Banks Aren't Bridges
The logic of the recent bank stress tests seems unassailable — bridges may fail, so may banks. But bridges are physical constructs. Centuries of research inform us about the maximum safe load and have allowed us to develop reasonable engineering equations relating failure to stress.
Banks are much more complex. Unlike a physical weight, the likelihood that a loan will become a stress factor depends upon overall economic conditions as well as the care with which that loan was granted. "Good" loans strengthen rather than stress the bank. And while models are used to evaluate both bridges and banks, economics remains a far less precise "science."
Models that seek to estimate fairly simple macrostatistics, such as the national deficit, have failed in recent years. Moreover, the people who select the parameters for these models can easily determine the results. Tests can be useful, but forcing subjective reality into the straitjacket of objective numbers may delude more than illuminate.
The current stress tests are an attempt to shift from market-based to political evaluations. Markets continually stress-test every financial institution, as investors or depositors consider whether banks will safely manage their funds. Such market stress tests have big advantages over politicized "stress tests": The market tester who gets it wrong suffers a direct economic consequence.
Determining the riskiness of an enterprise in a world of rapidly changing circumstances is not easy. Market participants can overestimate or underestimate those risks — but they benefit or suffer accordingly. In contrast, the government stress tester suffers little economic loss (or gain) from his errors.
Recent decades have seen a steady push by regulators to "objectify" reality — to create a world where fallible human judgments are replaced by precise numerical criteria. The current stress tests join the international banking accords known as Basel I and Basel II, as well as Sarbanes-Oxley, the accounting law passed in 2002. All are well-intended attempts to reduce financial risk, to create numerical standards to estimate future solvency. But hiding behind numbers — seeking to substitute accounting standards for judgment — is dangerous. And when those tests are designed and administered by political agencies, the dangers only increase.