The Effect of Social Security on Well-Being in Canada

Featured in print Bulletin on Aging & Health

The last fifty years were marked by a significant expansion of social security programs in most developed countries, including Canada and the U.S. During this period, many countries raised benefit amounts, reduced the earliest age of benefit availability, and introduced new types of retirement benefits such as spousal benefits and special means-tested benefits for the poorest beneficiaries.

Unfortunately, many of these same countries now face great fiscal challenges in their social security systems, as population aging and a largely pay-as-you-go financing system combine to create large projected deficits in the coming decades. Some countries have already enacted reforms to reduce the generosity of their social security programs and others are likely to follow suit.

If these fiscal challenges are met by reducing retiree benefits, how will this affect the well-being of retirees? Surprisingly, the economics literature has relatively little to say about this question, as much of the existing international literature on social security has focused on labor market distortions of the program. Breaking with this trend is a new study by researchers Michael Baker, Jonathan Gruber, and Kevin Milligan, Retirement Income Security and Well-Being in Canada (NBER Working Paper 14667).

The authors make use of changes in the Canada pension plan (CPP) over the past 35 years to identify the effect of social security benefits on retiree well-being. The authors look at numerous measures of well-being, including income, consumption, poverty, and happiness.

One challenge with this type of analysis is that social security benefits vary over time and across individuals for many reasons, not simply because of differences in social security program rules. For example, later cohorts may have more education and higher earnings and be better off in retirement as a result, so one would not want to attribute their improved retirement outcomes solely to increases over time in social security generosity.

To surmount this problem, the authors create simulated social security benefits. In calculating these benefits, they give all cohorts the same values for factors such as earnings, capital income, and family status, so that simulated benefits vary across cohorts due only to changes in social security laws. In one variant of this approach they also hold constant retirement ages, while in another they use the actual retirement ages of each cohort; the latter takes account of changes in retirement behavior that may be in response to changes in social security laws.

The authors first describe the Canadian social security system. Essentially, the program has four main elements - a flat-rate benefit for all individuals age 65 or older, a means-tested supplement to the flat-rate benefit, a spousal allowance for 60-64 year old partners of individuals aged 65 or older, and an earnings-related benefit available starting at age 60. Importantly for this analysis, the introduction of the Canadian social security system in the 1960s and 1970s and subsequent changes to it had heterogeneous effects on different birth cohorts. For example, the earnings-related pension was phased in such that those born before 1900 received no benefits while those born after 1910 received full benefits. Also, the introduction of the spousal allowance in 1975 created a sharp increase in benefits for eligible spouses relative to comparable families a few years older. Many other changes to these programs over time also affected cohorts differently.

The authors' first results show a statistically significant $0.76 to $0.83 increase in retirement income for each additional dollar of simulated social security benefits. One perpetual concern is that the social security program may "crowd out" the individual's own efforts to provide for his retirement income security through his savings, labor supply, and receipt of family transfers, so that $1 in government benefits will generate much less than $1 in additional retirement income. Here, though, crowd out is modest. The results for consumption are similar -$1 in simulated benefits is associated with a $0.66 to $0.80 increase in consumption.

The authors find that higher social security benefits are associated with significant reductions in relative income poverty (defined as having income below the 40th percentile of the income distribution for non-elderly families). In fact, their estimates suggest that virtually all of the drop in poverty for seniors during the 1971-2002 period, a 7.4 percentage point drop, can be explained by the increase in social security benefits for low-income seniors during this time. Interesting, though, there is no significant effect of simulated benefits on relative consumption poverty (similarly defined). The authors speculate that this may indicate that seniors use savings or other means to smooth their consumption while waiting to be eligible for benefits at age 65. Finally, the results for happiness are weak, quite possibly due to the more limited data for this measure.

In sum, the authors find strong evidence that Canada's social security programs have enhanced retiree well-being. Increases in social security benefits have raised retirees' income and consumption significantly and reduced relative income poverty. They caution, however, "extrapolating these results beyond the observed cohorts must be done with care. Policy changes may affect savings and consumption behavior of generations that had more time to alter their plans than for cohorts who were surprised by policy changes in their near-retirement years."

The authors acknowledge funding from the National Institute on Aging (P01 AG005842).