UPDATE 3-4-10: There was an error in the chart on the dealers’ “matched books” that got JPM’s 2008 exposure wrong. It has been corrected.
The broker dealers’ 10Ks are all up on Edgar now, so I was able to update my file on the banks’ credit derivative positions. What’s neat about the 10Ks is that all (except for Citigroup) reported not only the notional CDS sold and purchased, but also the dollar value of CDS bought that offset CDS sold. This latter piece of information is very useful, because when a market maker keeps matched books almost all of the contracts will have identical (“matched”) underlyings and will offset each other. Thus, we can use this data to see to what degree each bank is acting as a “matched book” market maker and to what degree each bank is taking proprietary positions in the credit derivatives market. (As a reminder of why I am looking at only five banks please recall that according to OCC data the five banks in the charts below hold 95% of credit derivatives in the US banking system.)
For each dealer the first two columns of the chart below look at the net notional value of credit derivatives bought divided by the total notional value of credit derivatives bought at the end of 2009 (blue) and of 2008 (red). Thus the first two columns are a measure of how much each dealer bank is using its role as a market maker in the credit derivatives market to buy protection.
For each dealer the third and fourth columns of the chart below use the banks’ reported level of offsetting derivatives to calculate the notional value of unmatched derivatives sold and of unmatched derivatives bought. These two numbers are added and then divided by the total notional value of credit derivatives bought and sold. For 2009 this fraction is green and for 2008 it is purple. Thus the second two columns give a measure of what fraction of each bank’s credit derivatives book is “unmatched”.
The most remarkable thing to note about this chart is that only JP Morgan — notably the bank with the biggest credit derivatives book of all — falls below 5% on both measures. Of JPMs $6 trillion credit derivative portfolio in 2009 only $98 billion is either excess protection on identical names or protection purchased on other names. The other $5.9 trillion in swaps are all matched trades.
In marked contrast, Bank of America bought and sold $2.8 trillion of credit derivatives (and thus has a $5.6 trillion portfolio and no net notional), but only reports $2.3 trillion in trades with identical underlying. This implies that the bank has half a trillion dollars of unmatched trades both bought and sold. It would be very interesting to learn what the explanation is for such a large number of unmatched trades — and how well Bank of America is managing these positions.
Morgan Stanley, like Bank of America, has a credit derivative portfolio with a remarkable number of unmatched trades. Even though MS has only $69 billion in net notional value of credit protection bought, once identical trades are taken into account, we find that MS has about $600 billion in credit protection purchased but not sold and well over $500 billion in credit protection sold without an offsetting position and credit protection sold on identical underlyings that exceeds protection purchased on these underlyings. Like Bank of America Morgan Stanley needs to explain why such a large fraction of its credit derivative portfolio is composed of unmatched trades and how it manages these trades.
Goldman Sachs has also taken large positions in the credit derivative market, although it is much more conservatively positioned than Bank of America and Morgan Stanley. It has $217 billion in credit derivatives sold that are not offset by purchases and $381 billion in credit protection purchased with no offsetting trades. Thus, Goldman has well over half a trillion dollars of unmatched credit derivative trades.
Off the top of my head I can see two problems with this huge overhang of unmatched trades. First of all like AIG most of the market makers seem to be using the credit derivatives market to expose themselves to losses that are multiples of their equity capital. While the expected losses on these contracts are of course significantly less than their notional value, the sheer size of the exposures is a problem. The chart below illustrates this by calculating the unmatched credit derivatives sold divided by equity capital.
Secondly, the fact that these banks are relying on the credit derivative market to sell them protection is also a matter for concern — because, after all, what private firm is there in the current economic environment that can be relied on to honor these contracts? The fact that the five dealer banks that account for 95% of the US CDS market have as a group bought $430 billion in credit protection from others is hardly comforting until we know who it is that sold the protection.