Thursday, July 18, 2013

Quantitative risk modeling and it's pitfalls

Risk refers to the potential quantifiable losses that can occur in the foreseeable future. The quantitative risk models are tools which have been designed with the aim of building consumer confidence by reducing and finally eliminating the risk factor in any investment scenario. However, it should be noted that these models do not provide for extremely unlikely losses. Thus, it often happens that in highly volatile environment these models do not work the reason being that the occurrence of extremely rare events are no longer a rarity. Case in the point is Long Term Capital Management which had a robust risk management system. However, when Russia defaulted in loans in ruble denominated currencies the whole US market and economy was impacted. This has lead to occurrence of a series of events that impacted the investment environment of United States adversely. The number of exceptions reported by banks such as UBS, Soceital General and Deutsche bank has increased exceptionally during the 2007 and 2008 collapse. It is important to understand the role of irrational exuberance and fear of losses in such events. When events proves that the investment decision has been right for a long period of time continuously irrational exuberance sets in. On the other hand the bubble bursts abruptly at the triggering of one negative event. This is because of setting of one unexpected event. The problem with most of the models is the inability to foresee losses by occurrence of such events. It would be wrong to say that models are not linked with the practical world however such losses cannot be predicted. According to Rene Stulz, the fact that a company has incurred huge losses after having robust risk models in place is not an indication of the failure of risk models. Risk models even if they are perfectly designed can at times fail to predict the future potential losses. Most of the risk models world over are based on the basic tenet that diversification reduces risks. Hence, investment companies possess assets that are negatively correlated. These negative correlated assets are identified using quantitative models which cannot be understood by individuals like us. However, it should be noted that when the whole market collapses the risk management models fail. This can be attributed to the fact that the falling market leads to margin calls which in turn lead to squaring of profitable positions in the market hence causing a further market wide collapse.  The important feature of these collapses is the abruptness with which they occur and their ability to make the possibility of occurrence extremely remote events a reality. When most of the models are based on long term historical data and collapses happen abruptly and are short term events it is not surprising that the models fail to capture risk. The success or failure of the risk models depends primarily on the ability of the modeler to capture both rational and irrational factors in his model. It is not important whether the factors are logical or illogical what is important is whether the factors are observable and statistically tested. Hence, it can be concluded that even though models work well in non- trending as well as bull markets they tend to fail in volatile and bearish market.

Author: Abhishek Sinha

Abhishek Sinha has approximately 8 year of experience in equity research, business research and consultancy. He has also had the privilege of managing a small portfolio of INR 3 million. However, his interest lies in teaching and "demystifying concepts." He has taught students right from the age of 3 years at PP1, to 40 years at executive courses and believes teaching is not about knowing the concepts; it is about relating the concepts to the audience. At present he is "gainfully employed" at Vignana Jyothi Institute of Management, Hyderabad; where he loves to teach finance to an enthusiastic bunch of management students. His hobbies include analyzing income statement, balance sheet and cash flow.> Google +

No comments:

Post a Comment