October 11, 2015

Observations on September 2008 Financial Crisis

Because September is the one-year anniversary of the worldwide financial meltdown, there were many good articles in the various newspapers and magazines over the last few weeks.

One particularly insightful article appeared in the September 14, 2009, issue of The New Yorker, entitled “Eight Days” by James Stewart. It highlighted three root causes that ended up building destructively on one another:

  • Excessively high debt leverage among many players in the financial services system;
  • Rapid asset sales driven either by panic or sophisticated profit-driven short selling; and
  • Provisions in credit agreements that triggered defaults based on accounting-driven asset valuations.

The high debt leverage (ratio of debt to equity), which ranged from 30/1 to 60/1 among major investment banks, created the first dangerous condition because it meant that investment banks had no flexibility to address significant downturns. Lehman Brothers filed for bankruptcy, but others were almost as vulnerable. Stewart referred to Merrill Lynch as the “firewall” because it was sufficiently stronger than Lehman that the Bank of America could be persuaded to acquire it, and give it some financial flexibility in the event of a severe downturn. However, with even 30/1 leverage, a need to compensate for the loss of as little as 4% of the total asset base of a firm would exhaust its equity capital.

Stewart’s article pointed out that all those who built up, managed, and regulated the financial system badly underestimated how quickly panic-driven behavior could spread and how big an effect it could have. The Lehman Brothers bankruptcy, which itself was triggered to some degree by asset value declines caused by panic selling, triggered demands the next morning from those with money markets assets to convert those assets into cash. Firms that have customers with significant money market assets generally cash out a very small portion of their money market assets at any given time, just like a bank expects withdrawals of only a small part of its checking and savings account demand deposits. When a significant percentage of account holders want cash all at once, whether from a bank or a money market account manager, the firms that hold those assets cannot deliver cash to everyone requesting it. The failure to meet cash demands destroys confidence in the system and causes even more panic. One firm alone had over $24 billion in cash demands made the day after the Lehman bankruptcy filing. That’s what was beginning to happen a year ago. Panic bred more panic, and threatened to destroy the fundamentals of our banking and brokerage systems.

Stewart also explained the problem with AIG, which resulted from credit guarantee agreements that required a minimum level of collateral to support them. The collateral values were determined by accounting-driven valuations. Accounting rules on “fair value” financial reporting are imperfect even as financial reporting tools. They make a series of artificial assumptions, one of which is that a firm will put up its entire asset base for sale at the end of every quarter, whether it needs the cash or not, and whether or not it is a good time to sell the assets. No firm would behave that way, and, therefore, the asset valuation is clearly imperfect.

However, when asset-value accounting rules are applied in a context in which collateral has to be maintained, the artificially low valuation becomes a much bigger problem. It operates as if the assets were being sold all at once. Given the panic selling that was going on for other reasons, asset values were unrealistically valued, but the unrealistic values mattered.

Something similar to this actually happened during the Great Depression in the residential housing market. In the 1930’s home mortgages were only five years in duration, so the average homeowner had to refinance his or her mortgage every five years, and homes were reappraised every time a refinancing was needed. As the residential housing market dropped in value, more and more homeowners found that they were getting smaller refinancings, and had to find higher amounts of cash to pay off the expiring mortgage. When they could not do so, their homes were repossessed and the mortgagees resold them at distress prices, which hurt every every other neighboring homeowner trying to refinance a mortgage. Think of AIG as the homeowner having to come up with collateral, and think of the accounting rules as triggering this requirement, even in the absence of a need for a sale. The federal government had to bail out AIG because other firms involved as direct participants would no longer have been able to secure credit, since the AIG guarantee would disappear. The whole credit marketplace would have cratered even worse than it did.

There are several lessons, other than the obvious ones of not rewarding high-risk behavior:

  • Leverage has to be managed downward to increase the flexibility with which firms can respond to crisis. A rapid downward movement in debt is not easy to manage, especially when we are trying to stimulate economic growth, but, over time, we cannot let firms assemble this much debt capital without a clear understanding of how high risk that capital is.
  • Panic selling will always be with us, but we need systems that create mechanisms to restore confidence quickly. The federal government has far more power today in the money market fund arena, and it has more ability to inject capital into the financial markets, but the issue with Lehman Brothers probably would not be solvable today, because the total financial risk of stepping in and guaranteeing the Lehman obligations was neither calculable nor limited at a level that permitted a comfortable taxpayer guarantee.
  • Imperfect accounting-driven valuations are a lousy way to measure the quality of a guarantee, and to trigger the requirement for more collateral. What made the AIG situation so bizarre was that it probably had enough cash to respond to the guarantees it would need to honor, when the underlying debt holders defaulted. What required AIG’s rescue was a rigid, formulaic collateral valuation change that was so rapid and large that it needed federal government help.

Clearly, we all continue to be humbled by the unintended consequences of what we created.