Supplementary evidence submitted by Dr Thomas Huertas

A note on models and judgment

This note further elaborates on my response to Mr. Thurso’s question regarding models and judgment at the hearing of the Treasury Select Committee on Tuesday 6 January 2015.

Models and judgment complement one another. Each has a role to play in risk management. Models are effective in determining how the value of an asset or liability should move in response to a change in a particular factor or set of factors. For example, the value of a government bond maturing in one year will decline, if the one-year rate of interest increases. The model determines the amount by which the value declines. Take for example a one year government bond pays £3 in interest (at the maturity of the bond) per £100 face amount of the bond where the one year interest rate is 3%. In this case the value of the bond is £100. If the interest rate changes, the value of the bond will change. Suppose the interest rate increases to 4%. In this case the value of the bond will fall immediately to £99.04. During the course of the ensuing year the bond accrues interest at 4% so that the bondholder receives the same £103 at maturity (£100 in principal + £3 in interest) that s/he would have received, had interest rates remained at 3%.

Judgment enters the picture in two ways: first, one has to form a judgment about the plausible range of changes in interest rates that might occur. Is this +/- 1% in either direction (i.e. a decline in rates to 2% or an increase in rates to 4%)? Or is the range of possible movements less symmetric (e.g. from a decline of 1% to an increase of 3%)?

Second, one has to form a judgment as to whether or not the model is complete, i.e. whether changes in the interest rate are the only factor that influences the price of the bond, or whether other factors, such as changes in the credit standing of the borrower or changes in the liquidity of the bond, also influence the value of the bond. If the latter is the case, one has to judge (possibly with the assistance of further models) whether or not those factors will change; the direction and magnitude of any change and the impact that such changes would have on the value of the bond.

In practice, models fell down during the crisis because they ignored market liquidity. The model was correct as far as it went – the maths were correct. What was lacking was judgment – judgment that other factors could be at work and that other factors such as market liquidity could change.[1]

8 January 2015