Friday, January 30, 2009

How Caisse's bet on quants went wrong

By Konrad Yakabuski in the Globe & Mail:

For four hours and eight minutes, amid the plush decor and sprawling New France tableau of the Quebec legislature's ornate Salon Rouge, Henri-Paul Rousseau gave much better than he got.

It was late 2007, and Mr. Rousseau, the man entrusted to manage Quebeckers' pension savings, had been called before an all-party parliamentary committee to explain why he had risked $13.2-billion of the provincial nest egg on a financial instrument so opaque and complex that only a coterie of elite mathematicians understood it. A financial instrument, what's more, that had turned out to be a bad investment.

The gathering had been billed as a rare public chiding of the head of the colossal Caisse de dépôt et placement du Québec, an opportunity for camera-loving members of the National Assembly to remind Mr. Rousseau that the Caisse was more than a pension fund manager comme les autres. Its $155-billion asset base, the largest single pool of investment capital in Canada, was Quebeckers' meal ticket and a means for them to achieve financial clout in a world where small-nation sovereign wealth funds, such as Norway's, were becoming big names.

The hearing did not go quite as planned. Mr. Rousseau batted off the legislators, a diverse bunch of nine that included a supermarket franchisee from the Montreal exurbs and a legal aide lawyer from Gaspé, with the all the force and self-assurance that his imposing six-foot-four frame and daunting intellect suggest.

One by one, he took their reprimands and remolded them into illustrations of his own genius. Rather than explain how he came to invest a whopping 8.5 per cent of the Caisse's assets in non-bank asset-backed commercial paper (ABCP) – the investment gone sour – he boasted of his own heroic efforts to rescue the entire $32-billion Canadian ABCP market. If the politicians were looking for Mr. Rousseau to eat humble pie, they could forget it.

“We are in control of the situation,” the Caisse chief executive officer persisted. “This will have been an event that will lower our batting average. But it's still very high, I can tell you that.”

The noxious ABCP would later lead to big writedowns, prompt a provincial election, and set off a chain of events that would leave the Caisse with a crushing cash shortage this past October.

By then, Mr. Rousseau had quit as CEO – for a job next door, at the palatial offices of the Desmarais family's Power Corp. of Canada – and it fell to his right-hand man and successor, Richard Guay, to deal with the fallout from the audacious investment strategies the duo had devised.

Those strategies helped take the Caisse to the top of its industry and nourished a long-held obsession – beating rival Ontario Teachers' Pension Plan. But the true costs of the strategies are only now about to be tallied. The Caisse is poised to release its 2008 results and they are virtually certain to be the worst in its 43-year-history. Investment losses of $25-billion or more will wipe out a huge chunk of the gains made during the Rousseau era. A reappraisal of Mr. Rousseau's batting average has already begun.

Almost no one came through the October market meltdown unscathed. But few endured the debacle that hit the Caisse. Somehow, its vaunted risk management practices had led it into a danger zone no pension fund is ever supposed enter: It ran out of cash.

The Caisse was not alone among pension managers in its quest to win higher returns to cover future pension obligations to an aging work force, not to mention industry bragging rights as top of class. It was not alone, either, in relying on mathematical models to manage risk – models that were supposed to predict the maximum potential losses on any investment and the likelihood of incurring them.

But compared with other Canadian institutions, the Caisse appears to have embraced the models more, or at least failed to implement sufficient checks and balances that would have protected it in case the models proved wrong.

Prove wrong they did, and the Caisse suffered accordingly when the market meltdown steamrolled over the models' assumptions. Now, the same practice is increasingly claiming other victims around the world. The Caisse's missteps provide a sobering window on how the meltdown of 2008 spiralled from a normal cyclical downturn into a much more damaging cataclysm.

The pride of Quebec

Created by the provincial government in 1965 in an act of Quebec nation-building, the Caisse has channelled Quebeckers' long-held desire for economic self-determination. Though it originally handled only the deposits of the Quebec Pension Plan, it has come to manage the assets of 25 provincial, municipal and sectoral pension and insurance funds, making it a global financial force and source of nationalist pride.

Starting in the late 1970s, the Caisse partnered with francophone entrepreneurs to lay the foundation of what would become Quebec Inc., a network of provincial institutions and private business empires whose aggressive tactics changed the Canadian business landscape. In the early 1980s, the Caisse even tried to engineer a creeping takeover of what was then Canada's biggest company, Canadian Pacific, by Paul Desmarais. But Ottawa stopped it in its tracks.

The Caisse's economic nationalism, however, has sometimes led to gargantuan blunders, from a botched leveraged buyout of the Steinberg supermarket chain to ill-fated forays into haute couture and Hollywood.

Mr. Rousseau, the youngest of eight children raised in small-town Quebec by a labour activist mother, vowed not to repeat those errors. With Quebec's population aging faster than almost any other in the developed world, the pressures on its pension fund managers to perform are intense. No one felt it more than Mr. Rousseau.

Though polar opposites in some ways, the testy, strapping Mr. Rousseau and diminutive, mild-mannered Mr. Guay were of like mind in their singular focus on returns. In their six years running the Caisse – Mr. Rousseau as CEO and Mr. Guay as chief investment officer, before succeeding his mentor last September – they played down the institution's nationalist mission, distancing themselves from former Caisse managers' stated bias in favour of local entrepreneurs. As former university professors – Mr. Rousseau in economics, Mr. Guay in finance – they were wonks at heart, and naturally receptive to the ideas of the geeky mathematicians who have increasingly replaced traders as the financial industry's biggest stars.

Complex statistical models devised by the mathematicians – or “quants,” as they're called – guided Mr. Rousseau and Mr. Guay as they sought out higher yields. They found them in ABCP, newfangled derivatives and a cornucopia of investments traditionally scorned by pension funds. It made the Caisse shine for a while, even enabling it to outperform Teachers in 2006, for the first time in nearly a decade, and again in 2007.

No one, however, could imagine Teachers running out of cash. But that is what happened to the Caisse in October of last year. Caught by a wrong-way bet on the Canadian dollar and margin calls on futures contracts, the Caisse – hamstrung by its excessive exposure to its ABCP, which had become completely illiquid – was grappling with a cash shortage. As rumours of the liquidity crisis spread, outsiders watched dumbfounded as the mighty Caisse cowered.

The statistical models, it turned out, had a fatal flaw. Using data series going back only a few years – a period of low volatility in asset prices – they failed to capture the risk of a market meltdown. Portfolio managers who relied on the models had developed a false sense of security as they sought out riskier, unconventional investments – ones without proven track records – in a bid to squeeze out ever higher returns. No one, it seemed, believed the models could be wrong.

The Caisse depended on the mathematical models when it bet so heavily on ABCP. But the models underestimated liquidity risks – the probability the Caisse would be unable to cash in its ABCP when it came due.

That is what happened in the summer of 2007 when investors became concerned about some of the assets behind the paper, such as shaky subprime mortgages and credit default swaps, risky derivatives whose holders bet on the likelihood a debtor will default. Those concerns prevented issuers of non-bank ABCP from selling new paper to pay back existing holders, the biggest of which by far was the Caisse.

Few imagined then just how vulnerable the ABCP exposure could leave the Caisse. But as markets tanked in October, 2008, it became clear that the commercial paper debacle could no longer be considered in isolation. With the toxic paper frozen, the Caisse was forced to liquidate other assets to meet margin calls on its huge position in foreign stock index futures.

In good times, ABCP and index futures produced higher yields than conventional debt instruments and stocks. But as October came around, it became clear that the unconventional investments also produced magnified losses in bad times. Unlike stocks and bonds, futures, for instance, are highly leveraged, so investors can lose much more than their initial investment.

Over the years, the Caisse had been ramping up its position in foreign index futures. By the end of 2007, its holdings tied to the relatively new derivatives had grown to 9.4 per cent of assets, worth $14.6-billion, from 3.4 per cent of assets in 1999.

The futures contracts allowed the Caisse to earn a return that mimicked that of foreign stock indexes, without incurring the transaction costs of buying the underlying stocks, all while putting up cash for only a small percentage of its overall exposure. The margin requirement, or amount it had to pledge, usually represented only about 5 to 10 per cent of the notional, or overall, amount by which it was exposed to the futures. (The amount of margin required is a function of volatility, the degree to which stock prices swing up and down.)

As global equity markets began posting near-double-digit movements almost daily in the fall, the Caisse found itself having to post more and more collateral against the futures with its clearing houses. Basic margin requirements on some index futures soared fourfold between early September and late October. And that sum did not include the increased margin required to reflect the massive declines in the stock prices that underlay the futures contracts, a process known as marking-to-market.

On top of it all, major ABCP holders such as the Caisse had to put up billions in additional margin on the frozen paper to prevent a complex restructuring of the ABCP market from collapsing. All told, the Caisse was forced to raise billions on short notice to meet the margin calls.

“They had a cash issue,” explains one Montreal derivatives expert. “Most funds would normally have twice as much margin deposited in the clearing house as required. But nobody provides for a 40 per cent decline in the market when they're setting up this kind of [futures strategy], nor do they assume that volatility is going to increase fourfold.”

Without enough of the high-quality assets demanded by clearing houses – for the most part, government Treasury bills – the Caisse was forced to sell stocks and other liquid holdings at October's distressed prices to raise cash. Had the ABCP on its books been rolling over normally, it could have readily cashed in the paper and bought T-bills. But the frozen paper was tied up in a complex restructuring.

Once again, the Caisse's risk management models had failed it. It made an already dreadful October even more trying for Mr. Guay and contributed to his departure on medical leave in early November, shortly after his 48th birthday.

In late November, as the rumours about the distress sale of stocks and other assets swirled, the Caisse publicly conceded that it had “adjusted its currency-hedging operations in the context of the Canadian dollar's instability and closed out certain futures contracts that could have created the need for additional capital in a down market.” It said the moves, completed in October, were designed to “protect depositors' assets and reduce the institution's total risk level.”

Caisse spokesman Mark Boutet refused to disclose the size of the October margin calls, though sources familiar with the situation pegged them at as much as $5-billion on the index futures alone. In an e-mail response to written questions, Mr. Boutet said the information would be released with the Caisse's 2008 results “if deemed appropriate.”

On Jan. 5, Mr. Guay resigned as Caisse CEO. He remains a strategic adviser to his temporary replacement, Fernand Perreault.

New financial tools

The entire global financial industry has been transformed in recent years by the intricate models that originated in the 1990s with a group of computer scientists and mathematicians at U.S. investment bank JPMorgan Chase & Co. The models, which fall under a broad category known as Value-at-Risk (VaR), use historical price movements and dozens of other variables to predict the extent and likelihood of potential losses on an investment.

If a typical poll claims accuracy 19 times out of 20, VaR could promise to be right 99 times out of 100. But it was useless in predicting what would happen when that one-in-100 moment arrived, much less a one-in-a-million meltdown.

Most institutions came to employ some form of VaR, but some placed more faith in it than others.

Those who did bank on it may, like the Caisse, be suffering the consequences now. Some experts believe that VaR has caused bigger losses at many financial institutions because it has a built-in bias in favour of highly leveraged investments. And since it assigns such low probabilities to nightmare scenarios, most institutions that relied on it never paused long enough to consider what would happen if the unthinkable arose.

“The failing is not in VaR but in the fact that many people stopped there,” says Leslie Rahl, president of New York-based Capital Market Risk Advisors. “The quantitative approach is about one-third of the puzzle. You would also have to take into account a vast number of qualitative factors.”

Ms. Rahl, a derivatives expert who has advised the Caisse on risk in the past, recommends that institutions regularly conduct a series of “stress tests” to determine how prepared they would be to deal with sudden strains on liquidity.

“One of the stress tests they should do is [to see] what would happen if margin requirements are increased. I don't know if the Caisse was doing that. Clearly, a lot of people were not,” says Ms. Rahl, who joined the board of Canadian Imperial Bank of Commerce in 2007 when the institution, hit hard by the subprime mortgage meltdown, moved to better manage its risk.

VaR's biggest critic is Nassim Nicholas Taleb, a professor of risk engineering at New York University, whose “black swan” theory is one of the hottest ideas in investment circles these days. Prof. Taleb asserts that VaR is nothing short of a fraud. It cannot, he says, predict the unpredictable, events that are as rare as a black swan – such as once-in-a-lifetime market meltdowns. So relying on VaR to manage risk is foolhardy.

That idea seemed almost heretical in investment circles prior to the recent crash, when VaR was almost considered infallible. Not any more.

“The black swan story is really popular these s,” says the Montreal derivatives specialist, who described the series of events hitting the Caisse in recent months as “a flock of black swans.”

Though the Caisse started using VaR before the 2002 arrival of Mr. Rousseau as CEO, it was during his tenure that it emerged as the institution's primary risk management tool.

“The era of Henri-Paul Rousseau was the VaR era at the Caisse,” says a former employee of the pension fund manager. “Everyone swore by it. They were convinced they could control risk with VaR.”

Mr. Rousseau, who declined to be interviewed for this story, also implemented a new VaR-based compensation scheme for portfolio managers. Managers could earn bigger bonuses if they earned returns that exceeded their VaR parameters. That, the former employee says, served as an incentive for managers to stuff their portfolios with ABCP, which had initially yielded higher returns than other short-term instruments. In 2006, the Caisse paid out a record $39.7-million in performance bonuses, a 55 per cent increase from the previous year.

In his e-mail response, Mr. Boutet described VaR as “one of the many tools that is used to manage risk at the Caisse […] Each of these tools has its own inherent strengths and weaknesses and must be used as such in an overall evaluation of risk.”

Mr. Boutet refused to provide a copy of a recent PricewaterhouseCoopers report the Caisse commissioned on its risk management practices “because it contains competitive information.” But he conceded that “certain areas for improvement were obviously identified” in the report.

He added: “Obviously, the current financial conditions of a historical nature have tested many of our practices and understandably we are taking at hard look at everything we do and making the appropriate changes as we identify them.”

That is a notable change in tone from the combative position expressed by Mr. Rousseau before the parliamentary committee, when he referred to an RBC Dexia study to insist the Caisse has a lower risk profile than its peers.

“We've created a lot more value and we've done it taking fewer risks,” Mr. Rousseau told Quebec MNAs. “It is false to say that we emphasized returns and forgot about risk.”

But experts familiar with the RBC Dexia study said it provided an incomplete picture. First, it only covered the years between 2005 and 2007, a period of historically low volatility in asset prices, hence low risk. Instruments such as ABCP had a short track record, so risk models could not accurately forecast how the paper would perform over time. For that reason alone, the experts say, the Caisse's risk profile was not nearly as conservative as Mr. Rousseau suggested it was.

“We know now that [before late 2008] we were in a period of low volatility when the underlying sticks of dynamite were lit but just hadn't gone off yet. So, if you generate for a period of time a return that reflects [ABCP] rising, but use low volatility as your common denominator, then of course you show up well. But it's the wrong measure for risk,” says one leading Canadian pension expert, who asked to remain anonymous.

Don McDougall, director of advisory services at RBC Dexia, agrees that the study should not be considered in isolation.

“There is no one perfect measure of risk and you have to look at a bunch of different risk measures over different periods of time,” Mr. McDougall says. “You can't cherry pick it for a particular period.”

As for the Caisse's decision to invest so massively in the non-bank ABCP, on which it took a $1.9-billion writedown in 2007, and which it recently conceded will require further provisions, the expert added: “It's the difference between understanding these investments in theory and having the visceral, intuitive sense about just what you do and don't do.”

Others are more generous toward the quants at the Caisse.

“If they're guilty of anything, it's that they were too smart,” says the Montreal derivatives specialist. “The guys that are losing now are the smart guys. They were trying to create a better mousetrap. But this is one time when it would have been better to be a follower rather than a leader.”

No comments: