A Demon of Our Own Design: Markets, Hedge Funds and the Perils of Financial Innovation
John Wiley & Sons
Review by Edward Chancellor
Here’s what happened. In the run-up to the crisis, a large number of hedge funds had been established. Many of them used debt to juice returns. The stock market had been far less volatile than usual, which encouraged some hedge funds to take on even more leverage. The smartest hedgies employed computer models to place their bets. Algorithms told them what stocks or bonds to buy and how that position could be hedged. Other models informed the hedge-fund managers about the risk in their portfolios and what was the most they could possibly lose, in a day or a month.
Suddenly, things started to go wrong. A hedge fund which had invested in some doubtful bonds started to bleed. As it turned out, several other hedge funds and investment banks held similar positions, also financed with borrowed money.
Their forced sales created unexpected linkages between apparently unrelated assets. The much-vaunted risk models turned out to be broken. The hedge funds’ hedges turned out to be full of holes. Wall Street and the rest of the financial world seemed trapped in a death-spiral.
Attempting to restore calm, the Federal Reserve injected cash into the banking system. Goldman Sachs joined a consortium to bail out a stricken fund. The Treasury Secretary, a former Goldman boss, was moved to comment that the economy was strong enough to withstand the tremors on Wall Street. Eventually, calm returned to the markets, but it had been a close shave.
Those are the salient facts of the crisis. But which crisis? The near-collapse of the hedge fund Long-Term Capital Management in the autumn of 1998, or the credit crunch which started this summer when two hedge funds, run by a Wall Street brokerage firm, ran into trouble?
The global financial system daily handles transactions worth trillions of dollars and should be robust enough to withstand relatively trivial losses on Russian bonds nearly a decade ago or the recent US sub-prime mortgage mess. Why this isn’t the case is the subject of this timely volume. Its author, Richard Bookstaber, has spent the last quarter of a century on Wall Street, observing various crises at close hand.
The events he describes, such as the 1987 crash and the unravelling of Long-Term Capital Management, are both familiar and rather better related elsewhere (see Roger Lowenstein’s brilliant When Genius Failed for the best analysis of LTCM). Nevertheless, this book contains many interesting insights which help explain the summer’s tumultuous events on Wall Street.
Bookstaber is one of the legion of clever mathematicians, physicists and engineers that has flocked to Wall Street over the last quarter of a century. The models built by these ‘quants’ now exert an extraordinary influence over the financial system. Hedge funds engaged in statistical arbitrage today account for nearly half of the turnover on the New York Stock Exchange. Computers now largely determine who receives a loan and who doesn’t. Banks and hedge funds alike have become slaves to their models.
The trouble is these models are flawed. The fact that LTCM employed the smartest quants in the business didn’t save the hedge fund from failure. Rather, its risk models were responsible for the crisis, just as similar quant models are to blame for both the 1987 stock market crash and this summer’s credit crunch.
So where exactly do the models err? For a start, as Bookstaber observes, they assume that markets are always liquid and continuous – that means people will always be able to find buyers for securities and that sales can take place without moving prices. This is true most of the time, but not during panics.
The assumption that markets are always open for business encourages hedge funds and others to take on debt, believing they can easily offload positions. But these very hedge funds have become major providers of liquidity.
So when they are forced to sell, liquidity disappears and markets go haywire. That’s what happened in the September 1998 and again this August, when several leveraged ‘statistical arbitrage’ funds, including two managed by the mighty Goldman Sachs, suffered large losses in a manner reminiscent of the LTCM affair.
There’s another problem with the models. They assume the future will resemble in large measure the past, and in particular that correlations between various securities will remain stable. The aim is to reduce risk by holding positions which are inversely correlated. For instance, when stocks go down, bonds generally rise; so a portfolio constructed of bonds and stocks should be less volatile than one constructed of stocks alone.
Of course, the bets placed by hedge funds are more sophisticated. But they are potentially more troublesome. During a panic all types of securities start moving in the same direction.
It’s like plasma physics, says Bookstaber, ‘as matter becomes hotter it becomes less differentiated. Turn up the heat more and more and the atoms themselves become melded into plasma… matter in its most uniform and undifferentiated state, no longer hydrogen and oxygen atoms, just a seething white-hot blur of matter.’
Statistical models tend to break down for a simple reason. Financial data is drawn from the actions of human beings, whose behavior is sometimes irrational and continuously changing. As a result, there are no ‘scientific’ laws of finance that hold true over time. LTCM’s models, writes Bookstaber, assumed the fund was playing a game of roulette, where the probabilities of loss could be accurately measured.
In fact, it was engaged in a game of poker, in which other players were trying to get a glimpse of its hand. John Meriwether, the former Salomon Brothers trader who founded LTCM, belatedly realized this.
‘The hurricane is not more or less likely to hit because more hurricane insurance has been written,’ Meriwether observed. ‘In financial markets this is not true. The more people write financial insurance, the more likely it is that the disaster will happen because the people who know you have sold insurance can make it happen.’
Perhaps because the Federal Reserve’s bailout was so successful, the lessons of LTCM were never properly learned. That’s why this summer’s crisis was more or less a repeat performance. The trouble is that hedge funds today manage roughly ten times more money than a decade ago and risk models are far more widely employed.
It’s true, these hedge funds don’t employ as much leverage as LTCM and the models employ more ‘stress-testing’ than previously. Nevertheless, the financial system went to the brink once again this August.
So what’s to be done? Bookstaber wants a brake on the pace of financial innovation. As things stand, investment banks are free to construct sophisticated debt bombs – such as the ‘synthetic collateralized debt obligations’ that contributed to the sub-prime disaster – without any regulatory scrutiny. In the modern credit system, many hedge funds act much like old-fashioned banks, borrowing and lending and making some money on the spread.
But there’s no reason why hedge funds, like banks, shouldn’t be obliged to keep sufficient reserves to survive a market panic without being forced to sell positions. Less leverage, of course, would mean lower returns for hedge fund investors. But it would also mean fewer crises for the rest of us.