Monitoring the Future: Regulatory Responses and Realistic Expectations

Remarks by Superintendent Julie Dickson
Office of the Superintendent of Financial Institutions Canada (OSFI)
to the Asian Banker Summit 2009
Beijing, China | May 12, 2009

Introduction

The topic of this session is systemic risk and counterparty risk – what have we learned so far about monitoring systemic risk and counterparty risks in the immediate future, and in the long term, as we seek to build a more stable financial system.

Unfortunately, one of the lessons learned to date is how little we really know about systemic and counterparty risk. The good news is that we now have a lot more information, and are zeroing in on possible improvements in order to make the global financial system more resilient.

Considerable work has been done in the past several months to take advantage of both fresh memories and a general "will to act", to push for bold steps. We have a multitude of good reports -- Financial Stability Forum (FSF) reports, G-20 reports, UK Financial Services Authority (FSA) reports, International Monetary Fund (IMF) reports...the list goes on. We have almost as many reports post-crisis as we did financial stability reports prior to the crisis. So you may have a hint as to what I really want to talk about: whether we are over-promising and whether we need to be more realistic about what can be achieved as a result of the lessons learned from the financial turmoil.

First, let's recap what went wrong in the financial sector.

We witnessed excessive leverage, unacceptable credit concentrations, product innovation and extreme complexity, compensation incentives within financial institutions that spurred risk taking inconsistent with risk tolerance, inadequate corporate governance, and pressures on regulators not to impede competition and not to impede innovation. All these factors, combined with inadequate due diligence, led to the assumption of risks that were poorly understood, and inappropriately priced.
The key environmental factor that discouraged greater prudence was that times were very good for so long -- and anything that hampered expansion was given short-shrift (look at all the warnings that were expressed in the years leading up to the bursting of the bubble).

When things did go wrong, why was it so much worse than the 'usual' downturn?

We had moved from the upswing in a business cycle to a very large bubble in the US housing market. That, together with the extensive interconnectedness of financial institutions and economies around the world -- thanks to years of globalisation, free trade initiatives, technology that enabled more sophisticated product design and trading -- meant that when the bubble burst, the ramifications would be much more extensive.

Macroprudential Orientation of Regulation and Supervision

There is a growing consensus that a key element in improving safeguards against financial instability is to strengthen the macroprudential orientation of regulation and supervision. This means that you need to focus on the financial strength of individual institutions, as well as the system as a whole. As noted by Claudio Borio1, in an excellent paper dated April 14, 2009, the macroprudential approach has implications for (1) monitoring of threats to financial stability and (2) for the calibration of prudential tools.

Let's take each in turn.

Macroprudential Monitoring

Monitoring more than just banks on a stand-alone basis makes a lot of sense. Borio points out the importance of common exposures across institutions and of possible generalized over-extension in balance sheets during economic expansions. Notable examples are unusually rapid increases in credit and asset prices and unusually low risk premia.

Of course we saw just this in the run-up to the turmoil. Why didn't the whole world act? Because many laid out a lot of reasons as to why the premiums might be low; they also noted the fact that the system had been resilient to disturbances.

Presumably the other factor at play was a general desire to prolong the good times, with all the advantages that are associated with growth. It is one thing to take away the punch bowl at a party when things start getting out of control, but it is another to slow things down in a global economy, with the spectre of job losses and slower growth. In other words, you may realize that taking away the drinks at a party is probably a very good thing, but not as many policymakers will agree that similar steps are a good idea when the economy is growing and the cost of credit is cheap.

I said earlier that there were many, many, financial stability reports done before the turmoil. Will the new emphasis on macroprudential monitoring mean that these reports simply get thicker, with no real impact? I think the additional monitoring, combined with financial stability being added to mandates of regulators and central bankers, as recommended by the G-20, will help. It will cause everyone to spend more time assessing such risks. It means that financial stability will get focus and will require a weighing against other policy goals, such as home ownership. This did not really happen prior to the crisis.

While more focus on financial stability will ensure that regulators, central bankers and government finance ministries are armed with more information, which should lead to more discussions than in the past, we should not oversell this. Some seem to be selling macroprudential monitoring as the panacea of permanent financial stability, but I think this is an elusive goal.

Permanent financial stability is elusive for a lot of reasons. Let's start with a powerful force: human nature.

As pointed out by Counterparty Risk Management Policy Group (CPRMG) III "financial excesses fundamentally grow out of human behaviour...which on the upside of the cycle, fosters risk taking and on the downside fosters risk aversion". The CPRMGIII report goes on to make a number of very basic recommendations: know the risk you are taking, determine a risk appetite and monitor it, have good corporate governance, and ensure that control functions have authority and independence from the business units, communicate well within the firm, among others. What were we doing before this crisis if we did not know the value of these recommendations? We keep learning that we should not assume risks that we do not understand, that we should be diversified, and more... and we keep relearning!

Why is it that people do not learn these lessons; is it because they have short memories? Memory does fade with time, but I would also suggest it is because powerful incentives are at play. Perhaps we should be looking at how these incentives can blind us to some basic common sense principles.
Looking at incentives requires us to look at a lot more than just bankers' compensation packages. It requires us to go down some paths that might be quite sensitive; many of them involve the depth to which the financial sector has pervaded our culture. Simon Johnson, in a recent article in The Atlantic2, talks about the great wealth that the financial sector created and the resultant political weight not seen in the US since the era of J.P. Morgan (the man). It talks about the flow of individuals between Wall Street and Washington, about a whole generation of policymakers, regulators and central bankers being mesmerized by Wall Street, the extension of Wall Street's seductive power to finance and economics professors and the resulting lack of academic independence of thought, and lastly the cult of finance seeping into the culture at large with books and movies that became best sellers and increased Wall Street's mystique.

So while every downturn produces lessons to be learned, the question remains how to, or whether it is even possible to, alter human nature and the strong influence of incentives.

But let's get back to macroprudential monitoring. Work has already begun in earnest on macroprudential monitoring. A very recent contribution is the IMF’s Global Financial Stability Report, released on April 21, 2009. It includes an Annex on "Assessing the Systemic Implications of Financial Linkages" and an Annex on "Detecting Systemic Risk".

There are lots of good ideas in this work. But tellingly, chapter 3 starts, "Systemic events are intrinsically difficult to anticipate." Chapter 2 talks of serious data issues, the fact that it is a complicated task, and concludes, "In sum, monitoring global systemic linkages will undoubtedly become increasingly relevant, and thus the development of reliable tools for the task should proceed expeditiously."

The message is that we have a long way to go and there is no silver bullet. That being said, forcing more monitoring, and more discussion on the results of that monitoring, will help expand our information base and hopefully will help counteract some powerful forces -- but these are very powerful forces.

Macroprudential Calibration of Policy Tools

Claudio Borio's second point is the need to calibrate policy tools so as to encourage the build-up of buffers in good times, so that they can be drawn down as strains materialize. This would help banks absorb shocks and might also restrain the growth of risk-taking in good times.

This is quite challenging. Proposals include tying a portion of a bank's capital requirements to macroeconomic indicators (like credit growth and asset prices in the economy). Some are now looking at whether we should focus more on bank specific indicators, like the risk premia a particular bank is charging, or growth in the earnings of individual banks. The idea is to lean against the wind from a macroeconomic perspective, and to build up cushions of bank capital in good times, which can be drawn down in bad.

Linking bank capital to macroeconomic indicators is extremely challenging because we cannot accurately predict business cycles. Part of the challenge is on the upside (where economic forecasters often get it wrong), and part is on the downside (where economic forecasters often get it wrong). This is not because forecasters are not smart; it is because we cannot accurately predict the future.

There are also lots of false signals. Changing bank capital requirements based on where in the cycle the economy is, and knowing when to start and then stop the increase in capital requirements, is a tall order. We would also need to know when to start decreasing the capital requirements.

As it is tough to take away the drinks at a good party, many have suggested that we should try to make the increase or decrease in capital requirements automatic.

This involves predicting business cycles with more precision than we have today, and today we have no track record to speak of. There have been endless attempts over many, many, years to develop better models to enable more reliable and accurate forecasting of business cycles. The work continues.

While much time has been spent trying to forecast business cycles, not as much time has been spent monitoring the build up to, and the bursting of, asset price bubbles. The hope is that a more reliable indicator can be found there. Thinking on this topic is in its infancy, and we clearly need to do a lot more thinking. Until very recently many have said it is impossible to lean against growing asset bubbles, now, central banks are re-thinking whether it is possible. Time will tell.

Here are some examples of what I mean about our lack of ability to predict regular business cycles with accuracy. The Banque de France3 has noted that, in the past, the quantitative methods used by experts tended to produce errors in analyzing current and future developments, particularly during major economic turning points. One of the most striking examples is the last US recession of 2001, which few economists had anticipated. In the case of France, neither the downturn in Q3 2000, nor the recovery in Q3 2003 was predicted by most forecasters.

Indeed, we have had a lot of trouble with this ‘cycle’ as well. We now see that once the bubble burst, it took time to figure out what we were dealing with and we got it wrong there too. When the bubble burst in the fall of 2007, many thought we were dealing with a liquidity crisis; several months later it was clear it was a solvency crisis. Further, for some months many felt this was a cyclical downturn; now it is clear that it is something much more.

In early September 2008, the median growth forecast in the US for the fourth quarter was 0.2 per cent, according to a survey by Blue Chip Economic indicators. The actual outcome was a 6.3 per cent annualized decline.

So developing a black and white rule based on a macroeconomic indicator that automatically increases capital requirements in an upswing, and then automatically decreases them in a downturn, is fraught with challenges. There are risks if we get it wrong on the upside, but also serious risks if we get it wrong on the downside (i.e. if capital requirements are reduced at precisely the wrong time).

Another question is: How can we implement a macroprudential orientation to regulation and supervision when we still do not have a consensus on how to do macro policy?

As noted by Business Week⁴ in its April 16, 2009 cover story, macroeconomists are divided as to how concerned they are about economic instability. One group, in the tradition of Keynes, worries about self-perpetuating economic declines that leave the economy in a deep trough it cannot escape. Members of this group say government needs to break downward spirals with the kinds of aggressive policies the US is following now -- cutting interest rates and raising government spending. Other economists have more confidence that the economy is self-equilibrating. They believe low interest rates and heavy deficit financing will be ineffective.

There are other issues we need to look at as well in assessing whether we can calibrate macroprudential policy tools in the area of capital. Today, different banks may have different sensitivity to the cycle. A bank with a solid through-the-cycle rating system will be in a better position in a recession as capital requirements will not spike up; these banks should have more robust cushions (especially with a strong pillar 2 approach). But if a bank knows that its risk-based capital requirement will be increased by a multiplier based on a macro indicator that it does not control, then it will have an incentive to manipulate capital by minimizing the base (because the base gets multiplied up). Thus there might be more of an incentive to have a less prudent, more point-in-time methodology.

Another key issue is which indicator to use as the basis for a multiplier. It could be a macro indicator like growth in the economy or it could be specific to banks (growth in their assets or earnings). We would need to think carefully about this. Global growth does not necessarily match domestic growth and even in a domestic economy, different regions can have very different rates of growth. Would a requirement, applied only to a big bank, to increase capital due to domestic economic activity lead to incentives to create more regional banks?

As well, usually when capital increases are implemented by regulators, considerable pre-testing is done to determine the impact on each bank and whether transition rules are needed. This would be hard to do if adjustments were automatic. This might also be an issue if a particular bank had problems that would be made more challenging by an automatic capital increase for systemic reasons, as opposed to reasons associated directly with individual bank risk.

It is clear that this global turmoil has been so serious that it calls for a fundamental re-think, and an examination of new ideas. It is understandable that an automatic capital increase idea is gaining popularity because it offers something we all want -- a moderation of cycles, stable banks, and perhaps even the prevention of bubbles. However, as Borio himself notes, both measuring system-wide risks and calibrating policy tools are far from straightforward. He also questions how much the build up of capital cushions should be based on rules, versus discretion, and what size of cushion would you need so that they could be credibly run down at times of potential stress.

It would be very nice to have a formula that would make capital neat and tidy, going up in periods of optimism and going down in periods of pessimism. But the tendency to react rationally as markets did when they demanded higher buffers and pushed up bank capital levels in 2008, is not always proof that they were wrong from the outset to do so. While some can say that this action induced a deeper recession, others can say that markets took a view on how deep the recession would be, and they were right.

Other Efforts to Make the System Less Pro-cyclical

It may be that the most promising avenues to explore are higher quality tier 1 capital; leverage ratios; loan-to-value ratios; through-the-cycle estimates under pillar 2; and loan loss provisioning. The first four were very important in terms of the Canadian banking system and allowed the system to withstand the stress of global market turmoil and also successfully raise private capital. They help to make the system less pro-cyclical or more counter-cyclical.

The pillar 2 approach, admittedly, is not ’automatic’ and many are concerned about the ‘will to act’ to ensure Pillar 2 is robust in good times. Many banks and supervisors were in the process of improving practices to meet Basel II expectations at the time the problems materialized. Pillar 2 suggests that banks should set target capital levels (higher than the minimum) that take account of their overall risk profile, operating environment, and are mindful of the stage of the business cycle.

Assessing pillar 2 is clearly challenging, but going forward supervisors will have more support if there is more global focus on macroprudential monitoring. The debate will be whether people will agree that the indicators are sending correct signals, but more debate of this kind would clearly better position supervisors and create more ‘will to act’.

Further, we have identified other weaknesses in regulation and supervision globally that we are dealing with now.

Namely, Basel capital rules did not keep pace with rapid innovation in the securitization market and in the credit derivatives market, and Basel capital rules did not properly pick up counterparty risk. Our supervisory programs did not really cover compensation systems, yet we have seen that internal incentive structures are as important as the risk control framework itself.

Supervisors relied on banks being able to transfer risk through their securitization products, but we learned that reputation risk is real, which means that banks really do not transfer any risk. Thus all off-balance sheet activity needs to be considered. Supervisory oversight of liquidity was lacking. All of these areas need to be addressed; this work has a high priority and much progress has already been made. At the same time I am sure that work will continue on proposals to develop counter-cyclical capital buffers that are automatic and go up and down at the right times. At the end of the day, while it may be that something can be developed, I am not as hopeful as many are.

Realistic Expectations

All of this suggests to me that we may need to be more humble in our aspirations. Such is the message of Kevin Warsh, Governor of the Federal Reserve System, in a speech dated April 6, in which he said that policymakers' objectives should be more humble: maximize sustainable economic growth while reducing the incidence, severity and economic fallout of future shocks⁵.

Indeed, the reality is that management of financial institutions will respond to the incentives facing them, leading to decisions which may, and often do, impose costs on society that the institution does not bear. Regulators are there to minimize those social costs.

But the real world is ever-changing, dynamic, and innovative, with no complete understanding on the part of macroeconomists and regulators of how the parts acting on their own will affect the system.

We are in a system of action and reaction with constant but not always successful attempts at anticipation. So we set out a framework, the framework is based on experience and lessons learned. Financial institutions act within this framework, but then something happens and we react and try to improve, institutions react and the cycle begins again.

As long as we are prepared to have private (albeit constrained) financial institutions and a world in which we have neither perfect information about the future, nor a complete understanding of human behaviour, we will continue in this action-reaction dynamic. By constantly learning (through acting and studying) we strive to design incentives and systems that will encourage a healthy competitive financial sector. Although as noted in a recent Financial Times article6, there is not, and cannot be, an economic theory for everything. Herd behaviour, asset mis-pricing and grossly imperfect information are realities that cannot be assumed away.

Conclusion

In closing, I return to the title of this session -- which is all about discovering where the next threat to the banks may be lurking. This is a bit like an iceberg --the most dangerous part lurks underneath the water, and it can be difficult to detect.

The biggest concern now is whether the bold actions taken by central banks and governments will have the desired impact and, if they do, whether the financial aid being delivered to the global economy and global financial system will have unintended consequences. Lots of questions can be asked: wealth has shrunk but debt has not -- can we cure a global debt problem with more debt? Will cyclical recoveries be like they have been in the past, or will they be truncated, with several false dawns, as a result of consequences of actions being taken today? No one has the answers to these questions.
As my remarks indicate, we also have to try to ensure that the regulatory and supervisory prescriptions we are developing do not create new risks for the banking system. This requires that we think through all of the measures we adopt and constantly try to assess unintended consequences.

I have appreciated the opportunity to speak to you today. My messages are:

• macroprudential monitoring is a good idea especially if we focus on what it tells us and do not become complacent;
• tying capital requirements to macro indicators is fraught with challenges and it is too soon to say that it is a viable policy option;
• a lot of concrete steps, can be taken to improve system resilience and this will have an impact (quality of capital, leverage controls, fixes to capital rules in specific areas such as securitization and counterparty risk; focus on liquidity; and focus on compensation plans);
• future risks to banking are very difficult to predict at this time; and
• we should not "oversell" our capabilities, given the challenges.

Thank you

Major Sponsors
Sponsors
Exhibitors
Technology Partners
Official Airline Network	Official Freight Forwarder	Official International Broadcast Partner
Media Partners
Supporting Partner
Industry Research & Intelligence Partners