05 June 2009

Why Did Credit Rating Agencies Screw Up?

Alex Blumberg and David Kestenbaum at National Public Radio's Morning Edition do a brilliant job at exploring why credit rating agencies erroneously rated subprime mortgage backed securities AAA when they were backed by junk with little historical data with a minimum of words or paraphrase. Attorneys could learn a lot from their persausive, get to the point, brevity.

They also, of necessity, overstate the case and oversimplify.

Yes, Standard and Poors wasn't interested in investing more in accuracy when they had 94% market share. Yes, the financial crisis doesn't have post-Great Depression precedents. Yes, the didn't understand the models they hired experts to predict risk with. Yes, they could even explain the model assumptions to a smart fund manager at Vanguard who asked the right questions and was belittled on a social level rather than given honest answers.

But, there are two other points that need to be made and are just barely made in the story.

First, S&P had no liability for making bad decisions, because bond ratings are considered "opinions" which aren't subject to legal liability, even though almost the entire bond market relies overwhelmingly on a bond ratings from a couple of agencies to the exclusion of almost all other evidence to estimate default risk, and regulatory agencies encourage this reliance. But, Standard and Poors gets paid because issuers ask for ratings, so they have little incentive to decline to rate or rate poorly issuers who aren't locked into the system. The market share statement goes to this arrogance, but doesn't explain the incentives that create it and would probably lead to an industry-wide systemic risk even if the industry were more competitive.

By comparison, certified public accounting firms that audit publicly held companies have massive liability for screwing up audited financial statements, which are considered statements of fact, rather than opinions.

To be clear, imposing massive liability on credit rating agencies isn't necessarily the solution. While it made a massive error of judgment in an entire industry, mortgage backed securities, S&P and Fitch have really pretty decent track records in the more traditional economy. Also, pinning down credit rating agencies after the fact in a particular case isn't easy; they predict that a certain percentage of bonds in each rating category will go bad, they don't predict which one of those issues will roll the dice and get unlucky. And, then there is the problem of counterparty risk. Credit reporting agencies are surprisingly small enterprises given their financial importance, and even minor mistakes on their part can lead to losses far in excess of their ability to pay judgments against them.

But, clearly some form of reform in the credit rating industry could do a great deal of good. At the very least their incentives ought to be neutral rather than biased towards calling debt safer than it actually is. Transunion (a consumer credit rating agency) would certainly look very different, for example, if its fees were effectively paid by debtors the way S&P fees are, rather than by creditors.

Second, while the report lets S&P argue that they were able to give mortgage backed securities with weak underlying investments high ratings because of the loss reserves that were made in those bonds, what was going on there doesn't really get across. Part of the loss reserves were simple -- some of the interest collected from families with mortgages was put in a bank account rather than paid to mortgage backed security owners to cover losses on bad loans.

But, the bigger problem was that part of the loss reserves came in the form of guarantees that losses wouldn't exceed certain amounts. The mortgage finance companies and companies that issued free standing bond guarantees called credit default swaps, in both cases, made promises to make good on losses in excess of expected amounts that gave credit rating agencies comfort.

After all, unlike the credit rating agencies themselves, the mortgage finance companies and credit default swap issuers (many of them investment banks who reinsured their risks in turn with companies like AIG in order to give beneficiaries of the guarantees comfort that "counterparty risk" wasn't great), stood to lose immense amounts of money if they had to make good on their guarantees. Mortgage finance companies could control that risk because they decided who got the loans that were bundled into mortgage backed securities. Investment banks and other credit default swap issuers had been around for a century making huge profits, had billions of dollars in equity, and were widely considered the smartest money on Wall Street.

The mantra of modern financial economics is that market participants with real money at risk make the smartest decisions. This is what, as much as their own hired econometric model experts, gave credit rating agencies comfort. If mortgage finance companies or investment banks had been balking, while S&P theoretical models were showing investments to be safe, S&P management would have been worried. But, that didn't happen.

While modern financial economics is and the Wall Street professionals who use it are very good at explaining short term microeconomic behavior, modern financial economics is still half baked when it tries to get a handle on macroeconomic cyclic patterns and systemic risk.

In hindsight, we know that both the S&P theoretical models, and the theoretically smart money in mortgage finance companies and investment banks, both got it wrong. Whole industries got it wrong. Almost every mortgage finance company or other financial institution (including the most exposed banks and thrifts) seriously exposed to the subprime and Alt-A mortgage market went out of business. Every major free standing investment bank in the United States ceased to be a free standing investment bank. AIG has escaped bankruptcy almost exclusively because the federal government bailed it out. The less regulated investor owned part of the market failed.

Not everyone got it wrong. Commercial banks, thrifts and credit unions overwhelmingly stayed out of the subprime and Alt-A markets. Vanguard stopped investing in mortgage backed securities when it couldn't get straight answers. The private mortgage insurance industry set aside reasonable reserves and made better underwriting decisions. Notably, institutions that had FDIC type regulation, effective state insurance regulation (which has reserve requirements for insurance products that must be backed by solid actuarial evidence), or consumer ownership (which reduces the incentive to take excessive risks) had very low rates of failure in this financial crisis, while institutions that took risks with other people's money had very high rates of failure in this financial crisis. The problem was not that no one could predict that trouble was on the horizon, but that not enough of the right people had enough of an incentive to avoid those risks. This strongly suggests that ownership or regulatory incentives to avoid excessive leverage and risk are the key elements of any regime that wants to avoid systemic risk.

1 comment:

Michael Malak said...

While the root cause of the housing bubble is the Federal Reserve's printing of money, the two immediate causes that the credit rating agencies were not aware of (and maybe or maybe not should have been -- separate discussion) are:

1. Fraud. The credit rating agencies took lending institutions' representation at face value, who in turn took borrowers' representations at face value. The institutions were the ones more directly perpetrating the fraud than the credit agencies, but when all institutions are doing the same thing, that is a systemic problem. Is it fair to fault credit rating agencies for failing to control an entire financial system? Probably not. Some immediately gravitate toward increased governmental oversight to fix systemic problems. Others such as myself fault government for system problems. In this case, it is the Federal Reserve, specifically, and fractional reserve banking in general, which should be treated as fraud rather than as codified in law as expected banking practice.

2. 10-year look-back. Wired ran an article a while back that Wall Street economic models kept predicting housing would continue to increase -- because they were using models that had only a 10-year look-back period. They should have had a 100-year look-back period. Here, the credit rating agencies are more at fault than in the case of mortgage fraud described above. The credit rating agencies should have analyzed the financial institutions models for veracity before oondoning the practice with high credit ratings.