October 2003     Vol. 4, no. 10

Family wealth across the generations

Raj K. Chawla and Henry Pold

For some, the accumulation of wealth is life's primary scorecard. For others, a cushion of savings and investments can help smooth out spikes and troughs in employment earnings or household expenditures. For most, building up sufficient assets to live comfortably in retirement is a cornerstone of family finances. Governments, too, support these goals by offering tax incentives for retirement and education savings, as well as exempting principal residences from capital gains tax.

Technically speaking, wealth is a stock—accumulated assets at a point in time—as opposed to a flow—regular earnings from a job, for example (Augustin and Sanga 2002). Lottery winnings and stock market bubbles aside, most wealth is accumulated over long periods by spending less than one earns and compounding investment returns on past savings. Over time, the vagaries of the economy can both help and hinder wealth accumulation, often with different effects for different types of families.

The period from 1984 to 1999 witnessed dramatic economic fluctuations, beginning with recovery from a recession in 1981-82 and including a plunge back into another in 1990-91. The 1981-82 recession was the more severe. Real gross domestic product (GDP) fell by 2.6%, unemployment rose to 11.0%, and the bank rate hit 17.93%. The 1990-91 recession, on the other hand, saw a GDP drop of 1.4%, unemployment of 10.3%, and a bank rate of 13.04% (Statistics Canada 2002). During recessionary periods, family incomes usually drop. The median pre-tax family income (in 1999 dollars) dropped by 3.7% during the 1981-82 recession and by 5.3% during the 1990-91 recession.  1 

The business cycle and other economic developments during the 1984-1999 period obviously affected family incomes and wealth. Real pre-tax family income increased steadily from 1984 to 1989, reaching $58,100. Mean incomes then dropped until 1997. In 1998, families recaptured their 1989 income level.  2  Meanwhile, unemployment plummeted from 11.3% in 1984 to 7.6% in 1999 and the bank rate from 11.31% to 4.92%. The steady drop in the bank rate may have encouraged more families to borrow for a home or to invest—total household credit increased from $161 billion in 1984 to $578 billion in 1999. Mortgages on owner-occupied homes accounted for almost three-quarters of the increase. House prices rose dramatically as the new housing price index (1992=100) jumped from 67.8 in 1984 to 101.0 in 1999.

A steadily improving economy also provided an impetus to invest in stocks. The volume of shares traded on the Toronto Stock Exchange jumped from 2.1 billion in 1984 to 29.3 billion in 1999; the price-earnings ratio peaked at 88.51 in 1993. Furthermore, to encourage saving for retirement and children's postsecondary education, amounts eligible for tax sheltering increased—from $7,500 in 1990 to $15,500 by 2005 for registered retirement savings plans (RRSPs), and $4,000 per year to a maximum of $42,000 (supplemented with a maximum yearly grant of $400 from the federal government) for registered education savings plans (RESPs).  3 

Not all families benefited equally. Among those who gained or lost, which asset and debt components were affected? How did the overall family balance sheet change? Did the overall distribution of wealth become more or less equal? To answer such questions, two approaches are possible. One is to compare wealth and its components by using 'similar' families from the 1984 and 1999 surveys (see Data sources and definitions). Groups can be defined by age or education of the major income recipient, type of family, or other characteristics of interest. The fundamental problem with this approach is that one would be comparing different families at different times (embodying the effects of demographic and economic trends). The only thing in common would be the classification characteristic and that, too, 15 years apart. Using age of the major income recipient, for example, the age-wealth profiles of families in 1984 and 1999 were similar (Chart A). In both years, wealth increased with age, reaching its peak during the pre-retirement years (the 55 to 64 age group) and then declining. The only observable difference was that 1999 mean wealth was equal or higher than in 1984.

Another approach is to use cohorts based on the age of the major income recipient (see Family cohorts). This method shows that over the 1984-1999 period, mean family wealth rose for families with a major income recipient under 45 in 1984 and dropped for those with one aged 45 to 54. The wealth held by each cohort in 1999 was its wealth in 1984 plus that added over the next 15 years. Not surprisingly, new wealth as a proportion of total wealth in 1999 decreased from the 1960s cohort to the 1930s cohort—from 89% to just 21% (Chart B). The 1920s and pre-1920 cohorts lost wealth over the period.

Asset and debt changes

1960s cohort

In 1984, this cohort of young families, nearly half of them unattached individuals and only one-fifth living in an owned home, had total wealth of $31 billion (Table 1). By 1999, 61% were living as two-spouse families, and 60% owned a home (Table 2).  7  Their mean income had increased from $28,000 to $54,300, so more were also able to save—the proportion with RRSP or RESP holdings jumped from 12% to 67%. Families in this cohort also expanded their portfolios into non-registered investments as the proportion with business interests rose by 16 percentage points and the proportion with stocks and mutual funds by 17 points. These three assets plus home equity accounted for 86% of their newly amassed wealth of $243 billion; home equity alone accounted for 35%. While this group accounted for 56% of new non-financial assets over the 1984-1999 period, they were responsible for just 18% of new financial assets.

1950s cohort

Families in this cohort added the most new wealth—$372 billion. Their mean income rose from $45,300 to $63,700. More owned a home (up 26 percentage points) or a business (up 13 points), and had savings in registered plans (up 44 points), or stocks and mutual funds (up 12 points). These four assets contributed 90% of the wealth amassed by this cohort between 1984 and 1999; home equity alone accounted for 33%. These families possessed 57% of the total new non-financial assets and 31% of financial assets.

1940s cohort

Mean family income in this cohort rose marginally, from $58,000 in 1984 to $65,000 in 1999. The rate of homeownership increased only 7 percentage points, as did business ownership. On the other hand, registered savings jumped 37 points, and stocks and mutual funds 13 points. Nearly 10% of homeowning families had discharged their mortgage, with the released funds likely saved in registered plans as well as stocks and mutual funds, which accounted for 55% of the new wealth for this cohort. This cohort owned more of the new financial assets (36%) than non-financial assets (23%). Of the added value of stocks and mutual funds between 1984 and 1999, these families accounted for 30%.

1930s cohort

Because many major income recipients in this cohort would have retired by 1999, mean family income fell from $61,200 to $42,500. The proportion living in mortgage-free homes went up 23 percentage points, and a small proportion had wrapped up their business activities. Some business equity was likely converted into financial assets. Registered savings plans, and stocks and mutual funds accounted for 125% of the total added wealth for this cohort.  8  One-fifth of the total additional value of stocks and mutual funds belonged to this cohort.

1920s cohort

Most families in this cohort had likely experienced the retirement of their major income recipient. Their mean family income dropped from $49,500 in 1984 to $36,100 in 1999, and mean wealth from $210,300 to $199,000. Some of the loss could be attributed to demographic change, since a portion of two-spouse families became unattached individuals (likely with the death of a spouse), and homeownership dropped by 5 percentage points. The proportion of families with no mortgage went up 10 points, and business activity dropped 10 points. Overall, the group lost $48 billion from its 1984 wealth, largely because of drops in business equity and registered savings plans.  9  On the other hand, they added $20 billion to their stock and mutual fund holdings—some 12% of the total additional value. Families in this cohort accounted for 26% of the drop in liquid assets over the 1984-1999 period.

Before 1920 cohort

Examining this cohort's change in wealth is tantamount to looking at the change in the wealth situation of families considered elderly in 1984. Even though their mean income moved up marginally from $25,600 in 1984 to $27,300 in 1999 (largely because of indexed government transfer payments) and mean wealth from $140,700 to $183,600, their aggregate wealth fell $125 billion.  10  Most of the loss in wealth (which excludes $4 billion held in annuity plans) reflected declines in home and business equity and financial assets (except stock and mutual fund holdings, which gained $5 billion). With aging, families tend to wrap up business activities. Some sell their home (either because of poor health or a lack of resources to maintain a home) and move into rental accommodation (40% in this cohort were renting in 1984 and 42% in 1999). Over time, this cohort also went through demographic changes as two-spouse families declined and unattached individuals increased (from 50% to 63%).

Wealth of 1984 cohorts in 1999

Although income and wealth are strongly associated, they do not necessarily move in the same direction for all families over time. For example, a family's income may drop in retirement, but its wealth may still increase because of a rising market value for their home. This may, in turn, result in a higher wealth-to-income ratio—an indicator of economic well-being. For each dollar of income, the pre-1920 cohort had $5.49 of wealth in 1984, rising to $6.72 by 1999. For the 1960s cohort, on the other hand, the wealth-to-income ratio moved from $1.15 to $2.04. The 1930s cohort had the largest increase—from $3.31 to $6.55. On the basis of the wealth-to-income ratio, the 1930s cohort appears to have fared the best.

The 1940s cohort had the highest mean wealth ($291,600) in 1999, the 1960s cohort had the lowest ($110,900). This pattern is consistent with the well-known relationship between wealth and life cycle—wealth is low for younger families and peaks in the pre-retirement years when major income recipients are in their late 50s or early 60s. Mean wealth in 1984 was highest for families in the 1920s cohort ($210,300) and lowest for those in the 1960s cohort ($32,300). Despite all the changes in asset holdings and demographics, the range in mean family wealth from the 1960s cohort to the pre-1920 cohort did not change much-$178,000 in 1984 compared with $180,600 in 1999.

Over the 1984-1999 period, the 1940s cohort made the greatest absolute gain ($153,900) in mean wealth, whereas the 1960s cohort gained the most in relative terms—244%. The sources of change in wealth differed between the various cohorts. For the 1960s cohort, most (71%) of the change arose from the rates of ownership of assets and debts, whereas for the 1930s cohort, it came from the amounts (86%) within asset and debt categories. These differences confirm that the process of building wealth by solidifying assets and reducing liabilities is much stronger during pre-retirement years.

Family balance sheets

The overall mix of wealth held by families changes as the major income recipient approaches retirement. The 1940s cohort had 66 cents of every dollar of assets in 1999 in non-financial assets (such as a home, vehicles or business equity) and 34 cents in financial assets, compared with 86 cents and 14 cents in 1984. For the 1960s cohort, non-financial assets constituted 84% of assets in 1984, which dropped to 79% by 1999. For the oldest cohort, the corresponding proportions were 65% and 53%. In fact, the pattern seems to be universal: As a family ages, non-financial assets drop as a proportion of total assets, and financial assets push steadily upward (Table 3).

For all cohorts, market value of an owner-occupied home was the major non-financial asset, becoming more valuable over time. Business equity was the second most important, but its representation in total asset holdings fell in all cohorts; the greatest drop, 16 percentage points, was experienced by the 1930s cohort. With the recession and economic recovery between 1984 and 1999, some families in this cohort wrapped up their businesses (the ownership rate fell from 20% to 14%) and likely converted part or all of the equity into financial assets. A similar pattern prevailed for the 1920s cohort.

The composition of financial assets also varied by cohort. For example, over the 1984-1999 period, the share of liquid assets declined for the 1960s cohort, while registered savings plans and stocks and mutual funds increased. In the 1920s and pre-1920 cohorts both liquid assets and registered savings declined as stocks and mutual funds increased.  11  For example, while stocks and mutual funds accounted for only 10% of the financial assets of the elderly in 1984, they jumped to 32% by 1999—a reflection of a booming stock market that encouraged investment in risk-bearing market equities. The proportion of holdings in stocks and mutual funds rose steadily across the four oldest cohorts—from between 1% and 3% in 1984 to between 9% and 15% in 1999.

An increase in the share of financial assets over a family's life cycle results not only from rising income but also from declining debts such as mortgages and other consumer loans. While the debt-to-asset ratio increased for the 1960s cohort—from 24% in 1984 to 33% in 1999—it declined for all older cohorts. Families in the 1950s cohort owed 30 cents for every dollar of assets in 1984, dropping to 19 cents in 1999. And debt was even less apparent among older cohorts—the 1920s cohort owed just 6 cents per dollar of assets in 1984, which they reduced to 3 cents over the 15-year period.

Wealth distribution

By following specific cohorts over time, the expected pattern is an upward shift in the wealth distribution  12  countered only by some erosion of assets near the end of the life cycle. So, for example, the proportion of families in the 1960s cohort with wealth of less than $50,000 (Table 4) fell from 85% in 1984 to 54% in 1999. Similarly, the proportion of the 1950s cohort with less than $50,000 dropped from 65% to 36%. These families improved their financial situation as more bought homes and engaged in business interests, thereby moving into higher wealth groups. Such upward shifts were less pronounced for older cohorts—20 percentage points for the 1940s, 6 points for the 1930s, and 9 points for the pre-1920. On the other hand, the 1920s cohort—the richest in 1984—witnessed shifts into the under $50,000 category and out of the $500,000 or more group.

This upward shift in the proportion of families with relatively high levels of wealth, with the exception of the 1920s cohort, corresponds to an increasing share of the wealth holding for families with wealth of $500,000 or more (see Millionaire families). Overall, the proportion of families with $500,000 or more in net wealth doubled between 1984 and 1999, while their share of wealth increased by almost 40%.

Distribution of wealth became more skewed

Since wealth accumulation moves families into higher wealth categories over time, the distribution of wealth may indeed become more concentrated among the richer members of a cohort. This results in a positive coefficient of skewness—the greater the coefficient, the more skewed the distribution. Similarly, one might expect a higher degree of skewness in older cohorts compared with younger cohorts at a point in time.

In 1984, the distribution was most skewed for elderly families (pre-1920 cohort) and least for young families (1960s cohort). This conforms to expectations, as do increases in skewness within the four younger cohorts over time. However, skewness dropped in the two oldest cohorts—quite dramatically in the pre-1920 cohort—indicating some countering influences later in the life cycle. As a result of these changes, there was no clear trend in skewness across age groups in 1999, the most prominent feature being a spike in skewness for the 1950s cohort.

Another characteristic of a right-tailed skewed distribution is that its median value is always less than its mean (which is affected by the extreme values). The median will move up if families move from lower to higher wealth groups over time. For the 1960s cohort, for instance, median wealth jumped from $3,100 in 1984 to $40,500 in 1999—a growth of 1,200%. Unattached individuals forming two-spouse families and increased home and business ownership were responsible for the gains. On the other hand, the median wealth of other cohorts (except the 1920s cohort) increased between 280% (1950s cohort) and 25% (1930s cohort). The median wealth in the 1920s cohort fell from $129,100 in 1984 to $125,000 in 1999 (-3%) as some families moved from an owned home to rental accommodation, wrapped up business interests, or liquidated some financial assets as the major income recipient, who was 55 to 64 in 1984, aged.

Wealth inequality decreased for some cohorts

Wealth inequality decreased most for the 1960s cohort. These families, mostly renters with relatively low incomes and wealth in 1984, improved their wealth situation by purchasing homes and starting businesses. The Gini coefficient, a measure of inequality, fell by 17% for this cohort (from 0.891 in 1984 to 0.740 in 1999). Inequality also dropped substantially for the 1950s cohort (-6%), and marginally for those born in the 1940s and before 1920. On the other hand, inequality increased among those born in the 1920s and 1930s. The former had much larger gains in financial assets, whereas the latter saw some families shifting from higher to lower wealth groups (likely because of moving to rental accommodation). As a result, overall inequality in the distribution of wealth dropped only marginally between 1984 and 1999, falling less than 1%, from 0.692 to 0.686.  13 

How does home equity affect wealth inequality—since such equity accounted for one-third of the total new wealth created by families between 1984 and 1999? For all cohorts, home equity reduced wealth inequality. In 1984, the reduction was smallest (-9%) in the 1960s cohort (largely because of the low rate of homeownership) and greatest (-19%) in the 1940s and 1930s cohorts. In 1999, the reduction was still smallest (-13%) in the 1960s cohort but it was greatest (-20%) in the 1920s cohort. Overall, the influence of home equity remained the same.

Summary

In the absence of the longitudinal data, this study examined changes in family wealth using the 1984 Survey of Consumer Finances and the 1999 Survey of Financial Security. Families with a major income recipient born in the 1960s gained most of the new wealth created between 1984 and 1999, largely because of demographic changes, home purchases, and business formation. On the other hand, cohorts born earlier than 1930 lost a portion of the wealth they held in 1984 (net of any savings in RRIFs).

The home remained the major asset held by families in all cohorts, but the percentage distribution of family assets varied both between and within cohorts. Financial assets as a proportion of total assets grew and liability decreased as families grew older. The younger cohorts carried most of the debt liability—largely attributable to mortgages.

The 1984-1999 period witnessed significant growth in stock market activity and changes in provisions of various tax-sheltered savings plans. Families in the 1940s cohort benefited the most, followed by those in the 1950s cohort. These two cohorts held almost two-thirds of the total additional savings in registered plans, and more than half the additional value of stocks and mutual funds. However, the wealth-to-income ratio for the 1930s cohort rose the most.

Although the distribution of wealth became more skewed among the younger cohorts, wealth inequality remained almost unchanged; it decreased for the 1960s, 1950s, 1940s and pre-1920 cohorts and increased for the 1920s and 1930s. Home equity generally reduced wealth inequality, but its effect was most pronounced for families in the 1960s cohort and least for those in the 1950s cohort.

Notes

1. Not all families are equally affected by recessions. For instance, families with relatively higher incomes and savings can make economic gains by investing their savings at the prevailing higher interest rates. On the other hand, incomes of those with government transfers as their major income source are protected since these are adjusted by the rate of inflation.
2. Since the late 1990s, families have experienced gains in post-tax income because of tax reductions introduced by the federal and several provincial governments.
3. It is beyond the scope of this article to detail all the developments in taxation of income, lifetime capital gains and dividends; rising levels of tax-sheltered savings; creation and administration of different trusts; and provisions to facilitate homeownership-all aimed at helping families to create more wealth.
4. The Survey of Financial Security interview questionnaire, Catalogue no. 13F0026MIE-2001001, is available free on the Statistics Canada Web site at www.statcan.ca/cgi-bin/downpub/research.cgi.
5. The more comprehensive concept of family wealth in the 1999 SFS includes savings in employer pension plans, registered retirement income funds, annuities, the value of contents of principal residence, and other collectibles and valuables. These five assets, excluded from the concept of wealth used in this paper, constituted 29% of 1999 family wealth (Chawla 2003, p. 23).
6. In 1984, there were 9,500,000 family units worth $1.2 trillion; by 1999, there were 12,200,000 units with net worth of $2.2 trillion. Some 1,800,000 families with a major income recipient aged 15 to 29 with wealth of $67 billion, as well as 800,000 families with a major income recipient who had immigrated to Canada after 1984 with wealth of $68 billion, were excluded from the 1999 data. The 9,700,000 families in 1999 represent only 1.7% more than the number of families in 1984. This difference between the two universes can be explained by the roundings/approximations used to bring the universes closer, as well as the dissolution of some two-spouse families in 1984 into lone-parent and unattached individuals by 1999.
7. During the 1984-1999 period, changes were introduced to facilitate homeownership among first-time home buyers. For example, individuals could withdraw from their RRSPs a maximum of $20,000, which they had to pay back into the system in annual instalments over a 15-year period commencing with the second year following the withdrawal. Another change involved lowering down payments to 5% so that a buyer could own a home by carrying mandatory mortgage loan insurance, which protects the lender but does not relieve the borrower of obligations under the mortgage contract.
8. Savings in registered retirement income funds (RRIFs) or other annuities providing a monthly or yearly cash flow have been excluded since these data were collected in 1999 but not in 1984. Of the total $68 billion ($64 billion in RRIFs), 25% was held by the 1930s cohort, 59% by the 1920s cohort, and 6% by the pre-1920 cohort. Since the largest share was held by families in the 1920s cohort, the change in their wealth without such savings should be interpreted with caution.
9. The amount in RRSPs had to be withdrawn after age 69 and used as income for consumption or investment or turned into RRIFs to draw annuity income. Since amounts held in RRIFs or other annuity plans have been excluded (to make family wealth comparable for the two surveys), the drop in savings held by this cohort in registered savings plans should be interpreted with some caution.
10. This situation would result if the proportionate drop in the number of families over time was more than the proportionate drop in aggregate wealth.
11. See note 9.
12. The change in the distribution of wealth by cohorts was studied in terms of four size groups of wealth (in constant 1999 dollars) rather than in terms of more conventional deciles/quintiles of families because the latter method would have used different thresholds of wealth for cohorts both within and between the time periods.
13. This conclusion is different than that drawn from these surveys by Morissette, Zhang and Drolet (2002). The difference is a result of the family universes used. Using a comparable concept of family wealth, Morissette, Zhang and Drolet used the full 1999 sample and calculated Gini coefficients for all families, excluding those in the top 1% and 5% of the wealth distribution. The exclusions from the 1999 sample of families with a major income recipient aged 15 to 29 or one who immigrated to Canada after 1984 resulted in a different Gini coefficient for 'all families' in 1999 (Table 4). This study does not calculate Gini coefficients on truncated samples.

References

• Augustin, Baudelaire and Dimitri Sanga. 2002. "Income and wealth." Perspectives on Labour and Income (Catalogue no. 75-001-XIE) 3, no. 11. November 2002 online edition.
• Chawla, Raj K. 2003. Comparability of family wealth estimates from the 1999 Survey of Financial Security and the 1984 Survey of Consumer Finances. Staff report no. 01-2003E. Statistics Canada, Labour and Household Surveys Analysis Division. Available through the author.
• Morissette, René, Xuelin Zhang and Marie Drolet. 2002. "Wealth inequality." Perspectives on Labour and Income (Catalogue no. 75-001-XIE) 3, no. 2. February 2002 online edition.
• Statistics Canada. 2002. Canadian Economic Observer: Historical Statistical Supplement, 2001/02. Catalogue no. 11-010-XIB. Ottawa.

The authors are with the Labour and Household Surveys Analysis Division. Raj Chawla can be reached at (613) 951-6901, Henry Pold at (613) 951-4608, or both at perspectives@statcan.gc.ca.

[ Highlights ] [ Main menu | Editor's corner | More news | Contact us | Survey information | Back issues ] [ Statistics Canada home page | In depth | Français ]

© Statistics Canada - Conditions of  use Published: 2003 10 16