From Risk to Uncertainty
Stress-test complacency will be a cause of the next financial meltdown.
Economic commentators have been increasingly using the word "uncertainty" of late. The context has included the business climate, the stimulate vs austerity debate, and forecasting the investment outlook across capital markets. Two examples:
"When businesses try to plan...or game out credit availability or the investment climate, they just don't know what it will look like. Uncertainty is a real killer." - Director of Government Affairs of the US Chamber of Commerce to the Financial Times
"Our current understanding of the future is extremely limited. There's an especially high level of uncertainty in forecasting Medicare." - Alan Greenspan commenting on the fiscal crisis of the US Government in the New York Times.
In financial markets, we typically think about "risk management" when we think about uncertainty. However, this was our core mistake in not anticipating the catastrophic "blast zone" of the recent financial meltdown. Our risk management models never contemplated the scale of the housing collapse, and most important, the interconnectedness that would allow the small US subprime market to infect the global financial system. It was also precisely the same mistake that led to the collapse of Long Term Capital Management in 1997, where spreads and volatilities moved out to unheard of levels, and correlations moved close to one (no more diversification benefits). The now infamous "100 year storm" or what financial risk managers call "tail risk" spoils all the best-intended work of risk managers. So called "fat tail" distributions imply that 100 year storms actually occur much more frequently, which matches our experience of repeated financial crises. Risk managers have always known of the limitation of Value at Risk ("VAR") models, and have responding by running extreme scenario analyses or "stress tests" dating back to pre-Long Term Capital days. Nevertheless, VAR remained the short hand metric to quantify expected loss. Probability based VAR models based on historical analysis of market price data are now seen by many, most notably Nassim Taleb, author of Black Swan, as inadequate or dangerous because of the false confidence they instill.
I have a personal connection to VAR models, having participated in the advent of using a VAR approach to aggregating risk at JPMorgan back in the early 1990s, and then spinning a company out of JPMorgan called RiskMetrics which became the industry standard for VAR-based analytic tools. (RiskMetrics was recently acquired by MSCI for $1.5 billion.) I served on the board of RiskMetrics for several years before it went public. One could say "guilt by association," but I prefer to think "balanced understanding," although the advanced mathematics has always been beyond me, as it is with most end-users of these models.
Today, sophisticated hedge funds run highly complex and comprehensive stress tests, and governments are requiring banks to run stress tests to provide the market with more transparency and comfort with their ability to weather the next 100-year storm. Not all stress tests are created equal. Stress scenario analysis are designed by humans using both judgment and mathematical algorithms. Both approaches are vulnerable to uncertainty they seek to quantify. They remain risk management tools, which presume that risk can be managed. Uncertainty implies risk that is not knowable and therefore not manageable.
Most important, stress tests are used as political tools. When I was at JPMorgan, whenever we ran a stress test that factored in what we considered to be real, worst-case scenarios, the outcome would be that the business would need more capital to stay in business. These scenarios are then typically rejected as "unrealistic" (in other words, "wrong answer"). At a firm level, true worst-case scenarios would most certainly imply the need for recapitalization, and if the firm is a "too big to fail firm" then individual firm recap implies another banking system rescue funded by taxpayers and the related economic collapse we are now all too familiar with. And that's just for the scenarios that we can imagine. Post crises balance sheets of the mega global banks have leverage of 12 to 15X (reduced form 30X before the crisis), off-balance- sheet leverage on top of that via derivative positions taking leverage over 20X I suspect, off balance sheet commitments, non-linear structural leverage imbedded in various products on its balance sheet, and funding mismatches between assets and liabilities. Such balance sheets cannot possibly "pass" a real stress test in my judgment without dependency on a central bank, not even the so called "fortress" balance sheet of JPMorgan.
A friend of mine runs a very sophisticated hedge fund. Their stress tests govern the risk they take. Their capital drawdown during the 2008-09 crash never breached half the loss predicted in their stress tests. And they hold cash equivalent to 200% of their stress test exposure, and equity far in excess of their cash. I don't think any major bank can make that claim. Certainly the financial industry as a whole cannot, and we now know no man (or bank) is an island. Instead, banks implicitly rely on access to the Fed's discount window for liquidity in extreme events.
Stress-test complacency by bank management, Boards of Directors, and regulators in an uncertain macro environment will be a cause of the next financial meltdown, just as VAR complacency contributed to the last one.
Our world is indeed increasingly uncertain and increasingly turbulent.
Just focusing on the US for a moment, we have unprecedented fiscal deficits at the local, state and federal levels, all brought on by the finance-triggered Great Recession, a structural fiscal imbalance driven by our entitlement commitments and military spending, structural underemployment worse than anyone alive has experienced, a trillion dollars of mortgages on the Fed's balance sheet it has no desire to hold, an insolvent Fannie Mae and Freddie Mac who also sit on nonlinear, interest-rate risk that will exacerbate and perhaps trigger a collapse in bond prices someday, deteriorated infrastructure, increasing climate-related crises with expectations that they will only increase, political and physical resource supply threats, particularly for oil, and growing social unrest as the middle class disappears in front of our eyes. Many believe we are heading into a double dip recession (or worse) which could drive real unemployment toward fifteen percent in the US at a time when our political will to respond is fatigued and we remain overextended and at war.
Turbulence is enormously more complicated than we imagine, as the great mathematician Benoit Mandelbrot, pioneer of fractal geometry and chaos theory warns us. Just what is the stress-test scenario we should ask our banks to run? The answer is, we do not know. We cannot possibly know. The only intelligent response is to structure our financial sector to be far more resilient than it is today. When you live in an earthquake zone, you build structures that can handle earthquakes. That resiliency, of course, has a cost, making it seem less "efficient". How to do that for the global, interconnected financial system is the topic for another day. But the message is that "efficiency", the goal of economics, is the wrong goal.
What we know from complexity science is that sustainable systems balance efficiency and resiliency. Efficiency has led us to gigantic global banks that are too big to fail, the financialization and commodification of markets, cultures, and, increasingly, the natural system called EARTH. The drive for efficiency has enabled a bonus culture on Wall Street that society detests but struggles to tame. A more resilient financial system would not afford such excess. It would hold up under stress, and better serve the needs of the real economy rather than threaten it.
The implications of a shift from a world of risk to a world of uncertainty are easy to underestimate. Being human, our behavior is far too slow and incremental. An appropriate adjustment would lead to outcomes that powerful people and institutions don't want to see happen, like changing business models of banks and restructuring the financial system to make it much more resilient (at the expense of efficiency). Higher capital requirements make banks more resilient and less "efficient" (ie lower return on equity). This should lead to lower bonuses if Boards of Directors do their jobs. This is why the banks will fight real Basel III capital reforms very hard. But capital buffers are just one structural reform our financial system desperately needs. Multiple smaller firms with a narrower focus create more system diversity and therefore more resiliency, but at a cost of efficiency. This is the tradeoff we must boldly seek. Instead we pass a "financial reform" bill in the United States and celebrate incremental change grounded in the naive risk management paradigm that presumes all "we" (now assumed to be the Financial Stability Oversight Council) need to do is better "manage" systemic risk. Hubris.
It is imperative that we reframe financial reform (and our financial management and investment strategies) as a response to moving from the paradigm of risk to the paradigm of uncertainty. As Mandelbrot would advise, such a reframing demands more than the incremental change we have seen, typical of a democracy. Financial Reform Act II calls for a bold new financial architecture designed for resiliency.