Notes

^1.

Registered pension plans (RPPs) are also referred to as occupational pensions or workplace pensions.

^2.

See Ostrovsky and Schellenberg (2009) for a more detailed discussion of data and sample selection.

^3.

From this point forward, the term husband is used to refer to both husbands and male common-law partners, while the term wife is used to refer to both wives and female common-law partners.

^4.

Couples with net annual self-employment incomes totalling less than $500 are retained in the sample.

^5.

This is done to reduce the size of the charts and tables and to make them clearer. The characteristics and outcomes of couples in the "other" category are not distinct, and the exclusion of the group from the discussion does not change the main findings.

^6.

In Q2, Q3, and Q4, the average earnings of husbands in wife-only pension couples are about $20,000 less than the average earnings of husbands in no-pension couples.

^7.

An alternative approach is to identify couples in which neither, one or both spouses are retired, on the basis of individual earnings. This yields four retirement categories: neither spouse; spouse A; spouse B; and both spouses retired; when combined with the five earnings quintiles and the four pension categories, this number of categories results in an unwieldy amount of data.

^8.

Wives who are younger than their husbands are more likely to continue working after their husbands' retirement.

^9.

On average, wives in the sample who have pension coverage are 0.4 to 0.8 years younger than wives who do not have pension coverage.

^10.

Couples in which only the husband has an RPP or in which both spouses have RPPs are the primary point of comparison throughout this discussion.

^11.

Readers should note that only 2% of couples in Q2 are those in which both spouses have an RPP. Hence, the mean and median replacement rates for this group refer to a small group and are based on a small sample.

^12.

The figures underlying these charts are presented in Appendix Table 1. Readers should note that charts show the percentage of couples in specific earnings replacement rate categories—40% to 49%, 50% to 59%, etc.—rather than the distribution of couples on a continuous scale from 1% to 100%. While a bar chart may be more appropriate given the categorical nature of the data, the line charts make it easier to compare the overall distributions of couples with different pension characteristics.

^13.

This statement pertains primarily to retired couples in which only the husband has pension coverage or both spouses have pension coverage.

^14.

The comparisons between groups are similar when replacement rates are calculated as: total income before tax 2006 / total income before tax 1989-1991.

^15.

Average annual earnings in 1989-1991 are about $20,000 to $25,000 higher among retired couples with no pensions than they are among those with pensions. This results in a larger denominator in the replacement rate calculation.

^16.

Again, couples in which the only the husband has an RPP or in which both spouses have RPPs are the primary point of comparison throughout this discussion.

^17.

These results pertain to retired men from the middle of the 1989-1991 earnings distribution (i.e., Q2, Q3 and Q4). The difference in median earnings replacement rates among men from Q5 is ten percentage points.

^18.

A commonly accepted method to take into account the number of persons residing in the household is to divide earnings and income figures by the square root of family size. Assuming that decline in family size is the same for retired couples as it is for all couples in the sample, such an adjustment would increase earnings replacement rates by 14%, on average.