The lessons of Basel 3 and the path ahead for Canada — Remarks by Carolyn Rogers, to the 2018 RBC Capital Markets Canadian Bank CEO Conference, Toronto, Ontario, January 9, 2018


Let me start by wishing everyone a Happy New Year.

It’s my first time joining you at this event but OSFI’s participation in this conference is a fairly long standing tradition. The chance to share our perspective and our priorities for the coming year with the people in this room is a great way to get the year started.

I understand this year’s conference is focused on three topics: mortgage lending and the impacts of OSFI’s B20 guideline; the impact of fintech on bank business models; and regulatory capital and the impact of Basel 3. My remarks today will focus on your third topic area: regulatory capital and the impact of Basel 3.

The lessons found in finalizing Basel 3

As I planned my remarks for today I looked back to the last speech we made to this event, which was two years ago, in January 2016. In that speech my predecessor, Mark Zelmer, opened by telling you that 2016 would be the year that the final chapter on Basel 3 was written. And when Mark handed me the reigns in August of 2016 that’s what he told me as well. And yet here I am today, two years later, talking about an agreement that was finalized just three weeks ago. What took so long, you might ask? It’s a fair question and there is probably more than one answer, depending on who you ask.

I’ve had the benefit of a few weeks of hindsight over the holidays to think about this and like anything that doesn’t go quite as planned, there are a few clear lessons to be learned. I thought these lessons might be of some interest to you as they offer a measure of Canadian perspective on the challenges we faced in finalizing Basel 3; more importantly though, these lessons will shed some light on what to expect as we prepare to implement the final reforms in Canada.

The importance of supervision

Lesson one is the value of supervision.

The Basel Committee is comprised of 45 member organizations, from 28 different jurisdictions. The members are a mix of central banks and prudential supervisors, so while our mandates vary, we share a common objective of promoting the safety and stability of the financial sector. That shared objective provides a lot of common ground, but there can also be different perspectives about how best to achieve the desired outcome.

These varying perspectives stem from many legitimate differences among the member organizations — things like the legal environment they operate in; the economic and political climate they face; the number and size of banks they oversee; and, importantly, their own experience: how they and their banks fared through the financial crisis and other significant challenges they’ve faced over the years.

What I have observed is that these differences impact the degree to which different jurisdictions rely on supervision as a complement to standard-setting or rules in regulating banks. Jurisdictions that readily rely on their supervisors to influence bank behaviour approach standard-setting differently than those who rely less on supervision as a key part of their regime.

Regulators who are skeptical about the role of supervision tend to prefer rules that leave as little room for supervisory judgment as possible. Those rules also leave as little room as possible for the judgment of the banks — hence these regulators are often also skeptical about models. For these jurisdictions, standards must do all or most of the heavy lifting.

This puts a heavy burden on the rules to do things that, in some cases, rules are not well designed to do. Rules are static and generally designed based on what we have learned in the past. Banks are not static and the past is not always a good model for the future. Rules work best when they are accompanied by strong judgment and a forward-looking view. In the world of bank regulation that is the role of the supervisor.

Timing is important

The second lesson I took from finalizing Basel 3 is that timing is important and delaying the medicine is rarely a good idea.

There were a couple of timing challenges that made finalizing Basel 3 difficult. There was the inevitable loss of momentum that occurred as the financial crisis moved further and further into the past.

There was also the challenge of varying degrees of progress in implementing previously agreed reforms. This meant the different jurisdictions at the Basel table weren’t all starting from the same point, and different starting points meant different impacts. Jurisdictions that had not implemented previous reforms or had afforded longer transition times had a much steeper hill to climb. Ultimately, this made it very challenging to agree to both the design of the final reforms, and to the transition plan.

Changes to regulatory standards must always be accompanied by a reasonable transition period — predictability is important and abrupt changes to capital ratios rarely produce a good outcome. But it’s also generally true that if a change to regulatory standards is worth making, it’s worth making sooner rather than later.

All standards have their limits

The third lesson I took was that all standards have limits and a shelf life.

Bank capital, as everyone here knows, is expressed as a ratio: capital as a function of risk-weighted assets. The immediate focus following the financial crisis was on the first half of the ratio — the numerator. There were some adjustments to the denominator as well, to capture risks that were clearly under calibrated, but improving the quantity and the quality of capital was seen as the first imperative to improve the ability of banks to withstand periods of stress. And on this objective, considerable progress was made. Banks now hold significantly more going-concern loss-absorbing capital than they did before the crisis and before the first round of Basel 3. And the introduction of bail-in and non-viability contingent capital ensures regulatory capital and other liabilities absorb losses, rather than taxpayers, when banks fail.

However, for the improvements to the numerator to have their full impact, changes were also needed to the denominator. Over time, too much unwarranted variability has found its way into the risk-weighting of assets produced by some banks’ use of models. This wasn’t a universal problem and it wasn’t a problem we saw in Canada. But that was, in large part, the problem. Unwarranted variability — variability that could not be attributed to legitimate differences in exposures or experience — meant capital ratios across global banks were less comparable and the playing field was less level. As a result, a foundational element in the calculation of regulatory capital had lost credibility. Completing this last round of reforms was necessary to restore that credibility and to restore a level playing field.

The lesson is to never be over confident in your rules and to be prepared to adapt them when they’re not having the intended impact. The environment will change, banks will adapt and risks will shift. Continual evaluation is necessary to make sure standards are having their intended impact.

International standards are also limited in their ability to deal with unique domestic risks or conditions. This is an ongoing challenge in Basel negotiations — the balance between setting and adhering to international standards that facilitate stability and a level playing field, and respecting domestic differences and the need for local discretion.

Earlier in my remarks I talked about the differences in the legal and economic environment and in the make-up of the financial sector that exists among the 28 different jurisdictions at the Basel table. This leads to some challenging discussions when it comes time to set standards around assets that are highly affected by these differences.

One clear example of this is residential mortgages. While mortgages are a fairly universal product, and fairly universally liked by banks as a business line, the differences in the risk they represent to banks is heavily influenced by the legal and public policy environment. Canada is a perfect example of this, where a publicly backed mortgage insurance scheme means that a portion of the credit risk for mortgages on our banks’ balance sheets is actually borne by the federal government and not the banks themselves.

Other members of the Basel Committee will also tell you why mortgages in their jurisdictions have unique features that need unique treatment. You can appreciate then, that the regulators around the Basel table, including OSFI, have strong views about how best to address mortgage risk in the risk-weight framework.

Differences like these and like the ones I mentioned earlier, lead to some lively debates and some compromises. We work hard to influence the direction of international standards so they are fit for purpose in the Canadian context and we have been successful at this on many fronts. But we have also learned to not hesitate to make adjustments to international standards where we feel they are necessary to meet Canada’s objectives.

Applying the lessons

With that brief reflection on the past, let me now turn to the future. Before I do though, just a quick recap of the three lessons:

  1. supervision is critical to any strong regulatory regime; rules only get you so far;
  2. timing matters and delays will eventually catch up to you; and,
  3. all standards have their limits and their usefulness can deteriorate over time; domestic context and ongoing evaluation is critical.

On to how we will take these lessons forward.

Making capital buffers more useable

OSFI is a principles-based regulator and as result we have always viewed supervision as a cornerstone to our regime. We maintain a core set of rules and standards, particularly when it comes to something as important as capital. But we believe in leaving room for our supervisors, who have the closest view of a bank’s risk profile, to apply their judgment.

Our steadfast reliance on the role of supervision reflects our equally steadfast view that it is a bank’s board and management that bear the ultimate responsibility to maintain a level of capital commensurate with its risks. Pillar 2 therefore plays an important role in our capital regime. Pillar 2 creates a capital buffer over and above Pillar 1 that is calibrated through a variety of exercises that reflect both the bank’s own view of its capital adequacy and the supervisor’s assessment. Its purpose is to capture risks that are not adequately captured in Pillar 1; judgment is therefore key.

Pillar 1 capital is more rules driven; more formulaic. With credit to Basel 3, those rules are now significantly more risk-sensitive and transparent and Pillar 1 now contains a variety of discrete buffers, designed to absorb different macro-economic and systemic risks.

The challenge we have now, the one we need to address, is to ensure the buffers imbedded in the improved capital regime will work as advertised. More specifically, we need to ensure that banks are appropriately incented to draw on their capital buffers in times of stress, rather than resorting to deleveraging or other actions that will have pro-cyclical impacts. The rules governing Pillar 1 may be less pro-cyclical but the question is whether the overall regime for bank capital will drive less pro-cyclical bank behaviour when the pressure is on.

The Countercyclical Capital Buffer — the CCyB — that was brought in as part of the first round of Basel 3 was specifically designed as a Pillar 1 buffer that could be switched on and off in response to material changes in credit growth leading to macro stress. As its name implies, its activation is meant to lean against the build-up phase of the credit cycle and then its deactivation, or reduction, in a downturn, would reduce the risk that constrained credit would undermine economic recovery and increase credit losses.

To date OSFI has not activated the CCyB in Canada, despite what some may characterize as clear signs of credit build-up and macro stress conditions. There are several reasons for this.

The first is that the CCyB extends the conditions attached to the Capital Conservation Buffer, including automatic restrictions on dividends, buy-backs and compensation, upon a breach. This isn’t a bad design feature: removing discretion to enforce capital preservation in times of severe stress, particularly given behaviours we saw globally during the financial crisis, makes a lot of sense.

At the same time, mandatory capital conservation actions could send a strong, negative signal to the market that could, in turn, increase pressure on banks’ funding and liquidity. Banks will, understandably, be highly incented to avoid these triggers and may instead look to the alternative, more pro-cyclical actions we want to avoid. Moreover, given the relatively small and concentrated Canadian market, it’s possible these negative pressures could quickly spill over to other banks. OSFI’s concern is that this design feature of the CCyB may make the buffer less useable, not more.

Another reason we have not activated the CCyB is that Canadian banks are currently holding sizeable Pillar 2 buffers, with some of that capital meant to mitigate the same risks captured by the CCyB. It is not clear to us that activating the CCyB would be a better alternative to these buffers.

The difficulty with capital buffers held in Pillar 2, however, is that there is less transparency around the drivers of Pillar 2 capital and the consequences for banks breaching Pillar 2 capital expectations. And, in the absence of transparency the market attaches its own, often worst case, assumptions on constraints to using these buffers that may also make them less useable than we want them to be in times of stress.

We know that OSFI’s expectations and incentives are not the only ones banks react to. And we also know that banks are not the only ones who interpret our requirements and extrapolate their impact. If we are going to ensure our capital regime works as intended we need to ensure it is well understood, not only by banks, but by investors and the broader market. It’s not enough that regulators expect that a bank will draw down its capital buffers in times of stress; the market must also expect it.

This year we plan to make adjustments to our capital regime to increase transparency. This may take the form of a modified version of the CCyB, or simply greater transparency on Pillar 2 buffers. We are still considering options and consulting with stakeholders. Our objective is not to add capital but rather to increase the likelihood that the capital that is already there will be drawn on as intended should we find ourselves in a stressed event in the future.

I am going to now turn to some specifics on our plans for implementing the final round of Basel 3 reforms.

Step one in the transition — a new output floor

The lessons on timing — the need to update standards that are no longer serving their purpose and the wisdom of sensible (but not protracted) transitions, will both guide the first step we are taking in implementing the final chapter of Basel 3 here in Canada. That step is to replace the current capital output floor.

The current output floor was implemented by OSFI in 2008. It’s a backstop metric whose purpose is to ensure that the level of capital held by banks that rely on models does not fall below a prescribed level. Banks are required to calculate their capital under the standardized approach and the IRB approach and disclose both results. IRB capital cannot be less than 90 percent of the standardized requirement.

The current floor was designed to be a temporary bridge from Basel 1 and Basel 2 and therefore uses a Basel 1 definition of capital and risk-weighted assets (RWA), and a relatively high calibration point at 90 percent. The thinking at the time was that the design and calibration would evolve as the remaining parts of Basel 2 were introduced. As we now know, Basel 2 implementation was overtaken by the financial crisis and subsequently by the Basel 3 reforms. The final chapter of Basel 3 will bring with it a new output floor, based on a more risk-sensitive standardized approach, but the transition to this floor is only proposed to begin in 2022.

The output floor we have in place now is showing its age. It applies a definition of capital and RWA that are more than 10 years old and significantly less risk sensitive, and it requires legacy systems at banks to be maintained for the sole purpose of calculating a backstop metric. Leaving it in place for another five years has operational risk and it also risks distorting the incentives we view as critical to a credible capital regime. Therefore, as an interim step, OSFI will be replacing the current floor with the more risk sensitive Basel 2 floor, calibrated at 75 percent. Specific details of the changes will be published on OSFI’s website shortly and transition will begin next quarter, and finish in Q4 of this year.

Next steps

With a new output floor in place that we expect to bridge us to the Basel 3 floor, our focus will be on mapping the rest of the transition plan to bring the final set of Basel 3 reforms into Canada. You can expect that we will approach this exercise much the way we have in the past.

We are strong advocates of the value of international standards in Canada but we have also never shied away from deviating from those standards where it makes sense in our domestic market. In some cases those deviations result in higher standards or tighter transition timelines. In other cases we have chosen not to implement certain standards, to modify the standard, or to delay implementation.

In making these decisions, our first question is always whether a standard will contribute to the safety and soundness of Canadian banks and to the credibility of, and confidence in, our own regime. Standards that meet this test will always receive priority treatment from us and transition timelines for these standards will be set based on the domestic conditions and context.

The second question we ask is whether implementing a standard in Canada will contribute to a more level global playing field. Standards that meet this test are also treated with importance by OSFI, but transition timelines are set largely based on international conditions and context; if other major jurisdictions delay, we are also likely to delay.

These same two questions will continue to guide the next round of implementation. I can tell you, for example, that there are elements of these final Basel 3 reforms that we view as important to strengthening the capital regime here in Canada. A more risk-sensitive standardized approach is one example and, in particular, a more granular approach to risk-weighting residential mortgage assets.

We also think the added constraints to models are sensible and will go a long way to dealing with the unwarranted variability problem. We have always relied on a strong supervisory regime to monitor models used by Canadian banks and to date we have not seen the variability problem found in other jurisdictions. But we intend to remain vigilant on this topic and we welcome these adjustments as a way to level the playing field and ensure models are used responsibly by all banks.

There are also elements in the final reforms that, in our view, may not contribute to safety, soundness and public confidence in the Canadian financial system. The protracted implementation timeline combined with a 50 percent output floor as a starting point does not, in our view, send strong messages of confidence.

An output floor is a backstop measure — we’re firm advocates of this. If it’s set too high and it ends up becoming the binding constraint, it’s not serving its intended purpose. But if it’s set so low as to never bind it’s also not serving its purpose; nor is it credible. 50 percent is not, in our view, a credible floor.

A 50 percent floor also creates domestic level playing field issues, or at least the perception of issues. The big banks here in Canada are all on the internal ratings-based (IRB) framework but the rest of the industry is on the standardized approach. There is already a perception that the IRB framework offers a capital discount to banks that are approved to use it. A 50 percent floor will enhance rather than combat that perception and that is not confidence or credibility enhancing.

And finally, a 10 year timeline for implementation is unnecessarily long. Canadian banks are well positioned to meet the final Basel 3 requirements. Delaying the start of the implementation by five years and then stretching it out over another five years does not send a clear message of safety and soundness. You can expect a shorter transition here in Canada.


I will stop here so as not to pre-empt the consultation I told you was forthcoming. My goal today was to set the stage for that consultation by sharing the principles that we use to frame these important decisions and the lessons that will guide us as we move ahead.

To summarize, you can expect the following things from OSFI in the year ahead:

  1. changes to the transparency of our capital regime focused on improving the usability of capital buffers;
  2. an updated, more risk-sensitive capital output floor that will bridge to the Basel 3 output floor; and,
  3. a start to consultations on the final round of Basel 3 reforms, guided by what we have learned about building a capital regime that is credible and contributes to the safety and soundness of Canadian banks, and to a level competitive playing field.

Thank you for the opportunity to share these lessons and our priorities for the New Year. I’d be happy to take a few questions.