Banks as hedge funds and hedge funds as banks

By: Casper de Vries, Erasmus University Rotterdam

A hedge fund is a privately unregulated investment pool that places counterbalancing bets on a variety of assets, often with considerable leverage. Similarly, banks are mostly heavily regulated incorporated intermediaries which go short in deposits and are long in loans, with a tiny capital base as well. Thus hedge funds play on a wider range of assets, are more arbitrage than intermediation oriented and are much less regulated than banks. The resemblance between the grasshopper and bolder hat, though, I find the more striking. Both species have gigantic short and long positions with a very small capital base. Not without reason do supervisors worry about the risks of these financials. Since the demise of LTCM, supervisors are in particular worried about the systemic risk for the banking sector posed by the hedge fund industry. It will be our contention that this worry seems misplaced as, if anything, the banking sector is the more risky of the two industries, with little systemic spill-over from the hedge funds to the banking sector. Hedge funds appear to deliver the protection their name suggests.

 

 

Regulation invites financial innovation. Hedge funds take on risks for parties who are otherwise constrained by the limitations placed by financial supervision (such as the Basle bank charters and mutual fund regulation). Hedge funds also deliver a public service through their arbitrage activities. By taking contrarian positions, they often provide liquidity when banks are constraining credit.  Consider the Credit Suisse/Tremont Hedge Fund index together with the index of the EU Banks in the figure below. When the banks became distressed at the end of 2001, the hedge fund index continued to rise. By providing loans and prime brokerage services to hedge funds, banks indirectly benefit from the returns generated by hedge fund activities. And lately, hedge funds also help to discipline inefficient banks behind the dikes in the chummy delta of Rhineland capitalism. Thus the banking industry and hedge fund sector are intertwined. Let us see what this does for systemic risk. To this end we first investigate the effects of short positions.

 

 

 

 

Short & long risks

 

In my previous brief on the case for macro prudential supervision, I discussed the interdependence of bank returns through their portfolio similarities. A cross-plot of different bank returns showed that the most extreme positive and negative returns occur jointly, something which would be very rare if the returns would be normally distributed with positive correlation. The higher than normal systemic risk stems from the fat tail nature of the underlying risk drivers. Plotting the returns of the embattled ABNAMRO bank versus the EU banks index reconfirms this picture. Note the numerous outliers. Since all banks hold similar exposures, the extremes occur in the first and third quadrant.

 

 

One of the more risky hedge fund strategies is the dedicated short bias, in which a fund holds an overall short position. Below we have cross plotted the monthly returns of the Tremont Dedicated Short Bias hedge fund index versus the EU Bank index. The interesting feature of this cross plot is that the numerous extreme realizations are now located in the second and fourth quadrant. Thus when the banks do badly, a dedicated short bias exposure provides a hedge. Contrarian investments deliver when long positions do perform badly.

 

 

Some simulations can give further insight behind the features of this negatively sloped cross-plot. Consider the next cross-plot containing 5000 draws from two independent Student-t distributed random variables X and Y, with three degrees of freedom. Note that the plot has the shape of a plus, rather than the cloud which would realize if X and Y would be generated from two independent normal distributions. Recall that the Student distribution is a fat tail distribution, which has tails that decay at a power rate rather than the much faster exponential rate of the normal distribution. The plus shape is driven by the extreme realizations. For a comparable normal based plot, see two figures down.

 

 

 

Suppose we form two portfolios from the independently simulated returns X and Y. One portfolio is long in both assets and reads X+Y, representing the loan portfolio of a bank. The other portfolio is the market neutral portfolio X-Y, resembling a market neutral hedge fund; it is long in the first asset and short in the second asset. Note that the two portfolios are uncorrelated. Nevertheless a cross-plot of the two artificial portfolio returns delivers the star shaped figure below. The portfolio returns are clearly dependent, albeit being uncorrelated, since given a high realization in the horizontal direction makes a high realization in the vertical direction much more likely than if the conditioning varaible is small. What is the case? The extreme realizations from the plus shaped figure have been rotated onto the two diagonals by the portfolio composition. Since either the X or the Y value is an outlier, the two portfolios X+Y and X-Y are dominated by either the contribution of X or Y, which shows up on both the diagonals.

 

 

 

For comparison we have also simulated the same portfolios by taking X and Y from normal distributions. This gives the cloud depicted below. Since the two portfolios are uncorrelated, which in a normal world implies independence, one obtains the same picture for the portfolios as for a cross-plot of X against Y. One lesson to take away from this is that in a fat tailed world, the correlation concept may not signal the dependency that is important for systemic risk.

 

 

 

Next we have constructed a portfolio X-2Y, which resembles the dedicated short bias strategy of some of the hedge funds. This is cross plotted against the portfolio with only long exposures X+2Y. Since this time the portfolios are negatively correlated the negative diagonal contains the more extreme outliers due to the scaling of the Y-return contribution.

 

 

Under normality the two portfolios X-2Y and X+2Y are also negatively correlated and this gives the negatively sloped disc in the next figure. But note that the former cross-plot is much more pointed in the upper left and lower right hand corners due to the fat tail nature of the pseudo random variables. This makes that the former plot better resembles the features of the cross-plot of the Dedicated Short Bias index against the EU Banks index.

 

 

Systemic risk

 

Now that we have a grasp of what short positions do in a world with fat tail risk, we can return to the question of systemic risk implications of hedge funds for the banking sector. The cross-plot of the ABNAMRO returns against the EU Banks index did not give the nice positively oriented disc which could be expected under normality. Rather, due to the fat tail nature of the banking risks numerous outliers are located along the diagonal. The dedicated short bias strategy versus the banking index has its outliers going in the opposite direction due to its short positions. Dedicated short bias is just one of the hedge fund strategies. Prudent banks have a diversified clientele portfolio. Hence, we can expect that banks have exposure to a mixture of different hedge fund styles. A cross-plot of the EU Banks index against the aggregate hedge fund index is therefore more representative for the systemic risks. Realizations in the four corners of the plot would be indicative of systemic risk. Fortunately, little evidence for this fear is present form the current cross-plot.

 

 

The first figure of this brief showed that the bank index and overall hedge fund index yielded similar total returns over the period starting in 1994. Perhaps this runs counter to the popular view of the hedge fund industry. Apart from the sometimes spectacular returns delivered by some of the hedge funds, there were also many deceptions and, moreover, many hedge funds are rather focussed on generating moderate returns without taking excessive risks.

 

One already sees from this first figure that the bank index was more volatile (implying a lower Sharpe ratio for the banking sector). The last cross-plot of this note confirms the picture that banks are the more risky of the two sectors, since the outliers in the horizontal direction are much more pronounced than in the vertical direction. This confirms that hedge funds concentrate on generating alpha, rather than picking up beta like most mutual funds. More pertinent to our discussion, the last graph shows there appears to be no basis for the fear of large spill-overs from the hedge fund sector to the banking sector. This is confirmed by the small fall-out from spectacular losses such as those of the Amaranth fund and the Long Term Capital Management failure. This is not to say that the hedge fund industry is without its risks. In particular the high failure rate due to fraud is something of concern. This requires that banks that have a relation with hedge funds should obtain adequate information. But for the time being it appears that hedge funds on average deliver what is traditionally ascribed to be the image of a reliable bank, while banks hop in the risk direction.

 

 

I am grateful to Oleg Tschernizki for preparing the data and discussion.

 

 

 

---