September 2004
Vol. 5, no. 9

Income is a major indicator of the economic well-being of families. But income measures only short-term inflows that affect current consumption and saving. Wealth, on the other hand, measures the surplus accumulated by families and thus provides a better indicator of long-term well-being. Wealth may be targeted toward long-term goals such as retirement, but it can also help families cope with income interruptions or handle unexpected expenditures.

Income is widely available from survey and administrative sources. Wealth, on the other hand, is much harder to measure and, as such, has been the focus of only infrequent surveys (see Data source and definitions).

Wealth has many components, some of which are measured more easily than others. The main division is between marketable components, which can be sold or transferred, and non-marketable assets, which have value only for those who hold them. Some marketable assets such as savings accounts, tax-deferred savings plans, stocks, bonds and mutual funds are readily measured. Others such as real estate, durable goods or business equity are seldom traded, and so their value must be estimated. The principal non-marketable asset is an employer pension plan. Employees or their survivors may draw benefits according to the plan, but its present value is not a tradable commodity and is complicated to estimate. After all assets are valued and summed, debt must be subtracted to arrive at a final measure of wealth.

This article explores the levels and components of wealth inequality in Canada. Many studies have detailed the effects of regional diversity on the distribution of income (Alasia 2003; Finnie 1998; Melvin 1987; Wilkinson et al. 2003; Beach 1996), so the main focus here is provincial variation in the distribution of wealth. Although wealth inequality is undoubtedly related to differing income patterns across the country, it also reflects patterns in the components of wealth: high residential property values in British Columbia, high rates of farm assets on the Prairies, greater pension assets in Ontario, and so on. As such, a multilevel decomposition technique is used to untangle the different effects. This technique highlights some aspects of wealth distribution that are relatively consistent across the country and others that are more specific to certain provinces and family characteristics.

Two in 10 families have virtually no wealth

Since income and wealth are strongly associated, one would expect families with higher incomes to have more wealth.¹ Indeed, several similarities in the provincial distribution of family wealth by pre-tax income deciles are apparent (Table 1).² First, families in the lowest decile had negative wealth. These families had more debts than assets, as in the case of younger or older families with small incomes, or families with businesses with negative net income.³ Families in the lowest two deciles held virtually no wealth.

Second, as expected, the share of wealth held by families rose as they moved up the income ladder. Those in the third and fourth deciles together held between 2% and 4% of all wealth; those in the top decile held the most, ranging from 42% in Nova Scotia to 52% in Alberta.

In seven provinces, families in the top income decile had mean wealth of more than one million dollars (the highest being $1.5 million in British Columbia). The gap in mean wealth between the top and bottom deciles was largest in British Columbia ($1.6 million), about 2.5 times that in Newfoundland and Labrador. On the other hand, the gap in mean income was much smaller, ranging from $166,000 in Alberta to $108,000 in Newfoundland and Labrador. Thus income is more equally distributed than wealth.

Theil's T statistic is a measure of inequality and can be used to decompose total inequality into 'between group' (for example, provinces) and 'within group.' It shows that income inequality was between 32% and 42% of wealth inequality.⁴ Wealth was much more unequally distributed among families in Alberta, British Columbia, and Quebec than in Ontario. Nova Scotia had the most equal distribution.⁵

Interprovincial differences account for little of total wealth inequality

Of total wealth inequality in Canada, 98% was attributable to inequality within provinces. The factors affecting family wealth inequality within provinces include homeownership status, business equity, financial asset components, employer pension plan savings, and mortgage and consumer debt.

Ontario, with 37% of all families and 41% of total family wealth, accounted for 41% of total wealth inequality, followed by British Columbia (14% of families and 16% of wealth) at 22%. Shares for Quebec and Alberta were 17% and 13% respectively. These four provinces, with 85% of all families and 88% of total family wealth, accounted for 93% of overall wealth inequality.

Wealth inequality by family characteristics

Besides financial assets and business equity, differences in income and homeownership contribute to family wealth inequality. As mentioned earlier, income and wealth are strongly associated, so any variation in family income is likely to result in a variation in wealth, both between and within income groups (Table 1).⁶

For instance, the gap in mean wealth between families with incomes under $25,000 and those with $100,000 or more was $491,000 in Newfoundland and Labrador (lowest mean wealth) compared with $764,000 in British Columbia (highest). On the other hand, the interprovincial range of mean wealth within income groups was $67,000 for families with incomes under $25,000 and $340,000 for those with incomes of $100,000 or more (Chart).

These within-income-group ranges show that family wealth across provinces is affected by other factors in addition to income, such as homeownership status, family type, and life-cycle stage. Within provinces, for example, from 20% to 34% of wealth inequality was explained by wealth differences between income groups, and from 16% to 38% by differences between homeownership groups (renter, owner with a mortgage, owner without a mortgage) (Table 2). Income and homeownership explained relatively more of the inequality than other characteristics such as business ownership, age of the major income recipient (used as proxy for life-cycle stage), or coverage in an employer pension plan.

However, given the high correlations between family income, homeownership, age of major income recipient, business ownership, and coverage under an employer pension plan, their individual explanatory powers cannot be added to derive the total inequality coefficient. To overcome this multicollinearity, it is necessary to recalculate the explanatory powers of between and within groups by classifying data by such characteristics taken together. The resulting between group's overall explanatory power is then split by each of the characteristics considered. To maintain statistical reliability, only a limited number of characteristics can be used at a time. The following discussion is based on income, homeownership and business ownership.⁷

Homeownership accounts for much of wealth inequality

More than half (53%) of wealth inequality in Ontario was explained by between-group inequality and the rest (47%) by within-group inequality (Table 3). The overall explanatory power of 53% could be decomposed into 15% for income, 28% for homeownership, 4% for business ownership, and 6% for their interaction. This decomposition shows that variation in wealth by homeownership explained more of the wealth inequality in Ontario than did variation by income group. A similar situation prevailed in Prince Edward Island, Nova Scotia, Manitoba, Saskatchewan, and British Columbia.

Variation by homeownership can be attributed not only to rates of ownership but also to wealth differences between renters, owners without a mortgage, and owners with a mortgage. The wealth of homeowners may, in turn, be influenced by local real estate values. In Newfoundland and Labrador, New Brunswick, Quebec, and Alberta, differences in wealth by income group were more important than homeownership. Business ownership remained in third place—with an explanatory power relatively higher for families in Alberta and Prince Edward Island.

With the ranking of family characteristics affecting wealth inequality across provinces established, one question remains unanswered: How is total wealth inequality distributed by levels of these characteristics? For example, in Quebec, income explained more than homeownership. Did the wealth of families in different income strata contribute equally to this inequality? In fact, the variation in wealth among families with incomes under $25,000 accounted for less than 1% of wealth inequality in Quebec, compared with 44% for those with incomes between $50,000 and $99,999 and 45% for those with incomes of $100,000 or more (Table 4). The corresponding shares in Alberta were 1%, 49% and 46%. More than half of wealth inequality in Ontario and British Columbia was attributable to families with incomes of $100,000 or more.

The relative contribution to total wealth inequality of families in rented dwellings was almost insignificant provincially, whereas the largest contribution was made by families living in mortgage-free homes. Similarly, families with a major income recipient aged 45 to 64 held the largest share of inequality, varying between 72% and 35% for eight provinces. The two provinces showing a different pattern were Newfoundland and Labrador, where elderly families had the highest contribution (50%), and Alberta, where younger families (major income recipient under 45) accounted for 45%.

Shares of total inequality by business ownership showed quite a contrast. In Alberta, where families had a higher rate of business ownership as well as a higher proportion of wealth in terms of business equity, the variability in holdings of families with a business accounted for 90% of provincial wealth inequality—compared with 57% in Ontario.

On the other hand, the variation in wealth of families with an employer pension plan accounted for 72% of wealth inequality in Newfoundland and Labrador compared with just 31% in British Columbia. Among families in most of the eastern provinces, coverage under such plans played an important role in accounting for wealth inequality, whereas for families in the western provinces, business ownership drove inequality.

Conclusion

Provincial economies differ considerably. These differences are in turn primarily responsible for the variation in family income across the country. However, the province with the highest mean income is not necessarily the province with the highest mean wealth. Other factors besides income influence family wealth. These include homeownership status, home values, financial assets, business ownership, other real estate, vehicles, coverage under employer pension plans, and possession of other durable goods.

Provincially, wealth was more unequally distributed than income and concentrated among families in the top income decile. Also, it was more unequally distributed in three provinces—Quebec, Alberta, and British Columbia.

Four provinces (Quebec, Ontario, Alberta and British Columbia) accounted for 93% of overall wealth inequality in Canada. A multilevel decomposition of wealth inequality by family characteristics such as income, homeownership, and business ownership showed that in six provinces, homeownership ranked higher than income in explaining inequality, whereas income led in the other four. Business ownership ranked third in all provinces.

The prevalence of income and wealth inequality is not new and occurs in almost all countries. However, the approaches to redistribution vary. For instance, Canada has a progressive income tax system, which allows the use of taxes and government transfers to reduce income inequality. Intergenerational wealth transfers, however, are generally taxed as income for the recipient. Canada has no direct wealth tax. However, income earned on financial assets is taxed, and municipalities levy property taxes on homes and other real estate.

On the other hand, the Canadian income tax system encourages personal savings and investment in a variety of tax-deferred savings plans.⁸ The objective of such incentives is to encourage families to save more for long-term goals such as retirement or children's education. However, families with higher incomes are more likely to use such tax-deferred plans since they are able to put money aside.⁹ Although such incentives may increase wealth inequality, investments in these tax-deferred plans must be converted into income at a later date and would be subject to taxation at the recipient's highest marginal rate.

Finally, some results indicate that family characteristics may be the strongest generator of wealth inequality. Renters and low-income earners tend to have compressed wealth distributions, as well as low average wealth. Homeowners and high-income earners, on the other hand, have not only higher average levels of wealth, but also greater variation in wealth. Excluding home equity, homeowners still have more than six times the mean wealth of renters.

Notes

1. Wealth usually rises with income. However, since wealth is accumulated over the life cycle, families with lower incomes during retirement may have much greater wealth than their younger counterparts with relatively higher incomes.

2. Tax exemptions based on a taxpayer's demographic situation (marital status, age, number of dependants), business status and investments made may affect the post-tax incomes of families across Canada. Pre-tax incomes, on the other hand, reflect the family's total income in a given year and are used for ranking families by decile groups.

3. In Newfoundland and Labrador, where mean income was lowest, 9% of families had negative or zero wealth, compared with 6% in Ontario and Alberta—provinces with relatively higher levels of incomes (Table 1).

4. Details can be found in Theil (1967), chapter 4; Allison (1978), and Bourguignon (1979). For its illustrative use, see Schwarz (1996), Cardoso (1997), Zyblock and Tyrrell (1997), and Frick and Grabka (2003). Also see Cowell (1985) for multilevel decomposition of Theil's Index.

5. In this paper, wealth inequality was studied using only Theil's coefficient because of its additive and decompositional properties. Other measures of inequality, including the Gini coefficient, log of variance of wealth, and coefficient of skewness were also used, but for brevity are not included here. A summary table containing results of these measures is available from the author.

6. Some of this high-income, high-wealth situation may be embedded in the diversity of provincial economies, resulting in varying incomes for their residents. Compared with Ontario (100), the index of mean wealth varied between 46 (Newfoundland and Labrador) and 108 (British Columbia). However, when mean wealth of families across provinces was recalculated on the assumption that Ontario's distribution of income prevailed in all other provinces, the gap in indices of mean wealth fell to 53 points. This shows that even if the distribution of income were the same across provinces, mean wealth of families in different provinces would still vary.

7. Even though the methodology allows a multilevel decomposition, it is still necessary to restrict the number of characteristics that can be used at a time in order to maintain the statistical reliability of conclusions. The use of five characteristics would have meant classifying families in each province into 144 cells—income (4), homeownership (3), age of major income recipient (3), business ownership (2), and coverage under a pension plan (2). This would have meant splitting the sample of 15,933 families into 1,440 (144x10) cells. Although only the results of incomelhomeownershiplbusiness ownership are shown here, outcomes of other combinations can be made available upon request.

8. These include registered retirement savings plans, registered retirement income funds, registered homeownership savings plans, and registered education savings plans. Also, reduced tax rates apply for investment income and dividends incomes and capital gains (after exhausting the lifetime exemption of $100,000 and $500,000 for qualified small business corporations and qualified farm property).

9. For example, of all taxfilers aged 25 to 64, only 4% of those with income under $10,000 contributed to registered retirement savings plans in 1999, compared with 74% of those with income between $60,000 and $79,999, and 78% with income of $80,000 and more. The Canadian Education Savings Grant program has recently introduced greater savings incentives for low- and middle-income contributors in the form of higher contribution match rates.

References

• Alasia, Alessandro. 2003. Sub-provincial income disparity in Canada: Evidence from 1992 to 1999. Agriculture and Rural Working Paper Series, no. 63. Catalogue no. 21-601-MIE. Ottawa: Statistics Canada.

• Allison, Paul D. 1978. "Measures of inequality." American Sociological Review 43, no. 6 (December): 865-880.

• Beach, Charles M. and George A. Slotsve. 1996. Are we becoming two societies? Income polarization and the myth of the declining middle class in Canada. Toronto: C.D. Howe Institute.

• Bourguignon, François. 1979. "Decomposable income inequality measures." Econometrica 47, no. 4 (July): 901-920.

• Cardoso, Ana Rute. 1997. "Workers or employers: Who is shaping wage inequality?" Oxford Bulletin of Economics and Statistics 59, no. 4 (November): 523-547.

• Cowell, Frank A. 1985. "Multilevel decomposition of Theil's index of inequality." Review of Income and Wealth 31, no. 2 (June): 201-205.

• Finnie, Ross. 1998. "Differences in earnings inequality by province, 1982-94." Canadian Economic Observer 11, no. 2 (February): 3.1-3.12.

• Frick, Joachim R. and Markus M. Grabka. 2003. "Imputed rent and income inequality: A decomposition analysis for Great Britain, West Germany and the U.S." Review of Income and Wealth 49, no. 4 (December): 513-537.

• Melvin, James R. 1987. "Regional inequalities in Canada: Underlying causes and policy implications." Canadian Public Policy 13, no. 3: 304-317.

• Schwarz, Johannes. 1996. "How income inequality changed in Germany following reunification: An empirical analysis using decomposable inequality measures." Review of Income and Wealth 42, no.1 (March): 1-11.

• Theil, Henri. 1967. Economics and information theory. Amsterdam: North-Holland Publishing; Chicago: Rand McNally.

• Wilkinson, Joe, Mehrzad Salem, Annette Laurent, Hans Messinger, and Brenda Bugge. 2003. Trends in provincial and territorial economic statistics: 1981-2002. Income and Expenditure Accounts Technical Series, research paper no. 43. Catalogue no. 13-604-MIE. Ottawa: Statistics Canada.

• Zyblock, Myles and Iain Tyrrell. 1997. "Decomposing family income inequality in Canada, 1981-93." Canadian Business Economics 6, no. 1 (Oct-Dec): 108-119.

Full article in PDF

Author
Raj K. Chawla is with the Labour and Household Surveys Analysis Division. He can be reached at (613) 951-6901 or perspectives@statcan.gc.ca.

You need to use the free Adobe Reader to view PDF documents. To view (open) these files, simply click on the link. To download (save) them, right-click on the link. Note that if you are using Internet Explorer or AOL, PDF documents sometimes do not open properly. See Troubleshooting PDFs. PDF documents may not be accessible by some devices. For more information, visit the Adobe website or contact us for assistance.