IFRS 13 Residential Mortgage valuation

Since a number of years OSIS has developed a suite of micro and macro models for mortgage default, prepayment and loss including multinomial logistic regressions, stochastic Markov Chains, structural dual trigger models and Bayesian Vector Auto-Regressions.

Since 2017 we have started to value residential mortgage portfolios in the light of IFRS 13. In our valuation approached we leveraged on our modelling experiences with IFRS 9 modelling and stress test modelling (CCAR) on Dutch residential mortgages and UK conforming and buy-to-let residential mortgages.

We have chosen for a bottom up approach where we calculate for each individual loan the probability of default, probability of prepayment, cure rate and loss given loss for point in time into the future. Further we take into account the current state of the economy to make this predictions point in time. Finally we allow the user to add its own stress test scenario’s looking at house prices, interest rates and change of unemployment to estimate the effects on the value of the portfolio.

We think is very relevant in a time where house prices have increase a lot and there is an increasing concern that the market is going to change.

On this page we will comment on new developments in mortgage analytics and valuations, explore the value of loan level data and report on our in-depth analysis of US and European residential mortgage loan pools.