Tuesday, November 29, 2016

Canada's Great Pension Debate?

In a response to Bernard Dussault, Canada's former Chief Actuary, Bob Baldwin, a consultant and former board of director at PSP Investments, sent me his thoughts on Bill C-27, DB, DC and target benefit plans (added emphasis is mine):
Bernard Dussault has circulated an article and slide presentation in which he has provided a endorsement of DB workplace pension plans coupled with an expression of concern about certain design features of DB plans that he sees as discriminatory. I share his preference for DB. But, I think his account of DB is incomplete and avoids certain issues and problems in DB that DB plan members, people with DB governance responsibilities and DB advocates should be aware of.

In his article “How Well Does the Canadian Landscape Fare?” Bernard says: “… DB plans offer better retirement security because they attempt to provide a predetermined amount of lifetime annual retirement income at an unknown periodic price.” The unknown nature of the price is unavoidable given the factors that determine the price that have magnitudes that cannot be foreseen such as: future wages and salaries, investment returns and longevity.
In context, two attributes of DB in its pure form are important to note. First, all of the uncertainty will show up in variable contribution rates and none in variable benefits. Second, the plan sponsor or sponsors have an unlimited willingness and ability to contribute more to the plan if need be.

The second of these attributes is, in principle, largely implausible. There simply are not sponsors who can and will contribute more without limit. Moreover, as I have noted in several publications, in practice when combined employer and employee contributions get up to the 15 to 20 per cent level, even jointly governed plans that were purely DB begin allocating some financial risk to benefits – usually by making indexation contingent on the funded status of the plan.

Moreover above some level, escalating pension contributions begin to depress pre-retirement living standards below post retirement levels. Even recognizing that the impact on living standards of a particular combination of benefit levels and contributions will vary from member to member in a DB plan (you can’t make it perfect for everyone), it is still desirable to try to avoid depressing pre-retirement living standard below the post-retirement level. The object of the workplace pension exercise is to facilitate the continuity of living standards and depressing pre-retirement living standards below the level of post-retirement living standards is not consistent with that objective. You can have too much pension!

The contributions that are relevant to the question whether contributions are depressing pre-retirement living standards to too low a level include both employer and employee contributions. This is because in most circumstances, the economic burden of employer contributions will fall on the employee plan members. This happens because rational employers will reduce their wage and salary offers to compensate for foreseeable pension contributions. There may be circumstances where an employer cannot shift the burden fully in the short term. But, in the normal case, the burden will be shifted. To the extent that required pension contributions are varying through time and being shifted back to the employee plan members, the net replacement rates generated by DB plans are clearly less predictable than the gross replacement rates.

Under the subheading “Strengths of DB plans”, Bernard’s slides include the following statement “investment and longevity risks are pooled, i.e. not borne exclusively by members, be it individually or collectively.” Having introduced the word “exclusively” the statement is probably correct. But, there are a variety of risks to plan members in DB plans. As was noted in the previous paragraph, there is a risk in ongoing DB plans that the pre-retirement living standards will be depressed below post-retirement levels. There is also a risk that a DB plans will get into serious financial difficulty and benefits will be reduced for future service and/or the plan will be converted to DC for future service. Both of these outcomes focus financial risks on young and future plan members as does escalating contributions. Finally, in the event of the bankruptcy of a plan sponsor, all members will face benefit reductions if the plan is not fully funded.

Bernard’s article and slides include reference to provisions in DB plans that he finds discriminatory. For me, the provisions on which he focuses raise a related issue.

The key factors that determine outcomes in all types of workplace pension plans are the same: rates of contributions, salary trajectories, returns on investment, longevity and so on. So what is it that allows a DB plan to provide a more predictable outcome? It is basically two distinct but related things: varying the contribution (saving) rate through time to meet a pre-determined income target; and, cross-subsidies within and between different cohorts of plan members.

In and of itself, the existence of cross-subsidies is not a bad thing. It is fundamental to all types of insurance and insurance is worth paying for. But, what DB plans could do much better than they do is to help plan members understand what the cross-subsidies are and how much they cost. This would allow plan members to decide what cross-subsidies are “worth it” and which ones are not worth it. My guess would be that within limits established by periods of guaranteed payments, members would accept cross-subsidies based on differential longevity in order to have a pension guaranteed for a lifetime. There may be less enthusiasm for a cross-subsidy from members whose salaries are flat as they approach retirement to those whose salaries escalate rapidly.

With respect to the plan features that Bernard has identified as discriminatory, my first and strongest inclination is to shine light on them so plan members have a chance to decide what is and is not acceptable.

The relatively predictable outcomes of DB plans in terms of the benefits they provide are clearly desirable. DC plans – especially those that involve individual investment decision-making and self-managed withdrawals – impose too much uncertainty on plan members with respect to the retirement incomes they will provide and demand excessive knowledge, skills and experience of plan members – not to mention time. As a renown professor of finance put it, a self-managed DC arrangement is like asking people to buy a “do it yourself” kit and perform surgery on themselves.

It is unfortunate however, that so much of the discourse about the design of pension plans is presented as a binary choice between DB and DC. There are several reasons why this is unfortunate.

First, the actual world of pension design is more like a spectrum than a binary choice. In Canada and across the globe, there are any number of pension plan designs that combine elements of DB and DC. Financial risks show up in both benefits and contributions.

Second, sometimes plans are managed in ways that are not entirely consistent with formal design features of plans. In the 1980s and 1990s when returns on financial assets were high and wage growth low, many DB plans ran up surpluses on a regular basis and these were often converted into benefit improvements. Many DB plans were managed as if they were collective DC plans (investment returns were determining benefits) with DB guarantees. The upside investment risk was not converted into variable contributions.

Third and finally, some plans that are labelled DB fall well short of addressing all of the financial contingencies that retirees will face. This is most strikingly true of DB plans that make no inflation adjustments. What is defined – in terms of living standards – only exists during the period immediately after retirement.

The difficulty in knowing exactly what we are referring to in using the DB and DC labels has not rendered the terms totally meaningless. As noted above, there are plans that are mainly DB but allocate some financial risk to the indexation of benefits. There are also a few grandfathered Canadian DC plans that include minimum benefit guarantees. The union created multi-employer plans have fixed rates of contribution like DC plans, pool many risks like a DB plan, but allow reductions in accrued benefits. The point is not to get stuck on the DB and DC labels but to understand how financial risks are being allocated.

The basic strength of DB plans in providing a relatively predictable retirement income is not diminished by the issues raised above. But, it is clear that for the well being of plan members and sponsors, the basic strength of DB has to be reconciled with acceptable levels and degrees of volatility of contributions. It also has to be reconciled with reasonable degrees of cross-subsidization within and between cohorts of plan members. With regard to cross-subsidies between cohorts, a regime in which accrued benefits cannot be reduced places all financial risk on young and future plan members. Target benefit plans try to avoid this problem by spreading the risk sharing across all cohorts as in the union created multi-employer plans.

Bernard’s article and slides touch on a number of regulatory issues. The only one of these that I will comment on is the prohibition of contribution holidays. This suggestion is put forward along with the use of realistic assumptions that err on the safe side.

In an environment where investment returns are consistently greater than the discount rate (e.g. the 1980s and 1990s), the practical effect of banning contribution holidays will be to build up surpluses that will significantly exceed what is required to protect against downside risks that plans may face. Rather than mandate the banning of contribution holidays, an approach that would create downside protection without building up excess surplus would be to mandate the adoption of funding policies that prohibit the use of surplus to reduce contributions or increase benefits until threshold levels of surplus have been achieved. The threshold levels should take account of the riskiness of the pension fund’s investments and the maturity of the plan. Funding policies of this general sort have begun to emerge in large Canadian plans and should be made to be common practice.
First, let me thank Bob Baldwin for sharing his thoughts on DB and DC plans. Bob is an expert who understands the complexities and issues surrounding pension policy.

In his email response, Bob added this: "(in a previous email he stated) my views were quite different from Bernard’s. I am not sure whether I should have said “quite different” “somewhat different” “slightly different”. In any event, they are attached. You would be correct in inferring that they cause me to be more open to Bill C-27 than Bernard is."

Go back to read my last comment on Bill C-27, Targeting Canada's DB Plans, where I criticized the Trudeau Liberals for their "sleazy and underhanded" legislation which would significantly weaken DB plans across the country. Not only do I think it's sleazy and underhanded, I also find such pension policy inconsistent (and hypocritical) following their push to enhance the CPP for all Canadians.

In that comment, I shared Bernard Dussault's wise insights but I also stated the following:
Unlike Bernard Dussault and public sector unions, however, I don't think DB plans can be bolstered just by prohibiting contribution holidays (something I agree with). I believe that some form of risk-sharing is essential if we are to safeguard DB plans and make sure they are sustainable over the long run. Target benefit plans are not the solution but neither is maintaining the farce that DB plans can exist with no shared-risk model.

[Note: To be fair, after reading my comment, Bernard sent me his proposed DB pension plan financing policy which "promotes true risk sharing at any level (ideally 50%/50%) between the plan sponsor and the plans members, in such a way that not only would both parties share the cost but also the 15-year amortization of surpluses and deficit."]

I take Denmark's dire pension warning very seriously and so should many policymakers and unions who think we can just continue with the status quo. We can't, we need to adapt and be realistic about what defined-benefit pensions can and cannot offer in a world of low or negative rates.

On that note, let me once more end by sharing this nice clip from Ontario Teachers' Pension Plan on how even minor adjustments to inflation protection can have a big impact on plan sustainability.

The future of pensions will require bolstering defined-benefit plans, better governance and a shared-risk model, which is why pensions like OTPP, HOOPP, OMERS, OPTrust, CAAT and other pensions will be able to deliver on their promise while others will struggle and will face hard choices.
This means while I firmly believe the brutal truth on defined-contribution plans is they aren't real pensions and will lead to widespread pension poverty because they shift retirement risk entirely on to employees and that the benefits of defined-benefit plans are grossly underestimated, I also firmly believe that some form of shared-risk must be implemented in order to keep DB plans solvent and sustainable over the long run.

I mention this because public sector unions think I am pro-union and for everything they argue for in regards to pension policy. I am not for or against unions, I am pro private sector, as conservative as you get when it comes to my economic policies and fiercely independent in terms of politics (have voted between Conservatives and Liberals in the past and will never be a card carrying member of any party).

However, my diagnosis with multiple sclerosis at the age of 26 also shaped my thoughts on how society needs to take care of its weakest members, not with rhetoric but actual programs which fundamentally help people cope with poverty, disability and other challenges they confront in life.

All this to say, when it comes to pension policy, I am pro large, well-governed DB plans which are preferably backed by the full faith and credit of the federal government and think the risk of these plans needs to be shared equally by plan sponsors and beneficiaries.

Now, Bernard Dussault shared this with me this morning:
I sense that the description of my proposed financing policy for DB pension plans deserves to be further clarified as follows:

My proposed improved DB plan is essentially the same as Bill C-27's TB plan except that under my promoted improved DB:
  1. Deficits affect only active members' contributions (via 15-year amortization, i.e. through a generally small increase in the contribution rate), and not necessarily the sponsor's contributions, as opposed to both contributions and benefits of both active and retired members under Bill C-27.
  2. Not only are contribution holidays prohibited, but any surplus is amortized over 15 years through a generally small decrease in the members' and not necessarily sponsor's contribution rate.
Therefore, my view is that if my proposed financing policy were to apply to DB plans, TB plans would no longer be useful. They would just stand as a useless and overly complex pension mechanism.
But Bob Baldwin makes a great point at the end of his comment:
Rather than mandate the banning of contribution holidays, an approach that would create downside protection without building up excess surplus would be to mandate the adoption of funding policies that prohibit the use of surplus to reduce contributions or increase benefits until threshold levels of surplus have been achieved. The threshold levels should take account of the riskiness of the pension fund’s investments and the maturity of the plan. Funding policies of this general sort have begun to emerge in large Canadian plans and should be made to be common practice.
Funding policies need to be mandated to prohibit the use of surpluses to reduce contributions or increase benefits until certain threshold elements of pensions are achieved.

Take the example of Ontario Teachers' Pension Plan and the Healthcare of Ontario Pension Plan, two of the best pension plans in the world.

They both delivered outstanding investment results over the last ten and twenty years, allowing them to minimize contribution risk to their respective plans, but investment gains alone were not sufficient to get their plans back to fully-funded status when they experienced shortfalls.

This is a critical point I need to expand on. You can have Warren Buffet, George Soros, Ken Griffin, Steve Cohen, Jim Simons, Seth Klarman, David Bonderman, Steve Schwarzman, Jonathan Gray and the who's who of the investment world all working together managing public pensions, delivering unbelievable risk-adjusted returns, and the truth is if interest rates keep tanking to record low or negative territory, liabilities will soar and they won't produce enough returns to cover the shortfall.

Why? Because the duration of pension liabilities is a lot bigger than the duration of pension assets so for any given drop (or rise) in interest rates, pension liabilities will soar (or drop) a lot faster than assets rise or decline.

In short, interest rate moves are the primary determinant of pension deficits which is why smart pension plans like Ontario Teachers' and HOOPP adjust inflation protection whenever their plans run into a deficit.

This effectively means they sit down like adults with their plan sponsors and make recommendations as to what to do when the plan is in a deficit and typically recommend to partially or fully remove inflation protection (indexation) until the plan is fully funded again.

Once the plan reaches full-funded status, they then sit down to discuss restoring inflation protection and if it reaches super funded status (ie. huge surpluses), they can even discuss cuts in the contribution rate or increases in benefits, but this only after the plan passes a certain level of surplus threshold.

In the world we live in, I always recommend saving more for a rainy day, so if I were advising any pension plan which has the enviable attribute of achieving a pension surplus, I'd say to keep a big portion of these funds in the fund and not use the entire surplus to lower the contribution rate or increase benefits (apart from fully restoring inflation protection).

I realize pension policy isn't a sexy topic and most of my friends love it when I cover market related topics like Warren Buffet's investments, Bob Prince's visit to Montreal, Trumping the bond market or whether Trump is bullish for emerging markets.

But pensions are all about managing assets AND liabilities (not just assets) and the global pension storm is gaining steam, which is why I take Denmark's dire pension warning very seriously and think we need to get pension policy right for the millions retiring and for the good of the global economy.

In Canada, we are blessed with smart people like Bernard Dussault and Bob Baldwin who understand the intricacies and complexities of public pension policy which is why I love sharing their insights with my readers as well as those of other experts.

It's not just Canada's pension debate, it's a global pension debate and policymakers around the world better start thinking long and hard of what is in the best interests of their retired and active workers and for their respective economies over the long run.

As I keep harping on this blog, regardless of your political affiliation, good pension policy is good economic policy, so policymakers need to look at what works and what doesn't when it comes to bolstering their retirement system over the long run.

Below, something that works for Ontario Teachers, HOOPP and other pension plans that have experienced pension shortfalls in the past is to adjust inflation protection when their plan is in a deficit.

It's not rocket science folks, in order to get stable, predictable pension payments for life, you need good governance and members need to accept some form of a shared risk model to keep these pension plans sustainable and viable over the long run.

No comments:

Post a Comment