Millions of Canadians have accumulated balanced in their capital accumulation plans (CAPs), ….. but converting them to DB-like lifetime income remains as problematic as ever. There are over $5 trillion in retirement assets, and of that a third is in RRSPs and RRIFs. Together with money purchase DC plans and DPSPs, about $2 trillion of retirement assets have very limited lifetime income options. Even though retirement income is in the definition for each of these arrangements under the Income Tax Act (income is the outcome!), practice is quite different from the legislated ideal.
It is useful to visualize the landscape of retirement income from capital accumulation plans as a 3-by-3 matrix, where different distribution channels are mapped against type of income guarantee:
First, let’s consider the three types of retirement income guarantee. Effective financial planning requires availability of both withdrawal pots and annuities (guaranteed or non-guaranteed). This is because different bequest preferences, different risk tolerances, different price sensitivities and even different individual mortality expectations. Most of Canadian CAP assets in retirement are in withdrawal pots, which are bound to run out of assets at some point, in part because there are minimum withdrawal requirements reaching 20% per year for ages 95 and older.
The only type of annuities practically available in Canada is a guaranteed life annuity. However, pricing for guaranteed annuities remains unattractive, both for objective and subjective reasons. Objective reasons for unattractive annuity pricing are:
- strong (=expensive) guarantees may be excessive for most retirees
- voluntary annuitization regime increases prices through adverse selection
- pricing continues to be opaque and distribution costly due to reliance on human agents/brokers
Subjective reason include:
- bequest motives
- persistent irrational stereotypes (pension = good, annuity = bad)
- misunderstood longevity risk (i.e. only planning for average longevity)
The fact that most retirement savings are in withdrawal pots is a significant public policy problem as it creates additional risks for unfunded social security (OAS/GIS). Retirees who run out of assets in retirement will fall back on the social safety net and may experience a reduction in their consumption level at a time when they are least capable of adapting to circumstances.
Low take-up rate for guaranteed annuities is not a structural industry problem. It is more of an inefficiency problem that can be solved over time, particularly when Canadians have better pricing benchmark in a form of non-guaranteed annuities. Judging by historical experience of two grandfathered Canadian plans (UBC Faculty Pension Plan and Saskatchewan Pension Plan), non-guaranteed annuities can result in much higher annuitization rate. If that observation is confirmed, there will be no grounds for Canadian life insurers to blame the “annuity puzzle” and do nothing to solve it.
Now, let’s turn to different distribution channels. We tend to think of employment-based plans as the starting point. However, majority of assets in retirement are in individual/retail accounts. This is because the Canadian retirement system is built on clear delineation of accumulation (during employment years) and decumulation (in retirement). A typical employer does not have the need to continue the relationship with their former employees after they leave work, including retirement. Most employment-based plans don’t have income distribution mechanisms (such as in-plan variable benefits or an associated Group RRIF).
As such, members of employment-based plans are usually pushed out of the employment-based plans after termination or retirement. In addition, a growing proportion of employment-based plans is represented by so-called Group RRSPs, which are legally a collection of individual contracts between subscribers and a financial institution. Although the sponsor has certain fiduciary rights and responsibilities, all government statistics treats them as retail account and there is no regulatory oversight with respect to retirement outcomes at the plan level.
Participation-based plans are practically non-existent in Canada. These are plans, where participation is not tied to employment. The 2 examples I know of are the Saskatchewan Pension Plan (SPP), which is open to all Canadians, and the VRSP in Quebec, which is similar to PRPP in other provinces but permits individual membership. Access to both plans is limited. VRSP is only available in Quebec, and financial institutions are not very interested in promoting these plans due to regulatory price caps. SPP has contribution limits, which make it necessary for many contributors to combine SPP with another plan.
On the balance, many of the potential solutions are either non-existent (such as non-guaranteed annuity for a retail account), or not accessible (such as any retirement income solutions within employment-based capital accumulation plans, or any participation-based plans that are not tied to employment).
How the access problem can be fixed? First, we need to understand how asset transfers work among different tax-deferred accounts under the Income Tax Act.
Simplified ecosystem of tax-deferred retirement savings accounts
In this diagram, the qualifying annuity is a special kind of life annuity from a licensed annuity provider. Transfers from tax deferred accounts to a qualified annuity are exempt from taxation under ITA s.60(l), so in effect the qualifying annuity is another tax-deferred mechanism, with the major difference that there is no possibility of transferring assets out of the annuity. Importantly, qualified annuities have certain restrictions on the variability of the periodic payments under ITA s.146(3)(b)(iii)-(v), which is a key factor when it comes to non-guaranteed annuities.
Single lines on the diagram are permissible transfers, and the double lines indicate that, where a RRIF governs a trust (such as a RRIF issued by a trust company), the trust can hold a qualifying annuity as an investment. RRIF and RRSP contracts held by insurance companies normally don’t have a trust wrapper, but their assets can still be transferred on a tax-free basis to purchase a qualifying annuity.
The non-guaranteed annuity (VPLA) was recently permitted for RPPs and PRPPs, which is a small step in the direction of solving the lifetime income access problem. The bulk of Canadian retirement savings is in RRSP and RRIF accounts, so access to VPLA relies on availability of transfer to either a money purchase RPP or to a PRPP. However, both RPPs and PRPPs are employment-based plans, so one needs to be employed by a company providing those plans. Also, both RPP and PRPP may or may not be willing to accept incoming transfers, depending on the position of the sponsoring entity. These are serious barriers that dramatically reduce accessibility of VPLA for Canadians with capital accumulation assets.
PRPP itself is practically not available despite being permitted since 2014. The key problem is that PRPP relies on the willingness of for-profit financial institutions to be asset aggregators for small employment plans in presence of onerous price controls. Given significant system development costs and marketing costs, making PRPP viable would require phasing in of price controls over time in proportion to rising asset level. Importantly, price controls can’t be a selling point to the public if the government relies on for-profit institutions to make it work, at least not immediately.
There are two main avenues for solving the access problem. Given the size and the structure of the retirement savings industry, I believe Canada needs both. Canadian financial industry exists in a close partnership with the state. For example, banking is dominated by 6 large banks who are protected from competition. However, the banks need to provide sufficient access to credit, and international comparisons indicate that Canada has done good job on that front in the last two centuries. The government has periodic legislative reviews and recharting of the banks as a proverbial stick to match the carrot. In the book “Fragile by Design”, Charles Calomiris and Stephen Haber wrote: “Widespread suffrage in Canada has given Canadian bankers strong reasons to follow the dictum “Pigs get fat, hogs get slaughtered””. Same logic applies to the life insurance business.
Fundamentally, Canada is a relatively small market with an export-oriented economy, and it evolved to have strong regulatory powers in combination with high market share concentration for providers in many sectors. There may be many problems with this picture, but one particularly relevant is that the legislators may not have good appreciation of the deficiencies in the way large financial institutions serve the domestic market, so they might not know when the stick should be applied.
One avenue for solving the issue of access to non-guaranteed annuities is to permit participation-based, not employment-based retirement savings pools that can act as decumulation-only VPLA aggregators. This is the idea advanced in the report on Dynamic Pension Pools (DPPs) published by the National Institute on Ageing, and it can be achieved by redesigning the PRPP regime or creating a parallel section of the Income Tax Act. These purpose-built lifetime income pools could be run by a trust established for the benefit of the participants by labour unions, industry associations or other non-for-profit institutions.
The other avenue is to expand the definition of “retirement income” for qualifying annuities to permit variability in payments caused by mortality experience. That solution would be run by for-profit life insurers, even though the insurers would not take any mortality risk and operate the VPLA segregated fund as an “administrative service only” (ASO) arrangement.
Both of these solutions should be pursued in Canada, since they enable different types of delivery agents. If we look at the Australian superannuation industry for inspiration, both for-profit and non-for-profit super trusts co-exists and can be successful and efficient.
It is important to add here that there should be strong subscriber protections for lifetime income pools since the structure involves irrevocable long-term commitments with uncertain outcomes. A financial institution running a lifetime income pool can exert significant influence on the choice of investment strategy and the underlying subadvisors (including a full slate of in-house subadvisors). From that perspective, all lifetime income pools, including pensions from RPPs or from VPLA pools, should have a regulator concerned with long-term member/subscriber protection. This is different from individual life annuities with fixed or indexed payouts, where there is little room for the financial institution to do anything harmful to the annuitant short of endangering the whole insurance company.
 There are some exceptions, such as certain RPPs having grandfathered built-in lifetime income options. Certain Group RRSP arrangements are paired with group annuity purchases or purchases of units group deferred annuities, but suffer from complexity, opaque pricing and reliance on a specific annuity provider.
 TFSA (Tax-Free Savings Account) is a tax-exempt rather than tax-deferred account. Although it has become a significant component of overall system of registered accounts, there are no permitted transfers between tax-exempt and tax-deferred accounts that can be made without tax consequences. Also, the stated regulatory objective of the TFSA is not retirement savings, but general savings (a personal emergency fund).