CHAPTER 11
Evaluating Value-at-Risk
Estimates: A
Cross-Section Approach
Raffaele Zenti, Massimiliano Pallotta and Claudio Marsala
11.1 INTRODUCTION
Since 1998, regulatory guiding principles have required banks with signifi-
cant trading activity to set aside capital to insure against extreme portfolio
losses. The size of the market risk capital requirement is directly related to a
measure of portfolio risk. Currently, in the regulatory framework, portfolio
risk is measured in terms of its Value-at-Risk (VaR).Also in the community of
asset management companies the quest for reliable risk management tech-
niques has grown in recent years. The concept of VaR is now widespread
among asset managers. This is an answer to the demand of sophisticated
investors, such as pension funds and foundations, and it is also a clear
response to the growing interest of asset managers for analytical tools that
give better control on their portfolios.
Therefore, an important issue for risk managers and regulators is whether
the VaR models in use are accurate enough. Verifying the accuracy of VaR
models requires backtesting and now there is a variety of tests that exam-
ine the validity of these models. Existing methods tend to absorb a large
amount of data and often show low power when used in small samples
(those typically available).
In contrast with previous research on assessing the accuracy of risk met-
rics, which has focused on back-testing models on a time-series basis, we
213