75-001-XIE
 
|
January 2003 Vol. 4, no. 1 |
Profiling RRSP contributors
Major changes to the Income Tax Act in 1990 allowed Canadians to increase their participation in registered retirement savings plans (RRSPs). Starting in 1991, eligible taxfilers were permitted to contribute more money to an RRSP during a given year and to carry unused 'room' forward to subsequent years. As a result, more people became eligible to contribute to RRSPs, and to contribute in greater amounts. Still, fewer than half of eligible Canadians make contributions. Increases in participation rates and contribution amounts have been documented for both individuals and families (Akyeampong 2000; Statistics Canada 2001a; Statistics Canada 2001b). However, relatively few determinants of RRSP participation have been established.
What makes one person more likely than another to contribute to an RRSP? Income has frequently been cited as the most important factor; however, when income is held constant, other factors emerge, including sex, age, and membership in an employer-sponsored pension plan (Palameta 2001). A host of other factors have yet to be investigated-for example, family variables, such as number of children and spousal income. A person with no children and a high-income spouse has a greater capacity to contribute to an RRSP than a person with the same income, several children, and a low-income spouse.
Incentives to contribute are less easy to predict. For instance, it is not clear whether investing outside registered plans makes a person more or less likely to contribute. On one hand, some people may consider RRSPs as alternatives to other savings vehicles and invest in only one or the other. On the other hand, people with a general propensity to save may see RRSPs as an added savings opportunity, without having to forgo other investments.
This article looks at some of the personal and family characteristics associated with RRSP participation using 1998 tax data from the T1 Family File (see Data source and definitions). Individuals with RRSP room were divided into three groups, based on presence of a spouse and whether or not the spouse also had RRSP room (see Groups and variables). Personal characteristics included income, sex, age, membership in an employer-sponsored pension plan, self-employment, and participation in non-registered savings and investment vehicles. Family characteristics included income disparity between spouses, spousal RRSP participation, type of marriage, number of dependent children, and presence of low- or high-income children 18 and over in the household.
Group 1 was evenly split between men and women (Table 1). The majority were from dual-earner families, and 42% had spouses in the same tax bracket. The second group was composed mostly of men. They had the highest incomes, ages, and number of children, and were also the most likely to be savers and investors. Most were sole earners in their families, and hence a majority (65%) had spouses in lower tax brackets. The majority of individuals in the third group were women. They were the youngest, and had the lowest incomes and the fewest children.
To see how different variables influence a person's likelihood of RRSP participation, a reference person with a specific set of characteristics was chosen. Changes in participation were calculated in comparison with the reference person. For example, an increase in income from $35,000 to $55,000 was associated with a 21 percentage-point increase in the likelihood of participation, everything else being equal (Chart A). Having a spouse in a higher tax bracket was associated with a 3 percentage-point drop in the likelihood of participation. Only statistically significant results are shown.
The influence of ...
Income
Not surprisingly, high income was associated with high likelihood of RRSP participation in all three groups. High income provides not only the means to contribute, but also the incentive, since those with high income bear the heaviest tax burdens. Data showing how RRSP participation rates rise with income have been published several times (Akyeampong 2000; Palameta 2001; Statistics Canada 2001b).
A higher-income spouse
Having a spouse in a higher tax bracket decreased one's likelihood of contributing in the first group (Chart A), but increased it in the second group (Chart B).
At a given level of individual income, having a higher-income spouse means higher household income, and therefore greater ability to spend on RRSPs. Yet, having a higher-income spouse with RRSP room actually reduced an individual's likelihood of contributing. Why so? Perhaps when both spouses have room, the priority is to use up the higher-income spouse's room, thus reducing the heavier tax burden first. For example, if a couple decided to contribute $2,500 to each spouse's RRSP, their combined tax saving would be more if the higher-income spouse contributed $5,000 ($2,500 to their own plan, $2,500 to the spouse's plan) than if each spouse contributed $2,500.
On the other hand, higher-income spouses with no room cannot make a contribution to reduce their own tax burden. However, they can increase the lower-income spouse's ability to make a contribution. Indeed, having a higher-income spouse with no RRSP room increased an individual's likelihood of contributing.
Age
Everything else being equal, younger people were more likely than older people to contribute in all three groups. The effect was most pronounced among single people; from age 30 to 50, the likelihood of contributing dropped by almost 10 percentage points.
These results largely confirm previous findings showing that except
in the lowest income brackets, persons aged 25 to 34 have the highest
likelihood of contributing (Palameta 2001).
Among single persons, younger people were significantly more likely to
contribute in all income brackets. 
5
Among people in the first group, the age effect was not significant for
income less than $20,000. In all income brackets above $20,000, younger
people were more likely to contribute. 6
Sex
Single women and women whose husbands had no RRSP room were more likely to contribute than their male counterparts.
When both spouses had RRSP room, husbands were slightly more likely than wives to contribute. It may be that instead of each spouse making a separate contribution, the spouse with the higher income-usually the husband-sometimes made contributions to both plans. If spousal contributions are made mainly by husbands, then the number of wives who have RRSPs may substantially exceed the number who make contributions.
Self-employment
Single self-employed persons and self-employed persons whose spouses had room were more likely to contribute than their employed counterparts, although the difference was not significant among those whose spouses had no room.
Why were self-employed persons more likely to contribute? The answer must be more than lack of a pension plan since the comparison groups in the charts were employees without pensions. Since self-employed workers tend to have more year-to-year income variability than employees, some may be using RRSPs as an income-averaging device. In good years, contributions serve to reduce tax burden, while in lean years withdrawals may be used to supplement income with a relatively light tax penalty. This hypothesis could be tested by examining whether the self-employed are also more likely to make RRSP withdrawals.
Private pensions
Those with employer-sponsored pension plans were more likely to contribute
in all three groups. However, previous findings show that in higher-income
brackets, people without pensions participate at higher rates (Palameta
2001). When the samples in this study were split according to income
bracket, similar results emerged. Among single people, those with pensions
were more likely to contribute only in income brackets below $20,000.
7
Those without pensions were more likely to contribute in the $40,000 to
$59,999 bracket, as well as in the $80,000-and-over bracket.
In the first group, those with pensions were again more likely to contribute
in income brackets below $20,000, while those without pensions were more
likely to contribute in income brackets above $30,000. 6
Savings and investments
People who reported interest and investment income (savers) were more likely to contribute than people who reported no such income. Also, investors (people who report dividends from taxable Canadian corporations) had a higher probability of participation than non-investors. This effect is particularly striking. In all three groups, an investor with an income of $35,000 was more likely to contribute than a non-investor with an income of $55,000.
These results support the theory that retirement savings plans are supplements rather than alternatives to other savings vehicles. People investing outside registered plans are also likely to participate in RRSPs.
A contributing spouse
An individual whose spouse had room was far more likely to contribute if the spouse also contributed-in fact, having a contributing spouse doubled a person's probability of participating (Chart A). However, decisions to contribute may depend on the income discrepancy between the two spouses. Couples may tend to behave as a single unit (both contribute, or neither contributes) when income discrepancy is low. However, as the discrepancy increases, the higher-income person may be more likely to make a spousal contribution, and the lower-income person more unlikely to contribute. Indeed, mean income discrepancy was close to zero in cases where both or neither contributed. Mean income discrepancy was 0.7 tax brackets when only one spouse contributed.
Marriage
Everything else being equal, people in a legal marriage were more likely to contribute than those in a common-law relationship. However, patterns of contribution differed between men and women.
When both spouses had room, men in legal marriages were more likely to contribute than men in common-law relationships. The pattern was reversed for women.
Among wives whose husbands did not have room, however, legal marriage was associated with a greater likelihood of contributing. Husbands whose wives did not have room also tended to contribute with greater likelihood in legal marriages, although the trend was not statistically significant.
Number of children
In all three groups, each additional child lowered the likelihood of contributing, especially for women.
When both spouses had room, having a child reduced a person's likelihood of contributing in all three groups (Charts A, B, and C). However, under separate groups for men and women, the effect was much more pronounced for women.
Whether a woman was married or single, having a child reduced her likelihood of contributing to a significantly greater extent than it did for a man (Table 2).
Children 18 and over in the home
The presence of adult children in the household tended to increase a person's likelihood of contributing, especially if the children had relatively high incomes ($30,000 or more). In all three groups, a 50 year-old with an adult child making $30,000 or more and living in the household was more likely to contribute than a 30 year-old with no children-even though younger people were generally more likely to contribute. Again, the effect on women was more pronounced.
Relatively low-income (under $10,000) adult children had a significant effect only in the third group (Chart C). At any given age, a single person with such a child in the household was more likely to contribute than a single person with no children. However, considering men and women separately, married women were affected by the presence of a relatively low-income adult child, while married men were not.
Adult children may increase their parents' capacity to contribute by providing extra income or by helping with unpaid work. The latter may explain why adult children tended to affect married women more than married men.
Summary
Factors particularly strongly associated with RRSP contribution include a contributing spouse and investments outside registered plans. This is true of both men and women at every level of income.
Personal income is another universal predictor of RRSP participation, but having a higher-income spouse was not always associated with a greater likelihood of contribution. In fact, both men and women were less likely to contribute if they had a higher-income spouse with RRSP room.
At most levels of income, younger people, whether married or single, were more likely to contribute than their older counterparts.
Women were generally more likely to contribute than men, except among married couples where both spouses had RRSP room. People legally married were generally more likely to contribute than people in common-law relationships-with the exception of women whose husbands had RRSP room.
Having children was associated with a decreased likelihood of participation, particularly among women. However, women's likelihood of contributing increased if they had an adult child in the home, regardless of the child's income.
Among employees, having a pension plan was associated with higher likelihood of participation in low-income brackets. However, at high incomes, those without pension plans were more likely to contribute. Self-employed persons were generally more likely to contribute than their employed counterparts.
Several important factors were not available from the data source and remain unexamined-notably education, wealth (net worth), and contributions to other registered plans such as RESPs.
Data source and definitionsThis analysis is based on a 2% sample of families from the 1998 T1 Family File (T1FF). The T1FF is derived from information reported on the T1 General Income Tax Return. Linkages are established between husbands, wives and children. The T1FF has been available since 1982. Pension adjustment (PA): For taxfilers whose employer provides a company pension plan, a PA is calculated according to a formula prescribed by the Canada Customs and Revenue Agency. The PA varies according to the amount contributed to the pension plan by the employer and the employee. The PA must be deducted from RRSP room. The PA deduction allows people without an employer-sponsored pension plan to make higher RRSP contributions than people with the same income whose employer provides a pension plan. For a limited number of high-earning employees, the PA is high enough to wipe out their RRSP room entirely. These individuals are excluded from the study. RRSP contribution: a normal contribution is one made within the limit set by the taxfiler's current RRSP room. In rare cases, such as some retiring allowance rollovers, taxfilers are permitted to make contributions that exceed their current RRSP room. However, rollovers are gradually being phased out, and most people with rollovers also make normal contributions. In 1998, less than 1% of RRSP contributors had rollovers only. RRSP room: the maximum RRSP contribution that can be deducted from income (for income tax purposes). RRSP room increases with earned income, including employment and self-employment income, business and rental income, and disability payments (minus employment expenses such as union dues, and business and rental losses). The maximum allowable annual new room is either a dollar amount or 18% of earned income, whichever is lower. In 1998, the dollar amount was $13,500. For those with an employer-sponsored pension plan, new room is reduced by the amount of the pension adjustment. Since 1991, any unused room can be carried over for use in subsequent years. |
Groups and variables The analysis was restricted to living taxfilers aged 25 to 64
with RRSP room. In families where both spouses met the eligibility
criteria, one was selected at random. The three groups were analyzed separately because decisions to contribute were likely based on different criteria. For instance, spousal characteristics are likely to influence a married person's decision, but obviously have no bearing on a single person's decision. For each of the three groups, individual RRSP participation was modelled as a function of: Income: the total before-tax income Income disparity 0 Taxable income = $0 1 Taxable income = $1 to $29,590 2 Taxable income = $29,591 to $59,180 3 Taxable income = $59,181 and above Income disparity ranges from -3 (spouse 3 tax brackets below) to +3 (spouse 3 tax brackets above). This variable was not used in Group 3. Age and sex of the selected individual. Self-employment: no wage or salary income, and more than 50% of
total income from self-employment. PA: A pension adjustment (PA) was taken to indicate an employer-sponsored pension plan. Saver: Interest and investment income, but no dividends from taxable Canadian corporations. Investor: Dividends from taxable Canadian corporations. Contributing spouse: This variable is only used in Group 1. Marriage type: Legal or common-law. This variable is not used in Group 3. Number of children: All children residing in the household. Adult child with income less than $10,000: Child 18 and over with total before-tax income less than $10,000 in the household. Adult child with income over $30,000: Child 18 and over with total before-tax income over $30,000 in the household. Logistic regression was used to examine the determinants of RRSP contribution. Within each of the three groups, in addition to models being run for the general population, separate models were run for men and women, and for people in different income brackets, to identify trends specific to these groups. Logistic regression estimates the probability that a particular outcome-in this study an RRSP contribution-will occur as a function of several explanatory variables. The association between each explanatory variable and the probability of contributing is examined while holding all other variables constant. In other words, the probability of contributing can be compared between individuals identical in every respect but one. For instance, a comparison can be made between men and women of the same age, with the same income, same number of children, etc. A Chi-Square statistic is computed for each explanatory variable to determine whether a change in the variable is associated with a significant change in the probability of contributing. Full results are available from the author. |
Determinants of RRSP participation over timeThe determinants of RRSP participation identified in this paper are likely to be enduring traits. Nonetheless, major legislative changes or changes in the economic cycle may affect certain groups' relative likelihood of participation. For example, a period of economic decline may affect people with children more than people without children. Because women's decisions to contribute are tied to the presence of children more so than men's, the result may be a relative decline in women's participation rates. It is difficult to test this hypothesis because the period since the last major legislative change in 1991 has been one of steady economic growth (Chart). RRSP participation rates increased steadily throughout the 1990s until 1998, the year from which the data in this paper are taken. The decline in participation rate coincides with the introduction of the Canada Education Savings Grant for registered education savings plans (RESPs). From 1998 on, RESP participants could receive up to $400 per child per year. It is possible that RRSP participation declined because some people switched to RESPs. If so, one would expect the RRSP participation rates of people with children-especially young people with children-to have declined the most steeply. Characteristics of RRSP contributors may also change over time as people age, and as some enter and others leave the population of eligible taxfilers. |
AcknowledgementsThe author wishes to thank Hubert Frenken, formerly of the Labour and Household Surveys Analysis Division, and Professor Robert L. Brown of the Department of Statistics and Actuarial Science, University of Waterloo, for their comments on an earlier version of this paper. |
Notes
- If both spouses were 25 to 64 with RRSP room, one was dropped from the analysis to maintain independent observations.
- Although the ability to contribute to an RRSP is a function of after-tax income, before-tax income offers the advantage of capturing the incentive to contribute, since RRSP contributions are deducted from before-tax income and thus serve to reduce individual tax burden.
- Income disparity was used instead of spousal income because the incomes of spouses are highly correlated. Including two highly correlated variables in a group may lead to collinearity problems.
- The Canada Customs and Revenue Agency does not make the distinction between incorporated self-employment and paid employment. Hence, in this study, incorporated self-employed taxfilers are considered to be employees, and only those unincorporated are defined as self-employed.
- The income brackets were: under $10,000; $10,000 to $19,999; $20,000 to $29,999; $30,000 to $39,999; $40,000 to $59,999; $60,000 to $79,999; and $80,000 and over.
- Individuals in the second group were not divided according to income bracket because the sub-samples would have been too small to obtain accurate results.
- The self-employed cannot have employer-sponsored pensions; hence, the groups compared are employees with pensions and employees without pensions.
References
- Akyeampong, Ernest B. 2000. "RRSPs in the 1990s." Perspectives on Labour and Income (Statistics Canada, catalogue no. 75-001-XPE) 12, no. 1 (Spring): 9-15.
- Palameta, Boris. 2001. "Who contributes to RRSPs? A re-examination." Perspectives on Labour and Income (Statistics Canada, catalogue no. 75-001-XPE) 13, no. 3 (Autumn): 7-13.
- Statistics Canada. 2001a. The assets and debts of Canadians: focus on private pension savings. Catalogue no. 13-596-XIE. Ottawa.
- Statistics Canada. 2001b. Retirement savings through RPPs and RRSPs, 1999. Catalogue no. 74F0002XIB. Ottawa.
Author
Boris Palameta is with Income Statistics Division. He can be reached at (613) 951-2124 or 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 01 24
|