“A neural network approach for the estimation of mortgage prepayment rates”
Venerdì 27 Maggio 2022, ore 16:30 - Sala Riunioni 701 e Zoom - Luca Sitzia (UniCredit SpA)
Abstract
Managing a Commercial Bank Balance Sheet entails a proper representation of liquidity aspects as well as of interest rate risks. While for trading items a “standard” pricing framework can be applied, some banking book positions need to be regarded from a different angle, considering the behavioural optionalities that are embedded both on funding and on lending sides – for instance, current accounts without a predefined duration, or loans that can be repaid before maturity. We propose an approach for the estimation of the rate of early repayment of loans which is based on a neural network technique. The model outperforms classical estimation methods, is based on a parsimonious choice of features, and explores the presence of a financial incentive to early repay floating-rate loans.