Discussion of Active Risk Models
Risk Management Research Series
7 Pages Posted: 14 Mar 2019
Date Written: September 24, 2018
Although investment managers and quants have deep understanding for risk, the common investors still have questions on what exactly the risk is: how to estimate it and how to define the sources of risk.
The risk theory started with Markowitz, passing though CAPM and developing Multi-factor risk models. However, in a multi-factor setting correlation is always a problem. This problem has been solved in two different ways:
First, developing Global Risk Attribute Model (GMFM). GRAM deals with correlation applying the technique called Cascading regressions. This technique basically removes the correlation between factors and allows us to use them directly in the equation.
Second, BARRA US E3 categorize exposure to multiple factors as a specific (active) risk. Here total active risk is separated in the risk due to other risk factors (Active systematic risk) and the company-specific risk (Active residual risk).
Based on the limitations of previous approaches we develop our own approach to the investment risk by separating it into three components. It is done so that the three main sources of risk are clear and from here the model can be used also for return attribution to any kind of fund. The three components are:
Benchmark risk – it represents the risk from exposure to the market. It is most used for appraisal of passive funds that sole goal is to track the benchmark. It consist of Passive risk and Market timing risk;
Beta risk – these are the risk of exposure to other types of systematic factors. For passive managers these betas should be the same as the market, if they are to replicate its performance. This kind of risk also is the base of Smart Beta strategies;
Active risk – if the goal of active fund manager is to get “pure alpha” all he needs to do is neutralize the exposure to the previous two types of risk, or very often his results will get mixed up with the returns generated by timing the other systematic factors. From here the active risk is deconstructed into pure tracking error and strategy risk.
Deconstructing the active risk is very interesting and has some very important nuances. First of all pure tracking error is the “desired risk” especially for active investors. After all you can’t get abnormal return if you do not let your portfolio deviate from the market. On the other hand the “strategy risk” measures what is the risk in your forecasting model. This is undesirable risk, and the main goal of this is to be able to compare strategies on how accurate they are. Such a definition of the active risk is needed in order for active managers to make informed decisions.
Keywords: Active Risk, Alpha, Beta, Residual Return
JEL Classification: G11, G12, C20
Suggested Citation: Suggested Citation