In 1997 Deutsche Bank, the parent of the UK asset manager Morgan Grenfell (now Deutsche Asset Management) paid £220m (e322m) in compensation to clients of funds run by one of Morgan Grenfell’s managers, Peter Young, who had lost money on investments in unlisted companies,
The bank was also fined a record £2m because Young’s funds were found to to be in breach of rules preventing more than 10% of the fund being invested in unlisted securities.
The episode could perhaps have been avoided if an alerting system or “smoke detector” had picked up the risk of a concentration of unlisted securities, says Yves de Naurois, the chief executive officer of Independent Risk Management Limited (IRML) a London-based risk monitoring and reporting company.
IRML, set up in last year by de Naurois and Peter Jeffreys, a co-founder of Fund Research (now part of Standard & Poor’s), has developed a web-based risk reporting system called Risk Radar. The centrepiece of this system is exception reporting – detecting changes in a fund manager’s portfolio risk profile and alerting the fund management company to the increased risk
Most alerts are false alarms, says de Naurois “Very often, like a smoke detector it’s not a real fire. It’s a cigarette that came too close to the detector. The cigarette, in our case, is data errors. The data that we receive from data vendors and custodians frequently has errors.
“So one of the first things we do when we have an alert is we check it for ourselves whether it’s real or whether it’s a data error.”
If there are real changes in a manager’s portfolio risk profile, these are often likely to be so small as to be almost undetectable. “For a portfolio manager to change drastically the risk profile of his portfolio overnight takes some pretty violent action, although we’ve seen it happen. Usually it’s a gradual process and something that can be picked up only by daily tracking of the data.”
De Naurois gained experience of risk reporting as a former head of global investment process for Citibank Private Bank. “Looking through thousands of client portfolios in the usual, prescribed way was monstrously expensive. It was also very inefficient, because you lost your ability to see things.”
An exception-based process was the only answer. “That is the way banks control their credit. There is absolutely no way they control everything all the time. What they do is create a number of indicators and if something breaks those guidelines that’s what gets their attention.”
De Naurois and his colleagues at IRML have applied this principle to fund management. “We felt we needed to have a relatively simple system to receive the data, run the portfolio against a number of risk models, and have a system send back reports and alerts.”
RiskRadar can track between 60 and 70 different risk characteristics, although IRML tends to concentrate on 12 or fewer. This is because experience has shown that typically between 80% and 90% of a portfolio’s total risk will derive from only four or five factors.
The system enables IRML, together with the fund management company, to set risk guidelines or pre-arranged levels of tolerance. Guidelines could be change in value at risk or in tracking error, concentrations, or any other measure of volatility. The exception report then identifies when a fund’s portfolio has fallen outside these guidelines.
Niall O’Connor, chief operating officer, says this rarely happens, “A happy portfolio is one where everything is nice and stable, and most of the portfolios should show pretty constant risk levels and pretty constant tracking error.
“Typically out of 100 portfolios you are going to have five to 10 where there’s a red flag saying there’s something going on here. We then look to find why the risk has changed.”
It is important to know what a manager is actually doing rather than what his benchmark suggests he is doing. There are cases, says de Naurois, where a portfolio manager chooses a benchmark that is representative of the returns he aims to generate rather than the investment policies he follows. This could be the case of a total return portfolio.
“A naive risk analysis of the portfolio against the benchmark would probably be meaningless. In this case we would want to have a proxy that is more representative of the actual policies implemented and would work with the product manager to build it , if it is not already in place.”
There may also be situations where a fund management company believes a portfolio manager is running far less risk than he actually is.
There was a recent case where a market neutral portfolio using plain vanilla options and futures was aiming to obtain a total return at a very low risk. This would fall under the non-qualified scheme definition. “The analysis revealed that while the risks were limited they far exceeded the risk profile,” says de Naurois
The problem lay in the complexity of the strategies rather than the instruments. “Unless you had spent a lot of time with this portfolio manager deciphering what he was trying to do, you would not have understood his real exposures. Risk techniques in this case allowed us to obtain relatively easily a good idea of the actual net exposures of the portfolio.
“So the notion that the portfolio manager was running a risk similar to cash was not exactly correct. It was closer to the risk of a balanced bond and equity portfolio. The fund management company addressed the situation with the manager and the risk was subsequently reduced.”
Managers who use complex derivative instruments may be running higher risks than their managements expect – or they may not. It is hard to know, says de Naurois, because complex derivatives are difficult to analyse. “We may encounter a very complex derivative instrument. All we can do in a first stage is to point to that derivative and say we cannot easily analyse it.
“You may see that as an admission of weakness on our part. But, per se, it is important information. We are alerting the fund management company to the fact that one of its portfolio managers has bought something which is not easily understandable.”
The regulator’s position is clear, he says. If fund managers intend to use complex derivative instruments they should be able to analyse and understand them.
“Asset managers are beginning to realise that when you buy derivative instruments, you also accept that, besides the cost of buying them from the investment bank, you have an additional cost of ensuring that you can understand them and price them independently of the investment bank.
Picking up data error, which accounts for 85% of alerts on RiskRadar, is an important part of IRML’s business, says O’Connor. “Recently we had a client who was asking how to set up an option because he had sent us a Nikkei warrant. We thought it looked a bit strange, and saw that the expiry date was 2049, which looked pretty unlikely. So we looked up the ISIN [International Securities Identification_Number] and found that it was just a Nikkei index tracker. Somehow it had been set up as a warrant in their system.
“Another client sent us an option where the ISIN was correct and all the parameters were correct but because our options routine spotted something was wrong with it very quickly, it became apparent that they had the wrong name of the company. It took them six weeks to spot it themselves.”
Exception reporting is not unique to IRML. Many other companies have developed their own exception reports internally. However, Peter Jeffreys, IRML director, places great store by the firm’s independence “Regulation of the retail side, UCITS III in particular, does require that directors are able to assess the risk profile of their portfolios independently,” he says.
He expects this requirement to spread to institutional business. “The trend, which is already embodied within retail fund regulation, is already being spread into best practice by IMA and other asset management companies for institutional portfolios. We fully expect that to be a major trend over the next two to three years.”