Wednesday, May 02, 2007

Trading & Derivatives Businesses of Banks

BMO Capital Markets, 2 May 2007

1. The Numbers Tell a Compelling Story Over the Past Decade

Banks do not disclose the profitability of their trading and derivative operations. However, it is possible to establish a crude estimate of profitability by combining data and various assumptions.

We estimate that the trading business of Canadian banks produces $2.0-2.5 billion in after-tax earnings annually, with an ROE in excess of 25%. They amount to between 10% and 20% of annual system-wide earnings and have been growing at a compounded annual rate of about 10% over the past decade.

Our analysis of the economics of trading is based on a review of trading revenues and regulatory capital over the past decade. Regulatory capital associated with the derivative and trading business is a reasonable proxy for invested capital and we have used a Tier 1 ratio of 10% of risk-weighted assets, comprised entirely of common equity. The major assumption we apply is an estimated profit margin on revenues. It is clear that the after-tax margin in the trading business is high (over 30%), reflecting the fact that there are few costs other than people and systems (and the hidden 'costs' of maintaining a solid credit rating and funding).

We have used TEB revenues, meaning the variations that can be caused by the tax rate are mitigated. This is not to say that banks do not 'locate' these businesses in offshore (low-tax) jurisdictions- but it is tricky to validate to tax authorities that the business is offshore without traders being physically located in those locations (which they generally are not).

While some will question our estimates of profitability of these businesses overall, these estimates are based on anecdotal evidence and discussion with senior executives over several years. What is absolutely clear in our mind though is that Trading and Derivative businesses are much higher ROE than traditional corporate lending activities, that there is more of a mark-to-market approach (rather than an accrual approach used in loan books) and that both exhibit volatility.

We believe that the high ROE in this business has been a contributing factor in producing more consistent earnings for the bank group through this down-cycle. For perspective, in 2002 overall bank system-wide ROE was 11.3%, well above the previous trough of 5.5% in 1992. Excluding the 31% ROE that we estimate was earned on the trading business, ROE for the bank group would likely have troughed 1 to 2% lower.

This, in our opinion, is one of the more powerful arguments in favour of banks expanding further in trading. The trading businesses do not appear to be correlated to the credit cycle. When bank earnings fell in 2002 (because of a spike in loan losses), trading earnings made up over 25% of the total earnings for Canadian banks.

Despite the material profitability of the business, Canadian banks do not appear to have been aggressively expanding their trading businesses. Trading revenues for the banking system have grown at a solid, but not aggressive rate. Indeed, in the 1st quarter of 2007, trading revenues across the entire system of $1.8 billion were a record, but do not appear to be meaningfully different from the levels achieved over the period 2001 to 2006. In fact, given the build-up in RWA over the past decade, trading revenues have actually been disappointing over the past two years.

As we have mentioned, from a longer-term perspective, trading revenues have compounded at about 10% over the past decade. Note that Canadian banks' earnings have grown in line with this.

For those who fear trading, it is safe to say that Canadian banks are no more dependent on trading than they were 10 years ago.

2. An Overview of the Businesses that Make Up Trading

When we say trading and derivatives, many perceive that the Canadian banks are operating in a business with gun-slinging, temperamental egomaniacs that are uncontrollable. Though stereotypes do certainly exist, this is a simplistic assumption and hides the fact that 'trading' really is an assortment of businesses that include cash and derivatives trading, proprietary and agency activities and numerous products across OTC and exchange-traded markets.

Though we view trading as a business distinct from origination and structuring, it is sometimes difficult to see where one ends and the other begins. As we discuss in the Operational Risk segment, the skills required involve understanding, measuring and controlling risks and exposures across a disparate group of businesses.

It is impossible to narrowly define trading and derivative businesses. First, they are evolving; second, they differ by bank; and, third, they are not necessarily defined consistently. As such, our attempt to 'segment' the business is fraught with risk, and is only meant as an indication of the scale of the businesses. We broadly believe that there are six businesses that make up the majority of Canadian bank trading books: Foreign Exchange, Interest Rate, Commodities, Equity and Index Derivatives, TEB activities and Credit Markets. Note that unless specifically stated, our segments include both the derivative and cash sides of the market.

Note that the proportion arising from these businesses changes over time and across banks. Investors should also be aware that trading 'begets' other businesses such as origination, structuring, underwriting and advisory assignments, and vice versa. For example, banks that have strong fixed income trading platforms are probably large underwriters of bonds.

It also goes without saying that for OTC products, the strength of the bank credit rating is an important variable in ensuring that Canadian banks are competitive against the global trading houses such as JP Morgan, Goldman Sachs and Citigroup.

Foreign Exchange: Canadian banks have historically been very strong in the foreign exchange derivatives business, with large swap books and trading operations to service clients in 'loonies' around the clock. The established, entrenched relationships with corporate and commercial Canada assures that Canadian banks have a disproportionate share of the flow in this business, and we see little reason why this would change over time. Importantly, Canadian banks probably do a better job in servicing mid-market issuers than do the global trading houses (because of commercial lending relationships)- and we would actually expect margins to be better on this business than for large corporates. We believe that Royal Bank is particularly strong in the Forex business and appears to have invested more than others. It has also developed strength in Aussie and Kiwi dollars.

Interest Rate: This is closely related to the foreign exchange business, and all Canadian banks also have large interest rate swap books.

With the reduction in the demand for funding from government issuers over the past decade, banks have had to deal with a reduced cash market, but with the development of the Maple Bond market in the past couple of years, this has once again become important. Similar to Forex, we believe that bank involvement in corporate and commercial Canada ensures good access to flow and provides a competitive advantage versus the global trading houses. If someone plans to swap a Canadian dollar or a Canadian interest rate, we believe that Canadian banks will be very competitive on pricing.

Commodities: The single most important change in this business was the growth and subsequent collapse of Enron as a major force in this market. In our simplistic mind, banks appear to have clear advantages versus corporate competitors because they have the regulatory oversight and systems to deal with businesses like this. Furthermore, we believe that Canadian banks, with tremendous involvement with issuers in energy and metals, are well positioned to compete. Scotia is a leader in the gold business and it is one of the 5 banks that participate in the London Gold Fix.

Equity and Index Derivatives: With the reduction of commission levels in the cash markets, the importance of the cash business has been reduced and derivatives have become the core of profitability for banks in their equity business. Increased complexity and demand by investors (pension plans, particularly) have resulted in banks packaging and trading more unique products for their clients.

Taxable Equivalent Basis (TEB) Activities: What began as an adjustment for banks for their holding of preferred shares has transformed into an arbitrage business resulting from the different tax treatment of securities held by various entities. In Table 3, we show the growth in the TEB adjustment over the past several years as evidence of its increasing importance as banks become more prepared to structure transactions to pass the 'smell test' regarding the tax treatment. Scotia has the biggest TEB adjustment (about 40% of Scotia's trading revenues are TEB).

Credit Markets: This business is growing substantially for three reasons. First, the development of the syndicated loan market has standardized much of the traditional loan market. Second, banks have become increasingly reluctant to hold loans to maturity. And, third, there has been tremendous growth in demand by institutional investors for credit risk. As such, there is more secondary activity and now a well-developed (if sometimes somewhat illiquid) derivatives market. Banks have always been in the business of evaluating credit risk, and this market is a natural for those with strong track records. In this business (as with the TEB activities), there is a large 'structuring' component with banks accumulating positions in advance of the creation of various products specifically designed to meet client needs.

While we have attempted to categorize the major segments that make up the trading business, we are not sufficiently bold (or naïve) to try to qualify how much of the business is 'agency' and how much is 'proprietary.' Broadly speaking, however, we believe that banks generally enter businesses to fulfill the needs of their clients, and we believe that 'most' trading revenues arise in support of client flows.

We note that over the past 10 years there has been tremendous volatility with the implosion of the technology bubble, a rapid revaluation of the Canadian dollar, the Russian Crisis, the collapse in Enron (which was a massive player in the derivative market) and dramatic change in commodity prices. Despite this, we have had only one 'loss' quarter for the entire industry over the past 40 quarters. It seems to us that if banks were taking on large amounts of proprietary risk, then there should have been more volatility.

3. Considering the Risks Associated with Trading Operations

There are several risks involved in trading operations, but we believe that the two biggest are volatility and operational risk. The former is easier to measure, and we believe that it is really only an issue to the extent it affects a bank's cost of capital. If volatility is too great, a large trading operation can negatively impact equity valuations, or debt ratings. If neither is an issue, then the ROE characteristics of the business certainly argue in favour of bank involvement in trading. We deal with the issue of volatility from a more mathematical perspective below.

The question of operational risk has come to the fore recently because of progress on Basel II, which will require specific capital for the potential risk arising from failed systems and processes. Having said that, the reality is that like measures of volatility, the downside is impossible to establish. Simply put, it is difficult to truly know how big a loss can occur. We discuss this matter and the issue of operational risk later in this section.

As we have shown previously, trading revenues don't run on a traditional cycle, and we believe are largely disconnected from the credit cycle. There are, however, 'event risk' issues that plague the business. There can be radical short-term volatility. The weakness in trading revenues in late 1998 coincided with the Russian crisis and the failure of Long-Term Capital Management (LTCM). All banks, with the exception of Royal Bank (the National Bank did not disclose its trading revenues in detail before 1999), experienced a noticeable decline in trading revenues in that quarter- two banks, CIBC and BMO, experienced large net trading losses.

If the risk from trading is volatility, then how can we evaluate the banking systems (and individual bank's) ability to manage this volatility? We consider two approaches: a traditional VaR-based approach and then a statistically based volatility metric coefficient of variation.

Before we discuss volatility, though, it is important to state the obvious: there is asymmetry in risk in the Trading and Derivatives business. Surprises are seldom of a positive nature, and the scale of surprises on the downside is materially larger than the scale of surprises on the upside.

i) Value at Risk (VaR) – What Is It, and What’s It Good For?

VaR, or Value at Risk, is the metric most used by investors and analysts to evaluate risk in the trading businesses of Canadian banks. The appeal of VaR is that it provides a single dollar amount that attempts to quantify the risk being undertaken. In many ways, the simplicity of the concept is its danger. Specifically, a 99% VaR (expressed in pre-tax dollars) is defined as the daily loss that the institution is unlikely to exceed with a 99% confidence factor. As important as understanding VaR itself, it is important to understand what it is not:

i) It is not the maximum loss that is expected. Indeed, that one percent of the time when VaR is exceeded, the loss can be meaningfully higher.

ii) It is not forward looking—it is based on the historical level of risk and the historical portfolio.

Probably the single biggest issue for us is the limited consistency across banks of their disclosure of VaR. Remember that for VaR to be calculated, a bank needs detailed systems and processes, and a depth of historical data to model the value at risk. For new trading businesses, the introduction of a VaR model actually “creates

• VaR. The irony is that the bank probably had higher risks before the model was introduced, but it simply wasn't calculated, and therefore wasn't disclosed. The good news is that regulatory capital (arising from derivative and market risk-weighted assets) is calculated whether or not a model exists. In fact, the regulator calculates RWA and capital from a basic approach unless models are approved. If the VaR models are deemed to be adequate, an advanced calculation based on the VaR model is used, which typically throws up less capital than the basic approach.

ii) Coefficient of Variation – A More Useful Metric

Another, and in our opinion better, metric for considering volatility in trading businesses is to measure the coefficient of variation for trading revenues. Coefficient of Variation (CoV) is a statistical measure of the relative dispersion of outcomes versus the mean. While this metric has its limitations (largely that it is backward looking), it can, at the very least, be calculated across the group consistently. Overall, we believe that banks with lower and more stable CoV are producing 'higher quality' trading results, as there appears to be less volatility in revenues over extended periods.

iii) Operational Risk – Impossible to Measure and Highlighted by the BMO Commodities Loss

Operational risk is defined in the new Basel II framework as 'risk of loss resulting from inadequate or failed internal processes, people and systems or from external events.' Our interpretation of this definition is quite broad, and includes all aspects of risk that banks take on in their trading and derivative books due to non-market level situations. In reality, this risk is probably more material than the classic market risk issue because it is so open-ended.

The BMO situation will become the cause célèbre for operational risk, and highlighted many of the risks of trading businesses. The reality is that there can be limited price discovery in some trading businesses. While 'marking-to-market' is a relatively simple exercise in cash businesses, for various esoteric products and derivatives there may be limited ability to verify price frequently.

As such, the 'carrying value' is based on models that incorporate prevailing market rates and prices (often established from a variety of market data points) and some valuation adjustment to cover variability in several factors, including volatility and liquidity.

When the models are flawed, and market conditions change, the scale of losses can be large, partially because the 'value' of the adjustments can be magnfied, but also because previously booked profit accruals need to be reversed. As products become more complex, the potential for this kind of problem is magnified.

4. Many Variations Across the Bank Group

Given inconsistencies in disclosures and the variations across business lines, it is difficult to truly compare trading performance across the bank group. One thing is clear, however; Royal is slightly more dependent on trading, and is better at it than its Canadian peers.

We believe that the best indication of 'productivity' in trading operations is to simply consider the amount of trading revenues that each dollar of trading capital generates. While we caution that this approach masks what could easily be variability across books, it is somewhat illustrative. As we show in Table 4, Royal has consistently generated more trading revenues for every dollar of risk-weighted assets, while the BMO generates the least.

Two variables are noteworthy here. First, the capital required is based on regulatory capital, which can be impacted by the degree to which regulators approve (or reject) models. From our perspective, this is a reality- and if a bank has good systems, it should be able to convince OSFI to approve the models over time. Second, BMO's poor performance is probably exaggerated by the fact that it 'houses' some of its trading operations in its accrual book and some of the revenues are not shown. We believe that even if this adjustment were to be made, BMO would still be an underperformer.

The strong performances of Scotiabank and National are also noteworthy. We would have thought that “the bigger the book of business, the better the margins,- as compliance, IT and management costs can be spread over a more diversified book. In reality, though, it appears as if execution is as important as scale, and BNS and NA should be commended. TD, largely due to unusually higher regulatory capital levels in the 2001-2004 period, has also been an underperformer; though it has improved.

Another factor to consider is that when banks enter new trading businesses, the capital often precedes the profitability. This is one reason why Royal appears to have lost its 'productivity' advantage, and probably explains why there have been some periods when trading has 'surprised' on the upside. The reality is that Royal's trading revenues could be somewhat higher given the level of capital (assuming a direct correlation).

Most investors have a preconceived idea of which banks have historically been more committed to trading than others. However, it is clear that this commitment varies over time.

We have already discussed volatility in the previous segment of this report, and we show our CoV by bank in Appendix B. It is interesting to note that banks that seem to be bigger and better in trading (RY and BNS) also appear to be more consistent. Without a doubt, Royal stands out on every metric: it has higher productivity of its trading capital, it has more dependence on it and its business appears to be more stable.

5. After an Extended Period of Caution, Banks Appear to Be More Aggressive

As we have shown in a variety of ways, banks in aggregate do not appear to have significantly increased their involvement in trading and derivative businesses over the long term, but this is changing. In many ways, it seems to us that they have universally agreed that the excess capital within the banking system can be profitably deployed into trading and derivatives. It will be interesting to see if the BMO problems dissuade other banks from continuing this shift: we doubt it.

As we have already mentioned, Canadian banks have not meaningfully increased their trading and derivative businesses over the past 10 years.

If a picture tells a story, then the graph of the make-up of bank risk-weighted assets clearly shows that banks do not appear to be shifting RWAs away from credit and toward Market and Derivative Risk. Given the higher structural ROE of the Market and Derivatives activities, one could argue that this is an opportunity lost.

This analysis does hide an important fact: that banks scaled back trading and derivative exposure following the Russian Crisis in 1998, and in the past two to three years seem to have rekindled their infatuation with it. The banks have ramped up their trading RWAs and notional derivatives in recent years, and in the past 12 months, growth has accelerated to over 25% for the former, and over 30% for the latter. One area that seems to have lagged is trading revenues, which raises the big question, are trading revenues positioned to grow significantly, or are banks investing in more and more marginal trading activities?

We are quite sanguine on this issue- we believe that trading revenues almost always lag increases in notionals and risk-weighted asset build-up (because of OSFI reluctance to approve models) and because banks need to invest before they get returns. In essence, we are saying that trading revenues for Canadian banks could grow meaningfully over the next one to two years.

Again, the Royal provides the most interesting insights. It has been growing RWA rapidly for the past three years, but until the last couple quarters, trading revenues have actually lagged. Given Royal's track record, one should assume that trading revenues could be meaningfully higher.

6. Conclusion – When Properly Controlled, Trading Appears to Be a Very Strong Business

We have been around for too long to simply accept that there are no threats to businesses which achieve unusually high ROEs over a sustainable period- particularly when the competitors in the business are large and well capitalized. There are, however, some competitive realities that, we believe, position banks well in the trading and derivative business.

First, trading is a rating-sensitive business, and Canadian banks with their strong balance sheets and debt ratings should be well positioned to compete. Second, many of the businesses that comprise trading are quite Canadian-centric. We believe that Canadian banks have a competitive advantage in trading securities and derivatives involving Canadian dollars and Canadian interest rates. They simply have a disproportionably large share of the flow. Third, the business appears to be scalable- the skills to manage and monitor foreign exchange traders are similar to those needed to oversee commodity, equity and credit traders. We believe that banks can leverage these core competencies.

Last, and most importantly though, we believe that the strength in the business is driven by franchises and relationships that are difficult to replicate. We believe that commercial and mid-market businesses in Canada have incumbent bank relationships that are very difficult for global competitors to access: they don't lend to these businesses, don't have the capacity to offer cash management and seldom have the interest in smaller relationships. Trade begets trade and Canadian banks seem to have some natural flow that ensures they can and will continue to be very competitive in most of these businesses.