Internet-mania first, then investors’ confidence crisis have added more and more volatility to the stock market. In this environment, risk management is crucial to keep any portfolio under control. But recently many ‘old’ tools and methods in the risk management industry have come under pressure and are questioned for their inability to fully predict unexpected crises. And money managers, who must demonstrate they take risk management seriously if they are to get pension funds business, are puzzled about which system is best suited for the task.
“It turns out that in order to fulfill his fiduciary responsibility a money manager must not only choose a risk model, but he must explain why he chose a specific one among all the available ones on the market and how he used it”, says Steven Bender, president of Advanced Portfolio Technologies (APT), a research firm that develops, produces and distributes leading-edge market-risk models and software-based tools.
APT started operating in New York 16 years ago and now has offices in other four countries, besides the US: London, Paris, Johannesburg, Seoul, Singapore, and Sydney. In UK APT’s risk models have just been adopted by the Financial Times for its new fund ratings, which analyses performance and risk for the retail fund market. A network of other European newspapers publishes quarterly a broad review of all European mutual funds based on APT's tools. Among American institutional clients, APT counts TIAA-CREF, Bank of America, Morgan Stanley, Deutsche Bank; outside the US, AXA, BNP, Pictet, Rothschild are clients.
“The reality is that a lot of money managers buy systems to manage risks, but they don’t use them or don’t even understand them”, goes on Bender. The majority choose systems that, like Barra’s one, first establish which risk factors must be taken into account to explain the behaviour of a stock; then they look at the history of the stock’s returns, they read it with the lens of “risk factors” and infer the future course.
“All those systems are based on what I call data fitting,” points out APT chairman John Blin, who prior to setting up the firm with Bender was a professor of econometrics at Northwestern University in Chicago and a senior vice president at the New York Futures Exchange. “No matter how many historical data you look at, you won’t understand really why a stock performed in a certain way,” says Blin. “Factors considered important today by financial market operators might be changing. Completely new variables break in, like during the ‘New Paradigm’ era, and you get left behind and get lost.”
On the contrary, the APT model doesn’t try to pick which risk factors are important: it looks only at stock prices, according to the ‘arbitrage pricing theory’ by Steve Ross (1976) and it finds variances and covariances among a huge number of assets. “The idea is simple,” explains Blin. “You assume the market knows how to make money selling certain (overpriced) stocks to buy other (undervalued) stocks. This arbitrage is embedded in prices. Stocks’ movements and inter-reactions are all you need to know to understand their volatility and to catch their risk level.” Then of course you need a very sophisticated set of computers and software for data mining: in APT’s database there are 70,000 stocks and 90,000 bonds worldwide, tracked on a weekly basis.