Determinants of the Elderly Share of Population: A Cross-Country Empirical Analysis

Abstract

COVID-19 has had a disproportionate impact on the elderly, who are over-represented among those who suffered severe illness or death. The obvious implication is that the share of the elderly in the population significantly affects the impact of COVID-19 on the overall health of a country. More generally, the elderly share has far-reaching economic and social ramifications. In this paper, we perform empirical analysis of cross-country data from 1970 to 2018 to identify the determinants of the share of the elderly—i.e., those aged 65 and over—in a country’s population. We find that the quality of health care, life expectancy, and female labor participation increases the elderly share while higher fertility and female education attainment lower the elderly share. In addition, we find that the share is higher for high income countries and countries in Europe and Central Asia.

Introduction

One of the major global structural shifts in recent decades has been population aging. Until quite recently, the phenomenon of a demographic transition toward older population structures—i.e., a growing share of the elderly in the total population—was largely limited to advanced economies. But more recently, many developing countries have also experienced aging. In some cases, the speed of aging has been rapid. For example, China’s population is growing older at such a speed that there are growing concerns that the country may grow old before it becomes rich. In response to such concerns, on 1 June 2021, Beijing announced that it will allow couples to have three children, loosening its historic family planning policies. The announcement came three weeks after China reported that its population grew at the slowest pace in decades. On the other hand, some developing countries in sub-Saharan Africa and elsewhere are still young. Nevertheless, on the whole, developing countries are also experiencing aging and the entire world is moving toward a greyer demographic landscape.

As a result, global population aging has emerged as one of the most important global challenges. An older population is generally viewed as detrimental to economic growth primarily because it implies a lower share of the working-age population—usually defined as those aged 15–64—in the total population. The effective supply of labor or the relative number of workers eventually shrinks, reducing the productive capacity of the economy. Aging is widely viewed as a major reason why economic growth has slowed down in recent decades.

According to various studies, along with low fertility rate, which tends to be associated with increased female education attainment and labor force participation, mortality, international assistance on health care and development, and immigration and emigration policy emerge as the major determinants of population aging. In the literature review section, we will review the existing studies on the determinants of population aging. A coherent policy for addressing population aging requires identifying its causes.

Concerns about the negative impact of older populations on economic growth are most pronounced in advanced economies but are also growing in fast aging developing economies such as China. The phrase “demographics is destiny” encapsulates the almost universally negative perception, which borders on conventional wisdom, of a negative link between aging and growth.

In addition to reducing the share of the working-age population, population aging can harm a country’s economic dynamism in other ways. For example, younger people tend to be more entrepreneurial since they generally have better grasp of new technologies and are more willing to take risks and start their own businesses. An iconic stereotype of young innovative entrepreneurs is that of a young Steve Jobs starting Apple with his friends in a suburban garage in California. An increase in the share of the elderly in the population thus has a negative impact on entrepreneurship (Liang et al., 2018). But the adverse impact of an older population is by no means confined to slower economic growth. For one, pension systems may become financially unsustainable since the ratio of retired beneficiaries to working contributors rises. But perhaps most significantly, a country’s health costs will rise as the population grows older since older populations tend to be less healthy than younger populations. This is due to an immutable biological fact—the physical and mental capacity of human beings decline when they grow old, and they become weaker and more vulnerable to disease.

The coronavirus disease (COVID-19) which precipitated a global public health and economic crisis since early 2020 highlights the vulnerability of older populations to disease. COVID-19 is a highly infectious disease which is caused by a coronavirus that was first discovered in Wuhan, China, in December 2019. The global pandemic brought the world to a halt in 2020 but the rollout of vaccines, which were developed in record time, is raising hopes about the end of the pandemic. One striking stylized fact about COVID-19 is its uneven health impact across age groups. More specifically, the elderly faces a disproportionately higher risk of severe illness or death from the pandemic. This explains why many rich countries, which tend to have older populations, have been hit hard by COVID-19 despite their advanced health infrastructures. The United States (U.S.) is a prime example. At the same time, many poor countries with youthful populations, such as sub-Saharan African countries, have weathered the pandemic relatively well despite having inadequate public health systems.

COVID-19 thus underlines the important role of population structure in a country’s health costs. More precisely, the share of the elderly in a country’s population influences the health impact of the pandemic on the country. At a broader level, the share of the elderly has a significant influence not only on a country’s health but also its economic and overall performance. As such, an interesting and significant question is “What are the determinants of the share of the elderly?” The central objective of our paper is to empirically investigate the determinants of the elderly share—defined as individuals aged 65 and above1—with cross-country data. That is, we will assess why the share is higher in some countries than others. The rest of the paper is organized as follows. Section 2 briefly reviews the relevant literature. Section 3 describes the data and empirical framework while Sect. 4 reports and discusses the empirical results. Section 5 concludes the paper.

Literature Review

According to various studies, low fertility, which tends to be associated with increased female education attainment and labor force participation, mortality, international assistance on health care and development, and immigration and emigration policy are the major determinants of population aging.

In particular, the process of population aging is significantly related to changes in fertility and mortality rate. (Teerawichitchainan & Low., 2021).

For example, an aging demographic structure is the result of both declining fertility—i.e., people having fewer children—and increasing life expectancy—i.e., people living longer. The literature generally points to fertility decline as the more significant factor. Weil (1997) shows that at least two-thirds of the increase in the U.S. population share of the elderly is due to fertility decline. Bloom et al., (2009, 2010b) show that fertility decline had a much larger impact than rising life expectancy on the age structure of a group of Asian countries in 1960–2005. The experience of the Asian countries, which grew rapidly for decades, is consistent with the international historical experience of family size dwindling as income level rises.

Bloom et al. (2009) provides good literature reviews of the sizable number of studies which point to female education attainment and female labor force participation as significant drivers of declining fertility. These studies include Andersson (2000) and DeCicca and Krashinsky (2016), which all confirm a strong negative relationship between female education and labor force participation on one hand and fertility on the other hand. Education enables women to participate in the labor market, which, in turn, enables them to earn more and invest more in their children. In addition, female education enhances the health of mothers and children, thus improving a woman’s physical capability to give birth and reducing the economic need for many children. Furthermore, Kim (2016) finds that the effect of female education on fertility may depend on a country’s development level.

Calot et al. (1999) argue that until twentieth century, mortality was the most important driver of population ageing but changes in fertility such as the baby boom and baby bust will be more important in driving the aging process in the middle of the twenty-first century. Van Nimwegen and van der Erf (2010) argue that low fertility caused population aging as well as population decline in Europe. They also point out that encouraging migration from outside Europe or within Europe would be an effective policy solution for tackling the aging challenge. Boz and Ozasari (2020) claim that longer life expectancy and lower fertility are the main causes of population aging. Their study finds the positive relationship between the health expenditure and population aging in Turkey.

Along with low fertility rate, the effect of mortality was considered as an important cause of population aging. According to the study of Murphy (2021) on 11 European countries from the nineteenth century to the present, among fertility, mortality, and net migration, fertility changes had a smaller overall effect on population aging than mortality changes. Horiuchi (1991) emphasized the distinction between direct and indirect effects of mortality changes on population growth which, in turn, affect the speed of population aging. Caselli and Jacques (1990) find that the mortality change had a significant impact on population aging. Their projection indicated that the proportion of the population aged 60 + in Italy would be affected slightly more bt mortality change than fertility change.

At the same time, immigration can also play a substantial role in determining a country’s age structure, especially if the age structure of the immigrants differs significantly from that of the host country’s current population (see Coleman, 2008). Aging rich countries such as the U.S. and Singapore can mitigate population aging by accepting generally younger immigrants from young developing countries. However, political economy factors hinder the cross-border migration of workers. In particular, there is often substantial political resistance from local residents against immigrants, out of fear that they will take away jobs and crowd out basic public services such as health care and education. Although the size and speed of population aging are largely determined by fertility rate and mortality rates, migration can have some impact too (Booth 2018; Kinsella and Phillips 2005).

Nagarajan et al. (2021) claim that along with well-known factors such as human capital development, female labor force participation, and economic growth, the population aging of less developed countries is significantly influenced by foreign aid on medical care and development.

Finally, a number of studies confirm a negative impact of population aging on economic growth. Bloom, Canning, and Fink (2010a) find that population aging reduces economic growth due to lower labor force participation rates and savings rates. They project that OECD countries, a group of mostly advanced economies, will experience tangibly slower growth during 2005–2050 as their population grows even older. Maestas, Mullen and Powell (2016) use predicted variation in the rate of population aging across U.S. states over the period 1980–2010 to estimate the economic impact of aging on state output per capita. They find that a 10% Increase in the share of the population aged 60 + reduces the growth rate of gross domestic product (GDP) per capita by 5.5%. One-third of the growth reduction stems from slower labor force growth while two-thirds is due to slower growth of labor productivity. Park and Shin (2011) and Kim et al. (2020) find that population aging has a negative impact on the economic growth of Asian countries. For example, Park and Shin (2011) project that population aging will subtract 0.785% from the growth rate of China’s per capita GDP in 2021–2030. For Thailand and Vietnam, the corresponding figures are 0.859% and 0.279%, respectively.

Data and Empirical Framework

In this section, we describe the data and empirical framework we use for our empirical analysis. We used data from World Development Indicators (WDI)2 for our empirical analysis. The regional and income category follows the World Bank’s classifications. Table 1 shows some relevant demographic indicators in different regions of the world. The share of population aged 65 and above is the highest in Europe and Central Asia, followed by North America. The next oldest regions are Latin America and the Caribbean, East Asia and Pacific, and Middle East and North Africa. The youngest region is Sub-Saharan Africa, followed by South Asia. The pattern of life expectancy is similar. Residents of Europe and Central Asia and North America live the longest whereas the residents of Sub-Saharan Africa and South Asia have the shortest life spans. Finally, life expectancy ranges broadly between 60 and 80 years in all regions except the Sub-Saharan Africa where mortality rate is significantly higher than the other regions.

Table 1.

Selected demographic indicators in different regions, 1970–2018¹ Source: World Development Indicators (accessed on 18 December 2020)

Region2	Share of population aged 65 and above (%)	Life expectancy3	Fertility rate4	Infant mortality rate5
East Asia and Pacific	5.46	67.64	3.38	35.92
Europe and Central Asia	12.34	73.23	2.07	17.25
Latin America and the Caribbean	6.23	69.76	3.24	35.19
Middle East and North Africa	4.18	69.64	4.19	34.5
North America	12.08	77.12	1.79	8.76
South Asia	3.91	60.06	4.58	79.98
Sub-Saharan Africa	3.40	53.3	5.72	85.24

The figures are measured as the average value from 1970 to 2018

Regional group refers to the guidelines of the World Bank at the https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups, see Appendix 2

Life expectancy at birth, total (years) indicates that “the number of years a newborn infant would live if prevailing patterns of mortality at the time of its birth were to stay the same throughout its life”

Fertility rate, total (births per woman): means “the number of children that would be born to a woman if she were to live to the end of her childbearing years and bear children in accordance with age-specific fertility rates of the specified year”

Mortality rate, infant (per 1,000 live births):“Infant mortality rate is the number of infants dying before reaching one year of age, per 1,000 live births in a given year”

Table 2 shows the same demographic indicators as in Table 1 for different income groups of countries. As expected, the share of population aged 65 and above is correlated with income level. The share is highest in high-income countries, followed by upper middle-income countries, lower middle-income countries, and low-income countries. Life expectancy is positively associated with income level. High-income countries enjoy the highest life expectancy and low-income countries suffer from the lowest life expectancy. On the other hand, both fertility rate and infant mortality rate are negatively associated a country’s income level. More specifically, low-income countries have the highest fertility rate and infant mortality rate, followed by lower middle-income, upper middle-income, and high-income countries. There is a negative relationship between income and both fertility and infant mortality.

Table 2.

Selected demographic indicators in different income groups of countries, 1970–2018¹ Source: World Development Indicators (accessed on 18 December 2020)

Income group2	Share of population aged 65 and above (%)	Life expectancy3	Fertility rate3	Infant mortality rate3
High-income	10.71	74.73	2.24	12.79
Upper middle-income	6.18	67.84	3.38	35.59
Lower middle-income	3.88	59.27	4.81	66.43
Low-income	3.19	51.64	6.05	99.91

The figures are measured as the average value from 1970 to 2018

We followed the World Bank’s income group classification at the https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups, See Appendix 3

High-income economies are those with a GNI per capita of $12,696 or more

Upper middle-income economies are those with a GNI per capita between $4,096 and $12,695

Lower middle-income economies are those with a GNI per capita between $1,046 and $4,095

Low-income economies are defined as those with a GNI per capita, calculated using the World Bank Atlas method, of $1,045 or less

Please refer to notes (3), (4), and (5) in Table 1 for definitions of life expectancy, fertility rate, and infant mortality rate, respectively

The basic empirical specification used for our analysis of the determinants of the elderly share of the population is as in Eq. (1).

$$\begin{array}{cc}\hfill \text{ELDERLY}\_{\text{SHARE}}_{\mathit{it}}=& \beta 0{}_{\mathit{it}}+\beta 1HEALT{H}_{\mathit{it}}+\beta 2FEMAL{E}_{\mathit{it}}\hfill \\ \hfill +\beta 3GDPP{C}_{\mathit{it}}+\beta 4DUMMIE{S}_{\mathit{it}}+\epsilon {}_{\mathit{it}}\\ \hfill \end{array}$$ 1

The dependent variable is the share of the elderly in the population. More specifically, we use two different measures for the elderly share—(1) population aged 65 and above as % of the total population of country i at time t and (2) population aged 65 and above as % of the total working population of country i at time t. To reflect the findings from the empirical literature on population aging, we include health- and demographics-related variables, denoted as HEALTH. These include fertility rate, infant mortality rate, life expectancy, and the number of physicians per 1,000 residents. We also include two female-related variables (FEMALE) which affect fertility, which, in turn, affects demographic structure. These are female educational attainment and female share of the labor force. We also included per capita GDP (GDPPC) since richer countries tend to be older than poorer countries. In addition, we include dummies (DUMMIES) for different regions of the world and different income groups of countries to see whether there are significant differences across countries in different regions and income groups. Lastly, ε refers to error term. Table 3 describes the dependent and independent variables used in the empirical analysis.

Table 3.

Dependent and independent variables

Variable	Description	Predicted sign1	Data source
Dependent variable
${POP\_65}_t^i$	Population aged 65 and above as % of total population of country i at time t		WDI
${POP\_65\_\mathrm{work}}_t^i$	Population aged 65 and above as % of total working population of country i at time t		WDI
Independent variables
Health and demographic variables
${\mathit{PHY}}_t^i$	Physicians per 1,000 people of country i at time t	( +)	WDI
${\mathit{FERT}}_t^i$	Fertility rate, total births per woman of country i at time t	(−)	WDI
${\mathit{LEXP}}_t^i$	Life expectancy at birth, total years, of country i at time t	( +)	WDI
${\mathit{MORT}}_t^i$	Infant mortality rate per 1,000 live births of country i at time t	( +)	WDI
Female variables
${EDU\_F}_t^i$	Educational attainment of female population (Average years of total schooling of country i at time t)	( +)	Barro & Lee2
${LBR\_F}_t^i$	Labor force, female share (%) of total labor force of country i at time t	( +)	WDI
Other control variable
${\mathit{GDP}}_t^i$	GDP per capita (constant 2010 US$) of country i at time t	( +)	WDI
Dummies
Dummy 13	Income group dummy		WDI
Dummy 24	Regional group dummy		WDI

A positive (negative) predicted sign means that the independent variable is likely to increase (decrease) the share of population aged 65 and above (%) in total population

We used the educational attainment dataset by Barro and Lee

We followed the World Bank’s income group classification, as in Table 1

We followed the World Bank’s regional classification, which can be found in https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups. The World Bank’s regions are East Asia and Pacific, Europe and Central Asia, Latin America and the Caribbean, Middle East and North Africa, North America, South Asia, and Sub-Saharan Africa

Table 4 shows the mean, standard deviation, minimum, and maximum of the independent and dependent variables used in the empirical analysis.

Table 4.

Statistical properties of variables.

Source: Authors’ calculation

Variables	Mean	Standard Deviation	Minimum	Maximum
${POP\_65}_t^i$	6.72	4.71	0.69	27.66
${POP\_65\_\mathrm{work}}_t^i$	16.71	10.35	0.94	56.17
${\mathit{PHY}}_t^i$	1342.24	838.74	1.00	2903.00
${\mathit{FERT}}_t^i$	3.71	1.93	0.90	8.46
${\mathit{LEXP}}_t^i$	65.83	10.78	18.91	85.42
${\mathit{MORT}}_t^i$	45.40	41.02	1.60	204.20
${EDU\_F}_t^i$	57.25	29.48	0.23	100.00
${LBR\_F}_t^i$	40.27	9.59	7.86	56.03
${\mathit{LogGDP}}_t^i$	8.44	1.53	5.09	11.86

Please refer to Table 3 for definitions of variables

Table 5 shows the correlation between the variables used in the empirical analysis. Due to high correlation between fertility, life expectancy, and mortality, we estimate (1) with and without mortality included. We report the estimation results without mortality in Table 6 in Sect. 4 and the results with mortality in the Appendix 1.

Table 5.

Correlation matrix.

Source: Authors’ calculation

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
(1)${POP\_65}_t^i$	1.00
${(2)POP\_65\_\mathrm{work}}_t^i$	0.97	1.00
(3)${\mathit{PHY}}_t^i$	0.76	0.74	1.00
(4)${\mathit{FERT}}_t^i$	− 0.72	− 0.67	− 0.73	1.00
$(5){\mathit{LEXP}}_t^i$	− 0.60	− 0.59	− 0.66	0.85	1.00
(6)${\mathit{MORT}}_t^i$	0.65	0.63	0.71	− 0.85	− 0.93	1.00
${\left(7\right)EDU\_F}_t^i$	0.67	0.58	0.68	− 0.71	− 0.68	0.66	1.00
$\left(8\right){LBR\_F}_t^i$	0.31	0.20	0.18	− 0.04	0.12	− 0.18	0.33	1.00
${\left(9\right)LogGDP}_t^i$	0.68	0.63	0.67	− 0.72	− 0.76	0.79	0.68	− 0.12	1.00

Please refer to Table 3 for definitions of variables

Table 6.

Random-effectsGLS regression result. Authors’ calculation

Source:

Variables	Population aged 65 and above as % of total population of country i at time t			Population aged 65 and above as % of total working population of country i at time t
	1	2	3	4	5	6
${\mathit{PHY}}_t^i$	0.003***	0.002***	0.002***	0.008***	0.005***	0.005***
	(0.000)	(0.000)	(0.000)	(0.001)	(0.001)	(0.001)
${\mathit{FERT}}_t^i$	− 1.042***	− 0.795***	− 0.884**	− 2.340***	− 1.876***	− 1.713*
	(− 0.149)	(− 0.239)	(0.289)	(− 0.351)	(− 0.554)	(− 0.67)
${\mathit{LEXP}}_t^i$	0.167***	0.285***	0.280***	0.294***	0.617***	0.635***
	(− 0.030)	(− 0.040)	(0.0365)	(− 0.008)	(− 0.100)	(− 0.009)
${EDU\_F}_t^i$	− 0.003	− 0.013***	− 0.008*	− 0.0152	− 0.0358***	− 0.0307***
	(0.004)	(0.004)	(0.004)	(0.009)	(0.008)	(0.009)
${LBR\_F}_t^i$	0.274***	0.236***	0.204***	0.432***	0.307***	0.260***
	(0.012)	(0.018)	(0.022)	(0.030)	(0.045)	(0.053)
${\mathit{LogGDP}}_t^i$	0.170	0.110	− 0.690**	− 0.410	− 0.810	− 2.483***
	(0.155)	(0.196)	(0.197)	(0.352)	(0.484)	(0.468)
East Asia and Pacific		− 3.885***	− 3.799***		− 10.290***	− 10.150***
		(0.272))	(0.270)		(0.605)	(0.610)
Latin America and the Caribbean		− 4.529***	− 4.014***		− 10.670***	− 10.070***
		(0.165)	(0.194)		(0.429)	(0.483)
Middle East and North Africa		− 3.576***	− 4.183***		− 8.914***	− 10.090***
		(0.477)	(0.471)		(1.230)	(1.198)
North America		− 1.741***	− 1.704***		− 3.760***	− 3.644***
		(0.279)	(0.290)		(− 0.710)	(− 0.750)
South Asia		0.410	− 0.365		− 1.860	− 3.235**
		(0.473)	(0.514)		(0.998)	(1.127)
Sub-Saharan Africa		− 1.000	− 0.572		− 1.81	− 0.970
		(0.949)	(0.870)		(2.311)	(2.142)
Low-income			− 3.180**			− 8.625***
			(1.035)			(2.422)
Lower middle-income			− 2.312***			− 5.243***
			(0.578)			(1.535)
Upper middle- income			− 2.452***			− 4.544***
			(0.236)			(0.577)
Constant	− 17.520***	− 21.270***	− 11.010***	− 19.610***	− 26.880***	− 8.780
	(2.315)	(2.220)	(2.531)	(5.741)	(5.402)	(6.146)
Number of observations	604	604	604	608	608	608
R-squared	0.74	0.83	0.84	0.64	0.76	0.76

We report robust standard errors in parentheses and rounded off the numbers to three decimal places. The actual standard errors for variable ${\mathit{PHY}}_t^i$ of models (1), (2), and (3) are (0.000243), (0.000252), and (0.000231) respectively

*** and ** denote statistically significant at the 1 percent level and 5 percent level, respectively

Regional and income dummy category follows the World Bank classifications. “Europe and Central Asia” and “High Income” are the base groups of the regional and income groups, respectively

Table 7.

Random-effects GLS regression results (including mortality)

Variables	Population aged 65 and above as % of total population of country i at time t			Population aged 65 and above as % of total working population) of country i at time t
	1	2	3	4	5	6
${\mathit{PHY}}_t^i$	0.003***	0.003***	0.002***	0.007***	0.005***	0.005***
	(0.000)	(0.000)	(0.000)	(0.001)	(0.001)	(0.001)
${\mathit{FERT}}_t^i$	− 1.230***	− 0.576*	− 0.676*	− 2.530***	− 1.249*	− 1.163
	(0.229)	(0.248)	(0.281)	(0.517)	(0.594)	(0.670)
${\mathit{MORT}}_t^i$	0.029	− 0.028	− 0.029	0.029	− 0.080	− 0.076
	(0.025)	(0.027)	(0.028)	(0.054)	(0.058)	(0.061)
${\mathit{LEXP}}_t^i$	0.204***	0.243***	0.236***	0.330***	0.502***	0.524***
	(0.034)	(0.034)	(0.035)	(0.082)	(0.090)	(0.089)
${EDU\_F}_t^i$	− 0.002	− 0.015***	− 0.010*	− 0.014	− 0.040***	− 0.034***
	(0.005)	(0.004)	(0.004)	(0.010)	(0.008)	(0.009)
${LBR\_F}_t^i$	0.278***	0.233***	0.200***	0.436***	0.297***	0.250***
	(0.012)	(0.019)	(0.024)	(0.032)	(0.048)	(0.057)
${\mathit{LogGDP}}_t^i$	0.228	0.107	− 0.674***	− 0.334	− 0.845	− 2.455***
	(0.149)	(0.200)	(0.194)	(0.343)	(0.491)	(0.456)
East Asia and Pacific		− 3.851***	− 3.782***		− 10.20***	− 10.10***
		(0.275)	(0.270)		(0.610)	(0.605)
Latin America and the Caribbean		− 4.537***	− 4.001***		− 10.700***	− 10.050***
		(0.169)	(0.197)		(0.441)	(0.478)
Middle East and North Africa		− 3.755***	− 4.359***		− 9.436***	− 10.570***
		(0.502)	(0.510)		(1.319)	(1.307)
North America		− 1.745***	− 1.716***		− 3.768***	− 3.669***
		(0.282)	(0.292)		(0.721)	(0.751)
South Asia		0.720	− 0.0730		− 0.990	− 2.477
		(0.654)	(0.647)		(1.374)	(1.377)
Sub-Saharan Africa		− 1.150	− 0.719		− 2.230	− 1.374
		(0.896)	(0.817)		(2.234)	(2.059)
Low-income			− 3.015**			− 8.164**
			(1.107)			(2.612)
Low middle-income			− 2.151***			− 4.856**
			(0.644)			(1.653)
Upper middle-income			− 2.430***			− 4.483***
			(0.240)			(0.592)
Constant	− 21.070***	− 17.940***	− 7.747**	− 23.070***	− 17.700**	− 0.505
	(2.539)	(2.396)	(2.681)	(6.598)	(6.168)	(6.443)
Number of observations	604	604	604	608	608	608
R-squared	0.75	0.83	0.84	0.64	0.76	0.76

We report robust standard errors in parentheses and rounded off the numbers to three decimal places. The actual standard errors for variable ${\mathit{PHY}}_t^i$ of models (1), (2), and (3) are (0.000243), (0.000252), and (0.000231), respectively

*** and ** denote statistically significant at the 1 percent level and 5 percent level, respectively

Regional and income dummy category follows the World Bank classifications. “Europe and Central Asia” and “High Income” are the base groups of the regional and income groups, respectively

Source: Authors’ calculation

Empirical Results

In this section, we report and discuss our empirical results.

Summary of Result

As our empirical estimation methodology, we used random effects generalized least squares (GLS) regression which is well suited for longitudinal or panel data estimation. A big reason we used the random effects empirical approach is that in the presence of an error term which will lead to an error covariance matrix that is not spherical, a GLS-type approach like random effects will be more efficient than ordinary least squares (OLS). GLS is a technique for estimating the unknown parameters in a linear regression model when there is a certain degree of correlation between the residuals in these regression model. In these cases, OLS can be inefficient and even give misleading inferences.

The estimated results, which are shown in Table 6, are mostly in accordance with our expectations. Both physicians per 1,000 people and life expectancy have positive effects on the elderly share of the population at the 1% significance level. The relative number of physicians is a proxy for the quality of a country’s health care system and a better health care system contributes to longer life expectancy, which increases the share of the elderly in the population. The effect of fertility rate is negative at the 1% significance level. If women have fewer children, the number of children relative to the population falls, which increases the elderly share of the population. The share of women in the labor force has a positive effect on the elderly population share. This is intuitively plausible since if women spend more time at work, they will have less time for having and raising children. All these findings are consistent with existing studies such as, Andersson (2000); Bloom et al. (2009), DeCicca and Krashibsky (2016), Weil (1997).

However, in contrast to previous studies (e.g., Andersson, 2000, Bloom et al., 2009; Caselli & Jacques, 1990; DeCicca and Krashibsky (2016); Horiuchi, 1991; Teerawichitchainan & Low, 2021) female educational attainment has a negative effect on the share of the elderly in a country’s population. The intuition behind a negative link is that better educated women are likely to have fewer children but invest more time and effort in raising each child. In other words, educated women are more likely to choose quality over quantity in having and raising children. However, upon closer thought, there are other channels besides lower fertility thorough which female education can increase the elderly share. For one, better educated women will be more knowledgeable about healthy lifestyles and thus make healthy lifestyle choices which will enable them to live longer. Such knowledge also enables them to make healthy lifestyle choices—e.g., healthier food—for their families. The end result is that healthier women and healthier families raise life expectancy and the elderly share of the population.

The estimation results indicate that the elderly population share of low-income, lower middle-income, and upper middle-income countries is lower than in high income countries, which was used as the base for interpreting the coefficients of the other regions in relative terms. The difference was significant at the 1% level for all three income groups. We dropped Europe and Central Asia from the regressions and use the region as the base to interpret the coefficients of the other regions in relative terms. According to the estimation results, the elderly share of all other regions was lower than Europe and Central Asia and the difference is significant for all regions at the 1% level, except Sub-Saharan Africa and South Asia which showed statistically insignificant results. The results for the random-effects GLS regressions which include the mortality rate as an explanatory variable are shown in the Appendix 1 and qualitatively similar to the results in Table 6.

Discussion

COVID-19 is an unprecedented global health crisis. Since the Spanish flu, also known as the 1918 influenza pandemic, there has not been any health shock which has had such far-reaching ramifications for the public health, economy, and society of all countries. Its profound impact is evident in the sharp contraction of international air travel and stringent local social distancing regulations which severely restricted the mobility of individuals. Although its effect extends beyond health into other areas, most notably the economy, COVID-19 is first and foremost a health crisis. Furthermore, the non-health effects of the pandemic—e.g., slowdown of economic activity due to lockdowns—are mostly rooted in mobility restrictions and other public health measures which seek to prevent the virus from spreading. As such, it is conceptually difficult to separate the health and non-health effects of COVID-19.

One striking feature of the health effects of the pandemic is its disproportionate impact on the elderly, who are much more vulnerable to the disease than the rest of the population. More precisely, those 65 and above account for a disproportionately high share of severe illnesses and deaths. The elderly generally has weaker immunity systems, which make them more vulnerable to the virus. Furthermore, they are more likely to suffer from pre-existing medical conditions which amplify the risk of falling seriously ill if they are infected. As a result, even advanced countries with advanced medical infrastructure, most notably the U.S., have been hit hard by the pandemic due to their relatively old populations. This points to a broader issue. The share of the elderly in the population has a significant impact on the health, economic, and overall performance of a country.

Older populations are associated with higher health costs, slower economic growth, and other socially undesirable outcomes. Given such importance of the elderly share of the population, a natural question that arises is why are some countries relatively young while others are relatively old? Put differently, what are the factors that determine the elderly share of a country’s population? To answer this question, we perform cross-country empirical analysis on a data set of 181 countries. Following the literature, we include a number of variables related to health and demographics as explanatory variables—e.g., fertility rate and life expectancy. In addition, we include variables which affect the decision of women to have children, namely female educational attainment and female labor force participation rate. We also include regional and income group dummies for the countries in our sample.

Conclusion

Our empirical findings are largely consistent with the empirical literature (e.g., Weil, 1997, Bloom et al., 2009; Andersson, 2000; DeCicca and Krashibsky 2016; Caselli & Jacques, 1990; Horiuchi, 1991; Teerawichitchainan & Low, 2021) and economic intuition. More specifically, we find that the relative number of physicians, life expectancy, and female labor participation all have positive and significant effects on the elderly share of the population. The relative number of physicians is a proxy for the quality of a country’s health care system. Individuals are likely to live longer in countries with better health care systems and this will increase the relative number of the elderly. As more women enter the workforce, they will have less time to have and raise children, and thus contribute to an older population. On the other hand, higher fertility and higher female educational attainment have negative effects on the elderly share. The positive effect of female education is counterintuitive since better educated women tend to have fewer children. One potential explanation is that better educated women and their families live longer because of better informed, healthier lifestyle choices.

Overall, our findings re-confirm a significant effect of health- and demographic-related variables as well as variables that influence the decision of women to have children. As such, our findings suggest that developing countries, especially fast-growing developing countries such as China, India and other Asian countries, will witness a steady increase in the elderly share of the population. The reason is that health- and demographic-related variables will change tangibly as a result of rapid economic growth and improvement of living standards. For instance, we can expect life expectancy to rise and fertility to fall. The natural policy implication is that developing countries, especially fast-growing ones, would do well to start preparing for a future in which the elderly population is larger in relative terms. For example, they may consider designing and implementing reforms to their pension systems to make their financial sustainability less dependent on favorable demographics.

Appendix

See Tables 7, 8 and 9.

Table 8.

Regional Group Classification.

Source: The World Bank’s regional group classification (https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lehnding-groups)

East Asia and Pacific
American Samoa, Australia, Brunei Darussalam, Cambodia, China, Fiji, French Polynesia, Guam, Hong Kong SAR, China, Indonesia, Japan, Kiribati, Korea, Rep. of, Malaysia, Marshall Islands, Mongolia, Myanmar, Nauru, New Caledonia, New Zealand, Northern Mariana Islands, Palau, Papua New Guinea, Philippines, Samoa, Singapore, Solomon Islands, Thailand, Timor-Leste, Tuvalu, Vanuatu
Europe and Central Asia
Albania, Andorra, Armenia, Austria, Azerbaijan, Belarus, Belgium, Bulgaria, Croatia, Cyprus, Denmark, Estonia, Faroe Islands, Finland, France, Georgia, Germany, Gibraltar, Greece, Greenland, Hungary, Iceland Ireland, Italy, Kazakhstan, Latvia, Liechtenstein, Lithuania, Luxembourg, Montenegro, Netherlands, Norway, Poland, Portugal, Romania, Russian Federation, San Marino, Serbia, Slovenia, Spain, Sweden, Switzerland, Tajikistan, Turkey, Turkmenistan, Ukraine, United Kingdom, Uzbekistan
Latin America and the Caribbean
Antigua and Barbuda, Argentina, Aruba, Barbados, Belize, Brazil, British Virgin Islands, Cayman Islands, Chile, Colombia, Costa Rica, Cuba, Dominican Republic, Ecuador, El Salvador, Grenada, Guatemala, Guyana, Haiti, Honduras, Jamaica, Mexico, Nicaragua, Panama, Paraguay, Peru, Puerto Rico, Sint Maarten (Dutch part), Suriname, Trinidad and Tobago, Turks and Caicos Islands, Uruguay
Middle East and North Africa
Algeria, Bahrain, Djibouti, Iraq, Israel, Jordan, Kuwait, Lebanon, Libya, Malta, Morocco, Oman, Qatar, Saudi Arabia, Syrian Arab Republic, Tunisia, United Arab Emirates
North America
Bermuda, Canada, United States
South Asia
Afghanistan, Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan, Sri Lanka
Sub-Saharan Africa
Angola, Benin, Botswana, Burkina Faso, Burundi, Cabo Verde, Cameroon, Central African Republic, Chad Comoros, Equatorial Guinea, Eritrea, Eswatini, Ethiopia, Gabon, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritania, Mauritius, Mozambique, Namibia, Nigeria, Rwanda, Sao Tome and Principe, Senegal, Seychelles, Sierra Leone, Somalia, South Africa, South Sudan, Sudan, Togo, Uganda, Zambia, Zimbabwe

Table 9.

Income Group Classification.

Source: The World Bank’s income group classification (ttps://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups)

High-Income Economies ($12,696 or more)
Andorra, Antigua and Barbuda, Aruba, Australia, Austria, Bahrain, Barbados, Belgium, Bermuda, British Virgin Islands, Brunei Darussalam, Canada, Cayman Islands, Chile, Croatia, Cyprus, Denmark, Estonia, Faroe Islands, Finland, France, French Polynesia, Germany, Gibraltar, Greece, Greenland, Guam, Hong Kong SAR, China, Hungary, Iceland, Ireland, Israel, Italy, Japan, Korea, Rep. of, Kuwait, Latvia, Liechtenstein, Lithuania, Luxembourg, Malta, Netherlands, New Caledonia, New Zealand, Northern Mariana Islands, Norway, Oman, Palau, Panama, Poland, Portugal, Puerto Rico, Qatar, San Marino, Saudi Arabia, Seychelles, Singapore, Sint Maarten (Dutch part), Slovenia, Spain, Sweden, Switzerland, Trinidad and Tobago, Turks and Caicos Islands, United Arab Emirates, United Kingdom, United States, Uruguay
Upper-Middle-Income Economies ($4,096 TO $12,695)
Albania, Algeria, American Samoa, Argentina, Armenia, Azerbaijan, Belarus, Belize, Botswana, Brazil, Bulgaria, China, Colombia, Costa Rica, Cuba, Dominican Republic, Ecuador, Equatorial Guinea, Fiji, Gabon, Georgia, Grenada, Guatemala, Guyana, Iraq, Jamaica, Jordan, Kazakhstan, Lebanon, Libya, Malaysia, Maldives, Marshall Islands, Mauritius, Mexico, Montenegro, Namibia, Nauru, Paraguay, Peru, Romania, Russian Federation, Samoa, Serbia, South Africa, Sri Lanka, Suriname, Thailand, Turkey, Turkmenistan, Tuvalu
Lower-Middle Income Economies ($1,046 TO $4,095)
Angola, Bangladesh, Bhutan, Cabo Verde, Cambodia, Cameroon, Comoros, El Salvador, Eswatini, Ghana, Honduras, India, Indonesia, Kenya, Kiribati, Lesotho, Mauritania, Mongolia, Morocco, Myanmar, Nicaragua, Nigeria, Pakistan, Papua New Guinea, Philippines, Sao Tome and Principe, Senegal, Solomon Islands, Sudan, Timor-Leste, Tunisia, Ukraine, Uzbekistan, Vanuatu, Zambia, Zimbabwe
Low-Income Economies ($1,045 OR LESS)
Afghanistan, Benin, Burkina Faso, Burundi, Central African Republic, Chad, Djibouti, Eritrea, Ethiopia, Guinea, Guinea-Bissau, Haiti, Liberia, Madagascar, Malawi, Mali, Mozambique, Nepal, Rwanda, Sierra Leone, Somalia, South Sudan, Syrian Arab Republic, Tajikistan, Togo, Uganda