Population aging may affect the aggregate saving rate by raising the fraction of the population in age groups traditionally associated with the drawdown rather than the accumulation of savings. It may also affect the economy’s average level of savings per capita, since individuals approaching, and shortly after, retirement age tend to have higher levels of savings than those at the start of their working career. The preceding chapter described both of these effects, which may affect the productive capacity of the economy. These effects may also, in turn, affect the rate of return that investors earn on their savings.
The average return on investments is a key determinant of the performance of funded retirement plans in both the private and public sectors. High rates of return enhance the power of private saving to provide for consumption and health needs in retirement. If investors can earn a real, net-of-tax return of 6 percent each year, for example, a dollar saved at age 53 will grow to 2 dollars at age 65. If the rate of return is 4 percent, however, a dollar must be saved at age 47 to double in value by age 65. With a return of only 2 percent, a dollar must be saved at age 30 to double by 65.
For many households, the ongoing shift from defined benefit (DB) to defined contribution (DC) plans has tightened the link between investment returns and retirement security. As discussed in Chapters 5 and 7, the returns earned by participants in DC plans, along with their contributions, determine their asset balances at retirement and directly affect postretirement living standards. In DB plans, the links between asset returns and participant benefits are weaker, since the firm or government offering the pension plan bears the risk of asset value fluctuations. When the assets in a
private DB plan perform poorly, shareholders in the sponsoring firm earn lower returns because the firm must make compensatory contributions to fund the plan. Taxpayers play the role of shareholders in public sector DB plans. Past and projected rates of return are likely to affect the willingness of firms and government to continue DB plans, or at least continue them at the same level.
Private wealth accumulation supports a substantial share of retirement spending for some older households in the United States. For those households with substantial wealth, the rate of return can be an important determinant of retirement income. However, not all households have much exposure to financial markets, and many retirees have low levels of both financial and physical assets. For these households, rates of return are likely to be relatively unimportant influences on their financial circumstances.
Poterba, Venti, and Wise (2011) report that in 2008, the median holding of all financial assets, including those in individual retirement accounts and other similar vehicles, among households headed by someone aged 65–69 was $52,000. The analogous statistic was $112,000 for married couples. Because the ownership of assets is concentrated, average assets per capita are substantially greater than the assets of the median household. For households near the median as well as for those with fewer financial assets, their financial asset holdings are smaller than the value of prospective Social Security benefits. Financial assets are also smaller than accumulated home equity for those who own a home and smaller than DB pensions for those who are eligible for such benefits. Because the wealth distribution is highly skewed, however, there are also many households with substantial asset holdings for whom changes in rates of return are consequential.
Many factors affect capital market returns, including global demographic trends, long-run trajectories of productivity growth, investors’ attitudes toward risk, and government tax and spending policies. Some analysts have suggested that population aging during the next few decades will drive down the general level of investment returns (Arnott and Chaves, 2011; Liu and Spiegel, 2011). They reason that as baby boomers approach the traditional age at which asset holdings peak, assets will be in plentiful supply and the equilibrium return on assets will decline. As The Economist reported (2006), some have even suggested that when baby boomers draw down their financial assets to pay for their retirement consumption, selling pressure may generate an asset market meltdown, a sharp decline in asset values. However, that scenario seems unlikely because it is inconsistent with forward-looking behavior on the part of financial market participants; it would require a sharp fall in asset prices in response to a predictable demographic event.
This chapter examines the ways in which prospective changes in the age structure of both the United States and the global population may af-
fect rates of return. The increasing globalization of world capital markets requires a focus on how global population aging will affect the global supply and demand for savings and the rate of return available to savers (see Box 8-1 for basic terminology used in this chapter). The committee begins by presenting a general framework that highlights the various ways by which global population aging may affect rates of return. Because other factor inputs, notably labor input, also affect asset returns, the committee also explores how changing population age structure may affect labor supply and the decisions of young households with regard to human capital acquisition. Rising investment in human capital by younger workers can potentially offset some of the rate of return consequences associated with population aging, since the effective labor supply from a small cohort of highly skilled workers can be comparable to that from a larger cohort of less-skilled workers. In evaluating how population aging may affect rates of return, the committee draws on empirical studies that have compared rates of return in the United States and other nations to various measures of demographic structure as well as on findings from simulation models that have been calibrated to reflect central attributes of the U.S. and the global economy.
Demographic structure is only one of many influences on prospective rates of return. While the committee does not attempt to evaluate all the other forces that may affect such returns, it does note that the large fis-
Throughout this chapter, a standard practice is followed and “savings” is used to denote a stock and “saving” to denote a flow. The stock of savings refers to the total amount of accumulated net assets that households, companies, and governments hold. The amount of saving in a given period, a flow, denotes the difference between income and consumption; the saving rate is the ratio of the flow of saving to the flow of income. An increase in the saving rate increases the supply of savings; an increase in desired consumption or investment increases the demand for savings. “Assets” can refer to any store of value, including physical assets such as land or plant and equipment, intangible assets such as patents, and financial assets that represent claims on the cash flows paid by companies, governments, or households. “Capital” here is used to denote the subset of assets that are used to produce goods and services. The capital-labor ratio, a common measure of the factor intensity of an economy, equals the ratio of the capital that is used in production to the amount of labor used in production. Labor in this context is a weighted sum, with the number of hours of labor input from different workers weighted by the workers’ skill level.
cal deficits projected for the United States and other developed countries, which in turn are substantially affected by population aging, could play a significant role. These deficits will raise government liabilities and absorb savings that would otherwise be invested in productive capacity, leaving the economy with slower growth and fewer resources to support an older population. One consequence of the government-induced increase in the demand for savings could be higher rates of return. As is noted elsewhere, the lack of a credible strategy and its timely implementation for reducing projected deficits raises the risk of investors losing confidence in U.S. Treasury debt, which could lead to a rise in interest rates and a decline in the value of Treasury bonds.
HOW DEMOGRAPHIC CHANGE AFFECTS EXPECTED
RATES OF RETURN: A FRAMEWORK
The balance between the supply of savings and the demand for savings determines the rate of return earned by investors. Households, businesses, and governments that accumulate savings hold financial claims such as bank deposits, corporate and government bonds, stocks, and deeds to houses and other real estate. Other households, businesses, and governments may demand savings to deploy in financing investment, personal consumption, or government consumption. They may issue financial claims, such as corporate bonds, to the suppliers of savers when they deploy these savings. An increase in the supply of savings lowers the expected rate of return to savers, whereas an increase in the demand for savings increases the return that savers can expect to earn. The decisions of savers about what share of current output to save each year, together with the value of savings that have been accumulated in past years, determine the aggregate supply of savings. The demand for savings, in turn, is determined by company decisions about how much capital to use in the production process, by government borrowing needs, and by household demand for credit.
An aging population can affect both the supply of and the demand for savings, and there are potentially countervailing effects. This makes it difficult to provide a clear-cut answer to the question, Are asset returns more likely to rise or to fall as a consequence of population aging? The committee concludes that the net effects of population aging on rates of return are likely to be modest.
This analysis is generally set in a framework in which global asset markets operate as an integrated whole, which means that assets can migrate freely to wherever they are expected to earn the highest rate of return. Migration of assets tends to equalize expected returns internationally: The global supply and demand for assets determines expected returns. This is
why the aging of the global population, weighted by the amount of assets that residents of various nations hold, rather than the aging of the U.S. population, is most relevant for predicting how aging will affect rates of return. While the committee focuses on the mobile assets case, if asset mobility is limited, then while global population dynamics may affect rates of return, domestic demographic factors may also matter.
Table 8-1 shows the evolution of the U.S. old age dependency ratio, the global old age dependency ratio in which population aggregates are simply added up across nations, and a weighted old age dependency ratio in which each country’s dependency ratio is weighted by its current and projected per capita gross domestic product (GDP) rather than its population in computing the global measure. The committee reports GDP-weighted values because it does not have detailed information on the aggregate asset holdings in each nation. The table shows that the global population today is older on a GDP-weighted basis than on an equal-weighted basis. It also shows that GDP weighting tends to reduce the disparity between the aging of the U.S. and the global population. The per capita GDP-weighted old age dependency ratio for the global economy rises more slowly than that for the United States for the next 20 years, but it catches up in the subsequent two decades.
Each year households decide how much to consume. When consumption exceeds their income, they must draw down savings to finance consumption, and vice versa. Companies decide whether to retain or distribute their earnings and how much to invest; and governments decide whether or not to save by collecting taxes in excess of current spending. Companies that seek resources to deploy plant or equipment in their business or to invest in new technology, households that wish to borrow because their desired consumption exceeds their current income, and governments that issue bonds because their tax revenues fall below their outlays determine the demand for savings.
TABLE 8-1 U.S. and Global Old-Age Dependency Ratios, 2010–2050
|Year||U.S. Dependency Ratio||Global Dependency Ratio (Unweighted)||Global Dependency Ratio (Per Capita GDP Weighted)|
SOURCE: Donehower and Boe (2012).
The standard theory of consumer behavior has clear implications for the amount of savings that households will choose to hold. The theory posits that, all else equal, households prefer more consumption to less, smoother consumption to a more volatile consumption profile, consuming sooner to consuming later, and greater certainty about their consumption path. For a given level of expected returns, the desire to smooth consumption over time causes households to hold higher levels of savings when they expect consumption growth to be relatively slow; conversely they save less or try to borrow against future income when expected consumption growth is more rapid. Impatient households postpone the accumulation of savings until they near retirement age. Risk aversion motivates higher levels of savings as a buffer against adverse economic shocks. While the effect of higher expected rates of return on the household saving rate is theoretically ambiguous, under many reasonable modeling assumptions it seems that higher returns draw forth higher saving.
An aging population can affect aggregate savings in a variety of ways. Because older households on average hold more savings than younger households, an aging population will tend to display rising savings per capita. The desire to smooth consumption leads households to save while working so that they can draw down their assets to finance their retirement years. Higher aggregate savings per worker can affect the marginal product of capital—the amount of additional output that is produced from an incremental deployment of capital investment—if it translates into a higher level of capital per worker. Economic models of production imply that when the number of workers per unit of productive capital declines, the marginal product of capital will decline. A lower marginal product of capital would translate into a lower rate of return on incremental investments, which could reduce the incentive to save and partly offset the aging-related increase in savings per capita.
Population aging may also affect household savings by affecting the expected path of productivity growth and thereby consumption growth. Productivity growth is the most important determinant of consumption growth over long horizons. The rates of output and consumption growth over substantial periods of time are roughly proportional to the growth rate of output per unit of labor used in production. Chapter 6 reviews the evidence of the effects of an aging population on productivity and concludes that there is likely to be a negligible effect of the age composition of the labor force on the level of aggregate productivity over the next two decades.
Another way by which population aging may affect savings is its effect on economic uncertainty. For example, households may increase their sav-
ings if the imbalances between promised social insurance benefits and the tax revenues available to pay them create doubts about whether benefits will be fully paid or whether taxes will be sharply raised. Aggregate savings could fall, however, if developing countries introduce more extensive social insurance programs for their older citizens that reduce their incentives for private saving. China is currently pursuing policies of this type.
Savings by Companies and Governments
Companies and governments also make decisions that affect the aggregate supply of savings. Companies save when they retain earnings instead of paying them out to shareholders. Retained earnings generally are invested in a firm’s own operations or in securities issued by other entities. Companies have a greater incentive to save when expected returns are higher and when there is more uncertainty about the availability of bank or investor financing. Because companies operate on behalf of the households that own them, household preferences are generally taken to be the fundamental determinants of company saving and investment decisions. The foregoing discussion of the effect of population aging on household savings therefore is suggestive of the forces that will indirectly affect corporate savings as well.
Governments save when current-year tax revenues exceed current-year spending, i.e., when they run budget surpluses. Often those surpluses are used to pay down government debt, but they also can be invested in real or financial assets. It is likely that increases in government saving are partially but not fully offset by reductions in household saving, so that government saving increases the overall supply of saving. The effect of government saving programs on household saving may depend on the nature of the tax and spending changes that are adopted. Except for a few years in the late 1990s, the U.S. federal government has been a dis-saver for most of the last three decades. Federal budget deficits have been particularly large as a result of the deep recession that began in 2007. Projections described elsewhere in this report suggest growing future deficits resulting from increased demands on social insurance programs.
Population Aging and the Demand for Savings
Companies use savings when they deploy various types of capital, such as land, real estate, equipment, and research and development capital, as inputs to production. They invest more when there are a greater number of profitable investment opportunities. When the return demanded by savers is low, a firm’s discount rate will also be low, and the number of projects that will generate returns in excess of this threshold, or that will meet a
present discounted value test, will be high. Population aging may affect a company’s demand for savings in several ways. First, as per capita labor supply declines and labor becomes relatively scarce, firms may wish to substitute capital for labor; this would stimulate investment demand. More capital-intensive production processes may also allow older workers to extend their stay in the labor force. A mitigating effect is that raising the capital:labor ratio will reduce both the marginal product of capital and the return on capital investment, all else equal.
Households demand savings when they borrow to finance purchases of homes or cars, to invest in education, or to finance consumption. The aging of the population may affect household borrowing demand. Older households on average borrow less than younger ones—they are less likely to want to buy a bigger house or to attend college, and they have less future earning capacity to borrow against. However, if financial product innovations create new ways for older households to use their home equity to finance consumption, it is possible that borrowing by older households in the future may be higher than it is today.
Governments borrow not only to pay for current spending, but also to invest in capital that is used in the production of government services such as transportation, education, and health care. An aging population per se is unlikely to have a significant impact on that aspect of savings demand. A more important way in which aging may affect government demand for savings is by its impact on social insurance programs and associated budget deficits, as noted above.
Asset Market Equilibrium and Expected Returns
Expected rates of return shift over time so as to equalize the supply of and demand for savings. The foregoing discussion underscores the many factors that influence supply and demand and their interactions with an aging population. The underlying determination of returns, however, is straightforward. An increase in expected returns increases the reward to saving. Whether this raises or lowers the supply of saving is theoretically indeterminate and has proven difficult to measure empirically. At the same time, higher expected returns reduce the demand for savings, because it becomes more difficult to find investment opportunities that earn a sufficient return.
To summarize this discussion, Table 8-2 indicates the main channels through which an aging population can affect expected returns. Because the various effects point in different directions, evaluating the net effect of population aging on expected returns requires using an economic model that makes it possible to compare the quantitative effects of different influences.
TABLE 8-2 Some Effects of Aging on the Supply of and Demand for Savings
|Link to supply of savings|
|Older households have higher savings||Increases supply|
|Policy uncertainty raises precautionary savings||Increases supply|
|Higher capital-to-labor ratio lowers return to savers||Decreases supply|
|Productivity growth effects||Uncertain|
|Link to demand for savings|
|Firms substitute capital for labor||Increases demand|
|Higher government deficits||Increases demand|
|Older households borrow less||Decreases demand|
|Productivity growth effects||Uncertain|
RISK AND RETURN
The preceding discussion implicitly assumed a single rate of the return to investors, but in practice, there are a variety of real and financial assets with different risk attributes and correspondingly different expected returns. For example, households can save by holding stocks or bonds, or by accumulating equity in their homes. Earlier chapters of this volume examine how the choice between such alternatives may be affected by population aging.
From a macroeconomic perspective, a key question is to what extent an aging population will affect the aggregate risk appetite of the capital market. If older investors are more risk-averse than younger ones, then as the population ages, the price of risky assets may decline, and the required risk premium might rise, relative to historical experience. Several studies (notably Bodie, Merton, and Samuelson, 1992; Jagannathan and Kocherlakota, 1996) have suggested that older households may be less tolerant of investment risk because they cannot offset adverse shocks to the value of their asset holdings by increasing their labor supply. The typical advice of financial advisers to reduce exposure to investment risk with age is consistent with that reasoning.
Several recent analyses suggest that the link between age and the tolerance for investment risk—and its relation to labor income—may be more complicated. For example, Benzoni, Collin-Dufresne, and Goldstein (2007) argue that labor income is a significant source of long-run market risk for young people, and hence they would be expected to be more averse to investment risk than middle-aged households whose lifetime labor income is more certain. They predict a hump-shaped pattern of risky asset holding
over the life cycle, which in aggregate would imply greater risk tolerance when a larger portion of the population is middle-aged. Changing tastes for housing and financial assets over the life cycle may also affect both the level of required returns and the risk premium for risky as opposed to safe financial assets. For example, Bakshi and Chen (1994) suggest that the demand for financial assets may increase as people age and their demand for housing diminishes; this could, in turn, affect financial market returns.
Empirical and Simulation Evidence on Demographic
Structure and Rates of Return
To address the quantitative effect of population aging on rates of return, a number of studies have compared historical returns on financial assets in different time periods or different countries that were characterized by different population age structures. Other studies have used simulation models that incorporate the supply and demand elements that were described above to analyze the size of these effects. Each of these research strategies has strengths as well as weaknesses. Empirical analyses rest on relatively little historical experience with the type of population aging that many developed nations are about to undergo. Simulation modeling depends on many assumptions about difficult-to-measure parameters such as discount rates, the elasticities of substitution between consumption today and in the future, and the extent of openness in cross-country resource flows.
The committee noted in Chapter 2 that if age-specific asset holdings were not affected by population aging, the change in population age structure between 2010 and 2050 would result in a noticeable increase in per capita asset holdings in the United States. Table 8-3 shows mean net worth for households, classified by age of the household head, from the 2007 Survey of Consumer Finances. Because of the skewness of the wealth distribution, age-specific means are typically much greater than medians. For
TABLE 8-3 Household Net Worth by Age of Household Head, 2007
|Age of Household Head||Mean Net Worth ($ thousands)|
SOURCE: U.S. Census Bureau (2012, Table 721).
the 65–74 age group below, for example, the median is $239,400, compared with $1,015,200 for the mean. For analyzing the total supply of savings, however, it is appropriate to focus on means. If it is assumed that these age-specific means and household headship rates will remain constant for the next 40 years, average per capita net worth (all ages, in 2007 prices) would rise from $225,900 in 2012 to $248,100 in 2050. This increase of 9.8 percent would be attributable to the changing age structure of the population. If we just consider the population in the prime working ages 20–64, the ratio of net worth to population aged 20–64 will increase more rapidly (21 percent) during this period because the population aged 20–64 will grow more slowly than the total population in the coming decades.
Making projections of future asset holdings based on current age-related profiles is challenging, because it is impossible to fully disentangle age and cohort effects on wealth accumulation. For example, it is difficult to determine whether a decline in assets between ages 60–64 and 65–69 reflects a movement along an age-wealth profile, or the fact that those who were in the latter age group experienced different labor market and financial market conditions and were consequently less wealthy than their slightly younger counterparts. It is also important to recognize that the wealth distribution is highly skewed, so that average asset levels can be quite sensitive to the holdings of a small group of households.
There are also important age-related patterns in labor supply. The Bureau of Labor Statistics (2012) reports that the fraction of the population aged 18–19 that was employed in 2011 was 36.9 percent. The fraction rose to 60.8 percent for those 20–24 and to 73.8 percent for those 25–34. For those between the ages of 25 and 54, the average was 75.1 percent, with relatively little variation across age groups. After age 55, however, labor market activity begins to decline, with an employment:population ratio of 68.1 percent for those 55–59, 50.8 percent for those 60–64, and 16.7 percent for those over 65. The rise in the employment:population ratio in the intermediate age group is more pronounced for men than for women, who traditionally took time out of the labor force for child raising and who had a lower peak labor force participation rate. Age-related changes in the aggregate labor force could affect the rate of return on assets.
Empirical Analyses of Past Returns and Demographic Structure
Data on population age structure and rates of return, both over time in individual countries and across nations, can be used to examine the correlation between demographic factors and asset market returns. While several studies identify a strong relationship between a particular measure of demographic structure and a particular set of asset market returns, others find little or no association. The research has used a number of different
measures of age structure, and some demographic measures do not seem to be related to return outcomes while others do.
Arnott and Chaves (2011), one of the latest studies in this tradition, examine the empirical relationship between population age structure and stock and bond returns in a number of developed countries. They allow for a relatively flexible relationship between age structure and returns and conclude that rapidly aging countries will experience substantially lower equity returns than other countries. The United States, however, is roughly in the middle of the country group that they study, with only modest return effects. Their analysis presumes that each country’s demography is related to its asset returns, in contrast to the committee’s focus on global aging as the key determinant of rates of return. Brooks (2006), in contrast, studies a number of nations and finds no robust relationship between age structure and asset returns. In fact, in countries with extensive stock market participation, such as Australia, Canada, New Zealand, the United Kingdom, and the United States, he finds that many households continue to accumulate assets well into old age. Poterba (2001) finds that measures of demographic structure have only a weak correlation with asset returns in the United States, with the strongest relationship observed between the price:earnings ratio and the share of the population in middle age. Geanakoplos, Magill, and Quinzii (2004) report that variation in the ratio of middle-aged to young households has predictive power for equity returns in the United States and several other nations. They also develop a simulation model that suggests a substantial decline in the price:earnings ratio for U.S. equities in the decades ahead.
The large variation of findings in the empirical literature is probably due to the relatively slow evolution of demographic variables in the recent past, which means that even when data are available for many years, there may be relatively little effective variation in the explanatory variables.
The simulation literature on the effects of changing demographic structure has included studies of a single economy, best interpreted as representing the global economy with fully integrated capital markets, as well as studies that recognize the different current and prospective demographic structures of various regions of the global economy. Most studies consider a single asset category, which can be thought of as all capital invested in productive uses. The return on such an aggregate capital measure would correspond to a weighted average of the returns that investors earn on stocks and bonds issued by corporations and on their investments in owner-occupied housing and other real estate. The simulation studies suggest that there may be a modest decline in rates of return—between 30 and 100 basis
points1—in response to population aging of the type that will take place in the United States and other developed nations in the next few decades.
A number of simulation studies have also considered how population aging may affect the equity premium—the difference between the expected return on risky assets such as corporate stocks and safe assets such as Treasury bills. Brooks (2004) suggests that when baby boomers retire, there will be an increase in the equity premium. This would translate into a decline in the value of corporate stocks and generate low returns for investors in this cohort. This result is driven by a large sell-off of equities by retiring baby boomers who want to hold less risky portfolios during retirement. Boersch-Supan, Ludwig, and Sommer (2007) reach the same conclusion. As with the effect of demographic change on overall returns, there is some disagreement about the potential effect of aging populations on the risk premium. Geanakoplos, Magill, and Quinzii (2004) and Brooks (2002) both predict a fall in the equity premium. Kuhle (2008) shows that whether the return on risky relative to less risky assets rises or falls depends on the relative price elasticities of the two asset classes. Existing empirical work does not provide definitive evidence on this elasticity.
FINANCIAL MARKET INTEGRATION AND
CROSS-BORDER FINANCIAL FLOWS
The discussion so far has assumed that cross-border financial flows equalize the returns to assets invested in different nations. In such an integrated global financial market, when the aging population in one nation leads to a rising supply of savings and in the associated physical capital:labor ratio in that country, households can invest in other nations and take advantage of the lower capital:labor ratio elsewhere to earn a higher return than the one that would be available if the domestic economy was an isolated entity. If the population of a small, “open” economy grows old but the rest of the world has a stable age profile, there may be very little if any effect on the rate of return earned by its residents—by investing abroad, they can continue to earn the prevailing global rate of return. The open economy setting implies that rate-of-return effects may be small, but it also implies that there may be substantial cross-border financial flows in response to changing demographic structure. These flows may be of independent interest as a macroeconomic phenomenon.
This section describes the patterns of U.S. financial flows that could emerge over the next three decades in a fully integrated world economy. In addition, it highlights the potential importance of restrictions on financial
1A basis point is a unit of measure equal to 1/100th of a percent (i.e., 0.01 percent), often used to describe the percentage change in the value or rate of a financial instrument.
flows when different countries are aging at different rates. The evidence on how global population aging will affect financial flows is based on analysis of historical experience and calculations using simulation models. Bryant’s (2006) analysis suggests that between the 1950s and the mid-1970s, demographic forces were a major factor behind flows of financial capital and direct investment from the Northern to the Southern Hemisphere. However, beginning in the 1970s, he finds that demographic change dampened, rather than augmented, these flows. His analysis suggests that the pattern of recent decades may persist for some time to come.
While the evidence on financial flows from simulation models is somewhat varied, several findings warrant attention. First, the paths of factor prices—the rate of return to capital and the wage rate—as well as aggregate variables such as assets, consumption, and investment, are considerably different under different assumptions about interregional financial flows. Attanasio, Kitao, and Violante (2007) and Boersch-Supan, Ludwig, and Winter (2006) find that population aging has significantly less influence on these macroeconomic variables in an open economy than in a closed economy. In these studies, developed-country wages increase less and rates of return decrease less under the open-economy than the closed-economy assumption. This affects the welfare of different generations differently, as shown by Krüger and Ludwig (2007) and Ludwig, Krüger, and Boersch-Supan (2007). Younger generations profit from the increase in wages, while older generations are likely to suffer from a modest decline in rates of return on assets. These patterns may be partly offset by government transfer programs that reallocate resources across age groups.
Financial flows have a moderating effect because they reduce the relative changes in capital:labor ratios. In these stylized models, initially, savings flow from fast-aging regions to the rest of the world, but this trend is reversed when households in aging economies start to draw down their saving. Yet these model-based predictions do not account for a number of other factors that may affect financial flows, and they also do not fit the recent experience of major financial flows from the developing world to developed nations like the United States. The recent experience underscores that other factors, such as cross-national differences in underlying saving rates, can play key roles. Prospective changes in social insurance programs in some developing nations could alter the demand for precautionary saving in those nations, thereby affecting their national saving patterns and financial flows.
Just as cross-border saving flows could play an important part in offsetting the impact of population aging in a given nation, cross-border labor flows could also matter. Immigrants tend to be younger than native populations, and if immigrant flows are large relative to existing populations, they could affect measures of the population age structure. The demographic
projections that underlie our analysis incorporate forecasts of future immigrant flows to the United States. Since the United States is already at the high end of the developed world in its immigrant inflow, it seems unlikely that the rate of immigration will increase enough in the future to substantially alter the rate of population aging (see Chapter 3).
The comparisons between simulation analyses with and without open capital markets underscore the significance of capital controls and other factors that might affect cross-border financial flows in determining the effects of population aging on rates of return. At the moment, the United States is not experiencing financial outflows and is instead a destination for financial flows. The prospect of any restriction on outbound investments therefore does not seem like a near-term possibility. Over the longer term, however, if global economic conditions shift and developed nations with aging populations assume a larger role as asset suppliers, public policies that affect financial flows could matter.
An additional factor that may affect cross-border financial flows is the global pattern of debt:GDP ratios. The trend in this ratio is quite different in developing and developed nations. The International Monetary Fund (2011) observes that the debt:GDP ratio in emerging market economies has fallen since 2006 and is projected to fall further by 2016, while the analogous ratio in developed countries is rising sharply. For the world as a whole, the ratio of government debt to output is expected to change relatively little from 2006 to 2016.
“HUMAN CAPITAL DEEPENING”: HOW MUCH
OFFSET TO LABOR FORCE DECLINE?
The foregoing discussion of how population aging affects the labor force treated the age-specific pattern of labor market activity as fixed, even though there are a number of margins on which adjustment is possible. These include an increase in labor force participation at old and young ages, an increase in female labor force participation, which is still below male labor force participation in most developed nations, and an increase in the quality of labor through more training and education (human capital deepening). Increases in the global supply of labor, whether through expanded labor market participation or through increases in the effective per capita supply of labor, would tend to raise rates of return. The migration of labor from rural to urban economies in developing nations such as China and India is one of the important factors that may continue to influence the effective level of global labor supply.
Both physical and human capital serve to transfer wealth over time and even between generations. Savings and human capital accumulation are therefore closely linked. Worker investment in human capital is a fac-
tor that may attenuate the effects of population aging on asset returns. For example, if laborers are scarcer in an older society than in a younger one, and if young workers observe rising returns to supplying labor and take actions to enhance the value that they can deliver to an employer per hour of work, the resulting deepening of human capital will partly offset the rise in the capital:labor ratio and the associated change in the marginal product of physical capital.
Some broadening and deepening of the labor force is likely in response to the wage changes that will be associated with population aging. Such reactions are observed in simulation results from models calibrated to the U.S. economy. They show that increasing capital intensity as a consequence of aging will increase human capital investment, which in turn increases the productivity of physical capital. Ludwig, Schelkle, and Vogel (2010) find that the strength of that mechanism will depend, among other things, on the relationship between private investments in human capital and subsequent after-tax, net-of-transfer program returns.
It is also possible that changes in the labor markets of other nations, particularly those in the developing world, could affect the productivity of capital invested in the United States as well as the wages of U.S. workers. There are few, if any, quantitative studies of the cross-border effects of human capital accumulation. Because human capital levels across Organisation for Economic Co-operation and Development (OECD) countries are already high and rather homogeneous, spillover effects are probably small and they are likely to be difficult to measure. On the other hand, further human capital deepening in emerging countries can, and probably will, play a significant role in determining the capital:labor ratio of those countries.
The committee is not aware of any macroeconomic studies calibrated to developing nations, but it is possible to use microeconomic studies to assess the potential consequences of increases in educational attainment in these countries. Psacharopoulos and Patrinos (2002) report that the social return on an additional year of secondary schooling is about 10 percent in rich countries, 13 percent in middle-income countries, and almost 16 percent in poor countries. Private returns are even higher. Average years of schooling are 9.4 for rich countries, 8.2 for middle income countries, and 7.6 for poor countries. This suggests that in many nations, there is still room for improvement in human capital and for associated reductions in the capital:effective labor ratio. Such human capital deepening would raise the marginal physical product of capital.
As economically important developing nations that are well integrated with the global economy, such as China and India, embark on growth paths with rapidly growing educational attainment, they will play an increasingly important role in determining the global capital:labor ratio. Restuccia and Vandenbroucke (2011) present some evidence on prospective catch-up
trajectories, suggesting that relatively rapid human capital deepening is possible in a number of developing nations. The trajectory of human capital acquisition in these nations will be a potentially important influence on the global rate of return to capital in the decades ahead.
Human capital investments in developing nations can moderate the decline in the rate of return available to U.S. investors as a result of global population aging. There are likely to be substantial differences across developing nations in the rate of human capital growth. In countries with relatively low birth rates and an aging population (e.g., China), the process of raising average human capital per worker by educating young workers can take a very long time. However, in countries such as India with rapid population growth and a large flow of young entrants to the labor force, raising the average education level in the population can take place over a shorter time period and potentially proceed for a longer time period.
Much of the foregoing discussion treats assets as though they are competitively traded in an integrated world economy. While that might be an apt description of the markets for stocks and bonds, it does not apply to assets such as land or owner-occupied housing. Such housing, in particular, looms large on the balance sheets of older households in the United States. Poterba, Venti, and Wise (2011) report that for many older households, Social Security and owner-occupied housing are the primary sources of retirement security. Housing equity may provide an important source of financial support in response to adverse shocks. One of the first studies of demographic variation and asset prices (Mankiw and Weil, 1989) focused on how a shifting age structure might affect housing demand and ultimately house prices. That research suggested that an aging population would lead to lower housing demand and falling house prices.2 While the two decades since that analysis have drawn attention to many other factors that may affect housing markets, a feature that distinguishes owner-occupied housing from stocks, bonds, and many financial assets is that it must be owned domestically. This results in a tighter linkage between a nation’s population age structure and the level of demand for such housing than, for example, between its age structure and the demand for corporate equity claims.
In nations with declining population numbers as well as aging populations, such as some countries in Europe, the demand for housing will
2The Mankiw and Weil analysis spurred considerable methodological discussion and critique in the 1990s (e.g., Hamilton, 1991; McFadden, 1994; Green and Hendershott, 1996) and continued to motivate research in subsequent decades (e.g., Federal Reserve Board of San Francisco, 2005; Takats, 2010).
decline in future decades. This will lead to a smaller total value of housing assets. This can occur through a decline in new construction, depreciation of the existing stock of housing, and/or a drop in the price of existing houses. The drop in house prices is a signal to builders to reduce the flow of new construction. Land prices are also likely to decline, since land, unlike residential structures, does not depreciate. The effects in the United States are likely to be more modest, because the U.S. population is projected to continue to grow through the next century. The effects on house and land prices are also likely to vary substantially across regions and even metropolitan areas.
Population aging in the United States, as well as global population aging more generally, is likely to have only a modest effect on rates of return. There is substantial uncertainty about the magnitude of this effect, as well as about the channels that will prove to generate the strongest effects. Some asset classes may be affected more than others. For example, owner-occupied housing in areas with rapidly aging populations may experience a decline in values, while land in the central business district of cities with rapid population inflows may rise in value. Cross-border capital flows and immigration could influence the ultimate effects. Moreover, there are close links between policies discussed elsewhere in this report, notably fiscal policy, and the long-run effect of population aging on rates of return.
There are important links between the financial markets in different nations, so the committee’s focus for studying population aging and rates of return is on an integrated global economy. In considering how U.S. population aging may affect the rates of return available to U.S. investors, therefore, it is important to examine a number of features in the global economy such as the prospective growth rate of currently emerging economies and the degree to which immigrant labor can move from savings-poor to savings-rich nations. Given the uncertainties in the many forces that could strengthen or attenuate the effect of population aging on rates of return, the committee concludes that it is reasonable to assume that such an effect will be modest. It recognizes that financial markets in recent years have become more volatile, and its conclusion in part reflects the view that future volatility is likely to be dominated by nondemographic factors. If volatility remains high, this could affect the asset allocation choices of older households. It is important to recognize, however, that there are scenarios in which rate-of-return effects could be substantial and in which they would have significant effects on the retirement income of the future elderly population. Volatile financial markets may increasingly challenge older households with the need to make sound financial decisions.
PHYSICAL RETURNS, ASSET PRICES, AND THE
TERM STRUCTURE OF FINANCIAL RETURNS
In a competitive financial market, the expected rate of return varies over time in a manner that equates the supply of, and demand for, savings. Savers directly invest in financial assets such as stocks and bonds that are issued by companies, they hold bonds issued by governments, and they also invest indirectly through financial institutions such as banks and mutual funds. Firms use the money raised from households and intermediaries to invest in physical assets—property, plant, and equipment—that are inputs into the production process. Over time, the income generated from such physical investments, net of expenses such as wages and maintenance, is used to pay interest and principal to bondholders and to pay dividends to stockholders or to repurchase their shares.
A simple example may help to illustrate the relationship between returns on financial assets, physical assets, and asset prices, and how those quantities are affected by the demand for and supply of savings. Consider a machine that produces $10 worth of output in excess of operating expenses every year forever. A firm finances the purchase of this machine by issuing a share of stock. The market value of the stock depends on the rate of return available to investors on similar investments. For example, if the expected rate of return is 5 percent over the next year and also for every year in the future on similar assets, then the price of the share would be $200. This is because the present value of a perpetual claim on $X, discounted at r percent per year, is $X/r. This price is consistent with the required return for investors: a 5 percent return on $200 is $10 per year.
If the cost of building another machine exceeded $200, no new machines would be produced because investors could make a higher return on alternative investments. Alternatively, if the production cost were less than $200, manufacturers would find unlimited demand for their products, since buyers could earn more than the 5 percent available on alternative investments. Thus, new machines would be put into production until the cost of building the units rose or the value of their output fell to the point where any additional machine built would earn 5 percent. Hence the forces of supply and demand for capital cause the expected return on any incremental unit of physical investment—the “marginal physical product of capital”—to equal the expected return on financial assets. The prices of existing financial and physical assets also adjust so that expected rates of return are equalized across investments. Those forces operate internationally; because investors seek the best investment opportunities at home and abroad, expected returns tend to be similar around the world. A more precise statement is that expected risk-adjusted returns are equalized; invest-
ments with higher levels of risk that cannot be reduced by diversification have higher expected returns than less risky investments. Tax differences across jurisdictions, and other market frictions, can also generate differences in rates of return across places.
The expected rate of return also depends on consumer preferences, which determine the supply of savings. When the aggregate supply of savings increases, for example because in an older society more people hold substantial retirement savings, the demand for existing assets rises. This puts upward pressure on asset prices and lowers expected returns, thereby encouraging greater investment in physical capital. In the foregoing example of a machine, if the expected rate of return were to fall unexpectedly from 5 percent to 4.5 percent, then the price of existing machines would immediately increase from $200 to $222 (= $10/.045). The higher price would encourage additional investment in new machines, which would continue until the cost of producing them rose or the value of their output fell to the point where the expected return on an incremental investment or the purchase of an existing machine was 4.5 percent.
Population aging is a predictable process that is unlikely to cause sudden changes in asset prices, but it may affect the time pattern of returns that investors expect to earn in the future. Investors may, for example, expect to earn different rates of return in different years. Continuing with the previous example, imagine that investors expect the rate of return on the machine to be 5 percent per year for the next 5 years, then to fall to 4.5 percent annually for the 5 years after that, and then to stay at 4 percent for the indefinite future, reflecting the higher amounts of per capita savings held by the older population at that time. The price of the machine would rise each year during the first 10 years, so that the sum of the $10 profit from the machine’s production, plus the capital gain, would generate the return demanded by investors. It would reach $250 (= $10/.04) in year 10. Table 8-A-1 shows the price path that would provide investors with a total return—the combination of the $10 profit and the associated capital gain on owning the machine—that would equal their required return. In year 9, for example, when investors require a 4.5 percent return, the price would begin at $248.80, and the profit of $10, plus the $1.20 appreciation of the machine, would generate a return of 4.5 percent: [(10 + 1.20)/248.80 = .045]. Note that the price path rises gradually over time and levels off after year 10, at which point returns remain constant forever.