A growing number of our clients are using LoanEDGE and iPrice to evaluate the relative profitability of everything from existing consumer rate sheet products to commercial lending deals. But the results can be misinterpreted unless we remind ourselves that a profitability model that looks forward, like LoanEDGE does, is not a prediction of results. In fact, if we are using the model correctly it won’t predict the exact outcome of any particular loan correctly, other than by luck. This is because these models are based on average behaviors and try to explain the overall behavior of a pool of loans or deposits – not the exact behavior of any particular borrower or depositor. We often make the mistake of thinking that we are forecasting the specific outcome (such as net income) with a loan model in order to determine exactly how to price it.
An institution budget is a forecast. It is attempting to describe what the balance sheet and income statement will look like at some point in the future. A profit model is not a forecast; it is a framework for comparing options. It shouldn’t be used like a forecast, because to do so you would have to evaluate every single non-commercial borrower independently and these models aren’t built to do that, nor do you have the time. A loan model is really helping you evaluate the overall value of a group of loans of some type compared to loans of some other type (or compared to something other than a retail loan, such as a wholesale market investment).
So what good is a model, if it won’t tell me how much we are going to make? For starters, a loan model is good for making pricing decisions that are not based on antiquated practices, or simplified views of what a competitor might be doing with rates. For example, many institutions fail to consider putting longer term fixed rate loans in their portfolios because of the myth that they have more interest rate risk than you can manage. A pricing model will show you how much interest rate risk you are carrying relative to other options, and then show you how you can adjust for that in the rate or fees charged.
Clearly a pricing model has to make assumptions about the market, consumer behaviors, and costs. We have to make guesses about market rates in order to evaluate wholesale alternatives. We have to make guesses about credit losses and prepayment activity – which are consumer driven activities. And we also have to make guesses about our operating costs and other expenses.
One of the observations we’ve made over the years is that for loan and deposit pricing purposes, some of these assumptions are more important than others and, therefore, some deserve more attention than others. Let’s look at the categories.
Market Rates
We all know the folly in trying to predict markets and rates. It can’t be done with any degree of accuracy that gives you any advantage. So what we do in a modern pricing application is evaluate today’s decisions using today’s yield curves. That’s it. The yield curve shows us what we can earn right now at any given duration point. For example we know what we would pay today to buy duration insurance in the wholesale market – via something like an FHLB advance rate. We take advantage of the fact that the current market yield curves represent the market’s evaluation of interest rate risk compensation along with relative preferences among wholesale options. We use that information in a loan model to determine funding cost assumptions and investment alternatives. Some people feel that a loan pricing model should attempt to describe what happens with different future rate environments, much as you do with an asset liability model that takes a prospective institution business plan through different possible rate scenarios. We don’t do that in an individual loan model because the results aren’t useful. That kind of analysis makes the mistake of trying to forecast a specific outcome for a specific loan, rather than looking at the relative value of a group of loans. In order to make sense of future rate possibilities, we would have to compare all possible newly funded loans at the same time – which is what an asset liability model scenario does.
