Employer pensions and the wealth of Canadian families

by Derek Messacar and René Morissette

[Release from The Daily] [Full article in PDF]

Skip to text

Text begins

Start of text box

Overview of the study

This study compares the wealth holdings of family units covered by workplace pension plans with those of other family units. It focuses on families and unattached individuals who had no significant business equity and whose major income recipient was aged 30 to 54 and employed as a paid worker. The paper also examines whether wealth differences observed between families with registered pension plan (RPP) assets and other families persist when key sociodemographic differences between the two populations are taken into account.

  • Excluding pension assets, family units with RPP assets had a median net worth of $210,600 in 2012. This compared with a median net worth of $64,000 among family units without RPP assets.
  • Family units with RPP assets were more likely than others to have characteristics that are conducive to wealth accumulation, such as higher incomes, higher levels of educational attainment and longer job tenure, among others.
  • In 2012, families with RPP assets were more likely to hold other types of assets than families with no RPP assets, including real estate equity (82% versus 56%), investments or RRSPs/LIRAs (79% versus 55%), or vehicles (91% versus 76%).
  • After accounting for differences in factors such as income and other characteristics, families with RPP assets were still more likely to hold investments or RRSPs/LIRAs than families with no RPP assets (by a margin of 8 percentage points instead of 24).
  • In 2012, differences in income and other observable characteristics between the two types of families accounted for approximately four-tenths of their difference in median net worth, and approximately one half of their difference in average net worth.

End of text box

Introduction

For most Canadians, the accumulation of private wealth during working years is a requisite for maintaining a comfortable standard of living in retirement. For many Canadians, employer-sponsored registered pension plans (RPPs) have been an important part of this process. Yet changes in the economic landscape over the past several decades have prompted some employers to move away from offering workplace pensions.Note 1

A key question in this regard is whether families who are not covered by RPPs accumulate as much wealth as their counterparts who belong to such plans. It may be that RPP members benefit directly from collecting employer pension contributions and, as a result, are able to amass greater private wealth than non-members. The lock-in provision of RPPs―the fact that money invested in these plans cannot usually be withdrawn before retirement―may also facilitate savings to the extent that it acts as a commitment savings device. In contrast, non-members may compensate for the lack of workplace savings by accumulating more wealth in other retirement accounts, stocks and bonds, or real estate equity than RPP members, which results in both groups having comparable replacement incomes in retirement irrespective of their pension coverage. Previous research on the extent to which workplace pensions raise net wealth or induce workers to reduce savings in other accounts has yielded mixed results. Some studies find that workplace pensions increase private savings,Note 2 while others find these plans simply redistribute wealth across accounts.Note 3

In order to provide new insights into this issue, this study compares the wealth holdings of families who are covered (or have been covered) by workplace pension plans with those of other families. Using the 1999 and 2012 waves of the Survey of Financial Security (SFS), the study focuses on families and unattached individuals who have no significant business equity (less than $1,000) and whose major income recipient is aged 30 to 54 and is employed as a paid worker (See Data sources, methods and definitions). Within this population, the study investigates the sources of the wealth differences between family units with RPP assets―those in which at least one person reports being a current or past RPP member―and those with no RPP assetsNote 4. Since the majority of families and unattached individuals with RPP assets are covered by defined benefit (DB) pension plans, a comparison of the wealth holdings of DB members with those of RPP non-members is also provided.Note 5 Throughout the article, wealth and income numbers are expressed in 2012 dollars using the Consumer Price Index (All Items) as a deflator.

Wealth in families with and without RPP assets

Irrespective of whether their pension assets are included, family units with RPP assets displayed greater wealth holdings than those without RPP assets (Chart 1). Excluding their pension assets, family units with RPP assets had a median net worth of $210,600 in 2012, more than three times the median net worth of $64,000 of other family units. Wealth differences between the two groups were significantly greater when pension assets were included. The median net worth in 2012 of families with RPP assets then rises to roughly $350,000. Although qualitatively similar patterns were observed in 1999, the wealth differences between the two groups were larger in 2012.

Chart 1 Median net worth, families with and without RPP assets, 1999 and 2012

Description for Chart 1

Wealth differences across the two types of families remain when families with RPP assets are classified according to whether their major income recipient has a defined benefit (DB) plan, a defined contribution plan or other type of plan, or was not covered by a pension plan at the time of survey collection (Table 1).Note 6 For instance, family units where the major income recipient had a defined benefit plan in 2012 had a median wealth of $381,200, or roughly $100,000 more than the median wealth of families whose major income recipient had a defined contribution plan or another type of plan. Of all DB members, those employed in public administration, education, health care and social assistance had the highest median net worth ($473,400).Note 7

Table 1
Median net worth, families with and without RPP assets, 1999 and 2012
Table summary
This table displays the results of Table 1 Median net worth Median net worth , Share of family units, 1999 and 2012 (appearing as column headers).
  Median net worth Share of family units
1999 2012 1999 2012
2012 dollars percentage
Family units with no RPP assets 55,413 64,000 39.1 38.7
Family units with RPP assets 208,225 353,140 60.9 61.3
Major income recipient has  
A defined benefit plan 221,718 381,160 44.8 38.3
A defined benefit plan outside public administration, education, health care and social assistance 212,215 292,989 26.0 17.9
A defined benefit plan in public administration, education, health care and social assistance 236,665 473,394 18.8 20.4
A defined contribution plan or other type 184,313 288,244 6.9 12.3
No RPPsNote 1 159,790 341,082 9.2 10.7

Accounting for differences in the characteristics of families with and without RPP assets

In an accounting sense, families with RPP assets may have greater wealth than families without RPP assets for two reasons. They may have a greater propensity to hold certain assets and, when holding a specific asset, the value of this particular asset may be greater. These differences might in turn arise―at least partly―because the two groups differ in terms of characteristics conducive to wealth accumulation such as education, family structure, income, and tenure with their employer.Note 8

This issue can be examined by comparing the characteristics of the two groups of family units. First, major income recipients in families with RPP assets were slightly older than those in other families (Table 2). Furthermore, those in families with RPP assets were more educated; more likely to be born in Canada and be in couples; more often unionized or employed in public administration, education, health care and social assistance; and had longer job tenure than their counterparts in families with no RPP assets.

Table 2
Descriptive statistics, families with and without RPP assets, 1999 and 2012
Table summary
This table displays the results of Table 2 Descriptive statistics 1999, 2012, Family units with RPP assets, No, Yes and Yes DB plan (appearing as column headers).
  1999 2012
Family units with RPP assets
No Yes Yes DB plan No Yes Yes DB plan
age
Average age of major income recipient 40.2 41.8 42.2 41.1 42.8 42.6
  percentage
Percentage of major income recipients who  
Have a high school education or less 40.9 27.1 26.6 38.4 24.2 20.6
Have a bachelor's degree or more 16.4 26.6 27.8 25.3 36.7 39.4
Are Canadian-born 71.8 82.4 83.3 69.6 81.0 83.7
Are unionized 14.6 55.6 67.3 11.5 48.6 65.6
Have 10 years of tenure or more with their employer 21.3 51.0 59.7 19.0 41.7 43.6
Are employed in public administration, education, health care and social assistance 10.2 34.2 42.0 13.0 36.9 53.2
Are in a couple 55.7 72.8 70.7 53.4 69.1 66.3
    2012 dollars
After-tax family income  
Average 48,946 74,513 73,536 60,092 91,064 88,880
Median 43,031 69,297 68,624 51,644 82,579 81,866
  number
Average family size 2.8 3.1 3.0 2.6 3.1 3.0
  dollars
Median net worth 55,413 208,225 221,718 64,000 353,140 381,160
  number
Sample size 2,048 3,370 2,437 1,228 2,346 1,478

For instance, more than one-third of major income recipients in families with RPP assets had a bachelor’s degree or more education in 2012. The corresponding proportion for their counterparts in other families was 25%. About 40% of major income recipients living in families with RPP assets had been with their employer for 10 years or more in 2012—twice the proportion observed among other major income recipients. Families with RPP assets also displayed higher income levels and were thus better-positioned to accumulate wealth through their savings than non-RPP families. To sum up, families with RPP assets differed from other families in several aspects, which may affect their annual savings and thereby their ability accumulate wealth. These findings will be taken into account below in multivariate analyses.

Families with and without RPP assets also differed in their propensity to hold various types of assets. Both in 1999 and 2012, families with RPPs were significantly more likely (by at least 25 percentage points) to hold registered retirement savings plans / locked-in-retirement accounts (RRSPs/LIRAs) or principal residence equity (Table 3). Families with RPP assets also held other types of assets more often, albeit to a relatively small extent in the case of assets held by the vast majority of families (e.g., deposits).

Table 3
Percentage of family units holding assets or debt, families with and without RPP assets, 1999 and 2012
Table summary
This table displays the results of Table 3 Percentage of family units holding assets or debt 1999, 2012, Family units with RPP assets, No, Yes and Yes DB plan (appearing as column headers).
  1999 2012
Family units with RPP assets
No Yes Yes DB plan No Yes Yes DB plan
percentage
Assets  
Deposits 84.5 91.0 91.0 92.1 92.7 92.2
Investments 20.2 41.3 42.4 16.2 28.8 28.1
RRSPs/LIRAs 55.7 81.1 81.9 50.7 75.3 73.7
Investments or RRSPs/LIRAs 58.8 85.0 86.0 55.3 78.7 77.9
Principal residence 50.8 76.9 77.7 50.5 79.0 78.6
Other real estate 12.3 17.0 17.2 14.3 19.6 20.1
Principal residence or other real estate 55.4 79.5 80.6 56.0 82.0 82.2
Vehicles 77.4 90.9 90.4 76.2 91.2 90.1
RPP assets Note ...: not applicable 100.0 100.0 Note ...: not applicable 100.0 100.0
Other assets 100.0 100.0 100.0 100.0 100.0 100.0
Debts  
Mortgage on principal residence 38.6 57.2 57.0 39.5 60.7 60.9
Other debt 71.0 77.9 77.5 73.7 80.9 82.6

For example, 82% of families with RPP assets had a principal residence or other real estate in 2012, compared with 56% for other families. Families with RPP assets were also more likely to hold investments or RRSPs/LIRAs (79%, compared with versus 55% among families with no RPP assets) and vehicles (91% versus 76%). On the debt side, RPP families were also more likely to hold mortgage debt (61% versus 40%) or other debt (81% versus 74%).

Differences in average asset values by RPP membership can also be examined for each category of assets. Doing so focuses the analysis on family units that have elicited some particular forms of assets since those who do not hold the corresponding category of assets are excluded from the calculations.

The results without controls indicate that for every category of assets, the average amount held by RPP families was larger than non-RPP families, both in 1999 and in 2012 (Table 4).

Table 4
Differences in average asset values between families with RPP assets and those with no RPP assets, 1999 and 2012
Table summary
This table displays the results of Table 4 Differences in average asset values between families with RPP assets and those with no RPP assets 1999, 2012, Control variables, No controls, Family income and Family income and others (appearing as column headers).
  1999 2012
Control variables
No controls Family income Family income and others No controls Family income Family income and others
dollars
Differences between family units with RPP assets and those with no RPP assets  
Deposits 3,398 -,653Note * -1,703Note * 6,225 545Note * -,661Note *
Investments or RRSPs/LIRAS 11,317Note * -5,439Note * -13,127Note * 32,268 -3,007Note * 1,295Note *
Net equity from real estate 26,814 1,275Note * 2,974Note * 29,646Note * -24,142Note * -4,709Note *
Vehicles 6,567 2,373 1,528 7,483 2,168 2,336
Other assets 10,645 2,769 1,134Note * 17,784 4,881 4,311Note *
Differences between family units whose major income recipient has a DB plan and those with no RPP assets  
Deposits 3,602 -88Note * -1,755Note * 6,475 1,354Note * 1,149Note *
Investments or RRSPs/LIRAS 9,140Note * -3,441Note * -14,578Note * 19,423 -9,513Note * 922Note *
Net equity on real estate 26,526 3,944Note * 6,110Note * 13,455Note * -33,183 -15,624Note *
Vehicles 6,714 2,709 1,851 7,167 2,246 2,660
Other assets 10,848 3,833Note * 2,218Note * 17,147 5,168 7,563

Hence, families with RPP assets who had financial investments (e.g., stocks and bonds) or RRSPs held, on average, roughly $32,000 more such assets than other families in 2012. Families with RPPs who had a principal residence or other real estate averaged roughly $27,000 more in net equity on real estate than non-members in 1999, and about $30,000 more in 2012.

To sum up, families with RPPs tended to hold broadly defined assets more often than other families, and, when they held such assets, they averaged greater amounts than those held by their counterparts with no RPPs in every asset category (Chart 2).Note 9 The same conclusion applied to families covered by defined benefit plans.

Chart 2 Average asset values, families with and without RPP assets, 2012

Description for Chart 2

The next step is to investigate the degree to which these differences in wealth are explained by differences in families’ observable characteristics. This can be done by using multivariate analyses that control for a number of factors, including: after-tax family income, family size, marital status, and region of residence, as well as the age, gender, education level, immigration status, union status, sector of employment, and job tenure of the major income recipient.Note 10

While differences in income between RPP and non-RPP families are expected to be the largest determinant of the wealth disparity, other factors may also help to explain differences in savings behaviour between the two types of families. For example, some individuals may be more likely to be union members out of a preference for non-wage benefits, including pensions. In addition, jointly controlling for earnings, education and tenure acts as a reasonable proxy for permanent income, a concept that better captures families’ ability to save than current income.Note 11

In the absence of controls to account for such differences, RPP families were at least 25 percentage points more likely than non-RPP families to hold RRSPs/LIRAs or a principal residence in both 1999 and 2012. This difference, however, became smaller when differences in family income and other covariates between the two groups were taken into account (Table 5). Specifically, when controls were added for family income and other characteristics, family units with RPP assets were about 10 percentage points more likely to have a principal residence than their counterparts with no RPP assets in 2012 (instead of 29 percentage points). Similar patterns were also observed for other categories of assets and debt.

Table 5
Percentage-point differences between families with RPP assets and those with no RPP assets in the propensity to hold specific assets or debts, 1999 and 2012
Table summary
This table displays the results of Table 5 Percentage-point differences between families with RPP assets and those with no RPP assets in the propensity to hold specific assets or debts 1999, 2012, Control variables, No controls, Family income and Family income and others (appearing as column headers).
  1999 2012
Control variables
No controls Family income Family income and others No controls Family income Family income and others
percentage point
Differences between family units with RPP assets and those with no RPP assets  
Assets  
Deposits 6.5 4.4 3.8 0.6Note * -0.5Note * -1.5Note *
Investments 21.1 12.7 9.4 12.6 5.8 3.9Note *
RRSPs/LIRAs 25.4 11.0 6.6 24.5 10.5 8.5
Investments or RRSPs/LIRAs 26.2 1.7 7.6 23.5 10.5 7.9
Principal residence 26.0 10.2 6.0 28.5 10.6 9.6
Other real estate 4.7 1.5Note * -0.4Note * 5.3 -1.3Note * -0.3Note *
Principal residence/other real estate 24.2 9.1 5.2 26.0 9.8 7.8
Vehicles 13.5 4.9 3.7 15.0 6.3 7.6
Debts  
Mortgage on principal residence 18.6 7.0 5.9 21.2 8.5 9.6
Other debt 6.9 3.8 3.8 7.2 4.9 4.7
Differences between family units whose major income recipient has a DB plan and those with no RPP assets  
Assets  
Deposits 6.5 4.3 3.6 0.1Note * -1.3 -3.7Note *
Investments 22.2 14.1 11.6 11.9 6.0 3.4Note *
RRSPs/LIRAs 26.2 12.0 7.0 23.0 9.3 7.1
Investments or RRSPs/LIRAs 27.2 14.0 8.4 22.7 10.2 6.7
Principal residence 26.9 11.1 5.2 28.1 10.3 10.8
Other real estate 5.0 2.1Note * -0.4Note * 5.8 2.2Note * 2.5Note *
Principal residence/other real estate 25.2 10.4 5.1 26.2 10.2 10.0
Vehicles 12.9 4.1 2.4Note * 13.9 4.6Note * 7.5
Debts  
Mortgage on principal residence 18.5 7.0 5.2 21.4 8.4 10.7
Other debt 6.5 3.5Note * 3.3Note * 9.0 7.2 8.7

Interestingly, controlling for only income yields almost the same effect as the full vector of control variables—implying that differences in family income are a major factor underlying the wealth differences between the two groups. This is expected, since employment income is a common resource for accumulating private wealth.

Nevertheless, controlling for observable characteristics does not fully explain differences in the propensity to hold investments or RRSPs/LIRAs or principal residence equity. After the inclusion of income and other observable characteristics, RPP families were still roughly eight percentage points more likely (in both 1999 and 2012) to hold investments or RRSPs/LIRAs than non-RPP families (see third and sixth columns of Table 5). Thus, significant differences in portfolio allocation between the two family types remained even after controlling for income and other characteristics. This suggests that families with and without RPPs either differ intrinsically in terms of saving behaviour, or end up having a different portfolio allocation because of the impact of workplace pension plans.

In contrast, differences in average amounts held in specific type of assets (among those who hold these types of assets) were mainly explained by income differences. For instance, families with RPP assets held, on average, $32,000 more assets in investments or RRSPs/LIRAs than their non-RPP counterparts in 2012; however, this difference became negative and no longer statistically significant after income differences were taken into account (see the fourth and fifth column of Table 4). Differences in net equity in real estate observed between the two groups in 2012 also became negative once income differences were taken into account. Similar results were found when comparing members of defined benefit plans with non-RPP members.

Together, the results of tables 3 to 5 indicate that the differences in family income and other characteristics conducive to wealth accumulation observed between families with and without RPPs accounted for differences in the average value of specific asset types (conditioning on holding these assets) to a greater degree than they accounted for differences in their propensity to hold specific assets. This raises the following question: To what extent do such differences in family income and other characteristics account for the differences in median and average wealth observed between the two groups? Wealth, or net worth, is defined as the overall value of assets, minus the total amount of debt held by the family.

Assessing the potential role of employer sponsorship

Although RPP membership may directly help families to overcome the costs and challenges associated with retirement planning, it is conceivable that workers who have greater predispositions for saving join firms based on pension coverage, or that firms choose to offer pensions based on the demands of their workers.Note 12 While these potentially greater dispositions for saving and the aforementioned workers’ demands cannot be observed in the data used in this article, it is possible to compute estimates of median and average wealth for the two groups that remove the effect of differences in family characteristics (such as income).

An ideal experiment to assess the potential impact of employer sponsorship on wealth formation would randomly assign family units into RPP coverage to assess its effect on net worth. In the absence of such an experiment, the approach used here is to ask: What would be the median and average wealth of family units with no RPP assets if their income and other characteristics were identical to those with RPP assets?

To answer this question, a decomposition method based on a re-weighting procedure is applied to family units with no RPP assets (Table 6).Note 13 The first two columns of Table 6 replicate the median net worth (or wealth) values shown in Table 1 (in addition to providing average net worth values). Table 6 indicates, for example, that the average net worth of family units with RPP assets was about $536,000 in 2012, compared with $191,000 for families with no RPPs.

Table 6
Actual and adjusted values for net worth, families with and without RPP assets, 1999 and 2012
Table summary
This table displays the results of Table 6 Actual and adjusted values for net worth. The information is grouped by Family unit with RPP assets (appearing as row headers), Actual data, Re-weighted data, Proportion of the difference explained, No and Yes, calculated using 2012 dollars and percentage units of measure (appearing as column headers).
Family unit with RPP assets Actual data Re-weighted data Proportion of the difference explained
No Yes No
2012 dollars percentage
1999  
Median net worth 55,413 208,225 131,263 49.6
Average net worth 122,097 300,416 222,078 56.1
2012  
Median net worth 64,000 353,140 177,500 39.3
Average net worth 190,926 535,564 358,885 48.7

The third column of Table 6 estimates the median and average wealth that families with no RPPs would have if their characteristics were the same as families with RPPs. Under this hypothetical scenario, family units with no RPPs would have an estimated average net worth of about $359,000. This finding means that about one-half of the difference in average wealth between the two groups can be accounted for by differences in income and other characteristics.Note 14 This suggests that employer sponsorship is still associated with greater wealth accumulation, but the magnitude of the effect is not as pronounced as the raw wealth comparisons would indicate. The explained portion was lower in the case of the median— about four-tenths of the difference in median net worth between the two groups can be accounted for by differences in income and other characteristics. These results are consistent with recent Canadian researchNote 15 that finds RPP contributions partially increase total savings, by approximately $0.50 per $1, but that some crowd-out between RPPs and RRSPs still occurs.

As well, significant wealth differences remained between families of DB plan members and families with no RPPs, even after the effect of differences in family income and other characteristics were taken into account (Table 7). Hence, in the case of both RPP families and families with DB plans, wealth remained significantly higher than non-RPP families, even across observationally equivalent families.

Table 7
Actual and adjusted values for net worth, families with DB plans and families without RPP assets, 1999 and 2012
Table summary
This table displays the results of Table 7 Actual and adjusted values for net worth. The information is grouped by Family unit with RPP assets (appearing as row headers), Actual data, Re-weighted data, Proportion of the difference explained, No and Yes
DB plan, calculated using 2012 dollars and percentage units of measure (appearing as column headers).
Family unit with RPP assets Actual data Re-weighted data Proportion of the difference explained
No Yes
DB plan
No
2012 dollars percentage
1999  
Median net worth 55,413 221,718 144,101 53.3
Average net worth 122,097 310,073 235,372 60.3
2012  
Median net worth 64,000 381,160 165,100 31.9
Average net worth 190,926 559,237 295,522 28.4

Another way to examine the effect of income and other characteristics on wealth accumulation is to examine the density function of wealth, which indicates the extent to which families are concentrated across the wealth distribution. Families with no RPPs, for example, had a relatively large degree of concentration around $30,000; families with RPP assets, in contrast, were more evenly spread across the distribution, with a certain degree of concentration occurring around $190,000 (Chart 3). If, however, families with no RPP assets had the same socioeconomic characteristics as families with RPP assets, they would have a density function that would move closer to that of families with RPP assets, but this would not entirely remove the differences between the two types of families. This supports the fact that socioeconomic differences explain a good deal of differences in wealth between families with and without RPPs, but do not explain them all.

Chart 3 Probability density function of net worth, families with and without RPP assets, 2012

Description for Chart 3

Conclusion

This paper investigated the potential role of employer sponsored pensions in the wealth accumulation process of working-aged Canadians with no significant business equity, by comparing the wealth of families with registered pension plan (RPP) assets with that of other families. The analysis first showed that family units with RPP assets had significantly more wealth than other families in both 1999 and 2012, even after excluding pension assets from the calculations. The propensity to hold assets was also higher for RPP families for all types of assets, including RRSPs/LIRAs.

A closer inspection showed that part of the reason for such behaviour is due to differences in observable factors, notably family income. However, even after controlling for differences in observable factors, families covered by registered pension plans still displayed higher wealth holdings than their counterparts who were not covered by such plans. Thus, a significant fraction of RPP contributions still pass through into greater wealth for reasons that cannot be explained in the data. Whether this finding reflects the causal impact of workplace pension plans on wealth accumulation or intrinsically different savings behaviour from RPP members and non-members remains to be determined. Nevertheless, this finding informs discussions regarding how changes in the availability of workplace pensions, both in Canada and internationally, may impact private wealth accumulation.

Derek Messacar is a researcher and René Morissette is Assistant Director in the Social Analysis and Modelling Division of Statistics Canada.


Notes

Related material for this article

Date modified: