Rebuilding the First Pillar of Canada's Retirement Income System

I'm here to start a conversation. I don't pretend to have all the answers. I offer examples to illustrate ideas.

For several decades I've heard people talk about the need to rethink Old Age Security, representing our suite of first pillar programs that support older Canadians in their retirement. And for several decades I've watched failed efforts to move that conversation forward. So, I'll take a moment to consider why this is failing and propose a slightly different (and hopefully better) way of rebuilding this first pillar. 

In what follows,  I think of there being two distinct objectives we are hoping to meet. First, Canadians want to recognize the past and current contributions of seniors to Canadian society. Second, we are concerned with alleviating poverty among seniors and would like to ensure they are able to achieve a reasonable standard of living. With two policy objectives in hand, it is often considered wise to have two distinct policy instruments to work with. I'll propose creating two new (but familiar-looking) programs to meet these objectives, which I will call the Canada Seniors Rebate (CSR) and the Canada Seniors Supplement (CSS).

Some background

Canada introduced our first Old Age Pension in 1927.  As a means-tested benefit available to individuals age 70 or older, only a few people were eligible. The benefits were also quite modest, up to $240 annually (or $4481 in today's dollars). We introduced Old Age Security in 1952 and since expanded elibility for benefits, with the near-universal Old Age Security (OAS) and the income-tested Guaranteed Income Supplement (GIS) available to people age 65 or older since 1967.  In 1975 we expanded the program further, creating an Allowance paid to some married people age 60-64 (with an older spouse age 65 and older) and again in 1985 to widows and widowers age 60-64. Over time benefit generosity also increased. By the summer of 2026, an unmarried 65 year old could expect to receive up to $22,508 in OAS and GIS benefits. 

The expansion of benefit eligiblity and generosity over the past century has driven remarkable reductions in senior poverty rates (which I discuss here). At only 5.5% (in 2023, based on our official poverty measure), this is clearly a Canadian success story. Our senior poverty rates are very low, by historical and international standards. We don't want to ruin that. 

But it is costly. In 2026, we are spending roughly $88.8 billion on the OAS suite of programs. We also spend several billion dollars on various tax expenditures with similar goals - $5.6 billion on the age credit, $1.4 billion on the pension income credit, and another $2.1 billion on pension income splitting. By any measure, nearly $100 billion is a lot of money and we want to make sure we are using it as effectively as possible.  I have the impression that most young people want our low-senior-poverty success story to continue, but as they find today's labour and housing markets incredibly challenging they'd also like to make sure we are using our tax dollars effectively.

Some basic features of OAS and GIS today

On the surface, OAS and GIS seem simple - we offer a lump sum benefit to most people age 65 and over, and add in a supplement to support those who are low income. But how much do people actually get? It gets complicated fairly quickly, with eligibility depending on marital status, age, individual and family incomes, the sources of income, the length of time one has lived in Canada, and more. For the purpose of illustrating some basic features affecting most older Canadians, I'm going to focus on an example of a 65 year old that is not married.

There are a few things to notice. 

First, the steepness of the green line tells us about the programs' contributions to marginal effective tax rates (or METR, ie. if you gain a dollar of income, how much gets 'taxed' back via benefit reductions and how much do you really have available to spend). For the lowest-income seniors, the GIS (top-up and regular benefit combined) is reduced by 75 cents for every dollar of income received. Regular GIS benefits are reduced by 50 cents for every dollar of income.  In some provinces that also have income-tested seniors benefits, some seniors will face METRs well in excess of 100%. I pay attention to this because it affects a person's incentives to work and save.  With this in mind, GIS allows seniors to earn some employment income before clawing back benefits. Income from the Canada/Quebec Pension Plan is subject to clawback, as is any private pension or other taxable income.

Second, this single 65 year old is completely phased out of the GIS when their income reaches $22,800. They receive the full OAS pension, however, until their income is over $93,454. Above that, benefits are reduced by 15 cents for every dollar of income.  

A failed proposal: targeting poverty alleviation

For several decades I have seen proposals to take the resources we have used for OAS and GIS and use them to build a program that better targets low-income seniors. This typically involves eliminating the near-universal component (OAS) and using those funds to create a more generous supplement - both in terms of the maximum benefit generosity, and in decreasing the benefit reduction rate so that more people receive the benefits as a supplement. We can think of this as similar to the creation of the Canada Child Benefits in 2016, which pooled the resources used for the Universal Child Care Benefit, the Canada Child Tax Benefit and the National Child Benefit Supplement to more directly target low-income families with children.  

For several decades I have seen proposals like this rejected. But, why? I tend to be a fan of such proposals and appreciate the potential to improve the lives of many low income seniors, so I have spent a lot of time talking to people about why it fails. Here's my thoughts.

First, people are incredibly protective of our first pillar programs. After all, they are big part of our success story and nearly everyone knows someone who relies on this system. This makes any suggestion to alter the system politically dangerous. To me, this suggests any proposal needs to be familiar, relatable, and easy (enough) to communicate. I expect any suggestion to simply eliminate OAS will be met with fierce opposition.

Second, I don't believe poverty alleviation is the only goal. I hear a second goal in the Canadian conversation: ensuring the sustainability of our social contract across generations with a recognition of everyone's contributions to society. The first pillar programs are part of our social contract, whereby younger people agree to support those who are older, with an expectation the next generation will do the same for them.  I hear older individuals who want their past and current contributions - with their jobs, their family, and with their community - to be recognized and clearly appreciated. So, we want to ensure that social contract survives and that people feel recognized. To me, this means we can't entirely eliminate the near-universal component of the program. But perhaps we can modify it.

Basic principles for rebuilding the first pillar

I proceed with a few principles that I think are sensible:

1. A policy proposal needs to be politically feasible. Otherwise I'm just talking to myself for fun.

2. We shouldn't spend more than we already are.

3. We should always be mindful of marginal effective tax rates to find the right balance between work incentives and program costs.

4. Always look for ways to improve on horizontal equity and address Canadians' sense of fairness.

5. Recognize that social norms and values change over time, and policy can be built with flexibility to adapt.

These may sound simple enough, but I know better. Really.

On METRs:

I have three concerns I am mindful of when it comes to benefit reductions related to higher income:

1. Earnings exemptions currently in place exist to improve incentives to continue working at older ages. We do not reduce benefits for the first $5000 earned, and only count 1/2 of the earnings between $5000 and $15000 as income for benefit calculation. 

2. Delaying CPP or QPP take-up after age 60, often associated with delaying retirement, results in correspondingly higher monthly benefits for retirees. For each year between age 60 and 65, the CPP adjustment factors increase benefits by 7.2% (QPP by 6-7.2%) and from age 65 to 70 the adjustment factor is 8.4%. However, for every dollar gained by delaying CPP, 50 cents (or more) is taken away from GIS recipients. This interaction of these programs adds to disincentives to work and/or delay one's pension and is not currently recognized in GIS benefit calculation.

3. There are several provincial programs to support low-income seniors. For example, Ontario GAINS offers a modest benefit with a 50 percent benefit reduction rate. when combined with GIS and other income-tested programs for seniors, METRs may exceed 100%.

How much do these METRs matter? We have evidence that the structure of our retirement income system creates disincentives to work at older ages that have a statistically significant impact on retirement decisions. (At the same time, where tax credits are offered to seniors that would, in theory, result in better incentives to work, we don't see a significant effect on employment. So, we keep studying things.) Importantly, METRs are not the only thing that matters.

Comparing CSR/CSS with OAS/GIS

The programs are clearly similar, but the degree to which they offer similar benefits depends on the parameters chosen and which individuals you have in mind. The figure below offers a summary that would help us focus on marginal effective tax rates, representing an (unusual) single 65 year old that does not have any earnings or CPP/QPP as income.

How is this really different?

The key feature in what I present here is that the CSR and CSS are designed to be distinct programs, each with its own policy parameters. Consider, for example, a proposal to increase the age for Old Age Security. Under current legislation, this would automatically raise the age of eligiblity for both the OAS and GIS. While I've seen broad agreement that it would be sensible for eligibility ages to adjust (at least a bit) with life expectancy, there remains a lingering concern for those seniors who find themselves in difficult circumstances at age 65. By separating the program parameters, we can manage this. For example, suppose we increased the age of eligibility for CSR to 67, but left our choices for the CSS minimum guaranteed income alone. Our examples from above would have their benefits adjusted (as in the next chart below). For those at age 65 with lower incomes, the CSR they are not longer eligible for is replaced directly with CSS benefits. Those with higher incomes would remain without any benefits until older ages. 

Looking at benefits across the income distribution, I expect people would express concern about the seniors in the middle income group - life with an income around $50,000 and even $75,000 is generally viewed as quite modest. Perhaps we can be more generous? If the concern is affordability and standards of living, we would focus on changing CSS. For example, we could reduce the CSS benefit reduction rate to only 30% so that more people with modest income will get CSS support (as below). But my rough modelling suggests that simple change eats up any potential savings when compared to OAS and GIS.

So, what?

I'm not a politician. I don't belong to a political party. I don't know if anything I've suggested would successfully make it past the political feasibility test. I'd want to go through several rounds of analysis and consultations before suggesting the parameters chosen are the best options. 

But I think it's worth talking about. I see a desire among younger people to seriously think about the first pillar income support we offer seniors. I'm concerned about the current system's sustainability. Recent proposals to expand benefits have been not been viewed positively, rather they are met with demands to recognize the challenges facing young people. I see seniors who recognize those challenges. Today, there seems to be an opportunity and taste for rebuilding our first pillar and I think we should try.

Just one more thing... About those 60-64 year olds

We currently offer the  Allowances to some 60-64 year olds with low income as part of our first pillar programs. The benefit amount is roughly equivalent to the OAS and GIS offered to single seniors over 65. But eligibilty is tied to a person's marital history. If you have an older spouse (65 or older) or are widowed, you are deemed worthy of assistance. But if you have a younger spouse (under 65), were divorced, or simply never married, you are expected to be able to fend for yourself (or seek out other supports). Do we think it's easier to find a job if you have a younger spouse? Why would someone widowed several years earlier need more support than someone recently escaping a bad marriage? I have no idea. But I expect if you discussed the eligibility criteria in the 1970s, with fairly different social norms and values in place, it would not have been questioned. It's 2026. If it were up to me, I would phase out the Allowances entirely and replace it with a simple Survivor benefit to supplement incomes of widows and widowers of deceased pensioners until they are eligible for their own CSS benefits. 

And finally, some code for reference

In case anyone wanted to work with this, here's my Stata code for calculating CSR and CSS benefits. If you want to work on something, think harder about where/when to use family vs. individual income

program define CSSCSRben

****************************

*Note: prior to running this program, must load a dataset

*  with variables for age, married 0/1, earnings, CPPben, CPPage , other

*********************************************** 

* A.  Enter key policy parameters.

*** Step 1: Choose Minimum Guaranteed Income for each age 65+.

*   I've chosen:

*   Age 65 single slightly above current OAS/GIS maximum benefit as the baseline:

scalar basemgi = 1900*12

*  Allow this to increase by one percentage point per year age 66-75

*  No further increases at 75

gen MGI=basemgi if age==65 & married==0

replace MGI = 0.75*basemgi if age==65 & married==1

* Increment by 1% per year of age over 65 :

replace MGI = basemgi*(1+.01*(age-65)) if inrange(age,66,75) & married==0

replace MGI = 0.75*basemgi*(1+.01*(age-65)) if inrange(age,66,75) & married==1

* No further increments after age 75:

replace MGI = basemgi*1.1 if age>=75 & married==0

replace MGI = 0.75*basemgi*1.1 if age>=75 & married==1

*** Step 2: Choose the monthly amount and eligibility age for the CSR

* I've chosen:

scalar basecsr = 400

scalar agecsr = 65

*

*** Step 3: Choose parameters that define income exemptions and benefit reductions

* I've chosen:

* first earnings threshold (fully exempt):

scalar earn1 = 5000

* second earnings threshold (1/2 exempt):

scalar earn2 = 15000

* percent of CPP benefit exemptions (per year of age of initiation of benefits after 60)

scalar rateCPP = 0.036 

* additional exemption for integration with provincial programs:

scalar addexempt = 2208 /*ON GAINS is $92 per month, clawed back 50% rate for first 184 per month*/

* Threshold above which CSR Recovery tax applies:

scalar thresCSR = 93454

*rate at which we reduce CSR for any income:

scalar rateCSR = 0.25

*rate at which we reduce CSS for any non-exempt income:

scalar rateCSS = 0.50

************************************************

************************************************

* B. Calculate CSR benefits

gen CSR=max(basecsr*12-rateCSR*max(earnings+CPPben+other-thresCSR, 0), 0) if inrange(age,agecsr,999)

replace CSR=0 if age<agecsr

* recovery tax is applied later

************************************************

* C. Calculate CSS benefits

* C.1. Determine non-exempt income

*   Note: for couples, this is family income and we're calculating individual benefits

* Earnings:

gen clawearn = 0 if earnings<=earn1

replace clawearn = 0.5*(earnings-earn1) if earnings>earn1 & earnings<=earn2

replace clawearn = (earnings-earn2) + .5*(earn2-earn1) if earnings>earn2

* CPP benefits for delayed take-up:

gen clawcpp = CPPben - rateCPP*(CPPage-60)*CPPben

* apply provincial programs avoiding high METRs when finding total income:

gen clawincome = max(clawearn + clawcpp + other - addexempt, 0)

* C.2. Calculate CSS based on MGI applied

gen CSS = max(MGI - CSR - rateCSS*clawincome, 0) if married==0

replace CSS = max(MGI - CSR - rateCSS*clawincome/2 , 0) if married==1

************************************************

end