Introduction

The last decades have seen an increase on income inequality. With the actual income distribution, the richest top 1% of the population accumulates more of the 50% of global income, with a corresponding Gini index of 0.85.¹ Where the Gini index values are inside the range of cero to one. A Gini index equal to zero implies that everyone has the same income. A Gini index equal to one implies that only one person on the world holds all income. During this time, health systems in North America have experienced an increasing income inequality, alongside with increasing private health expenditures. Most of the private health expenditures are conformed by out of pocket expenses. These conditions hinder health access for income and health vulnerable groups. In what follows Tables 1 and 2 display some of these dynamics for the period under study.

Table 1 shows that health expenditure per-capita in The United States is 1.87 times larger than in Canada, and 15.39 times larger than in Mexico. The government and private health expenditure decomposition exhibits that private health expenditure is larger in The United States and México, tan the government health expenditure. In Canada government expenditure is larger than the private health expenditure. The country that expends the most in health is the United States with 15.45 of its total gross domestic product, it is followed by Canada with 9.92% and Mexico with 5.55%. The health expenditure per-capita in decreasing order are The United States (7,730 US$), Canada (4,133 US$), and Mexico (502 US$).

Table 2 baseline is the % total for North America. Therefore, all entries are equal to 100. The government percentage on total health expenditure are above the 50% for North America region, for the five years under observation. A similar pattern is shared by Mexico and The United States. In contrast, Canada exhibits a government health expenditures percentage of about 70%. Thus, the figures for Mexico and The United States are inferior to Canada.

It is important to mention that Canada in the two decades under analysis, the government health expenditure has been reduced from 75% to 69%. Perhaps this trend speaks of a reduction in the participation of the Canadian public health expenditure, and at the same time it has been replace with an increasing participation from the private sector in this type of expense. In contrast, Mexico and the United States has been increasing the government health expenditure in the two undergoing decades. The smallest private percentage on total health expenditure is for Canada with figures close to 30%, although increasing from 25% in 2000 to 31% in 2019. Mexico and the United States exhibits figures around the 50% levels for the five years under consideration, albeit a decreasing participation of private percentage in total health expenditure.

There are not data for the pandemic years to date (April, 2023) in the World Development Indicators, World Bank data base. Perhaps new dynamics would be established with the massive government financial interventions in health sectors, as well as in research and development of vaccines, treatments, and cash stimulus. According with some references the pandemic outbreak caused by the SARS-CoV-2 has placed a setback on the health systems⁹.SARS-CoV-2 has declined global life expectancy and had placed setbacks in áreas including childhood immunizations and infectious diseases treatments such as malaria and tuberculosis¹⁰ . This author mentions some mitigating measures from The United States to palliate SARS-CoV-2 effects: The United States global health funding has increased since the onset of the pandemic. Through regular appropriations, but mostly through emergency supplemental funding. Most of this funding has been specifically used to support global access to the SARSCoV-2 vaccines, testing, and therapeutics. Continuing With this author, central to this funding strategy is the goal of strengthening health systems to achieve multiple objectives across global health programs. He says that such a system requires a robust primary care infrastructure buttressed with a cadre of community health workers and supported by universal health coverage: crosscutting activities such as surveillance, diagnostic laboratory capacity, and access to clean water, sanitation, and hygiene, particularly in health care facilities. Optimal health access depends on addressing the prevention and treatment of a wide spectrum of illnesses². This view is wider than the previous ones, as it considers health as a holistic process.

One of the immediate impacts of the SARS-CoV-2 is that global life expectancy has been declined by around two years¹¹. Other impact of the SARS-CoV-2 is the reduction of breast diagnostic services and breast cancer treatments, which would undoubtedly increase woman mortality rate¹². Life expectancy in The United States declined nearly a year from 2020 to 2021, the lowest level since 1996¹⁰. Also, identified substantial geographical and socioeconomic inequalities in cervical cancer globally, with a clear gradient of increasing rates for countries with lower levels of human development¹³. The council on foreign relations released a seminal study on the emerging crisis of noncommunicable diseases, and recommended The United States leadership in several areas, including primary and secondary prevention of cardiovascular disease, tobacco control, human papillomavirus vaccination, and cervical cancer screening¹⁰. Other priority areas include nutrition, physical activity, obesity, and integration of mental health into primary care¹⁰. Even though some of these areas are addressed in current United States programmatic efforts, non-communicable diseases fail to obtain the level of attention that they deserve to support health gains in low-income countries.

It is also important to consider that health access depends on organization and financial support from the health system. Therefore, government health expenditure is key to provide universal access. In this way those that cannot afford health services with out of pocket expenditures can gain access to the health system and improve their health status.

Is worth mentioning that health is a multidimensional phenomenon. It is affected for environmental and economic factors. However, not all these factors have a proper record. Thus, several studies that analyze health status simplify its factors list to only economic factors^14,15,16 . Other studies have suggested a relationship between an increasing income inequality with health indicators poor performance, like a low life expectancy and an increasing child mortality rate¹⁷.

It is important to mention that health markets exhibit failures like asymmetric information and externalities. These markets are characterized by natural monopolies (public goods), and intellectual property rights monopolies (patents). With market failures out of pocket health expenditures provides health access for those with sufficient income to pay monopoly prices. If health access is left to market incentives, income and health vulnerable groups would not have a proper medical attention. This could imply social inequalities in individual development. To avoid this type of social inequalities, government intervention is necessary to correct health market failures and provide universal health access.

One way in which government can intervene in health markets is through public health expenditure. Public health expenditure has been growing in the last decades given two new dynamics. One, an increase in life expectancy that overwhelms pension systems. Two, new technologies, drugs and treatments increasing costs^3,18. During the pandemic SARS-CoV-2 years the cost of financing research and vaccinates has been vast. Research and development from the 15 largest pharmaceutical companies record about $138 billion of US dollars in 2022. This last figure represents a 43% increase with respect to 2017¹⁹.

Several studies have pointed out that universal health Access is desirable for vulnerable groups for reasons of income or specific health conditions⁵. When out of pocket expenses are not enough to secure a good standard in health status for the vulnerable groups, public health expenditures should be ideally a complement to them. This public policy view made necessary an assessment of the public health expenditures and health status related with efficiency and income inequality regarding health market failures alleviation.

The objective of this paper is assessed the nexus of health status, economic growth and the Gini index in North America and its countries using a panel model. Health status is gauged by life expectancy and public health system by government health expenditure. Economic growth is gross domestic product percentage change. Income inequality is represented by the Gini Index. The gross domestic product percentage change would allow the obtention of an economic growth measure. The Gini index is a measure of income inequality. The North America region is composed by three countries: Canada, Mexico and The United States. A panel model is implemented to distinguished individual fixed effects and individual random effects in this region. The period under analysis is 2000-2019 with annual data. Recent years are not available in the Work Development Indicators data base (Retrieving date April, 2023) to show what had happened in health systems regarding the SARS-COVID-19 outbreak. It is important to know how the health status relates with key economic indicators to help assessing the effectiveness of the health system and try to direct resources thorough a suitable policy where they are needed the most.

Materials and method

The summary statistics of each variable used in the panel model are reported on Table 3. These variables are life expectancy at birth total (years), domestic general government health expenditure (% of gdp), Gini index, gross domestic product (gdp constant 2015 US$), for each country that conforms the North America geographical area: Canada, México and The United States. The statistics reported are the mean, standard deviation, coefficient of variation and kurtosis.

The mean life expectancy at birth for males and females (total) measured in years is larger in Canada with 80.90, than in Mexico (74.28) and The United States (78.06). The coefficient of variation (the ratio between standard deviation and the mean) for life expectancy is bigger for Canada and The United States (0.01, respectively), than in Mexico 0.00. These figures implies that there is more data variability around the mean for Canada and The United States, tan for Mexico. The kurtosis of 3.34 indicates that the data have more outliers in Mexico. The United States and Canada kurtosis are 1.95 and 1.70, respectively. Thus, the country with less data outliers is Canada. So, Canada is the country with the biggest life expectancy, less outliers, and greater variation of the data around the mean in North America.

The domestic general government health expenditure as percentage of the gross domestic product is larger in The United States with 7.45%. It is followed by Canada (7.03%) and Mexico (2.65%). The largest data variability around the mean is also for The United States, with a coefficient of variation of 0.13, follow by Mexico (0.12) and Canada (0.08). Mexico exhibits a kurtosis of 2.39 meaning it has the biggest outliers, it is followed by The United States (1.88) and Canada (1.59). So, The United States is the country with the largest health expenditure as percentage of the gross domestic product and coefficient of variation.

The average of the Gini index is bigger in Mexico (49.06), than in The United States (40.86) and Canada (33.49) for the period 2000-2019. The biggest dispersion of the data measured by the standard deviation is for Mexico (1.67) followed by The United States (0.44) and Canada (0.29). The largest dispersion of the data around the mean is for Mexico (0.03) followed by The United States (0.01) and Canada (0.01). The largest outliers in the data are for Canada (3.70) followed by Mexico (2.95) and The United States (2.25).

The largest gross domestic product at constant 2015 United States trillions of dollars on average for the entire period of 2000-2019 is for The United States (16.59), followed by Canada (1.42), and Mexico (1.04). Similar pattern is followed by the standard deviation, the biggest one is for The United States (1.82), next Canada (0.16) and last Mexico (0.13). The coefficient of variation indicates that Mexico has the biggest dispersion of the data around the mean (0.13) followed by The United States and Canada, both with a value of 0.11. Regarding the kurtosis, the biggest outliers are in The United States (2.12), followed by Canada (2.01), and finally Mexico with (1.79).

During the period under analysis there are 19 years of data where The United States have a higher health care expenditure. Life expectancy in The United States is lower than in Canada, who in turn spends less in health care. In 2019 life expectancy in Canada is 82 years on average for women and men, while for the same year life expectancy in The United States is 79 years total. The lowest life expectancy is exhibited for Mexico with an average of 74 years for 2019.

The highest health expenditure is exhibited by The UnitedStates for the period 2005-2019. In 2019 health expenditureis 8.5% of The United States gdp. Canada has a healthexpenditure on 2019 of 7.5% of its gdp. The country thatexpends less in health is Mexico with a 2.8% of its gdp forthe 2019 year.

The United States spent 17.8% of its gross domestic producto on health care in 2016¹⁴. The data analyzed here points out that for 2016 the corresponding figure for government health expenditure (private health expenditure is excluded) is 8.5%. The United States total health expenditure as percentage of the gross domestic product is 15.45% on average for the period 2000-2019. The figure provided probably considers the total health expenditure as the sum of private and public health expenditures. Prices of health care labor and goods, including pharmaceutical and administrative costs, appeared to be the major drivers of the difference in overall spending between The United States and other high-income countries^14,21.

“… The United States spends more on health care than any of other OECD countries spend, without providing more services than the other countries do^”

³. The United States provides health insurance from 20 million to an additional 25 million people ²⁰. This increase on the number of people with health insurance might imply that health care in The United States is becoming more equitable.

The biggest income inequality is experienced by Mexico, with an average of 49.06 of the Gini index for the whole period under analysis. There is a tendency to diminish income inequality in this country: in 2000 it has a value of 53 (in a scale of 0 to 100) and in 2019 it was 47. So, it has decreased by about five percentage points. The next North America country that exhibits a higher Gini index is The United States, although its variation (0.01 coefficient of variation) during the two decades under analysis is nil. The country with less income inequality is Canada with an average of 33.09 for allthe 2000-2019 period, also exhibiting small variation duringthese nineteen years (0.01 coefficient of variation). TheUnited States gdp is the highest in the region (16.59 trillionsof dollars). On average for the period under analysis is 12times bigger than Canadá gdp (1.42 trillions of dollars), and16 times bigger than México gdp (1.04 trillions of dollars).

A panel model is proposed to assess the nexus between health, economic growth, and the Gini index in North America. Canada, Mexico and The United States conform this geographical área and the corresponding panel model. The analyzed period is 2000 to 2019 with annual data. This period is chosen given the most recent available data in North America. The panel modeladvantages consist of studying individual fixed effects andindividual random effects for each country as well as for theentire North America region. At the same time, statistics areobtained for individuals and the overall regression, conveyingmore information. The individual fixed effects are countrydummies for the intercept, and the individual random effectsare country dummies for the slopes. It is important to mention,that the econometric results of this panel model allow makingstatistic inference, and therefore provides a sound ground forpolicy making.

In what follows the panel model is represented by the following two equations:

logleit= β0it+ β1it loggdpit+ β2it logginiit+ u1it (1)

logheit= β0it+ β1it loggdpit+ β2it logginiit+ u2it (2)

where le _it represents life expectancy at birth total in years, gdp _it stands for the gross domestic product in constant 2015 United States dollars, stands for the Gini index, heit is the domestic general government health expenditure as percentage of the gross domestic product, where represents individual countries (i=1 for Canada, i=2 for México, i=3 for United States), at time t t=200,…,2019 β0it is the intercept estimator for individual, i and time t, β0 is the intercept for North America, β1it is the slope estimator for individual i and time t with respect to gdpit, β1 is the slope for North America with respect to gdp, β2it is the slope estimator for individual i and time 𝑡 with respect to giniit, β2 is the slope for North America with respect to gini, 𝑢 1𝑖𝑡 is the error term for individual 𝑖 and time t of equation (1), u2it is the error term for individual 𝑖 and time 𝑡 of equation (2) log expresses the logarithm operator. The error terms are assumed to be niid (normally independent and identically distributed) ~N (0,I1).

Without effects hypothesis:

H0:β0=0 and β1=0 and β2 =0

H1: β0≠0 and β1≠0 and β2≠0

Where H0  is the null hypothesis. H1  is the alternative hypothesis. The alternative hypothesis assumes that the intercepts and slope are different from zero. The null hypothesis implies that there are no differential effects in the intercept and the two slopes in equations (1) and (2) in North America. The country and time dimensions do not change in this case.

Individual fixed effects hypothesis:

H0: β0i=0

H1:β0i≠0

The null hypothesis assumes that there is no change in the intercept at a country level. Here the individuals are the countries. The alternative hypothesis assumes that the intercept is different for each country. The time dimension does not change between countries, but its variation is considered within each country. Therefore, the subscript does not appear in the null and alternative hypothesis.

Individual random effects hypothesis:

H0:β1i=0 and β2i=0

H1: β1i ≠0 and β2i ≠0

The null hypothesis assumes that there is no change in the slopes at a country level. The alternative hypothesis assumes that the slopes for each independent variable (gdp, and gini respectively) are different for each country. The time dimension does not change in this case, but is variation is considered between each country. Therefore, the subscript does not appear in the null and alternative hypothesis.

Results

The unit root test is important because it assures that the estimated regressions will be not spurious. According with the Levin-Lin-Chu unit root test results in Table 2 all panel variables tested are stationary. This is because the null hypothesis that panels contain unit roots are rejected in three or more options of the Levin-Lin-Chu unit-root test. Thismeans that the autoregressive polynomial of the characteristicfunction roots of each variable tested stay inside the unitcircle. Therefore, these roots do not scape to infinity, whichcould happen with unstable and volatile panel variables. Unitroots inside the unit circle assures parsimony in the economicbehavior of the analyzed panel variables, and therefore isstationarity, which is a desired property to assure sampleestimators convergence with the true population values.

Equation (1) is estimated with three variations in effects in the panel model: without effects, individual fixed effects, and individual random effects. The corresponding estimators are reported next in Table 3. The first column reports without effects, the second column reports individual fixed effects and the third column individual random effects. The estimation method is Ordinary Least Squares (OLS).

From Table 4 the gini estimate is negative in the first columna reported. This negative sign implies that for one percent that gini increases the life expectancy will be reduced in 0.21% of a year for all North America countries. The estimate in column one (-0.21) is decomposed in the within estimator (-0.14) column two, and between estimates (-0.20, 0.12 and -0.31) column three for each individual country. In columna three the between estimator for Canada indicates that for one percent that gini increases the life expectancy will be reduced by 0.20%. For Mexico for one percent that gini increases life expectancy will be increased by 0.12%. The estimator for Mexico exhibits a positive sign perhaps for the time/spatialcorrelation that this country has with Canada and The UnitedStates. Further investigation to separate this correlation canbe done with an error component model. However, this investigation escapes from the scope of this study and may bean interesting subject for future research.

For The United States for one percent that gini increases the life expectancy would be reduced by 0.31%. Some individuals in The United States get excellent care, but not everyone, and that even though The United States is a great country it tolerates profound disparities in health²⁰. Such health disparities seem to be a great social injustice³.

“… the state if health in the country all the more concerning, at least in relative terms. The U.S.’s life expectancy at birth -77- is close to three years below the OECD average, and is marred by ever-present racial and ethnic disparities”

^22.

In column two the constant is investigated for each country with the aid of individual fixed effects. The time dimensión does not change between countries, but its variation is considered within each country. For the case of Canada, the initial average value of life expectancy with respect to time is 1.99 for the period 2000-2019, the highest in North America. It is followed by Mexico with a figure of 1.87, and later by The United States with 1.76. These results implies that Canada have better life expectancy even before the period of 20002019.

Regarding the gdp estimate, it is almost nil in the first columna reported (0.00). This estimate implies that gdp changes does not influence life expectancy in North America. However, to investigate further the effects of gdp per country, the estimate in column one (0.00) is decomposed in the within estimator (0.06) column two, and between estimators (0.09, 0.04 and 0.09) column three. In column three the between estimate for Canada indicates that for one percent that gdp increases the life expectancy will be increased by 0.09%. For Mexico for one percent that increases life expectancy will be increased by 0.04%. For The United States for one percent that gdp increases the life expectancy will be increased by 0.09%. So, the countries with have a better advantage from an economic growth in increasing life expectancy are Canada and The United States. Without a panel model, the impact of the gross domestic product over life expectancy would be nil. However, the panel model investigates the effects in each country, which makes a richer analysis and allows an idiosyncrasy study for each North America country.

From Table 5 it can be read that the gini estimator are negativein the first column reported (-2.37). This negative signimplies that for one percent that gini increases the domesticgeneral government health expenditure will be reduced in-2.37% on average per year in North America. The estimate in column one (-2.37) is decomposed in the within estimate(0.01) column two, and between estimates (0.78, -0.07 and-5.61) column three. In column three the between estimatefor Canada indicates that for one percent that gini increasesthe domestic general government health expenditure will be increased by 0.78%. This estimator perhaps is contaminatedwith a spatial correlation with the other two countries, wherean error component model could dilucidated this. In the case of Mexico for one percent that gini increases the domestic general government health expenditure will be reduced by 0.07%. For The United States for one percent that Gini increases the domestic general government health expenditure will be reduced by 5.61%. Perhaps, the greatest elasticity of The United States of gini with respect to health expenditure responds to an inarticulate response from health policy makers to an increase in income inequality.

“However, despite its higher spending, The United States performs poorly in áreas such as health care coverage and health outcomes”

^14,18,23,24. A proposal is made by these authors to revert these health outcomes:

“Efforts targeting utilization alone are unlikely to reduce the growth in health care spending in The United States; a more concerted effort to reduce prices and administrative costs is likely needed”

¹⁴.

In column two the constant is investigated for each country with the aid of individual fixed effects. The time dimension does not change between countries, but its variation is considered within each country. For the case of Canada, the initial average value with respect to time is 1.94 of the domestic general government health expenditure. It is followed by Mexico with a figure of 0.96, and later is The United States with 1.99, the highest in North America. These results implies that The United States have the highest domestic general government health expenditure even before the period of 2000-2019 has started.

Regarding the gini estimate it has a coefficient of 0.21 in column one. For each one percent that gini increases, the domestic general government health expenditure would increase in 0.21% for North America on average for the period under study. However, to investigate further the effects of Gini per country the estimate in column one (0.21) is decomposed in the within estimate (0.76) column two, and between estimates (0.63, 0.71 and 1.36) column three. In column three the between estimate for Canada indicates that for one percent that gini increases the domestic general government health expenditure will be increased by 0.63%. For Mexico for one percent that increases the domestic general government health expenditure would be increased by 0.71%. For The United States for one percent that gini increases the domestic general government health expenditure would be increased by 1.36%. So, the country with the biggest elasticity towards public health expenditure is The United States. The biggest expenditure in the world among the richest countries is made by The United States ^3,20. These authors explanations rely on The United States health market structure and income levels. In descending order, it is followed by Mexico (0.71) and Canada (0.63).

Discussion

Given the descriptive statistics and the panel model econometric results, health status is negatively influenced by an increase in income inequality in North America. The results suggest that an increase in economic growth (gross domestic product percentage change) would ameliorate life expectancy and government health expenditure. Therefore, health status is negatively influenced by an increase in income inequality and a reduction on economic growth. Other empirical análisis using time series and cross-section data had found similar results as in this document about the dynamics of health and economic indicators ^{25,26,27,28,29}. A panel analysis finds that financial inclusion significantly reduces income inequality in developing countries³⁰. Also, in an empirical panel analysis it is found that poverty levels are reduced through economic growth in the emerging world, and it recommends policies working in improving the distribution of income where governments could improve the health of the income and health vulnerable groups with the increase of economic growth, and thereby increasing life expectancy³¹ .

The country that expends the most in health care is The United States with the biggest elasticity, follow by Canada and Mexico. The country that reduces more life expectancy by one percent of gini increase is The United States, followed byCanada and Mexico. These results match some authors insides, since it is known that The United States has the biggest healthexpenditures in OECD countries and at the same time it hasprofound health disparities and social injustices^3,20.

Conclusions

All variables that conform the panel model are stationaryaccording with the unit root tests. The estimates reported inTables 4 and 5 suggest that an increase in income inequalityreduces life expectancy measured in years. At the same time, an increase in income inequality reduces government healthexpenditure measured as percentage of the gross domesticproduct for the North America region. For its part, an increasein gross domestic product increases health expenditure and lifeexpectancy at birth, for this same region.

The North America region average results (without effects) are described in the previous paragraphs. Country results points out the presence of heterogenous characteristics in their health status. These characteristics are studied with a panel model that allows to decompose the North America estimators at a country level. In general, each country behaves as the North America region but with different elasticities. The country with a higher initial average value in equation (1) and (2) constant is Canada and The United States, respectively. These results imply that Canada has better life expectancy before the period analyzed. Also, they imply that The United States has a bigger general government health expenditure even before the period of 2000-2019.

The countries that have a better advantage from an economic growth for increasing life expectancy are Canada and The United States. It seems that the greatest elasticity of gini with respect to health expenditure is in The United States (-5.61). This elasticity probably responds to an inarticulate response from health policy makers (given to an excessive expenditure in transactional costs) to an increase in income inequality.

It seems that the health status improves with a reduction in income inequality and a greater public health expenditure. Therefore, those policies that increases income inequality and reduces public health expenditure seems to be advocates of a reduction in health status, population welfare and economic growth.