Saturday, August 11, 2007

Repo Man

There is much concern about the lack of liquidity of mortgage based bonds, that is, the endogenous thin-ness of markets for such securities. It is certainly true that most concerned people have no idea that this is the problem. I think it might be useful to explain one of the meanings of the abbreviation "Repo". I don't really mean reposession as in repo man although foreclosure of mortgages is the basis of the financial storm. I refer to "repo accounts" opened by hedge funds at investment banks.

Hedge funds manage to obtain immense leverage without any credit rating at all. I think they still do it this way. They open "repo" accounts at investment banks. In such an account the hedge fund is notionally long some securities (that is the bank owes the fund the value of those securities) and short other securities (the fund owes the bank the value of those securities). Part of the deal is that the amount the bank owes the fund must always be positive. Thus the bank doesn't have to worry about the fund defaulting, going bankrupt, being some hackers with a pc and a modem. The bank, in contrast, has deep pockets (lots of money) so it can certainly pay the fund what is owed if the fund or the bank decides to close the account. Once upon a time, given the fact that prices change fast and suddenly selling a bunch of assets drives down their price (that is in bankerspeak assets are not infinitely liquid) banks required that the net value of the repo account (amount the bank owes the fund that is value of the assets the bank owes minus value of the assets the fund owes the bank) was at least 2% of the gross long position (total value of the assets which the bank owes the fund). The bank closes the fund sells and buys assets and transfers the resulting money to the fund if the fund doesn't respect this limit. If the net value of the repo account to the fund falls below 2% of the gross long position, the fund must transfer highly liquid safe assets (basically cash) into the account.

The requirement that repo accounts have positive value (plus 2%) is necessary given the fact that the bank has no way of knowing the financial soundness of the hedge fund (the fund will typically have risky accounts open at many banks). It also reates financial fragility. A decline in the value of assets for which the fund is long may force it to reduce its gross long positions, that is liquidate, that is sell. This creates positive feedback and is dangerous.

When there is a risk of such an event, markets can freeze up, become thin, cease to be liquid. That is no one wants to buy or sell the assets. It is very important that the problem is not just that people have decided that the value of the assets is much less than they thought. In that case, the price just falls and trading continues. The problem is that the asset price becomes very unpredictable, because it depends on possible distress selling by hedge funds. No one wants to buy because the asset value could tank, no one wants to sell because it could recover.

Hedge funds manage to make huge returns without huge risk by hedging, that is going long and short assets with highly correlated returns so the variance of the value of their repo account is much less than the variance of the value of the long position (and of the short position). They can't, even in principle, hedge against the risk that they will have to liquidate (unwind) their account. The theorists who proved that hedge funds are not risky (Black and Scholes) assumed that the hedge fund's transactions have no effects on prices (Scholes was a founder of Long Term Capital Management and lost most of his shirt).

There is another very deadly problem with REPO accounts when people stop buying and selling assets. The whole scheme depends on knowing the current prices of the assets. When trading volume is high one just looks at the current transaction price. When trading stops, one must rely on the list price, the price at which investment banks offer to buy or sell the asset.

Ah investment banks. The same entities which are betting against the hedge funds in the REPO account. If no one wants to buy an asset, the investment asset can say it is worth very little. If no one wants to sell it the bank can say it is worth a lot.
When markets freeze up because hedge funds might have to close REPO accounts, the banks can decide how much they owe the hedge fund. This would tend to be zero not 2% of the gross long position wouldn't it ? Thus, when people fear that the hedge fund will have to close an account and drive down the price of assets for which it is long and up the price of assets for which it is short, the investment bank can make money by listing made up prices so that this happens.

Nothing about worsening economic prospects, increasing fundamental risk or general badness elsewhere is required. Long Term Capital Management (LTCM) was driven under principally by an increase in the value of options to buy European stock market index funds. Such an increase would normally correspond to good economic news about European profits. In this case it went this way. LTCM bet that such options were overpriced and sold short. They were in trouble because of the Russian default. People thought they might have to liquidate their short position (buy options to cover their promise to pay banks the value of the options) this drove the price of the options up. More importantly, the uncertainty about what LTCM would have to do prevented people from trading. This allowed investment banks to offer to buy the options for a huge price without actually having to fork over the money. This forced LTCM to liquidate, transferring the options at the huge price to the banks.

The lesson ? Markets don't freeze up because of bad news, they freeze up due to risk. Possible future forced closure of hedge funds' REPO accounts creates such risk. When markets freeze up investment banks can make the current price whatever they want and force hedge funds to close their REPO accounts. This can happen whenever there is a shock of any kind. It does not require bad news.

No comments: