential consumption, or forgo care. Evidence from low- and middle-income countries nevertheless showed that insurance does not provide uniform financial protection; concomitant outcomes depend on scheme design, benefit coverage, provider’s access, administrative capacity, and accountability (E​z​e​ ​e​t​ ​a​l​.​,​ ​2​0​2​3; R​a​h​m​a​n​ ​e​t​ ​a​l​.​,​ ​2​0​2​2). Accordingly, financial literacy matters not only for enrolment but also for household risk governance because it may help households assess coverage, costs, exclusions, and the reliability of available schemes.

Several studies have examined the relationship between financial literacy and health insurance uptake, primarily in developed countries. For example, C​l​a​r​k​ ​e​t​ ​a​l​.​ ​(​2​0​1​4​) found that higher financial literacy correlated with greater health insurance coverage among U.S. adults. A cross-country study by Z​h​e​n​g​ ​e​t​ ​a​l​.​ ​(​2​0​2​5​) recommended exploring underrepresented regions and emerging trends, with implications for expanding health insurance coverage and promoting healthcare equity. The role of health insurance in improving household healthcare utilization is widely debated in many developing countries because poverty, high user fees, and out-of-pocket payments strain household incomes (K​i​t​o​l​e​ ​e​t​ ​a​l​.​,​ ​2​0​2​3).

Household financial literacy is critical for users to make informed financial decisions, including health insurance consumption. In Tanzania, where health insurance coverage is relatively low, knowing the role of financial literacy could inform efforts to improve health insurance uptake and strengthen households’ financial risk protection (F​i​n​a​n​c​i​a​l​ ​S​e​c​t​o​r​ ​D​e​e​p​e​n​i​n​g​ ​T​a​n​z​a​n​i​a​,​ ​2​0​2​3​b; L​o​t​t​o​,​ ​2​0​2​0​a). Despite evidence from developed countries, causal understanding of how financial literacy influences health insurance uptake in Sub-Saharan Africa remains limited, particularly in Tanzania. This paper examined how two dimensions of financial literacy, i.e., numeracy and borrowing skills, were related to health insurance enrolment in Tanzania, encompassing heterogeneity by rural–urban residence, employment status, and age.

The structure of this paper is organized as follows: Section 2 reviews the literature. Section 3 presents the methodology, including data sources, key variables, and models. Section 4 discusses the empirical results and further analyses. Finally, we conclude the paper with insights into future research directions.

The CHF was introduced to provide health insurance to rural populations and workers in the informal sector, while the NHIF targets formal-sector employees. Despite these initiatives, enrolment rates remain low, with significant disparities in coverage between urban and rural areas (B​o​r​g​h​i​ ​e​t​ ​a​l​.​,​ ​2​0​1​3). The government’s efforts to expand health insurance coverage, such as the CHF and the NHIF, have had limited success, thus highlighting the need for complementary strategies to enhance uptake (B​o​r​g​h​i​ ​e​t​ ​a​l​.​,​ ​2​0​1​3; M​a​c​h​a​ ​e​t​ ​a​l​.​,​ ​2​0​1​4).

Tanzania advocated community-based health insurance (CBHI) in the 1990s as part of the broader health sector reforms to improve healthcare financing and increase access to health services. Two main schemes were established: the CHF, which targets rural households and the informal sector, and Tiba Kwa Kadi (TIKA), which focuses on individuals in urban areas. Both schemes receive central government support through matching grants. However, challenges such as low enrolment and weak management propel the government to transfer CHF administration to the NHIF in 2009 in order to enhance operational efficiency and expand coverage. Subsequent to this transition, the NHIF implemented several reforms, including improvements to the benefit package and administrative systems. Nevertheless, low participation in CBHI schemes presents a persistent challenge in Tanzania, as well as in many other low- and middle-income countries (K​i​g​u​m​e​ ​&​ ​M​a​l​u​k​a​,​ ​2​0​2​1).

The Organisation for Economic Co-operation and Development (OECD) defined financial literacy as knowledge and understanding of financial concepts and risks; it included the skills, motivation, and confidence to apply such knowledge and understanding to make effective decisions across a range of financial contexts, to improve the financial well-being of individuals and society, and to encourage participation in economic life (L​u​s​a​r​d​i​,​ ​2​0​1​9). Meanwhile, L​u​s​a​r​d​i​ ​&​ ​M​i​t​c​h​e​l​l​ ​(​2​0​1​4​) argued that financial literacy predominantly affected individuals’ competence to manage finances, save, invest, and borrow effectively.

Financial literacy is prevailingly recognized as a key determinant of financial behaviour and risk-management choices (L​u​s​a​r​d​i​ ​&​ ​M​i​t​c​h​e​l​l​,​ ​2​0​1​4). It could elevate households’ ability to compare premiums and benefits, interpret exclusions, understand risk pooling, and assess insurance’s value in the long term. Several studies have linked financial literacy to health-related financial behaviours. Evidence from China demonstrated that higher financial literacy was positively associated with both the probability of holding life insurance and the premiums paid, thus suggesting that financial knowledge could demolish information barriers in making complex insurance decisions (W​a​n​g​ ​e​t​ ​a​l​.​,​ ​2​0​2​1). Likewise, household evidence indicated that financial literacy could reduce poverty vulnerability partly by encouraging participation in commercial insurance and improving the choice of formal financial channels (W​a​n​g​ ​e​t​ ​a​l​.​,​ ​2​0​2​2).

Previous evidence collected from developing and low-income economies, however, is not uniformly positive. In Senegal, B​o​n​a​n​ ​e​t​ ​a​l​.​ ​(​2​0​1​7​) discovered that a standalone insurance-literacy module did not prominently increase health-microinsurance take-up. In contrast, interventions that cut down on entry costs substantially raised adoption, especially among poorer households. This finding suggested that knowledge might be necessary but insufficient when liquidity constraints, trust, product quality, and administrative barriers remained binding. In urban Ethiopia, education, knowledge of the benefit package, and the ability to mobilize funds were associated with willingness to join CBHI. However, affordability was considered to be a major constraint (D​e​k​s​i​s​a​ ​e​t​ ​a​l​.​,​ ​2​0​2​0). Recent Ugandan evidence further illustrated that digital literacy was positively associated with insurance inclusion and operated partly through the adoption of insurance technology, emphasizing the growing importance of digital capability in low-penetration insurance markets (K​i​w​a​n​u​k​a​ ​&​ ​S​i​b​i​n​d​i​,​ ​2​0​2​4).

The broader household-risk literature clarified the reasons for promoting insurance participation. A systematic review and meta-analysis encompassing low- and middle-income countries reported that CBHI generally increased healthcare utilization and reduced out-of-pocket and catastrophic expenditure. Nevertheless, the strength of financial protection varied by context and scheme design (E​z​e​ ​e​t​ ​a​l​.​,​ ​2​0​2​3). A related scoping review concluded that findings on financial risk protection remained geographically concentrated and often measured only catastrophic expenditure or impoverishment, leaving gaps on coping strategies, forgone care, and low-income-country settings (R​a​h​m​a​n​ ​e​t​ ​a​l​.​,​ ​2​0​2​2). Thus, existing research supports a relationship among literacy, insurance participation, and household resilience, but it also confirms that translating knowledge into enrolment depends on household resources and institutional conditions. Financial literacy positively impacts purchase intentions, and health insurance literacy positively influences intentions. Research on geographical areas remains concentrated in major economies like China, the United States, and India, while smaller or lower-income countries receive insufficient attention (Z​h​e​n​g​ ​e​t​ ​a​l​.​,​ ​2​0​2​5).

The growing literature satisfactorily addressed three research questions. First, many studies relied on evidence from life insurance, microinsurance, or commercial insurance outside Tanzania. In contrast, comparatively few studies examined household health-insurance enrolment using nationally representative evidence from a low-income East African country. Second, previous research often used a single composite variable to measure financial literacy and may have overlooked that numeracy and borrowing skills influenced insurance enrolment differently. Last, average effects might conceal inequalities in access, since rural location, employment arrangement, and age affected access to providers, income stability, exposure to employment-based insurance schemes, and the ability to use financial education. Tanzania is unique because health insurance coverage remains limited and divided into formal and community-based schemes, where people in rural areas or those informally employed face particular challenges in accessing and affording insurance (D​u​r​i​z​z​o​ ​e​t​ ​a​l​.​,​ ​2​0​2​2; K​i​g​u​m​e​ ​&​ ​M​a​l​u​k​a​,​ ​2​0​2​1). This research will contribute to the literature by exploring how numeracy and borrowing skills influence insurance enrolment separately.

The study employed a cross-sectional analysis using data from the FinScope Tanzania 2023 Survey (F​i​n​a​n​c​i​a​l​ ​S​e​c​t​o​r​ ​D​e​e​p​e​n​i​n​g​ ​T​a​n​z​a​n​i​a​,​ ​2​0​2​3​a). This survey was a nationally representative demand-side survey of adults (16 years or above) living in Tanzania. The survey was conducted through public-private sector collaboration with the Ministry of Finance and Planning, Tanzania and Zanzibar; the Bank of Tanzania; the Financial Sector Deepening Tanzania (FSDT); the National Bureau of Statistics (NBS); and the Office of the Chief Government Statistician, Zanzibar (OCGS). The survey targeted 10,005 individuals and achieved a 99% response rate, with 9,915 interviews. The survey covered 30 regions in mainland Tanzania and Zanzibar. This study was conducted at the household level; the data were collapsed to 5,298 households, restricting respondents to heads of household (Hhs) with knowledge of health insurance. The data obtained include household demographics, financial behaviour, and health insurance status.

To measure financial literacy, this paper adopted three standard sets of questions found in numerous surveys in Tanzania and other countries. The questions were based on two concepts: numeracy and borrowing skills, including the capacity to calculate interest rates and understand risk diversification (L​u​s​a​r​d​i​ ​&​ ​M​i​t​c​h​e​l​l​,​ ​2​0​1​4). L​o​t​t​o​ ​(​2​0​2​0​b​) also measured financial literacy using interest rates, discounting, and borrowing. This study was based on a two-question concept, in which households were interviewed with questions to test their financial literacy, including numeracy and borrowing. Table 1 below describes the variables in this study.

This study investigated the relationship between household financial literacy and health insurance consumption in Tanzania using a probit model. The probit model suits binary dependent variables, as it assumes a standard normal cumulative distribution function for the probability that the dependent variable is a binary indicator of health insurance enrolment (1 if the household is enrolled in any health insurance scheme, 0 otherwise).

The specific probit model for this study is written as:

where, Y_i represents the health insurance status of Hh i (1 if enrolled in health insurance, 0 if not enrolled on the health insurance). FL_i: Financial literacy score of Hh i, X_i is a set of control variables, including household education level, gender, location, income, employment, and age. ε_i is the error term. The coefficient for financial literacy (FL_𝑖) indicates how changes in financial literacy score affect the probability of being insured. These coefficients represent the effect of the respective explanatory variables (age, gender, education, income, employment status, and urban/rural residence) on the probability of being insured.

Table 2 indicates that 5,298 households are Hhs in the sample. This is the total number of observations used in the analysis. The results indicated that only about 11.9% of households in the sample had health insurance, suggesting low health insurance coverage in Tanzania compared with other countries like Ghana. A​d​j​e​i​-​M​a​n​t​e​y​ ​&​ ​H​o​r​i​o​k​a​ ​(​2​0​2​3​) identified that only about 40% of the population was enrolled in the National Health Insurance Scheme (NHIS). Numeracy skills, based on principal component analysis (PCA) scores, are relative to a sample mean of zero; a sample mean of 0.021 implies that households in the sample have a very slight positive deviation of 0.021 standard deviations from the population benchmark. Whereas, 9.6% had financial literacy based on borrowing skills, especially knowledge of interest rates. Therefore, the sample's average financial literacy level was quite low. As indicated, 59.2% of Hhs in the sample were married, and 62.9% of Hhs in the sample were male. Approximately 20.7% of Hhs had at least a secondary education, and only about 9.2% were currently employed. This result highlighted the need for financial literacy programs in Tanzania.

Multicollinearity Test

The study opted for a probit model to analyze the relationship between health insurance consumption and household financial literacy in Tanzania, while controlling for households’ demographic and socio-economic factors, including age, gender, education, income, employment status, and urban/rural residence. Before running probit estimations, the researchers conducted diagnostic tests to check for multicollinearity, including the variance inflation factor (VIF) and the pairwise correlation matrix. As shown in Table 3, the VIF indicated the extent of multicollinearity. The VIF test shows that all variables have VIF values below 10, suggesting that multicollinearity is not a major concern in this model. Similarly, the correlation coefficients are found mostly below 0.3 on average, indicating no serious multicollinearity problems when these variables are included in our estimation together.

Model Specification Test

We analsyzed whether the model includes all necessary variables and none that are unnecessary. Auxiliary variables were defined as the predicted value (_hat) and squared predicted value (_hatsq). For a model to be specified appropriately, _hat must be statistically significant. Table 4 shows that _hat is statistically significant at the 1% level (p = 0.000), confirming that the model is well specified.

Table 5 presents results from a probit model examining the relationship between health insurance coverage and financial literacy, alongside other control variables. The analysis revealed a statistically significant positive relationship between financial literacy and health insurance coverage across various model specifications. In column 1, with no basic controls, the results were positive at the 1% significance level, except in column 4, where borrowing skills is significant at 10%. Specifically, the coefficient on numerical skills was 0.09, indicating that Hhs with numerical skills were more likely to enrol in health insurance coverage. As shown in column 2, the coefficient on borrowing skills was 0.221, indicating that Hhs with borrowing skills were more likely to enrol in health insurance coverage.

The analysis included all other control variables; the results were positive at the 1% significance level. Column 3 shows that the coefficient on numerical skills was 0.09, indicating that Hhs with numerical skills were more likely to enrol in health insurance coverage, consistent with the result in column 1. In column 4, the coefficient on borrowing skills was 0.158, indicating that Hhs with borrowing skills were more likely to enrol in health insurance coverage. The positive association between health insurance and financial literacy was consistent across all model specifications, even after controlling for other factors. This finding is consistent with K​o​p​p​l​i​n​ ​(​2​0​2​4​), who reported that financial literacy significantly increased the probability of holding health insurance and reduced adverse health-related financial behaviours. Similarly, H​u​b​b​a​r​d​ ​(​2​0​2​4​) concluded that financial knowledge was associated with a greater likelihood of obtaining health-insurance coverage and a lower likelihood of accumulating medical debt. Evidence from India also concurred that financial literacy was one mechanism through which digital financial participation promoted voluntary health-insurance enrolment (S​e​n​g​u​p​t​a​ ​&​ ​R​o​o​j​,​ ​2​0​2​5). Therefore, the stability of the coefficients across the model specifications suggested that financial literacy was an important correlate of health-insurance enrolment in Tanzania.

To ensure the robustness of our findings, several heterogeneous analyses were conducted to test the robustness of the findings. The analysis included heterogeneous effects by urban and rural locations, employment status, and age categories.

We examined the robustness of the findings by understanding the relationship between health insurance enrolment and financial literacy separately for urban and rural households, along with other control variables. Table 6 indicates that numeracy skills are more strongly associated with health insurance enrolment in urban than in rural areas, while borrowing skills are significant only for urban households. This pattern may reflect closer proximity to insurers and health facilities, greater exposure to formal financial products, more reliable information, and more frequent monetary transactions in urban settings. Rural households may understand basic calculations, yet remain unable to translate that knowledge into enrolment because of irregular incomes, distance, limited provider networks, and fewer suitable products. The weaker and less stable association for borrowing skills is also plausible because knowledge of interest and credit does not directly imply an ability to evaluate insurance benefits, exclusions, or provider quality; it may matter only where credit markets and insurance payment options are accessible. These results demonstrated that financial literacy, measured through numeric skills and borrowing skills, positively correlated with health insurance enrolment. The analysis revealed a distinct disparity between urban and rural areas. Urban settings showed a stronger influence of financial literacy types on health insurance enrolment, particularly numeric skills, while borrowing knowledge had a smaller and less consistent influence. In contrast, these variables had less impact on rural areas. Socio-economic factors such as age, education level, and employment status are significantly positive predictors of health insurance coverage. Gender differences exist, particularly in rural areas where male Hhs are less likely to enrol in insurance coverage.

Table 7 tests robustness by examining the relationship between health insurance enrolment and financial literacy, based on how numeracy and borrowing skills interact with health insurance across Hhs’ employment status, along with other control variables. Numeracy skills are shown to be positively related to enrolment across all employment status groups, while borrowing skills do so only for full-time and privately employed Hhs. Numeracy skills refer to the ability to make decisions; they enable premium budgeting, cost/benefit comparison, and understanding contribution schedules regardless of employment status. Borrowing skills, in turn, become relevant under formal borrowing facilities, constant money flows, linkages to payrolls and transactions with banks. Such factors are characteristic of full-time and private rather than part-time and unemployed households. Instability in borrowing skill estimates could also reflect conceptual differences between interest-calculation knowledge and knowledge about insurance, which involves probability concepts and claim uncertainty. Consequently, programmes should include numeracy and insurance education along with suitable products and contributions schemes for workers with non-standard income. These results showed a stronger role for numeracy skills across all Hhs’ employment statuses. This aligns with other studies such as Z​h​e​n​g​ ​e​t​ ​a​l​.​ ​(​2​0​2​5​), which has a more consistent and stronger impact on health insurance enrolment than on borrowing skills. Fully employed households likely benefit most from financial literacy because of structured work benefits. Part-time workers and unemployed individuals benefit less because they have limited access to employer-provided options. Education has consistently boosted health insurance enrolment across groups, thus underscoring its critical role in financial decision-making. Older individuals are more likely to enrol in health insurance, but the effect of age also varies by employment status.

The study further tested the robustness of the findings by examining the relationship between health insurance enrolment and financial literacy, based on how numeracy and borrowing skills interact with health insurance across different household age groups, along with other control variables. Table 8 suggests that numeracy and borrowing skills were associated with enrolment among non-elderly Hhs but not among elderly Hhs. One possible reason is that younger, working-age adults may use financial knowledge more actively because they have longer planning horizons, greater exposure to education and digital or formal financial services, and more opportunities to compare and purchase insurance. Among older households, enrolment may depend more on affordability, health status, family support, eligibility conditions, trust, or previous experience with providers. These constraints could weaken the conversion of financial knowledge into action. Borrowing skills again display a less stable relationship because credit knowledge is not equivalent to insurance literacy and may facilitate enrolment only when households have access to credit and sufficient repayment capacity. The age-specific findings therefore supported targeted communication and simplified enrolment for older households, while retaining numeracy and insurance education for younger groups. The proposed mechanisms are plausible interpretations and require direct testing in future research.

The study also examined the mechanisms through which financial literacy could influence insurance consumption. Table 9 identifies three channels as dependent variables in the analysis: insurance works, which measures the effectiveness of the insurance in place; insurance benefits, which represent the benefits or perceived value of insurance; and insurance costs, which refer to the costs associated with insurance. Each outcome is modelled using a similar set of independent variables, including region dummies.

In column 1, the dependent variable is how insurance works. The results showed a positive and significant coefficient of 0.126 for numeracy skills at the 1% significance level. This means Hh with higher numeracy skills are more likely to understand how health insurance works. With higher numeracy skills, households can better interpret the intricacies of insurance policies and related premium and risk-pooling calculations.

In column 2, the dependent variable is insurance benefit. The results showed a negative coefficient of -0.052 for numeracy skills at the 1% significance level. Higher numeracy skills were associated with lower health insurance enrolment, based on insurance benefits. This finding can be interpreted as follows: households with higher numeric skills recognize that insurance policies involve a lot of information and therefore tend not to exaggerate their knowledge. Alternatively, households with lower numeric skills can have an exaggerated sense of knowledge due to subjective bias.

In column 3, the dependent variable is insurance cost. The results recorded a negative coefficient of -0.042 for numeracy skills at the 1% significance level. A negative coefficient means that higher financial literacy is associated with lower insurance costs. Higher numeric skills greatly decrease the probability that households claim that they cannot afford health insurance. It can be assumed that people with higher numeric skills manage their family budgets better and can weigh risks and benefits when buying insurance.

One potential source of endogeneity in estimating the effect of financial literacy on household health-insurance enrolment is that financial literacy may be correlated with unobserved characteristics that also influence insurance decisions, such as financial decision-making capacity. To address this concern, the study used a two-stage least squares (2SLS) instrumental-variable approach, using a binary indicator equal to one if household members are involved in financial decisions about purchasing goods and services, and zero if otherwise, as an instrument for financial literacy. Table 10 presents the results of a 2SLS estimation used to address potential endogeneity in the key variable, financial literacy (numerical skills), when examining its effect on enrolment into health Insurance. Column 1 presents the first-stage regression that estimates numerical skills using instrumental variables for financial decision-making and controls. For involvement in financial decision-making, the result showed a positive coefficient of 0.252 at the 1% significance level. This means that involvement in financial decisions, such as purchasing goods and services, spending and saving money, increases households’ numerical skills and improves financial literacy.

Meanwhile, the results showed that older Hh had lower financial literacy, while married, male, and employed Hh tended to have higher financial literacy. Hh with higher education had higher financial literacy, suggesting that higher education improved financial literacy and that employed individuals had higher financial literacy. First stage F-statistic = 87.84: A high F- statistic indicated that the instruments used were potentially strong predictors of numerical skills in respect of financial literacy, thus fulfilling a crucial requirement for 2SLS estimation.

Column 2 reports the second-stage regression results. This stage used instrumented numerical skills to estimate its effect on enrolment in health insurance, while using the same control variables. For numerical skills, the results showed a positive coefficient of 0.21 at the 1% significance level. This indicated that a one-unit increase in instrumented numerical skills was associated with a 0.21-unit increase in the health insurance outcome. This suggested that higher financial literacy led to desirable health insurance outcomes through improved decision-making and a deeper understanding of insurance options. The instrument's validity was supported by a strong first-stage F-statistic and Hansen J- statistic, suggesting that the endogeneity problem in numeracy skills was effectively addressed.

This study investigated the relationship between household financial literacy and health insurance consumption in Tanzania, to address research gaps identified in recent policy reports and academic literature. Our analysis of nationally representative data revealed three major findings: first, financial literacy significantly increased the probability of health insurance enrolment, with effects ranging from 9% (numeracy skills) to 22% (borrowing skills). Second, the interrelationship operated primarily through improved understanding of insurance mechanics and reduced cost barriers. Third, effects were heterogeneous, while stronger in urban areas, among formally employed individuals, and within younger households.

These findings extended financial literacy theory to the health insurance context in low-income settings, showing that financial literacy affected not only traditional financial behaviours (saving and investing) but also health-related financial protection decisions. The heterogeneous effects aligned with behavioural economics perspectives and suggested that structural constraints mediated the translation of knowledge into behaviour. With only 11.9% of Tanzanian households currently insured, our results underscored that financial literacy interventions could meaningfully increase coverage. However, standalone education programs might have limited impact unless integrated with affordability measures (premium subsidies), structural improvements (rural service delivery), and trust-building initiatives. Through a governance and risk-management lens, these findings position health insurance as more than a healthcare-financing product: it is a household mechanism for pooling risk and limiting exposure to unpredictable medical expenditure.

Financial literacy represents a modifiable determinant that, when coupled with structural reforms, could accelerate progress toward universal health coverage and financial protection for vulnerable households. Financial literacy could strengthen demand-side governance by enabling households to evaluate premiums, benefits, exclusions, and provider reliability. However, effective protection also requires transparent scheme rules, accountable administration, accessible complaint and claims procedures, and dependable service delivery. Evidence from low- and middle-income countries shows that financial-protection effects from insurance vary by context and scheme design. Offices such as the Tanzania Insurance Regulatory Authority should coordinate literacy initiatives with affordability measures, improvements of rural access, and institutional safeguards rather than treating them as standalone solutions. Such integration would make insurance uptake more likely to translate into genuine resilience against medical expenditure shocks and reduced household financial vulnerability.

This study has few limitations. The cross-sectional design prevented analysis of literacy and insurance dynamics over time. While instrumental variables addressed endogeneity, unobserved factors may persist. Financial literacy measures, though based on established frameworks, might not capture all relevant dimensions in the Tanzanian context. Improving health insurance coverage in Tanzania requires addressing both supply-side constraints and demand-side behavioural barriers. Future studies should employ longitudinal designs to track behavioural change, experimental approaches to test integrated literacy-insurance interventions, and mixed methods to explore cultural and trust factors. Research should also examine the role of digital financial literacy as mobile insurance platforms expand.