Operational Risk -- Alpha Deception
Eric Falkenstein 10/8/03
Alpha is the term applied to a strategy, or person, who can generate a return greater than what is accounted for by risk. It is the ultimate free lunch, a risk-adjusted residual. Like pro-forma earnings, however, it is liable to manipulation.
Business line managers almost always overstate their alpha, mainly to inflate their perceived value and therefore compensation. A good example are traders. Most traders are really brokers, executing trades for others and making money via a spread. They are providing liquidity, relying crucially on their organization’s network of contacts and brand value to funnel customer orders to them. They have discretion in the precise tactics of execution, essential for finding incremental advantages, but this also then sows confusion as to how their activity adds value.
When I was a risk manager at a large bank I remember many traders who portrayed themselves as speculators, price takers who made money using market savvy, as opposed to buying for a fraction less than they sold all day. In reality they were liquidity providers with privileged access to retail trading flow, but through bullying and their use of an imposing trade vernacular, they were able to propagate the belief they where big alpha alchemists worthy of hefty compensation.
Not all deception is deliberate. Even traders I know who trade for themselves don’t comprehend their true edge. This is a natural self-serving delusion, like the documented fact that 90% of all drivers think they are above average. For example, many traders who have privileged access to retail flow think that while they get some edge from this, most of their edge, in their mind, is from ‘reading the tape’ or some other vague intuition.
This can be a benign misconception, in that if a market maker takes a few random positions every day, chances are they won’t affect his aggregate profits much over time, even if they selectively dominate his descriptions of his activity. Thus delusional trading ability can be a low-cost method to keep a trader focused on the market, and help them develop a richer context with which to chat up customers. Lots of myths have such indirect net benefits.
The problem with alpha deception is twofold. First is that firms overpay these guys, because people in charge of compensation assume revenues are inseparable from the individual. Since experience contradicts this assumption (within bounds, clearly some people are better than others), there must include some unspoken collusion at various levels, which is understandable given that the trader’s boss faces the same situation with his boss, and so on.
Few people are in a position to break the green wall of silence as there is little guarantee his direct boss won’t simply appropriate any savings as his revenue (via executive cost cutting), while alienating colleagues (in case you think such large unspoken conspiracies are fanciful, note the case where NASDAQ traders avoided odd-eights quotes until the mid 90’s after academics pointed out it was a collusive practice). Most of these high-priced traders can and should be replaced by people making less than half what they are being paid.
In fact, I remember one trader who made around $5mm year, who left in disgust at being told his compensation would be a mere $1mm during a bad year. His very pedestrian assistant, who previously made ‘risk manager’ money, was promoted to desk head and the desk functioned no differently than before at a fraction of the cost. Win for the shareholders.
The second issue is that the effort to inflate alpha misleads risk managers, senior management, and investors. Often the most egregious alpha exaggerators argue internally that their risk assessment is severely understated using conventional metrics. For example, I once reviewed a desk making $10mm a year, which had a mere $50k daily VaR, a fact they vehemently disputed.
The problem was that this fact would imply a ridiculous Sharpe ratio of around 25, too high to plausibly be due to investing skill. They did not want upper management to realize the degree to which they were simply acting like bank tellers, processing what customers bring forth, because tellers don’t make nearly as much as traders.
A firm creates confusion in producing sophisticated risk reports for business lines that are really not crucially dependent upon subtle short-term investment strategies, and misleads people inside and outside the firm into thinking profits are driven by something they are not. I once sat in a meeting where our head of trading noted proudly to our CEO that his group made money over a period in which the stock market fell precipitously. This was disingenuous because 1) they didn’t even trade or hold equities and 2) they didn’t have any significant net positions per course of regular business. The CEO nodded approvingly, I winced. This was bad on multiple levels.
It is important to estimate a daily VaR for each desk regardless of its business edge, if only to manage rogue trader risk (this way you catch him once he puts on the position). But clearly one shouldn’t spend too much time on these things when the risk is primarily operational. A rough approximation of the VaR (say, within a factor of 3) is really all you need for these cases.
An interesting capital allocation problem remains. For many activities in large institutions there is no alpha and there is little conventional risk, just a predictable cash flow due to franchise value. Most methods for allocating capital are based on estimating some statistically based adverse scenario, and this does not work well when applied to a franchise returns.
For these annuities, such as income to an internal FX or swap book, the cash flow can only go to zero, not much below, and therefore ‘risk’ is seemingly insignificant. They key to allocating capital to these activities is the key to the capital allocation problem in my opinion, and I will leave that for another post.