Abstract

Risk Management for Investment Funds on a Value-at-Risk (VaR) Basis

The present article analyses the suitability of the value-at-risk (VaR) approach for assessing risks in the field of capital investment as well as for regulating the use of derivatives according to the DerivativeV (method). The most important results of this analysis are as follows: The VaR represents a relatively simple reference number easy to communicate. However, it is problematic that the VaR is only ascertained on the basis of the number of loss-exceeding cases to be expected, whilst the dimension of the amounts lost does not exert any influence on risk values. Another drawback is that the VaR describes just one point in the probability distribution of market value changes. For risk ascertaining purposes, holding periods should take account of the great lengths of time customary in the field of capital investment in order to avoid possibly counter-productive short-term risk management policies. Moreover, uniform holding periods should be used for VaR ascertainment and for backtesting. A lower confidence level is preferable to a higher one, as a matter of principle, for permitting a more accurate estimation of the VaR. An at least one-year period with historical data must also be regarded as too short for the field of capital investment because of its limited historical data base. Model calculations show that, although the VaR pertaining to long holding periods has been substantially overestimated on the basis of various approximation equations, the coverage of real losses has been better as a result. Finally, it is demonstrated that investment management firms can make substantial investments in options and invest even their whole special assets in the case of specific market conditions in accordance with the requirements of the DerivativeV (method). This permits the conclusion that the VaR approach may also be used for purposes of the capital investment field and that the DerivativeV (method) may be applied in a meaningful manner as a matter of principle. Since all the other unidimensional risk measurement systems are affected by comparable deficits as well, the task is to uncover such deficits and to apply accompanying rules for getting such deficits under control, if appropriate, by requiring stress tests to be made, for instance