Accounting for Derivatives and the Rare Event
“New York’s Citigroup reported an $8.3 billion loss”…
“Bank of America said on Monday that its quarterly profit declined 77 percent”….
A loss is an accounting loss attributed to the company. But most of these banks SOLD the toxic investments that they made. What is going on here?
These losses, admit most banks, are stemming from the abrupt fall in demand for CDO derivates, especially mortgage-backed securities. I tackled some of these points in an earlier post. The question is, why are banks losing money? The original intent of this mess was to unload default risk onto investors who wanted to bear it.
The basic structure is as follows. A mortgage broker makes 100 loans. They sell the loans to a large bank. The bank sells them to an off-balance sheet special purpose vehicle (SPV) and sells shares in the SPV while their (in-house) investment bank collects fees for underwriting the shares. In a “normal” world – if the shares lose value, the investor who jumped in loses everything.
Along comes another derivative toy. Along with the shares, many banks (notably Bank of America and Citigroup) sold put options along with the shares in their SPV companies. Unlike this off-balance sheet “company”, put options are a direct contract between the investor and the bank and are treated as such.
Suppose the shares in the SPV are sold for $100, and the put option is sold with a strike price of $90. If the value of the CDOs ever plummets (like the mortgage market), the investor has the right to sell the shares back to the bank for $90. The investor is on the hook for losses of at most $10, the bank is on the hook for losses of at most $90. Of course, the bank collects a fee for allowing itself to be on the hook for this potential downside. In “normal” times this is not a problem – markets go UP, right? The odds of a 10% drop in the market should be relatively slim.
The problem is that the odds are not that slim. And in fact, the odds implicitly used by accountants and financial folk do not allow for these huge downsides. As a result, when using Black-Scholes or a similar method to value these written puts, the reported downside exposure is seriously limited. In fact, it is incorrect – the recorded value of written options tends to be too high in practice because the valuation techniques underestimate potential volatility.
When banks attached this put option to the SPV shares, they collected the upfront fee and slept easy, thinking that it was free money. When the unexpected happened and prices plummeted, they had to buy back these instruments at waaaaaaaaaaay above fair value and book the difference as a loss. It is easy to say in hindsight, but conventional models of derivative behavior underestimate the risk. We use the wrong statistical models (like Black-Scholes) because they are simple to work with. The world is a lot more volatile than our current reporting methods assume.
In a nutshell, what does this mean? It means that, for most companies (who use derivatives to hedge), the value of their options are undervalued when marked to market. For financial institutions, or anyone else who WRITES options, the value is overvalued when marked to market. This has implications for the current debate on mark-to-market, which I’ll hopefully get to in my next post.