Retirement incentives and decisions across the income distribution: Evidence in Canada

In this paper we build on earlier work in Milligan and Schirle (2023) by evaluating the retirement incentives embedded in Canada’s retirement income system from the perspective that the system may differentially affect individuals located in different parts of the income distribution. There are several factors that were important to account for in this regard. First, we observe a shifting longevity-earnings gradient, whereby individuals with higher earnings tend to live longer than those with low earnings, and life expectancy has increased over time. Second, we illustrate how benefit eligibility (with some components clearly linked to a person’s work history) varies by lifetime earnings decile. These factors may be considered opposing forces—while larger social security benefits are available to individuals with lower earnings in their work history, those from the top of the income distribution tend to enjoy longer lives over which they may receive benefits. Overall, we see greater Social Security Wealth among individuals from lower deciles.

Our results show that the structure of Canada’s social security programs leads to very different incentives to continue working at older ages. The implicit tax rates on continued work tend to be higher for workers from lower-earning deciles. Considering changes to actuarial adjustments in the Canada Pension Plan associated with early pension take up, these implicit tax rates on work at older ages fell substantially after 2011. The patterns are slightly different for individuals from lower deciles (for whom incentives to continue work at age 60 improved substantially more than at ages 65 or over) and upper deciles (for whom incentives to continue work after age 65 were most clear). Our regression estimates confirmed the important effects on retirement behaviour, with substantially larger effects for individuals in lower deciles. These effects are greater for women than men, who tended to have lower career earnings. In simulations, we show that changes to the actuarial adjustment after 2010 had some impact on retirement rates by lowering the implicit tax on work. The overall redistributive effect of these induced retirement changes was fairly small, however, as the actuarial adjustments brought the system closer to actuarial fairness.

Draft (October 2024)