Economic growth is a result of efficient utilization of resources. It is a symbol of prosperity and economic development. Many indicators stimulate the growth of an economy among them human capital is the most important one. It is helpful to accelerate the output level, boosts labor's productivity, and enhance the efficiency of available resources.

Health is a basic element of human capital which not only uplifts the worker's capability but also improves output. Health is an indispensable for growth of an economy. A healthy body and mind are most important in performing daily life activities, and healthy person is able to enjoy the life without being dependent upon others. Health expenditures also improve food productivity and information about diseases. Improvement in health in industrial states raises economic growth up to 40%, and an increased rate of mortality causes a low growth level in developing countries (Arora, 2001).

The rising mortality rate is a sign of poor health facilities, as a result national income declines and vice versa. Infant mortality rate is declined that was 99.7 (per 1000) in 1990 and 28.2 in 2016, and the GDP per capita was 399.48 US$ in 1990 and 1029.57 US$ in 2016 in Bangladesh. The child mortality is also reduced in Pakistan from 106.2 (out of 1000) in 1990 to 64.2 (per 1000) in 2016, and the GDP per capita was 741.80 US$ in 1990 and 1179.41 US$ in 2016. Figure 1 shows an inverse affiliation among growth and infant mortality in MIC over 1991-2016. (World Bank report, 2017)

This section carries the review of previous studies on the nexus between health, education and economic growth. According to Mayer (2001), improvement in health facilities enhances productivity and income growth for age group of 1950-1990 in Latin America. GLS technique is applied using fertility rate, Govt. consumption, primary education and life expectancy. Carrion-i-Silvestre (2005) also exposed the association between healthcare expenditure (HCE) and GDP for OECD countries.

Similarly, Lago-Penas et al. (2013) found that HCE are feeding to GDP in 31 OECD countries over 1970-2009. It is also exposed that private share in health expenditure are more sensitive to GDP as compared to total expenditures. Mladenovic et al. (2016) tried to explore the relationship between HCE and GDP for 28 European members over 1974-2015 by applying adaptive neuro-fuzzy technique. The empirics show that HCE has progressive role to growth rate predictions.

Bhargava et al. (2001) explained that adult survival rate is an increasing factor of growth in low income countries. Saida and Kais (2018) also found a causality from health expenditure to GDP per capita in Africa over 1990-2015. Narayan et al. (2010) examined co-integration among health and economic growth for five countries over 1974-2007. Results were in line with the previous studies proving health as an important ingredient of growth. Gong et al. (2012) also summarized that health improvements are the source of growth for China over 1978-2003.

The empirical report of Swift (2011) confirmed a co-integration association among life expectancy and GDP in 13 OECD countries for two hundred years. Results exposed that a 1% improvement in life expectancy rate causes a six percent rise in GDP. Mahumud et al. (2013) also inspected that life expectancy increases the real income of people in Bangladesh. The study expressed that life expectancy is greater in females than males in the previous one and half decades.

Arora (2001) described the nexus among health and growth in 10 industrial economies for one century, and showed that economic growth raises 30-40% by improvement in health. Public spending on health also improve food productivity and information about diseases. The findings also expressed that increasing rate of mortality causes a low growth rate in developing countries.

Health is not the mere accelerator of growth, reverse also holds true. In a study on a sample of 148 countries for 1970-2010, Linden and Ray (2017) found per capita GDP as a basic reason of life expectations. Goode et al. (2014) investigated that household income has a strong impact on children health in China. Lu et al. (2017) discovered that GDP per capita is negatively correlated with mortality rate in 38 Chinese provinces for 2002-2014.

Katrakilidis et al. (2016) found a one-sided causation from income to infant mortality in Greece for 1960-2012. The results showed that economic growth improve health quality. Hooters' and Posnett (1992) ensured that GDP is an important factor of health spending and found a positive link among them in a relatively large sample of OECD countries. Results suggested that OECD countries panel is not an identical group.

A bidirectional affiliation among HCE and GDP in 20 OECD countries for 1970-2009 is examined by Amiri and Ventelou (2012). Similarly, Chaabouni et al. (2016) exposed a health-growth causal correlation for a panel of 51 states over 1995-2013, and Piabuo and Tieguhong (2017) also found the same nexus for CEMAC (the Central African Economic and Monetary Community) countries.

Education is a main character of economic growth. The study of Benos and Zotou (2014) shown that mostly studies used education attainment and ignored the quality of education, by considering 57 studies from previous literature. They have also guessed that qualitative measures are not used due to missing data. Mercan and Sezer (2014) explained a positive role of educational expenses to economic growth for Turkey during 1970-2012. Afzal et al. (2011) found twofold causality among education and growth in Pakistan for 1970-2008. The study expressed that an effect of higher education is stronger as compared to other education levels. A progressive role of higher education on growth is also found by Mariana (2015) in Romania for 1980-2013.

A study of Hongyi and Huang (2009) exposed that both health and education are accelerating factors of economic growth in China for 1978-2005. Ogundari and Awokuse (2018) expressed that health is more important than education for growth for a panel of 35 countries over 1980-2008. The findings of Bloom et al. (2004) pointed out that schooling and health are the main indicators of growth. They have also shown that a 1% improvement in life expectancy raises the output by 4 percent.

Frimpong and Adu (2014) inspected the significant character of schooling and health to raise growth in Africa over 1970-2010. The results also expressed that health has greater influence on growth as compared to education. The study of Lenkei et al. (2018) expressed the positive influence of primary and secondary education on economic growth in fourteen Asian economies for 1960-2013. The empirics exposed that tertiary education is negatively correlated with growth because of brain drain. Abdullah (2013) confirmed the accuracy of data on education for Malaysia and found a negative impact of education on growth due to unproductive use of education.

The results of secondary and tertiary education on economic growth are vary from country to country. Our purpose is to indicate the impact of life expectancy, infant mortality, secondary and tertiary education on economic growth in 76 MI countries.

A healthy body and mind are most important in performing activities of daily life, and healthy person is able to enjoy the life without being dependent upon others. A health production function describes information about health status of a nation. It explains the association between inputs and outputs in a particular time. An econometric model is formulated to find health and economic growth nexus. We follow the aggregate production function of Weil (2007) as,

Y=AK^α(Lω)^β ---------------------- (1)

where, Y output, A represents labor productivity and technology, Kis capital, and L is labor,ω is human capital per capita. Human capital of a worker carries better health and education which can be written as,

ω = S^γH^δ ------------------------ (2)

where, S and H represent schooling and health respectively, γ is the share of schooling and δ is the share of health towards human capital. Taking natural logarithm of Eq.2

lnω= γlnS+δlnH --------------------- (3)

Taking natural logarithm of Eq.1, the aggregate production function is as

lnY=lnA+αlnK+β(lnL+lnω) -------------- (4)

By substituting Eq.3 into Eq.4

lnY=lnA+αlnK+β(lnL+γlnS+δlnH) ----------- (5)

lnY=lnA+αlnK+βlnL+βγlnS+βδlnH ----------- (6)

The empirical model for panel data analysis from derived equation can be expressed as,

y_it= α₀+α₁k_it+α₂l_it+α₃s_it+α₄h_it ------------- (7)

where, the terms i is used for MI countries and t is used for time period from 1991-2016, yis used for log of economic growth, k is natural logarithm of capital, l is labor, s is lo of education and h used for natural log of health. The term α₀ is intercept, α₁ is used for elasticity of capital with respect to growth, α₂ is used for elasticity of labor, α₃ for schooling, and α₄ is the elasticity of health. Health is measured by life expectancy (LE) and infant mortality (IM), and included secondary and tertiary education as proxy of schooling.

Various methods and techniques are applied to inspect the association between human health and economic growth by incorporating education. The study uses the panel data analysis. There are many advantages of panel data: (i) competency of the parameters estimates is high due to high sample variability and the degree of freedom (ii) panel data helps to control the effect of missing and unobservable variables (iii) it helps to control individual heterogeneity that is it allowed to control unobservable variables (iv) it reduces chance of multi-colinearity and (v) it produces more accurate estimates as compared to cross-sectional and time series analysis. The basic techniques, panel OLS, Random Effect (RE) and Fixed Effect (FE) models are used for panel data analysis. We have used RE and FE models for MIC.

The study applied the Hausman test for picking an appropriate technique between RE and FE models. The hypothesis 'RE model is suitable' is tested against the alternative that is 'FE model is good'.

The key assumption of pooled OLS estimation technique is that intercepts and slope coefficients do not vary across cross-sectional units over time. Variations across countries over time are captured by error. Pooled OLS estimation technique is appropriate if cov(u_it, X_it)=0. The pooled regression model can be written for Eq.7,

y_it= α₀+α₁k_it+α₂l_it+α₃s_it+α₄h_it+μ_it ----------- (8)

The panel OLS takes common intercept and coefficients for all cross sections. This notion collapse the association among indicators. This is the reason to move towards RE and FE models.

FE model is used to determine the effect of coefficients that changes over time. This method assess the relationship between dependent and independent variables for each cross sectional unit, as every unit has some distinctive features which may have an impact on independent variable. FE model controls the impact of these time invariant features by allowing variability of characteristics in intercept. Now the proxy variables of education and health are added into Eq.7 and illustrated it according to the FE model,

y_it= α_i+α₁k_it+α₂l_it+α₃sec_it+α₄le_it+μ_it --------- (9.1)

y_it= β_i+β₁k_it+β₂l_it+β₃ter_it+β₄le_it+ε_it ---------- (9.2)

y_it= γ_i+γ₁k_it+γ₂l_it+γ₃sec_it+γ₄im_it+ζ_it ----------(9.3)

y_it= φ_i+φ₁k_it+φ₂l_it+φ₃ter_it+φ₄im_it+έ_it -------- (9.4)

Here, the intercept of each cross section vary due to the exclusive characteristics of each country. The slope coefficients show elasticity of capital, education, labor and health with respect to economic growth.

Random effects approach is appropriate if variation across entities have influence on dependent variable. It allows considering time invariant parameters. The key assumption of RE approach is that intercept of each cross-sectional unit is randomly drawn with constant average value of intercept. Thus Eq.7 with the proxy variables of education and health in form of RE model is,

y_it= α₀+α₁k_it+α₂l_it+α₃sec_it+α₄le_it+ú_i+μ_it ------ (10.1)

y_it= β₀+β₁k_it+β₂l_it+β₃ter_it+β₄le_it+υ_i+ε_it ------- (10.2)

y_it= γ₀+γ₁k_it+γ₂l_it+γ₃sec_it+γ₄im_it+ν_i+ζ_it ------ (10.3)

y_it= φ₀+φ₁k_it+φ₂l_it+φ₃ter_it+φ₄im_it+δ_i+έ_it ------ (10.4)

Or

y_it= α₀+α₁k_it+α₂l_it+α₃sec_it+α₄le_it+ω_it -------- (11.1)

y_it= β₀+β₁k_it+β₂l_it+β₃ter_it+β₄le_it+ψ_it -------- (11.2)

y_it= γ₀+γ₁k_it+γ₂l_it+γ₃sec_it+γ₄im_it+η_it -------- (11.3)

y_it= φ₀+φ₁k_it+φ₂l_it+φ₃ter_it+φ₄im_it+ώ_it -------- (11.4)

where ú_i is cross-sectional error component, u_it is time series and cross-sectional error and ω_it is an addition of ú_i and u_it. In contrast to FE, RE does not create degree of freedom loss. This method is suitable when correlation between independent variable and random error term is zero. FE model is more suitable when independent variables are correlated with error term.

This section contains the data, descriptive statistics and correlation among variables. Various variables are used to determine the relationship among indicators, and data is used over the period 1991-2016. The dependent variable include economic growth and measured by GDP per capita in literature (Siddique and Majeed, 2015).

The independent variables include infant mortality and life expectancy as health, and both are repetitively used in the literature (Linden and Ray, 2017; and Akram, et al. 2008). Schooling is measured by enrollment in secondary and tertiary education that is also used in economic literature (Akram, et al. 2008; and Ljunge, 2016). Capital is measured by Gross capital formation (% of GDP) which is utilized by Siddique et al. (2016), and labor force participation rate (above 15 age) is also independent variable (Siddique and Majeed, 2015). The data on all variables is taken from WDI (2018).

The study is concentrated on the panel of MIC (WDI 2018). We used three panels separately, the first is UMIC, the second is, LMIC, and third is MIC. UMI economies are those in which 2016 per person GNI was $3,956-$12,235. LMI countries are those in which 2016 GNI per capita was below $3,956.

This section carries the summary of descriptive statistics of used variables. Table 1 shows the maximum, minimum, standard deviation and mean values of data for MI countries. The maximum value (9.59) of economic growth belongs to Venezuela, RB and minimum value (6.06) belongs to Cambodia. The average economic growth is 8.00.

The maximum life expectancy (4.37) belongs to Costa Rica, minimum value (3.82) is for Swaziland, and mean values of life expectancy is 4.22. The maximum value of infant mortality (4.74) belongs to Nigeria, minimum value (1.06) is for Belarus, and mean values of infant mortality is 3.21. The bottom of Table 2 contains the descriptive stats for UMI countries and LMI countries separately.

This section explains the correlation among the used variables for three panels over 1991-2016. The correlation for MI countries (full sample) is shown in Table 2. Table 3 shows that correlation between variables for UMI countries, and Table 4 holds the correlation for LMI countries. The direction of relationship among variables are almost same for three panels. There is a positive correlation among life expectancy and economic growth. Capital, secondary and tertiary education are positively correlated labor and infant mortality is negatively correlated with growth.

The results of empirical analysis between health and economic growth for three panel datasets are discussed in this section.

Hausman test is applied to choose the best technique between RE and FE model. The main hypothesis (RE model is suitable) is tested against alternative (FE model is good) hypothesis. Results are displayed in Table 5.

The results for MI countries show that FE model is suitable for life expectancy with secondary education while RE is good for tertiary education equation. The FE model is appropriate for infant mortality with secondary and tertiary education. In case of UMI countries, the FE technique is reasonable. For the panel of LMI countries the FE approach is suitable for all cases except infant mortality with tertiary education.

Table 6 shows the empirical results of 76 MI countries (full panel) over 1991-2016. The study used four equations to assessment the nexus between education, health and economic growth.

Life expectancy has the coefficients 3.02 and 2.15 which shows a 1% increase in life expectancy rate causes 3.02% and 2.15% rise in economic growth. The result is consistence with Bloom et al. (2004). Infant mortality has adverse relationship with economic growth which shows that increasing rate of infant mortality decreases economic growth in MI countries. The coefficients of infant mortality are -0.61 and -0.57, which explain that a 1% increase in mortality rate declines growth rate by 0.61% and 0.57%, respectively. Lu et al. (2017) also discovered that GDP per capita is negatively correlated with mortality rate in China.

The results express that education is playing a vital role in increasing economic growth. The coefficients of secondary education are 0.39 and 0.14 which illustrate that a 1% rise in secondary education enrollment shows a 0.39% and 0.14% increase in economic growth. The enrollment in tertiary education is an indication of economic growth. The coefficients 0.29 and 0.13 express that a 1% percent rise in tertiary education is a sources of an intensification in economic growth by 0.29% and 0.13%, respectively. Afzal et al. (2011) found that the influence of higher level of education is stronger on growth as compared to other education levels. A progressive effect of higher education on growth is found by Mariana (2015).

Capital is an accelerating factor of economic growth which is consistent with economic literature (see, for instance, Siddique and Majeed, 2015). The coefficients of labor show an inverse association among labor and economic growth. Siddique and Majeed (2015) also found a negative sign of labor in economic growth model.

Hausman test confirmed that FE model is appropriate for the panel of UMI countries for all equations. Table 7 shows the empirical results of FE model for 42 UMI countries over 1991-2016. Life expectancy has a positive influence to economic growth. The coefficients are 3.53 and 2.23 which illustrate that a 1% increase in life expectancy rate causes an intensification in economic growth by 3.53% and 2.23%, respectively. Mahumud et al. (2013) also inspected that life expectancy increases the real income in Bangladesh.

The findings confirm that decreasing rate of infant mortality increases economic growth. The coefficients of infant mortality are -0.62 and -0.58 which explain that a 1% increase in infant mortality declines 0.62% and 0.58% in economic growth for UMI countries. The results are consistence with the findings of Lu et al. (2017).

The coefficients of secondary education (0.46 and 0.13) illustrate that a 1% rise in secondary education enrollment shows a 0.46% and 0.13% increase in economic growth. The coefficients of tertiary education are 0.35 and 0.14 which express that a 1% percent rise in tertiary education is a sources of 0.35% and 0.14% increase in economic growth respectively. The study of Lenkei et al. (2018) also expressed the positive influence of schooling on economic growth in Asia.

The results show that capital is an increasing determinant of economic growth. Furthermore, results indicate an inverse link between labor and income growth which is significant in two equations and insignificant in others. It is also observed in the literature (Omri, 2013).

Economic growth is a result of efficient utilization of resources. It is a symbol of prosperity and economic development. Various indicators stimulate the growth of an economy among them human capital is the most important one. It is helpful to accelerate the output level, boosts labor's productivity, and enhance the efficiency of available resources. The RE and FE methods are applied to find the results for 76 MI countries, 42 UMI countries and 34 LMI countries over 1991-2016. The study used four equations to estimate the nexus between health, education and economic growth.

Life expectancy is an increasing factor of economic growth in MIC (full panel). Child mortality has negative impact on economic growth which indicates that a decreasing rate of mortality increases economic growth. The results show that health is an indispensable for economic growth of an economy. The results are expressed that enrollment in secondary and tertiary education playing a progressive role in increasing economic growth. Capital is positively correlated with economic growth while the labor has negative effect on economic growth.

The findings of UMI and LMI countries expose that life expectancy and capital have positively correlated with economic growth, while labor and infant mortality are negatively correlated with economic growth. The findings also express that secondary and tertiary enrollment have positive effect on economic growth.

The study concluded that the impact of life expectancy, infant mortality, enrollment in secondary and tertiary education on economic growth is stronger in UMI countries as compared to LMI countries. Better health facilities and improved educational institutes in UMI countries are contributing more towards economic growth and development. The study recommended that economies should focus on education and better health facilities to improve economic growth especially in lower income countries.