Institutionals failing to use risk models to gauge correlation
GERMANY – Most German institutional investors are in favour of applying mathematical risk models to their portfolios, but only a handful use the models for complex themes such as correlation or alternative investments.
Nearly 80% of respondents to Union Investment's latest risk survey of more than 100 German institutional investors said the use of mathematical risk models was 'very important' or 'extremely important' for decision-making.
Only 15% were sceptical about the use of such models or rejected them outright.
Union said the use of risk models was "especially widespread" in traditional asset classes, such as fixed income portfolio (73%), equities (57%) and money market funds (40%).
The figures dropped considerably for alternative investments, with 13% of respondents using risk models for asset-backed securities, 12% for private equity and just 4% for hedge funds.
Professor Daniel Rösch from the University of Hannover, one of the authors of the study, attributed the findings to the "complexity involved in using models to map these asset classes".
He said complexity might also be the reason why only 38% of respondents use risk models to forecast correlations, while 90% said risk models played a key role in mapping the effects of diversification.
Instead, risk models are applied to predict key risk indicators (61%), volatilities (59%) and investment returns (57%).
Alexander Schindler, a board member at Union, said he was "concerned" about the figures, as portfolio diversification was growing and increasing the importance of a "systematic evaluation of correlation effects".
Rösch added: "The interdependencies between risk types and risk drivers, as well as their measurement, will attract growing interest in the future.
"The crises of recent years have shown, however, that risks are mutually dependent and can reinforce each other."
Schindler said stricter regulation would be a "contributing factor" here, as it forced institutional investors not only to analyse individual risks "in isolation", but also to understand and evaluate them as "part of the bigger picture".
Overall, one-quarter of respondents conceded that their risk models needed improvement.
According to Union, nearly two-thirds of models are developed in-house – albeit largely with external support – whereas roughly one-third of the models developed relied either mainly or totally on external expertise."