Buttonwood
The coming storm
Feb 17th 2004
From The Economist Global Agenda
SourceBanks are risking ever more of their own money in search of returns. Have they really learned nothing?IN THE autumn of 1998, Buttonwood was at a conference organised by Credit Suisse First Boston in—appropriately enough—Monte Carlo, when Allen Wheat, the then head of the investment bank, stood up after dinner and delivered a breathtaking mea culpa. Some sort of apology certainly seemed in order given the huge sums the bank had just lost from extravagant punts on Russia in particular and financial markets in general. The bets went spectacularly wrong after Russia defaulted, financial markets went berserk, and Long-Term Capital Management (LTCM), a very large hedge fund, had to be rescued by its bankers at the behest of the Federal Reserve. CSFB eventually admitted to losses of $1.3 billion, though the bank’s official figures and the numbers bandied about by insiders were somewhat at variance. To cut to the chase: had they Mr Wheat’s balls, Buttonwood thinks that the bosses of many a big bank will be making a similar speech before the year is out.
The reason is simple: the size of banks’ bets is rising rapidly the world over. This is because potential returns have fallen as fast as markets have risen, so banks have had to bet more in order to continue generating huge profits. The present situation “is not dissimilar” to the one that preceded the collapse of LTCM, says Michael Thompson, a strategist at RiskMetrics, a consultancy that specialises in the very risk-management models that banks use. Like LTCM, banks are building up huge positions in the expectation that markets will remain stable. They are, says Mr Thompson, “walking themselves to the edge of the cliff”. This is because—as all past financial crises have shown—the risk-management models they use woefully underestimate the savage effects of big shocks, when everybody is trying to wriggle out of their positions at the same time.
Even the banks themselves admit that they are taking more risk. Though they do not divulge the size of their positions, or in which markets they are concentrated, the degree by which those positions have grown can be gleaned from the risk-management models that all the big banks use (which are released in their financial statements). So-called value-at-risk (VAR) models determine the amount of capital that banks must set aside against their trading positions, and purport to show how many millions of dollars a bank might lose should markets turn against it. If its VAR is rising, a bank is, in effect, taking more trading risk—and VARs have been climbing for just about all of the banks that dabble seriously in financial markets. The VAR at Goldman Sachs, which is known on Wall Street as a hedge fund with an investment-banking business on the side, has more than doubled. One of the bank’s senior traders was even told recently that he must take still more risk.
Rest assured that he is far from the only one being told this at Goldman Sachs, or anywhere else for that matter, even though it was only a few years ago that many banks specifically eschewed punting as a good way to make money. Earlier this month UBS, a big Swiss bank, said that “with markets and investor sentiment starting to improve” it would gradually increase credit and trading risks. Even the likes of Citigroup, which stopped explicitly trading for its own account a few years back, and HSBC, a bank that used to think of trading as rather common, both announced recently that they too are increasing the amount of trading they do with their own money. Having previously scaled back its own trading, CSFB is also now increasing the amount of money it devotes to trading, though it claims that it will no longer “bet the ranch”. Allied Irish Banks, which you might have thought had had more than its fair share of trading fiascos, having lost nearly $700m thanks to activities of John Rusnak, one of its foreign-exchange traders, is trying to hire another 20 traders in Dublin.
VAR crash
Of itself, VAR is not the best guide to the huge size of banks’ current positions. In simple terms, these models assess the amount of risk that a bank is taking by looking at the volatility of the assets it holds and the correlation between them (the less correlation the better). In that way, banks can see how much they might lose were these bets to sour. A cynic would say that such models thus purport to measure an uncertain relationship between lots of uncertainties.
Crucially, if markets become less volatile, banks can pile on more positions and still have the same VAR. With the exception of Treasuries, markets have indeed become much less volatile—volatility has roughly halved in many financial markets over the past year-and-a-half; equity markets are now less volatile than they have been for a decade. Roughly speaking, if markets are half as volatile, banks’ positions can be twice as large for the same amount of capital. But since VARs have in fact risen, some banks’ positions are probably three times what they were in the autumn of 2002.
Or at least the ones they have on their balance sheets. For banks have been increasing their trading exposures in other ways, too. The most notable is via direct investments in hedge funds, often those set up by traders who used to work for the banks themselves. Chemical Bank, now part of J.P. Morgan Chase, started the trend 15 years ago. Now, almost all big banks invest their own capital in hedge funds. Citigroup may have shut down its “proprietary” trading operations five years ago (temporarily, it now transpires) but it invested a few hundred million dollars of its money in a hedge fund set up by those proprietary traders. Deutsche Bank recently invested $1 billion in a hedge fund run by its erstwhile traders. J.P. Morgan Chase is thought to be the most generous in doling out its cash, but CSFB, Goldman Sachs, Lehman Brothers and BNP Paribas together also invest hundreds of millions of dollars of their shareholders’ money in hedge funds.
In total, banks have invested many billions of dollars in such funds. The reason, apart from an understandable desire to invest money with good traders, is that the money invested in this way is counted as an investment, and not as a trading position, so is not included in the banks’ own trading books. Most of the money that banks invest has gone into hedge funds that specialise in bonds and other sorts of fixed-income instruments. Like the banks, hedge funds have been leveraging up their exposures to markets.
All of which is splendidly profitable, as long as markets behave themselves. But the strategy puts banks and hedge funds alike at huge risk if markets suffer a severe shock—a far more common occurrence than banks allow for. Their models (and, yes, hedge funds use VAR models as well) assume a certain level of losses for moves of a given magnitude. The problem comes for the tiny number of crises when markets move much more and, to add insult to injury, banks’ assumptions about the diversity of their portfolios are shown to be wrong. In other words, the models, says one regulator with a chuckle, are of least use when they are most needed.
By regulatory fiat, when banks’ positions sour they must either stump up more capital or reduce their exposures. Invariably, when markets are panicking, they do the latter. Since everyone else is heading for the exits at the same time, these become more than a little crowded, moving prices against those trying to get out, and requiring still more unwinding of positions. It has happened many times before with more or less calamitous consequences.
It could well happen again. There are any number of potential flashpoints: a rout in the dollar, say, or a huge spike in the oil price, or a big emerging market getting into trouble again. If it does happen, the chain reaction could be particularly devastating this time. Banks and hedge funds have increased their exposures most to those markets that they are least able to get out of. Think, if you will, of the extraordinary rise in the price of emerging-market debt and junk bonds. “I used to sleep easy at night with my VAR model,” said Mr Wheat in his speech in Monte Carlo. Suffice to say that he suffered a sleepless night or two when that model was found wanting—and that bank bosses could be in for many a sleepless night this year.
Read more Buttonwood columns at www.economist.com/buttonwood