Barbican Consulting Limited
Earnings at Risk
Earnings at risk
Earnings at risk (EAR) is a risk management tool that's used by banks and mutual firms to manage the risks associated with Net Interest Income, (NII). It's used more by management than dealers for reasons that may become apparent.
How does it work?
In order to determine the EAR you take the current balance sheet. You then project how it will change over consecutive months typically using one year time horizon.
This data is then fed into your EAR model. This is normally a bespoke or in-house system that contains the risk management algorithms you intend to use. Next you import the current yield curve.
You could now calculate the NII you expect to earn for the next year. If you had no other income you would have an estimate of the gross profit before costs. But this assumes interest rates remain static. Your one earnings figure would not have been subjected to risk! To overcome this, EAR projects the path of future interest rates and their effect on your NII. How does it do this?
You add interest rate volatility (a measure of risk) to your EAR model. This allows the EAR model to project an interest rate path. Using this path as a base for your funding cost the model will calculate your monthly and hence annual estimated NII.
This process is then repeated several hundred times. Each time slightly different interest rate paths are generated and they in turn lead to different NII outcomes. The purpose of this simulation is to obtain a better distribution of potential NII outcomes. So which is the correct one?
None of them are "correct" they are all estimates based on market parameters. Some NII estimates will be high, some low and most will be in the middle. It is the distribution of the NII that you are interested in. It shows you the risk you potentially face.
What does the distribution tell you?
It tells you how you can be affected by different interest rate paths.
If the NII estimates have a wide distribution the lower quartile presents a significant risk to the firm. This is the earnings at risk.
It is argued by those who use this technique that this is very useful information. Executives can compare the lower estimates with the average and their budget NII.
How do you get a wide distribution?
A wide distribution shows you that your NII is subject to variance as a result of changes in interest rates. That risk can result from three things.
First you may have unhedged balance sheet exposures. As interest rates change these unhedged positions affect your NII.
Second interest rate volatility may have increased. This increases the potential for the yield curve to change in the simulation process and amplifies the risk you have.
Third some products in your balance sheet may have option like characteristics. For example some mortgages have a tendency to pre-pay as interest rates increase. If you have correctly assessed the product characteristics in your EAR model these products behave as short (sold) option positions. Rising interest rates and/or increased volatility will therefore reduce your NII.
Should limits be applied to the earnings at risk?
For many firms NII is the main income source (fee based income is by way of comparison small). This means that the board would be well advised to consider how it should protect the NII from being too low. A limit or limits for interest rate risk would normally be used for this purpose.
Limits based on the interest rate gap, basis point value, value at risk and recently stress testing are commonly used. Whether an additional EAR limit is appropriate should be subject to debate. Under consideration will be whether the structure of the firm's balance sheet warrants EAR and the strengths and weaknesses of EAR outlined below.
The strengths of EAR
EAR is more than a snap shot of risk. It attempts to model risk based on the future balance sheet, changes in interest rates and volatilities. In this respect EAR is dynamic. It shows you how your earnings could realistically be affected by changing rates and the characteristics of the products you offer. A static gap analysis will not do this.
This has a further advantage. It promotes valuable discussion around what risks you face, what causes those risks and whether those risks need to be reduced in accordance with the firm's risk appetite.
The weaknesses of EAR
The greatest weakness of EAR is that it is a "black box". Valuable debate and discussion around EAR depends on three assumptions.
First the data that goes into the model is correct.
Second the model and algorithms are correct.
Third the executives using EAR understand the reports generated.
In practice, it is not uncommon to find at least one or more of these assumptions can be overlooked.
Executives who believe this not to be the case may wish to reflect on the following questions:
What causes your EAR to change month-by-month?
How accurate are the product assumptions (for example pre-payments)?
Can you account for significant differences?
Are the most extreme NII figures important to you?
How do you justify any EAR limits you use?
Does the model contain historic or implied volatility?
How does the model stress test interest rate risk?
How do you model attrition rates?
On balance EAR is helpful but it should be used in conjunction with other risk measures in order to paint a picture of the firm's risk.
Why don't dealers use EAR? Perhaps it's because things that don’t have simple explanations are often mistrusted. And if you managed all your risk to a one year time horizon you could end up losing your shirt. Caveat utilitor (user beware).
First Published by Barbican Consulting Limited 2010