A Country-Level Empirical Study on the Fiscal Effect of Elderly Population Health: The Mediating Role of Healthcare Resources

Demographic shifts towards an aging population are becoming a significant fiscal challenge for governments. Previous research has explored the fiscal consequences of the expanding elderly population, but the impact on the elderly’s health quality is less mentioned. The balanced relationship between elderly population health and public finance is a major concern of the global political agenda on the aging society. This article used cross-country panel data from 2000 to 2019 to examine the fiscal effect of the elderly health burden and the mediating role of healthcare resources. The results are demonstrated: The elderly health burden has a negative impact on fiscal balance, especially in aged society and longevity countries. Moreover, the mediating effect of healthcare resources is significant, whereby various forms of healthcare resources such as funds, labor, and facilities all have significant effects. Thus, the conceptual framework of elderly population health, healthcare resources, and public finance is confirmed that the elderly health burden specifically leads to the growing consumption of healthcare resources, which reduces the fiscal balance. It is concluded that reducing the elderly health burden and improving healthcare resource efficiencies are two feasible strategies to enhance fiscal sustainability.


Introduction
Population aging is becoming more widespread and accelerated. The global population aged 60 years and over was only 205 million in 1950, increasing to 810 million in 2012, and it is estimated to reach two billion by 2050. By then, the proportions of people aged 60 and over will be 10% in Africa, 24% in Asia and Oceania, 25% in Latin America and the Caribbean, 27% in North America, and 34% in Europe [1]. On the other hand, the proportion of people aged 65 and over increased from 7% to 14% in France in the space of 116 years, compared to only 68 and 46 years in the United States and the United Kingdom, respectively. Surprisingly, it is expected to take only 26 and 21 years in China and Brazil, respectively [2]. Population aging is often associated with health and medical services and is a major driver of growth in social welfare spending [3]. Due to the natural situations of physiological decline and decreased resistance, the elderly inevitably behaves more demand in healthcare and medical services [4]. As known to all, pool health and disease burden can bring a multitude of unfavorable macroeconomic consequences [5]. Along with the rising population aging and healthcare expenditure, many governments have been under more and more fiscal pressure. Japan, one of the most aging countries in the world, has nearly tripled its share of spending on medical care, nursing care, healthcare, and other social security areas from 17.5% in 1990 to 32.7% in 2014 [6]; a similar situation as in Europe [7]. As can be seen, the health conditions of the increasing elderly population are expected to be a challenge to public finance, and thereby it should be attached importance to figure out the fiscal effects and mechanisms for policy-making. age working population and the old-age retired population, respectively, according to the general model [19,20]. Thus, N t denotes the number of young people surviving in period t, N t−1 denotes the number of old people surviving in period t, and the total population P t is calculated by the following formula: If the population growth rate is n, then the relationship between the number of young people and the number of old people in period t can be written as: Therefore, let α t denote the proportion of the elderly population in period t, which can be expressed as: Meanwhile, Equation (3) can be converted to: A balanced fiscal system is assumed, in which the fiscal revenue generated by the young-age working population exactly offsets the fiscal expenses used by the old-age retired population. With the fiscal revenue of each young person denoted as T and the fiscal expenses of each elderly person denoted as S, the fiscal revenue generated by the young population in initial period 0 fully offsets the fiscal expenses of the elderly population in the previous period, which can be expressed as follows: According to Equation (2), Equation (5) can be transformed into: According to Equation (4), Equation (6) can be transformed into: Thus, α 0 denotes the share of the elderly population in base period 0. Let D denote the fiscal balance. Then, D 0 denotes the fiscal balance in period 0, where its value is zero due to the revenue balancing the expenses. D 1 denotes the fiscal balance in period 1, which can be expressed as the difference between the fiscal revenue generated by the young population in period 1 and the fiscal expenses spent by the elderly population in period 0.
According to Equations (2) and (6), Equation (8) can be expressed as: According to Equation (4), Equation (9) can then be expressed as: Assuming that the health burden per capita of all society, denoted h, is constant, Equation (10) can then be transformed into: In Equation (11), since N 0 , T, α, h > 0, the sign of D 1 depends mainly on the difference between α 0 h and α 1 h. Therefore, α 0 h denotes the elderly health burden in base period 0. When α 0 h < α 1 h, indicating a rise in the elderly health burden, the fiscal balance D 1 is negative, denoting a fiscal deficit; when α 0 h > α 1 h, indicating a decline in the elderly health burden, the fiscal balance D 1 is positive, denoting a fiscal surplus; when α 0 h = α 1 h, the fiscal balance D 1 is zero, indicating that the fiscal balance is maintained. Above all, it can be seen that the elderly health burden is inversely related to the fiscal balance.

Mechanism Analysis
Considering the relationship between the elderly population and public finance, there is a widespread belief that a growing elderly population will drive up the consumption of healthcare resources and cause an ever-widening federal deficit [21]. Health demand is a continuous social demand, which is an important factor in promoting fiscal sustainability and economic development. Healthcare resources are the general term of social resources consumed in the process of providing healthcare services, which is one measure of the capacity of the healthcare system [22]. According to the literature, there are generally the resources of funds, labor, and facility invested in the healthcare domain. First, one of them, the elderly health burden requires spending health funds and thereby reducing fiscal balance. It was found that health care costs were significantly higher in the older age group than in the younger age group from the evidence of Australia [23]. In terms of hospitalization costs, the frequency and cost of hospitalization were significantly higher in the elderly group than in other age groups [24]. There was a positive effect of the elderly population proportion on the fiscal deficit because population aging leads to a dramatic increase in medical costs and puts healthcare insurance funds under great pressure. On the other hand, the human resources for health are central to overcoming health crises in some pool countries and for building sustainable health systems in all world [25]. According to OECD countries, one in every ten jobs was in health or social care in 2019, which have employed the hugest workforce in history [26]. In addition, more health facilities have increased fiscal expenditure and thereby make a reduction of financial balance [27]. In summary, the elderly health burden causes more consumption of healthcare resources, which have an adverse impact on public finances. Therein, various forms of resources such as funds, labor, and facilities all have reduced the fiscal balance. Thus, the theoretical mechanism is characterized in Figure 1. Assuming that the health burden per capita of all society, denoted h, is constant, Equation (10) can then be transformed into: In Equation (11), since N0, T, , h > 0, the sign of D1 depends mainly on the difference between 0 ℎ and 1 ℎ. Therefore, α 0 h denotes the elderly health burden in base period 0. When 0 ℎ < 1 ℎ, indicating a rise in the elderly health burden, the fiscal balance D1 is negative, denoting a fiscal deficit; when 0 ℎ > 1 ℎ, indicating a decline in the elderly health burden, the fiscal balance D1 is positive, denoting a fiscal surplus; when 0 ℎ = 1 ℎ, the fiscal balance D1 is zero, indicating that the fiscal balance is maintained. Above all, it can be seen that the elderly health burden is inversely related to the fiscal balance.

Mechanism Analysis
Considering the relationship between the elderly population and public finance, there is a widespread belief that a growing elderly population will drive up the consumption of healthcare resources and cause an ever-widening federal deficit [21]. Health demand is a continuous social demand, which is an important factor in promoting fiscal sustainability and economic development. Healthcare resources are the general term of social resources consumed in the process of providing healthcare services, which is one measure of the capacity of the healthcare system [22]. According to the literature, there are generally the resources of funds, labor, and facility invested in the healthcare domain. First, one of them, the elderly health burden requires spending health funds and thereby reducing fiscal balance. It was found that health care costs were significantly higher in the older age group than in the younger age group from the evidence of Australia [23]. In terms of hospitalization costs, the frequency and cost of hospitalization were significantly higher in the elderly group than in other age groups [24]. There was a positive effect of the elderly population proportion on the fiscal deficit because population aging leads to a dramatic increase in medical costs and puts healthcare insurance funds under great pressure. On the other hand, the human resources for health are central to overcoming health crises in some pool countries and for building sustainable health systems in all world [25]. According to OECD countries, one in every ten jobs was in health or social care in 2019, which have employed the hugest workforce in history [26]. In addition, more health facilities have increased fiscal expenditure and thereby make a reduction of financial balance [27]. In summary, the elderly health burden causes more consumption of healthcare resources, which have an adverse impact on public finances. Therein, various forms of resources such as funds, labor, and facilities all have reduced the fiscal balance. Thus, the theoretical mechanism is characterized in Figure 1.

Empirical Strategies
First, the baseline model in Equation (12) was established on the basis of the analysis above, where Balance represents the fiscal balance, Burden represents the elderly health burden, and Z represents a set of control variables that affect the fiscal balance. Meanwhile, γ and δ are the coefficient terms of the regression variables, respectively, α is the intercept term, and ε represents the residual term.
Coefficient estimation may be biased due to the fact that the heterogeneity of the sample and the disturbance brought about by the time series are not taken into account in the baseline model; hence, two-way fixed effects regression (2FE) was further constructed in Equation (13), where u and y represent the geographic region and the time of year in which the country is located, respectively, and the subscripts i and t denote the specific geographic region and year.
Lastly, to test the mechanism of the elderly burden and the fiscal balance, the mediating effect models in Equations (14) and (15) were constructed, where Mediation denotes the mediating variable, and the remaining variables have the same meaning as above. Referring to Baron and Kenny's method of mediating effects analysis [28], if the coefficients of the core explanatory variables in the main regression γ 0 are significant, then the coefficients are estimated from the mediating effects of γ 1 and θ, where γ 1 represents the effect of the elderly health burden on the mediating variable, and θ then the effect of the mediating variable on the fiscal balance. If both coefficients are significant, this indicates that there is a mediating effect, whereby the elderly health burden can affect the fiscal balance through the mediating variable; if at least one of them is not significant, Sobel testing is needed to determine whether a mediating effect exists.

Variable Descriptions
The dependent variable was Fiscal Balance, which was measured as a function of the ratio of the net operating balance to GDP (Balance). The net operating balance was calculated simply by the difference between public finance revenue and expense. A positive value of this indicator denotes a fiscal surplus. Conversely, a negative value of this indicator denotes a fiscal deficit. In addition, the ratio of net borrowing and lending to GDP (Finance) was used as an alternative indicator for robustness testing. The data of the net operating balance and net lending and borrowing were obtained from the Government Finance Statistics (GFS) published by the International Monetary Fund.
The independent variable was elderly health burden. There are many indicators to measure population aging, including the proportion of the elderly population, mortality rate, life expectancy, and life/course ratio [29]. In this paper, we adopted disabilityadjusted life years (DALYs) as the measurable indicator of the health burden [30]. Thus, the elderly health burden was measured as a function of the disease burden rate of the elderly population (Burden), which is the DALY ratio of the population aged 65 and above to that of the total population. According to the World Health Organization (WHO), DALYs are the total number of healthy life years lost from onset to death, including both years of life lost due to premature death and years of healthy life lost due to disease-induced disability. In addition, the old-age dependency ratio (Aging), which is the ratio of the population aged 65 and over to the working-age population, was used as an alternative for robustness testing.
The data required for the calculations were taken from the Global Health Data Exchange (GHDx) database.
Healthcare resources represented the mediating variable in this study, and indicators were set to measure this from the perspective of resources such as healthcare expenditure [31][32][33], human resources [34], and hospital beds [35], using three indicators, namely, healthcare expenditure rate (THE), physician ratio (PHY), and hospital bed ratio (BED), to represent the financial resources, labor resources, and facility resources invested in the healthcare domain, respectively, as a comprehensive measure of healthcare investment. The data of the mediating variable indicators were obtained from the World Health Statistics (WHS) database established by the World Health Organization.
With reference to control variables of previous studies, the factors affecting the fiscal balance were controlled from multiple perspectives, such as social, economic, and environmental [36][37][38]: (1) Urbanization (URB), measured by the ratio of the urban population to the total population of the country; (2) openness (TRD), measured by the ratio of the total import and export trade to the country's GDP; (3) foreign direct investment (FDI), measured by the ratio of net foreign direct investment to the country's GDP; (4) forestation (FOR), measured as the ratio of forest area to the country's total land area; (5) growth (PRG), measured by the annual growth ratio of GDP per capita. The data for the control variables were obtained from the World Development Indicators (WDI) database published by the World Bank.
In all, Table 1 reports the selection and description of variables, indicators, and data sources.

Data Explanation
This study selected panel data of 45 countries for the period of 2000-2019 as the sample for analysis; in the process of data collation, the data completeness of the indicators for the explanatory and core explanatory variables was the priority criterion, and those samples with abnormal indicators and serious missing data were excluded. Then, interpolation was used to fill in the sporadic missing data for individual years. The statistical characteristics of the main variables are reported in Table 2.

Basic Regression
During the basic regression analysis, changes in the relationship between the elderly health burden and the fiscal balance were observed by gradually adding control variables, and the results are reported in Table 3. Column 1 shows that the estimated coefficient of the elderly health burden without the inclusion of any control variables was −0.065, which had statistical significance at the 1% level. The results from Columns 2 to 6 show that the estimated coefficients of the elderly health burden ranging from −0.128 to −0.114 did not change much when control variables were gradually added, all of which were significantly negative at the 1% statistical level. It is indicated that the elderly health burden had a significant dampening effect on the fiscal balance and that countries or regions with a higher elderly health burden had a weaker fiscal balance. Considering the control variables, the results from Column 6 show that the estimated coefficients of all control variables were significant at the 1% or 5% statistical level. Among them, urbanization, openness, forestation, and GDP per capita growth had a significantly positive effect on the fiscal balance, indicating that a higher urbanization rate, more open trade, a greener environment, and faster GDP per capita growth lead to a better impact on public finance. In contrast, the effect of foreign direct investment on the fiscal balance was significantly negative at the 5% statistical level, indicating that a higher share of foreign investment leads to a lower fiscal balance. In addition, after controlling for the time and geographic effects of the sample countries, the effect of the elderly health burden on the fiscal balance remained significantly negative.

Endogeneity Concerns
To control for endogeneity issues of the model, this paper used lagged term as the instrumental variable for further dynamic regression analysis by system generalized method of moments (GMM) [39]. Table 4 shows the regression results of the system GMM estimation. Column 1 shows that the estimated coefficient of the first-order lagged term of fiscal balance is significantly positive at the 1% significant level, indicating that the current period of fiscal balance is positively correlated with the previous period. Moreover, the regression coefficient of the elderly health burden was significant at the 5% level, which indicated a negative-driving effect on fiscal balance. In addition, to avoid measurement error, we replaced the variable measurements to re-run the regressions by taking the ratio of net borrowing and lending to GDP (Finance) and the old-age dependency ratio (Aging) as an alternative. As the results are shown in Columns 2 to 4, the core explanatory variable remained significant except in Column 4. Finally, according to model specification tests, it is also demonstrated that moment conditions employed in the system GMM are appropriate-errors are uncorrelated of order 2 (AB test p-value > 0.1), and there are no overidentifying restrictions (Hansen test p-value > 0.1), thus indicating that the selection of instrumental variables and the setting of the model are reasonable.  Note: *** p < 0.01, ** p < 0.05, and * p < 0.1; standard errors in parentheses.

Robustness Tests
The results of various robustness tests are reported in Table 5. First, given the cyclical nature of the fiscal budget and the possibility of inter-year effects on the fiscal balance, it was necessary to examine whether the current period's level of the elderly health burden had an effect on the next period's fiscal balance (Balance t+1 ). The results of Column 1 show that the regression coefficient of the elderly health burden was significantly negative at the 1% statistical level, consistent with the conclusion above.
Second, we replaced the OLS regression model in the basic regression with a logistic regression model, redefining the explained variable fiscal balance as a dummy variable (Balance_dummy), whereby a value of one was assigned for a fiscal surplus, and the value of zero was assigned for a fiscal deficit. According to the logistic regression results in Column 2, the findings did not change substantially from the conclusions above.
Third, selecting the restricted samples. In order to exclude the influence of sample outliers on the research findings, we conducted robustness tests by adjusting the sample groupings. On the one hand, the public finance situation was not only affected by the domestic economy but was subjected to shocks from global macroeconomic events. Thus, we retested the regression results after excluding the sample data during the financial crisis of 2008-2010, as shown in Column 3. On the other hand, sample data with too extreme fiscal surplus or deficit rates were excluded. The regression results are reported in Column 4 using the sample data with a fiscal balance rate in the range of ±15%. The results show that the regression coefficients of the elderly health burden were both significantly negative at the 1% statistical level, indicating that an increase in elderly health burden still significantly reduces the fiscal balance, posing a negative impact. The results remained robust. Note: *** p < 0.01, ** p < 0.05, and * p < 0.1; standard errors in parentheses.

Heterogeneity Analysis
Considering that country heterogeneity possibly affects the basic result, we have performed a heterogeneity analysis from grouping by OECD member, aged society, and life expectancy. The results are reported in Table 6. First, from the subgroup results of whether OECD members in Columns 1 and 2, we can see that the elderly health burden has a significantly negative effect on fiscal balance at the 1% level, in both groups. Second, according to the United Nations, an "aged society" is defined as a country whose share of people aged 65 years or more has reached 14% of the total population. Thus, we classified the countries with the elderly population proportion less than 14% as being in the Pre-aged group, and more than 14%, the Aged group [40]. From the results presented in Columns 3 and 4, it is clear that the elderly health burden has a significant negative fiscal effect at the 1% level in aged society countries, but not in pre-aged society. Third, according to the rank of the sample countries by the mean value of life expectancy from 2000 to 2019, the median is 79.41 years old. Countries with an average life expectancy below 79.41 years old are classified as being in the Normal group, and countries with an average life expectancy above 79.41 years old are classified as being in the Longevity group. The results presented in Columns 5 and 6 show that the elderly health burden has negatively affected the fiscal balance in the Longevity group at the 1% level, but it is not significant in the Normal group. Overall, the reason for the significant fiscal impact of the elderly health burden in aged society and longevity countries may be due to the deep population aging and high life expectancy, which means a larger proportion of the elderly population over 75 or 80 years old, who are likely to incur greater costs in dying [41]. This is because the two years before dying is a period of high healthcare costs for the elderly [42,43], with the greater probability of higher spending on healthcare as death draws nearer [44].

Mediating Effect
Further, to recognize the mechanism of healthcare resources, the mediating effect was further analyzed. According to the research design, healthcare resources were measured in various forms, including funds, labor, and facilities. They were tested one-by-one in accordance with the model and analysis steps of the mediating effects to verify the existence of the three transmission mechanisms, and the results are reported in Table 7. First, in terms of healthcare funds, the results in Columns 1 and 2 show that the regression coefficients of the elderly health burden on healthcare funds, and of healthcare funds on the fiscal balance, were statistically significant at the 1% level. Thus, it is straightforward to conclude that a mediating effect exists without the need for Sobel testing. Column 1 shows that the effect of the elderly health burden on health funds was significantly positive, while Column 2 shows that the effect of healthcare funds on the fiscal balance was significantly negative, indicating that a rise in the elderly health burden increases the health expenditure and, thus, reduces the fiscal balance.
Second, in terms of healthcare labor, the estimates in Columns 3 and 4 show that the two-stage regression coefficients of the elderly health burden on healthcare labor and of healthcare labor on the fiscal balance were statistically significant at the 1% level. Thus, a mediating effect existed. Column 3 reveals a significant positive effect of the elderly health burden on healthcare labor, while Column 4 reveals a significant negative effect of healthcare labor on the fiscal balance, demonstrating that an increase in the elderly health burden necessitates increased healthcare labor, thus reducing the fiscal balance. Third, in terms of healthcare facilities, the regression coefficients of the elderly health burden on healthcare facilities and of healthcare facilities on the fiscal balance in Columns 5 and 6 were statistically significant at the 1% level; hence, a mediating effect existed. Column 5 shows a significant positive effect of the elderly health burden on healthcare facilities, while Column 6 shows a significant negative effect of healthcare facilities on the fiscal balance, indicating that an increase in the elderly health burden necessitates increased healthcare facilities for support, which decreases the fiscal balance.

Discussions
This paper empirically investigated the fiscal effect of elderly population health and its mediation mechanism through cross-country panel data, and the main findings were as follows: (1) The regression results of the fixed effects model showed that the impact of the elderly health burden on the fiscal balance was significantly negative, suggesting that an increase in the elderly health burden adversely affects the fiscal balance; (2) the heterogeneity effects have been identified in different countries, the elderly health burden performed a significantly negative influence on the fiscal balance in aged society and longevity countries; and (3) the results of the mediation effect tests showed that a rising elderly health burden leads to an increase in healthcare resources, which in turn has a negative impact on the fiscal balance.
Taken together, such results keep consistent with most of the previous studies [45,46]. It is explainable that when the disease burden of the elderly population in society rises, the social demand for healthcare increases. Healthcare is the main aspect of public services, which is related to the livelihood of the country; thus, the government has a responsibility to provide basic healthcare services. Accordingly, the budget structure is adjusted to increase healthcare resources, forming a crowding-out effect [47]. This can also be analyzed from the perspective of supply and demand, whereby an increased elderly health burden leads to a rise in a society's healthcare on the demand side; hence, the government has to increase the fiscal investment to meet the growing health demand on the supply side [48]. This analysis confirms that an increase in the elderly health burden leads to an increase in healthcare resource consumption, which in turn leads to a decline in the fiscal balance.
The limitation of this study is mainly that it lacks a detailed analysis of healthcare resources. Actually, healthcare resources encompass a wide range of components, and in this paper, only three terms of healthcare resources were discussed, which may not be fully representative, so the final research conclusions have less completeness and guidance for fiscal policies on population aging. Future research is expected to cover this limitation and strengthen the conclusions of this study.

Conclusions
International experience shows that rising healthcare expenditure can be a dangerous shock to fiscal sustainability, whereby higher welfare is often associated with higher debt [49]. Previous studies have focused on the impact of the elderly population quantity on public finance, but the elderly health quality is rarely considered. This study, from a new perspective of disease burden, has demonstrated that the elderly health burden, which could be a new indicator of population aging, has a negative effect on fiscal balance, and the mediation mechanism of healthcare resources on the fiscal effect of elderly health burden has been confirmed. Therefore, it is beneficial for governmental fiscal balance to reduce the disease burden of the elderly population and improve the utilization efficiency of healthcare resources.
Based on the above findings, some policy recommendations could be considered. The first one is to consider "aging and health" as the top priority in fiscal policies. It is necessary to encourage the innovation of health financial products such as pension funds, chronic disease medical funds, and long-term care insurance to cover health shocks and risks. The second is to improve the efficiency of healthcare resources to respond to population aging. "Value-based healthcare" [50] is supposed to be effective for elderly healthcare to optimize higher health outcomes and lower health costs. The third is to promote to elderly people a sense of "active health" in order to keep healthy and reduce disease burden. Health literacy for the elderly should be improved in order to inform them on how to conduct health management correctly and actively.  Data Availability Statement: Publicly available datasets were analyzed in this study and are attainable from the corresponding author upon reasonable request.