The Effect of Health Expenditure on Life Expectancy

Many factors contribute to increasing levels of health spending in all developed countries. Those determinants of health expenditures include GDP growth, life expectancy, infant mortality, medical advancement, technological improvement, public funding, population ageing, alcohol consumption, tobacco consumption, and so on [1].

The population's health is under the influence of two types of health systems: 1) the Social Security Systems and 2) the National Health Services and their resources. The relationship between resources and outcomes is important to assess whether a country has a functioning health system [2].

Various authors such as Lange and Vollmer [3], Mohanty and Behera [4], and Raitano [5] analysed and explained the importance of the issue of life expectancy, health expenditure, GDP, economic growth and other variables of interest in this research.

Economic growth can give governments the means to ensure a better public health system and invest in an infrastructure that directly or indirectly affects health [3]. Healthy people can serve as a major driver for economic activities and the development of a nation. In an emerging economy, government plays an important role in providing affordable and accessible health services to its poor and needy population. In general, public health practices are necessary to implement large-scale reductions in mortality rates [4].

According to Raitano [5], in the economic policy debate, an ageing population will lead to large increases in age-related public spending - mainly pensions and health care. Most per capita health care costs are concentrated in the last years of one's life (so-called "death-related" costs). The main results obtained from studies such as those by Ahn et al. [6], Batljan and Lagergren [7], Breyer and Felder [8], Felder et al. [9] on health economics show that there is an interaction between age, proximity to death, and health care expenditures. Based on these studies, one can note that age alone is not a good predictor of increased health care spending, but proximity to death should also be used as a predictor of this spending.

Authors such as Elola et al. [2], Raitano [5], Baltagi and Moscone [10], Lange and Vollmer [3], Phi [1], Mohanty and Behera [4] have analysed the effects of health expenditure on the national life expectancy of OECD countries. The main importance of conducting this study lies in the fact that health expenditures and the national life expectancy are very important determinants of the health status and sustainable economic development of a nation. Therefore, to achieve sustainable economic development, countries must ensure a sustainable health system and an efficient allocation of government spending in this sector. Given that OECD countries have a high level of GDP per capita and have experienced positive results in the health sector in recent years.

Economic theory and the findings of various studies mentioned in the literature review below show that an increase in health expenditure is expected to result in positive effects on the quality of the health sector in terms of the population's longevity. Therefore, a study to examine the correlation between national life expectancy and health spending is necessary to help policymakers and governments in OECD countries understand the effectiveness of health spending on national life expectancy. The data obtained can be used by authors, policymakers and governments as a benchmark (for comparison) to analyse the effect of national health expenditures on national longevity.

This study accounts for exogenous factors such as economic growth, population, infant mortality rate, potential years of life lost, deaths from cancer and suicide rate; as control variables. These indirectly affect the effectiveness of health expenditures as a function of the national life expectancy of OECD countries. The novelty of this scientific study, when compared to the studies by other authors, is the inclusion of independent variables, which give a special focus to the academic discussion on the various challenges and problems faced by the health systems of OECD countries and which affect the effectiveness of health expenditures on national life expectancy.

In recent years, OECD countries have faced various challenges in their health systems, such as increasing infant deaths, potential years of life lost, deaths from cancer, suicide rates, etc. These health system challenges have an indirect impact on the receipt and allocation of health expenditures by the governments and policymakers of OECD countries. Therefore, taking into account the impact of these exogenous factors on the health system and the continuous population growth and economic growth in these countries adds value to the data and literature on the effectiveness of health expenditures on national life expectancy.

This study will provide new scientific evidence regarding the impact of health expenditures on the national life expectancy of OECD countries, taking into account the controlling effect of some independent variables. Moreover, compared to studies of other authors, this study incorporates a wider coverage.

Given that this study deals with the effects of health expenditures on national life expectancy, panel data analysis was applied from a sample of 37 OECD countries and 518 observations during the period 2005-2018. Initially, the econometric model was built where national life expectancy was taken as a dependent variable. In contrast, health expenditures and financing, GDP per capita and productivity, population, infant mortality rate, potential years of life lost, deaths from cancer and suicide rate were taken as independent variables. For data processing, we used the Microsoft Excel application and Software for Statistics and Data Science (STATA) and carried out Multiple Linear Regression, Fixed Effect, Random Effect, Hausman - Taylor Regression, GMM Model - Arellano Bond Estimation, Generalized Estimating Equations (GEE Model) as well as linear trend analysis - the historical and comparative methods.

The main hypothesis of this study is:

H1: Health expenditure has a positive impact on the national life expectancy of OECD countries.

H0: Health expenditure does not positively impact the national life expectancy of OECD countries.

To prove the validity of the hypothesis of this study, we have constructed the following econometric model:

LEBit = β0 + β1 HEFit + β2 GDPcapit + β3 POPit + β4 IMRit + β5 PYOLLit + β6 DFCit + β7 SRit + γit  

Where:

LEB – Life Expectancy

HEF – Health Expenditures and Financing

GDPcap – Gross Domestic Product per capita and Productivity (GDP per cap.)

POP – Population

IMR – Infant Mortality Rate

PYOLL – Potential Years of Life Lost

DFC – Deaths from Cancer

SR – Suicide Rate

γ – stochastic variables (other factors not taken into account in the model)

i – code and t – time period.

Table 2. Description of the variables included in the econometric models

Source: Data processing by authors (2021)

Table 2 shows the specification and description of the variables of the econometric model of this study. We take the LEB as the dependent variable. This represents the statistical age at which a person is expected to live based on various actuarial data. National life expectancy as a variable is widely applied in the financial industry in life insurance, pension planning and social security benefits. In OECD countries, national life expectancy calculations are performed by national statistical agencies based on very large amounts of actuarial data.

The main independent variable in this study is HEF. This includes all expenditures for providing health services, family planning activities, nutritional activities, and emergency health assistance. This category of health expenditures does not include the provision of drinking water and hygiene. Thus, HEF is the main independent variable data on final consumption expenditures for health care goods and services from the functions of the health sector, various financial schemes, cost structures of health insurance provision and the formation of capital in the health care system.

The second independent variable in this study is gross domestic product per capita and productivity. This indicator is a global measure applied to measure the prosperity of nations, and various economists use this indicator together with GDP to analyse the prosperity of a country based on its economic growth. The governments of different countries apply the analysis of GDP per capita to understand how a country's economy is growing along with its population.

Other control variables that are important in explaining what effect health expenditure has on the national life expectancy of OECD countries are population, infant mortality rate, potential years of life lost, deaths from cancer and suicide rates.

The third independent variable in this study is the population of OECD member countries between 2005 and 2018. The main reason the population is taken as an independent variable is to analyse the controlling effect of the number of inhabitants within a country that impacts the correlation between national life expectancy and health expenditure.

The infant mortality rate is the other independent variable included in this econometric model, defined as the number of deaths of children under the age of one, expressed per 1,000 births. The various international factors influencing infant mortality rates are due to variations between countries in registration practices for premature infants. According to OECD estimates, the United States and Canada are two countries that record a much higher percentage of babies weighing less than 500g, with a low probability of survival, resulting in a fairly high level of reported infant mortality per 1,000 births.

Potential years of life lost (PYOLL) is the other independent variable included in this econometric model as a controlling and exogenous variable. This variable represents a summary measure of premature mortality. This variable provides statistical data on deaths that occur at a young age and how to prevent them. The calculation of the potential years of life lost is done by summing up all the deaths that occur at any age and multiplying this value by the number of years left to live up to a limit of the chosen age. According to OECD estimates for health statistics, this limit is set at the age of 70. However, it is worth noting that this limit is not fixed since each OECD member country can set a specific standardised value for each year.

Another independent variable considered in this study is the number of deaths caused by cancer. According to OECD, more than 100 different types of cancer are known, and the risk of developing these diseases increases with age. The mortality rate is based on the number of cancer deaths recorded in a country and the size of the respective population. This variable is reported annually, and cancer mortality is measured per 100,000 inhabitants (total), per 100,000 men and per 100,000 women.

The last independent variable in this study is the suicide rate. This variable is defined as the deaths deliberately initiated and performed by a person with full knowledge or expectation of its fatal outcome. Based on OECD estimates, statistical data on this variable, which is compared between different countries, are influenced by several reporting criteria, including how a person intends to commit suicide, who is responsible for completing the death certificate if a forensic investigation is carried out and the provisions on the confidentiality of the cause of death.

Table 4 presents the descriptive statistics for the variables included in this study, while Table 5 shows the Pearson correlation coefficient values between LEB and other independent variables (HEF, GDP per cap., POP, IMR, PYOLL, DFC, SR).

In Table 5, we notice that all the variables incorporated in the econometric model are correlated. LEB has a positive average correlation with HEF (R = 0.5149), which means that an increase in health expenditures and financing increases the value of national life expectancy.

There is a very weak value of the Pearson coefficient between national life expectancy (LEB) and the value of GDP per capita (R = 0.2324). This means that GDP has a positive effect on increasing the value of national life expectancy. Even the correlation between national life expectancy (LEB) and population (POP) is very weak (R = -0.0327), indicating a negative linear correlation between these two variables. Therefore, with an increase in POP in OECD countries, we will have a decrease in the value of LEB.

National life expectancy (LEB) and infant mortality rate (IMR) are, on average, negatively linearly correlated. Therefore, an increase in IMR will result in a decrease in the value of LEB. There is a very high negative linear correlation between national life expectancy (LEB) and potential years of life lost (PYOLL). With an increase in (PYOLL), we will have a decrease in LEB value. The linear correlation between national life expectancy (LEB) and deaths from cancer (DFC) is weak and negatively correlated, implying that with increasing (DFC), we will have a decrease in the value of LEB. The same applies to the suicide rate, as national life expectancy has a weak negatively correlation with suicide levels. This means that with increasing suicide rates (SR), we will decrease the value of national life expectancy (LEB).

As noted above, to test the hypotheses of this study, we applied six statistical tests using the Stata application. Specifically, linear regression, random effect, fixed effect, Hausman - Taylor Regression, GMM Model - Arellando Bond Estimation and Generalized Estimating Equations (GEE Model). The following are the econometric results for the hypothesis of this study and our aim to determine the impact of health expenditures on the national life expectancy in OECD countries.

$\begin{aligned} L E B_{i t}=\beta_0+& \beta_1 H E F_{i t}+\beta_2 \ln G D P c a p_{i t}+\beta_3 P O P_{i t} \\ &+\beta_4 I M R_{i t}+\beta_5 P Y O L L_{i t}+\beta_6 D F C_{i t} \\ &+\beta_7 S R_{i t}+\gamma_{i t} \end{aligned}$

$\begin{aligned} L E B_{i t}=85.17565 &+0.1929712 H E F_{i t} \\ &+0.1833448 \ln _{i t} \boldsymbol{D P c a p}_{i t} \\ &-0.0078948 \boldsymbol{P} \boldsymbol{O P}_{i t}+0.1077461 \boldsymbol{I M R _ { i t }} \\ &-0.0010156 \boldsymbol{P} \boldsymbol{O L L _ { i t }} \\ &-0.0186612 \boldsymbol{D F C}_{i t}+0.0189861 \boldsymbol{S R}_{i t} \\ &+\gamma_{i t} \end{aligned}$

Results in Table 6 show that all variables included in this econometric model are significant (p-value of less than 0.05). We, therefore, use the GEE model to interpret the effect of health expenditures on national life expectancy in OECD countries?

β0 - If all other factors are constant, then the value of national life expectancy will be 85.17 units.

β1 - If health expenditure and financing (HEF) increases by 1 unit keeping other factors constant, then national life expectancy (LEB) will increase by 0.193 units. This statement is statistically significant (p-value = 0.000 < 0.01).

Based on this result and the trend in the analysed period (2005-2018), we can conclude that an increase in health expenditures and financing in OECD countries positively impacts the value of national life expectancy. To better explain this impact of health expenditure on national life expectancy, we also consider the economic growth of OECD countries. Thus, if the value of GDP is high, countries can and tend to spend more on the health sector, contributing to the increase in the value of life expectancy in the OECD area. Therefore, (β2) if GDP per capita increases by 1%, then national life expectancy (LEB) will increase by 0.0018 units. This statement is correct since the significance value (P-value = 0.032 < 0.05) is at the level of statistical significance. The empirical results that explain the correlation and effect between health expenditure and national life expectancy are consistent and comparable even to the studies by other authors noted in Table 1.

β3 - If the population (POP) increases by 1 unit keeping the other factors constant, then the national life expectancy (LEB) will decrease by -0.0079 units. This statement is correct since the significance value (P-value = 0.001 < 0.01) is at a level of statistical significance.

Developed countries can spend more on the health sector, leading to lower mortality rates. During the period 2005 to 2018, the trend shows that an increase in population results in a national life expectancy increase, but not significantly. This is because there have been variations in population sizes and per capita income over the years. Increases in the population in the OECD countries result in income increases. However, these increases are not enough to influence the spending by these countries on the health sector and consequently have an impact on the increase of the value of national life expectancy. The GEE model test shows that in OECD countries, the population has negatively influenced the value of national life expectancy.

Table 4. Descriptive statistics for the variables included in the econometric model

Table 5. Correlation analysis for the variables included in the econometric model

Table 6. Econometric results and empirical findings of the study

Explanation: P-values are shown in parentheses: *** indicates statistical significance at the level of 1%; ** indicates statistical significance at 5% level and * indicates statistical significance 10%.

β4 - If the infant mortality rate (IMR) increases by 1 unit while keeping the other factors constant, then the national life expectancy (LEB) will decrease by -0.1077 units. This statement is correct since the value of the significance (P-value = 0.000 < 0.01) is at the level of statistical significance. We can conclude that an increase in the infant mortality rate has reduced the value of national life expectancy in OECD countries, but fortunately, during the analysed period, we noticed that the infant mortality rate has decreased and consequently increased the value of life expectancy. Even this decline in infant mortality rates in OECD countries over the past decades had declined by an average of 28 deaths per 1000 live births in 1970 to less than 7 in 2002.

β5 - If the potential years of life lost (PYOLL) increases by 1 unit while keeping the other factors constant, then the national life expectancy (LEB) will decrease by - 0.0010 units. This statement is correct since the value of the significance (P-value = 0.000 < 0.01) is at the level of statistical significance. From this, we can conclude that in OECD countries during the analysed period, the potential years of life lost has decreased. Based on our econometric results the declining trend of deaths from cancer in OECD countries for the period under analysis has had a positive impact on increasing life expectancy at birth. Therefore, (β6) if deaths from cancer increase by 1 unit while keeping other factors constant, then national life expectancy (LEB) will decrease by -0.0186 units. This statement is correct since the value of the significance (P-value = 0.000 < 0.01) is at the level of statistical significance.

β7 - If the suicide rate increases by 1 unit while keeping the other factors constant, then the national life expectancy (LEB) will increase by 0.0189 units. This statement is correct since the value of the significance (P-value = 0.000 < 0.01) is at the level of statistical significance.

In general, it is usually concluded that suicide is a factor that affects the decline in life expectancy. However, our results show that with an increased suicide rate life expectancy at birth increased due to the increase in the average age of the death of the victims of suicide. The negative effect of the increase in suicide on national life expectancy was greatly reduced as a result of the increase in the average age of these victims. Thus, with the increase in the average age of suicide victims in OECD countries, national life expectancy has increased. However, without this effect, suicides would have reduced the increase in national life expectancy during the analysed period (2005 to 2018).