Conventional risk management as it is understood today covers many different disciplines and purveyors of this art have many different backgrounds. Doug Hubbard in his book “Failure of Risk Management” classifies risk managers into essentially four categories and calls them the four horsemen of risk.
It would be helpful to recap his taxonomy here before we discuss challenges that traditional risk management faces. So here are the four horsemen:
- “War quants” – primarily engineers and scientists from nuclear and aerospace industries
- Economists and financial analyts
- Management consultants
Hubbard particularly dings on the last group (apparently he was part of it himself) calling their art the soft-sell and snake oil salesmanship! Each group of course employs particular methodologies for measuring and managing risk. But they all suffer from one or more of the following three problems or challenges
Challenge 1: Subjectivity
The flavor of risk management preferred by last group above is the so-called “Scoring” method. You basically calculate the risk of an event by scoring its likelihood (sometimes subjectively) and multiplying it by its severity (usually subjectively). The most frequent defense for this method is that “real data is hard to come by”.
Challenge 2: Models are unreliable and cannot be quality-controlled
Every model needs to make some assumptions in order for it to be implementable. This is fine, but the reliability of models falls off dramatically when some of the basic assumptions that go into building them fall apart. For example, the classic Black-Scholes model for option pricing makes the assumption that equity returns are normally (or predictably) distributed. The failure of this model was dramatically demonstrated more than 10 years ago by the collapse of Long Term Capital Management. Furthermore models are rarely tested to see if they really work.
Challenge 3 (really more of an excuse): Quantitatively sophisticated methods are impractical
Most business problems are by nature complex. Which means that there are numerous interacting agents which influence an outcome that usually has low predictability. When someone claims they do “tactical” complexity management – which basically means treating each problem as unique – this allows them to amplify the effect of the FUD (Fear-Uncertainty-Doubt) factor and gloss over quantitative methods that can be objective.
Smart use of advanced business analytics tools can overcome the subjectivity challenge. Overcoming the mindset that quantitative techniques are impractical (challenge 3) will require democratizing business analytics. The second challenge is an over-reliance on failed models, which will also be overcome as analytics becomes more mainstream and people realize the true benefits and pitfalls of models.
Are there any other challenges that conventional risk management faces today?
Originally posted on Fri, Feb 18, 2011 @ 11:48 AM