That Stein speech February 11, 2013 at 9:42 am
The recent speech by Jeremy Stein received a lot of attention for its suggestion that monetary tools might be used in addressing credit market-overheating. That is an interesting argument, but I don’t want to deal with that today. Rather, I want to look at Stein’s comments on collateral transformation:
Collateral transformation is best explained with an example. Imagine an insurance company that wants to engage in a derivatives transaction. To do so, it is required to post collateral with a clearinghouse, and, because the clearinghouse has high standards, the collateral must be “pristine”–that is, it has to be in the form of Treasury securities. However, the insurance company doesn’t have any unencumbered Treasury securities available–all it has in unencumbered form are some junk bonds. Here is where the collateral swap comes in. The insurance company might approach a broker-dealer and engage in what is effectively a two-way repo transaction, whereby it gives the dealer its junk bonds as collateral, borrows the Treasury securities, and agrees to unwind the transaction at some point in the future. Now the insurance company can go ahead and pledge the borrowed Treasury securities as collateral for its derivatives trade.
Of course, the dealer may not have the spare Treasury securities on hand, and so, to obtain them, it may have to engage in the mirror-image transaction with a third party that does–say, a pension fund. Thus, the dealer would, in a second leg, use the junk bonds as collateral to borrow Treasury securities from the pension fund. And why would the pension fund see this transaction as beneficial? Tying back to the theme of reaching for yield, perhaps it is looking to goose its reported returns with the securities-lending income without changing the holdings it reports on its balance sheet.
There are two points worth noting about these transactions. First, they reproduce some of the same unwind risks that would exist had the clearinghouse lowered its own collateral standards in the first place. To see this point, observe that if the junk bonds fall in value, the insurance company will face a margin call on its collateral swap with the dealer. It will therefore have to scale back this swap, which in turn will force it to partially unwind its derivatives trade–just as would happen if it had posted the junk bonds directly to the clearinghouse. Second, the transaction creates additional counterparty exposures–the exposures between the insurance company and the dealer, and between the dealer and the pension fund.
What this makes clear is that high collateral quality thresholds can be deeply unproductive, in that they encourage collateral transformation activities without enhancing safety. Equity markets professionals have long been used to the idea that low quality highly volatile assets often make good collateral providing that the haircut is big enough (and there isn’t substantial jump risk). That is surely right. Low quality assets can be a lot safer than high quality ones in that their haircuts are less procyclical: they are always big. I’d much rather post 200% of the value of my exposure as BBB bonds and know the haircut was highly unlikely to ever be more than 50% than post 105% of AAAs with the risk that my haircut quadruples at the first sign of trouble. (Assuming good segregation and bankruptcy remoteness of the overcollateralisation anyway.)