Total Observation = N_C * T* Sample per Country/Year

where N_Crepresents number of countries and T represents the number of years.

Total observation = 20 * 23 = 460

Jamaica and other 19 countries with the North America were selected. Their selection was based on the data availability and assurance that this countries have balanced data set for all the variables employed in this study. The countries and their respective capital cities are reported in Table A1 underAppendix.

The conceptual framework of this study is built on [1], which is considered a household with the following resource allocation. An individual with a fixed income I which is allocated between healthcare consumption, H and non-healthcare consumption, NH.Note that the goal of a rational consumer is to maximize overall utility subject to the resource constraint, I.

The constrain is expressed in equation as follows:

where ph is the cost of healthcare, pnh represents the price of non-healthcare. If the consumer receives the subsidies d, (where d ≤ s), the budget constraint is then modified as shown in equation (2). Equation (3) explains the interaction between OOPE and healthcare subsidy (hsd), where f represents a function, and x connotes other variables that affect OOPE

OOPE = f (hsd, x) (3)

Empirical Framework

To investigate the nexus between user fees in health OOPE and healthcare subsidy in some selected countries in North America, the healthcare financing and protection framework is suitable. Through this framework, we explore how various health financing sources such as external funding, government financing and personal expenditure interact and determine access to healthcare and financial protection [15]. This framework shows how a rise in in external support can reduce OOPE and enhance affordability and healthcare utilization. By estimating this model, the framework gives more information on how well-structured health-care financing can reduce economic burden on households, most especially in a low and middle income countries and enhance better access to health care services. Therefore, from equation (3) we have the dynamic and static panel model specification is as follows

where OOPE which denotes out-of-pocket health expenditure is the dependent variable, EXE,External health expenditure per capita (current US$), (GHE) connotes Current health expenditure per capita (current US$). These two variables are used as proxy for healthcare subsidies. Others are control variables, which are; Life expectancy at birth (LEXP) Population of female above the age of 65, (POPFML), Population of adult above the age of 65 (POPADT) and mortality from cardiovascular diseases and others (MCVDOTHERS) all which we obtained from world development indicators (WDI)

Table 1 presents the descriptive statistics of all the variables employed in this study. As shown in Table 1, we have 460 observations. The mean of user fees in health about $157 per capita, with the minimum value of 5.116 per capital and maximum value $637.824 per capita. Similarly, external health expenditure has an average value of $8.288 per capita, while the minimum and maximum values stand at $0.004 and $465.077 per capita, respectively. In the same vein, the average, minim and maximum values of government health expenditure, are $431.4, $8. And $1472.0 per capita, respectively. Gross domestic product has a mean value of $2.69e per capita, with a minimum and maximum values of 3.33e and 4.60e current US dollar, in that other. The average years of life expectancy in the study area is about 72 years, minimum, 45.6, maximum 80.4 years, while the standard deviation is 45.1 years. Furthermore, Table 1 also presents the mortality rate of cardiovascular diseases and other diseases, with an average mortality of 17.5%, while the minimum and maximum values stand at 8.9% and 34.8%, correspondingly.

Table 1. Descriptive statistics of the variables 8.9% and 34.8%, correspondingly.

Our correlation matrix presented in Table 2 shows that health subsidies proxy by EHE and GHE are positively correlated with User fees in health which is proxy by OOPE. To determine the presence of multicollinearity the independent variables, the study employed variance inflation Factor (VIF) as reported in Table 3. The result shows no significant problem of multicollinearity. This confirms the level of robustness of our model.

Table 2. Correlation analysis.

Table 3. Variance inflation factor (VIF).

The result of stationarity test is reported inTable 4. This is conducted when you have a relatively large “t”, that time frame. Following [16] which is specified as follows:

where time trend ( γ 1 i t ) and individual effects ( γ i ) are incorporated in the model. Δ y i t is the series for cross-section iii and time t. µi is the error term. Note that one important source of heterogeneity in this model is deterministic component since the coefficient of the predicted variable is constrained to be homogeneous across all units in the panel. Our findings show that the test statistics are below the critical values at the 1% level of significance at level except for GHE which falls below the critical value at first difference.

Table 4. Levin-Lin-Chu panel unit root test.

Trend Analysis between OOPE and Subsidy in Jamaica

Figure 1presents the trend analysis between OOPE and external health expenditure in Jamaica between 2000 and 2022. Figure 1shows that external health expenditure is far lower than the proportion of OOPE in the years under review. Figure 1also shows that there are inconsistencies in the trend of OOPE. Notably, in the years 2001 and 2008, the variable depicts unusual spikes. On the contrary, external health expenditure reveals a steady rise except for 2013 when the graph nosedived and then steadily rose until 2022.

Figure 1. OOPE and health external health expenditure in Jamaica.

Similarly,Figure 2presents the analysis and interaction between OOPE and government health expenditure in Jamaica. Unlike external health expenditure, government health expenditure is higher than OOPE in Jamaica. Additionally, the trend shows an upward trend within the year under review. This implies that the Jamaican government is improving on her health system on a yearly basis. As it has been previously, observed, OOPE maintains a steady rise but for 2009 when there was a slight drop before it picked up again.

Figure 2. OOPE and government health expenditure in Jamaica.

Table 5 presents the empirical results of this study. The fixed effect as well as random effect results were estimated through Stata 17. The test of choice of preference was conducted Hausman test. The result of the fixed effect was preferred due to the value of the chi square value which shows the level of significance at 1% significant level. The results of F-statistics and the corresponding level of significant are 392.11 and 0.000, respectively. The value of R² suggests that the model is fit.

From Table 5, there is a positive relationship between both External health expenditure, government heath expenditure and user fees in health. Specifically, a percent change in EHE and GHE leads to a change in household’s OOP in health by 0.0004 and 0.4638, respectively. In other words, OOPE is not health subsidies elastic. Furthermore, as would be expected, a 1% change (increase) in life expectancy reduces the OOPE in health by 1.3440. Similarly, there is an inverse relationship between change in the population of adults and out of pocket health expenditure.

Table 5 further shows that there is an inverse relationship between cardiovascular and other diseases mortality rate and out of pocket health expenditure.

It also shows that a 1% increase in CVDS and other diseases mortality rate reduces OOPE by 0.2285.

Table 5. Static panel data.

The results of the dynamic panel is reported in Table 6. The results of the Sargan test and the values are 22.6174 and 1.0000, respectively. The result shows that the instrument is not valid since the p-value is insignificant. Similarly, we report the report the p-value of AR1 and AR2 as 0.000 and 0.1364. The rule of thumb says that the p-value of AR2 to be insignificant for the model to be declared free of autocorrelation.

The direction of coefficients of our subsidies shows that one of the proxies is positive while the other has a negative relationship. The positive relationship of GHE implies that a 1% change (increase) in EHE increases OOPE by 0.5236%. Conversely, the negative association between OOPE and EHE shows that a % change (increase) in EXE reduces OOPE by 0.0025%. Though not in conformity with a priori expectation, they are both significant at 1% level of significance. Similarly, GDP is also positively correlated with OOPE.

Table 6 further shows that a 1-year change (increase) in life expectancy of individuals in the region under consideration increases the amount of OOPE by $0.6659. Similarly, a one 1% change in the household adult population reduces their OOPE by $0.1830. Additionally, a 1% change (increase) in mortality rate due to cardiovascular and other diseases reduces the amount of OOPE by $0.2743.

Table 6. GMM dynamic panel result.

The study considered the effects of the external health expenditure and government health expenditure (healthcare subsidies) on OOPE in Jamaica and some selected countries within North America. We conclude that only external health expenditure has an expected impact on OOPE while GHE did not have expected outcomes on OOPE. It implies that GHE is not sufficient enough to incite impacts on user fees in health in the study area. Similarly an increase in GDP does not have potential to provoke a fall in the amount of OOPE. This also suggests that absolute growth in GDP does not automatically translate to improvement in health expenditure or health outcomes.

Our findings also show that there is a statistically significant correlation between CVDs and other diseases and, OOPE, implying that a reduction in cardiovascular mortality reduces the amount of households OOPE. These findings conclude that since only external health expenditure reduces OOPE, government should work towards the achievement of universal health insurance coverage for her citizens.

There are lot of missing figures in some other independent variables that affect OOPE. This has led to reducing the number of variables, so as not to have unbalanced data set.