Reengineering Pricing Versus Reprising Gaps in Asset/Liability Management

Stefan Doytchinov
Commercial Bank BIOCHIM, Sofia

Nikolai Mincoff
Systems Optimisation Laboratory, Technical University of Sofia

Abstract

Gap Analysis, Dynamic Simulation Analysis and Net Asset Value Sensitivity are most common used analytical techniques in order to measure financial institutions' sensitivity to volatile interest rates. Recently several authors pointed out that an accurate interest rate risk analysis should encompass not only the gap reprising schedules but also "correlated" risks such as a basis risk, prepayment risk, early withdrawal risk as well as a cap risk.

With our paper we try to open a discussion on modelling capabilities of advanced information technologies such as fuzzy logic and neural networks, recently we applied to construct a flexible interest rate risk support tool for a financial institution. Fuzzy logic provides a modelling morphology to emulate linguistic attributes and interpolate rules associated with interest rate experts' judgement. On the other hand, the adaptation capabilities of the computational neural network add powerful futures for pattern classification and volatility approximation. We capsulate these tools in a common fuzzy-neural architecture, designed to support the interest rate risk of the institution's asset/liability mix.

Such a fuzzy-neural reengineering approach of the gap pricing process creates the opportunity to prepare hedging tool for forward gaps, to take into account managers' opinions as well as to reduce the interest-rate risk by rule-supported matching the duration of asset products from both balance sheet parts.

Keywords: Pricing, interest rates, gap analysis, reengineering, fuzzy logic, neural networks.