FRBSF Economic Letter
1997-19 | June 27, 1997
Since 1978, the share of households’ financial assets held in depository institutions has declined steadily from about 39% to about 17%. Previous research suggests that an important contributing factor may be the shifting age structure of the population. In this Economic Letter, I investigate this possible link and find that there appears to be more than just a simple demographic shift behind the relative decline in deposits. This leaves open the possibility that changes in preferences and financial innovation affecting the supply and demand of deposits and other financial instruments have led to the reallocation of household asset portfolios.
A pure measure of all households’ financial assets is not available. However, the Board of Governors’ Flow of Funds Accounts (FFA) does report household financial assets combined with the financial assets of nonprofit organizations, and, according to available estimates, they make up roughly 95% of total assets in that combined category. Therefore, I refer to household and nonprofit assets simply as household assets.
The components of household financial assets are deposits and currency (domestic time and savings deposits, domestic checkable deposits and currency, and foreign deposits); money market mutual funds (MMMFs); stocks and bonds, including those held through mutual funds; and assets held through bank pension funds and life insurance policies. Also included are assets held in trust in bank personal trust accounts. The FFA reports checkable deposits and currency together. Currency likely is a very small percentage of household financial assets, and I therefore refer to household deposits and currency as just household deposits. Following previous research on the share of stocks and bonds in household financial assets, the above definition of household financial assets excludes non-corporate equity, security credit, and miscellaneous assets.
At the beginning of 1978, deposits constituted the largest single category of household financial assets (38.8%), followed by trusts, pensions, and life insurance (32.7%), stocks and bonds (28.4%), and finally MMMFs (0.2%). (Percentages do not add to 100 due to rounding.) Between 1978 and 1996, household deposit growth slowed relative to the growth of stocks and bonds, and, by the first quarter of 1996, deposits and stocks and bonds had reversed their positions in the household asset ranking: stocks and bonds accounted for 42.3% of household assets, while deposits accounted for only 16.9%. The shares of trusts, pensions, and life insurance and MMMFs had grown, too, to 38.1% and 2.7%, respectively. Clearly, however, the net changes in asset shares indicate a marked shift to stocks and bonds from deposits.
Economist Donald Morgan (1994) investigated the increase in the share of stocks and bonds in household assets and hypothesized that this phenomenon may be closely linked to the upward shift of the age profile of the population that has taken place in recent years. As Morgan points out, young households are starting families and borrowing to buy houses and goods. He posits that when young households channel funds to savings, they are likely to invest in safe, short-term assets such as bank deposits, which can be converted readily into cash. But, as households age, their investment preferences change, and they are willing to accept less liquidity and the greater short-term volatility of stocks and bonds in return for their higher long-term returns.
Consistent with this view, Morgan finds evidence for the 1953 to 1993 period that suggests that the proportion of total household assets that is held in stocks and bonds may indeed be highly correlated with the percentage of the working-age population (at least 16 years old) that is at least 35. This in turn suggests that changes in the proportion of household assets that is held as deposits also are correlated with demographic shifts.
Figure 1 shows data similar to those used by Morgan. The figure shows, for the years 1952 to 1996, the percent of household assets that is held as deposits, the percent of household assets that is held as deposits or MMMFs, and the percent of the working-age population that is not in their middle years, that is, not between the ages of 35 and 65. In recent years, the percent of the working-age population under 35 has decreased, and the percent of the working-age population under 35 or at least 65 also has decreased.
The age cohorts used here differ from the “under 35” and “35 or over” groupings used by Morgan. I single out those 65 or over as well as those under 35, because, as the household survey data presented in the next section indicate, both of these groups stand out as having relatively high shares of deposits and MMMFs in their asset portfolios. As discussed above, young people may tend to shun stocks and bonds because they have limited means but face large and frequent expenditures associated with establishing a new household. Retired people may have a high demand for liquidity because they have lost the relatively certain return of their labor income.
According to the FFA, the household sector first started holding MMMFs at the beginning of 1974. I include MMMFs in Figure 1 because their risk and liquidity characteristics are similar enough to those of deposits that households would treat them as at least partial substitutes for deposits. Therefore, there likely is a stronger relationship between the percent of the population that is either relatively young or relatively old and the percent of household assets that is held as either deposits or MMMFs than there is between the population variable and the deposit share alone. Note from Figure 1 that, as mentioned above, the growth of MMMFs can account only partially for the decline of the deposit share; including MMMFs with deposits delays the sharp decline but does not come close to eliminating it.
It appears from Figure 1 that there may indeed be a correlation between the percent of the population under 35 or at least 65 and the percent of household assets held as deposits or MMMFs. Over a 44-year period, the two variables have moved through two downswings and one upswing together. However, if there was one relationship for the first 34 years or so, there may have been a different relationship since about 1986, when the two series, as depicted, began to diverge. This possibility and certain other statistical ambiguities associated with the two series mean that, despite the appearance of a possibly meaningful correlation, it remains quite unclear from the FFA data whether the decline in the share of the population not in their middle years could have caused any significant portion of the decline in the deposit plus MMMF share in recent years.
In order to investigate further whether recent demographic shifts could have accounted for the shrinkage of deposits, we can turn to individual household survey data from the Board of Governors’ Survey of Consumer Finances (SCF). SCF data are available for a much shorter time period than FFA data. However, for the relatively short time period available, the SCF, like the FFA, does show a decline in the proportion of household assets being held as deposits or money market mutual funds. For example, according to the SCF, between 1983 and 1992, this proportion fell from 29.2% to 24.5%. (The FFA shows a sharper decline between 1983 and 1992, from 35.8% to 24.4%.)
Figure 2 shows the proportion of household assets that is held as deposits or MMMFs for various age groups for two years in which the SCF was conducted, 1983 and 1992. Focusing on 1983, Figure 2 confirms that households headed by middle-aged individuals (35-65) on balance have a lower proportion of deposits and MMMFs to assets, compared to households headed by younger (under 35) and older (at least 65) individuals. Given this pattern, the increase in the percent of the population in their middle years should have contributed something to the decline in the deposit and MMMF share of household assets for the population as a whole.
By making certain assumptions, we can estimate how much of a contribution. Holding constant at their 1983 levels the percent of household assets that is held as deposits or MMMFs for the various age groups and the relative per capita wealth of middle-aged and non-middle-aged households, the 1983 to 1992 decrease in the percent of the population that is not middle-aged would have accounted for only 16 basis points of the actual 1983 to 1992 SCF decline of 470 basis points in the deposit and MMMF household asset share.
What else happened between 1983 and 1992 to reduce the overall deposit and MMMF household asset share? As seen in Figure 2, according to the SCF, middle-aged households markedly reduced their deposit and MMMF asset shares between 1983 and 1992. This reduction played a much larger role than the population shift because per capita wealth for middle-aged households is much higher than for non-middle-aged households. In other words, the changes in the way middle-aged households allocate their assets have mattered more than the relatively small increase in the percentage of middle-aged households.
Although aggregate data suggest that recent shifts in the age profile of the population may have made an important contribution to the decline in the deposit share of household assets, individual household survey data indicate that demographic factors may have played a relatively minor role. What appears to be more important are the changes over time in middle-aged household portfolios. Such changes may be due to a decrease in the demand for deposits by banks that are increasingly moving assets off of their balance sheets through securitization and a shift in individuals’ preferences that is not due to aging. In addition, improvements in computer technology and telecommunications have contributed to a decline in the relative cost of non-deposit investments through discount brokerage firms and mutual funds and thereby to the decline in the deposit share of household assets over the past 19 years.
Morgan, Donald M. 1994. “Will the Shift to Stocks and Bonds by Households be Destabilizing?” Federal Reserve Bank of Kansas City Economic Review. Second quarter, pp. 31-44.
Opinions expressed in FRBSF Economic Letter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System. This publication is edited by Sam Zuckerman and Anita Todd. Permission to reprint must be obtained in writing.
Please send editorial comments and requests for reprint permission to
Attn: Research publications, MS 1140
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120