Attending a banker's conference run by Sendero Corp., the Scottsdale, Ariz.-based company that specializes in developing asset-liability management models for banks, I was intrigued by two points.
First, about 1,200 banks now use this company's models.
Second, the models are being completely revised right now.
Asset-liability models are complicated - but look how many banks are using them! Now, even some of the young math whizzes at this conference told me that much of the discussion was over their heads. Would the chief executive of a small or medium-size bank to take time out to study the math behind these models?
A call to Elliot Rosen, Sendero's president, convinced me that you don't need to know anything about asset-liability models to get the best out of them.
Basically, I learned from Mr. Rosen that what most users of asset- liability models want is simply a "what if" formula.
They want to be able to plug in any interest rate scenario, plug in their own portfolios, and then have the model spin out for them what risks they face on market value and income under any possible set of circumstances.
Forewarned is forearmed. And a banker taking the interest rate forecast of a major economics firm or the implied future money rates as indicated by the shape of the yield curve can then determine what gaps will develop in a portfolio, what its market value will be, what runoff of assets will develop through mortgage prepayments, and what will happen to income and liquidity levels if particular portfolio-switching steps are taken.
The program works like a flight simulator; you get a chance to crash electronically while learning to handle the real thing.
To hear Mr. Rosen tell it, even I - who never went beyond geometry and algebra - could understand enough to get benefits from an A-L model.
But if using one is so simple, despite the complexities built in, why does Sendero feel it necessary to tear down the whole structure and start over again?
For one thing, a Windows version of the program was needed, Mr. Rosen explained. But in addition, he said:
*New derivatives products must be measured and valued for risk.
*Regulatory requirements force financial institutions to pay more attention to the market value at risk due to changing interest rates.
*New products, like collateralized mortgage obligations, have been developed, with cash flows that are difficult to model in an integrated, seamless fashion.
*Models must be revised to handle the increasing optionality in balance sheets. (Banks are being offered derivatives with options such as caps and floors that help manage risk, if used properly. Good asset-liability management models must include the effect of these opportunities.)
As for derivatives, Mr. Rosen agrees that no bank can get into trouble if it uses them solely as a modest revenue enhancer - for example, for moving up from the yield available on Treasuries to that available on government-guaranteed paper of similar maturity. It is only when derivatives are used as an easy way to gamble on interest rate movements that users have gotten into trouble.
Mr. Rosen convinced me that any community banker can use an asset- liability model to predict what effects uncontrollable events would have, minimize their risk, and generally be ready for whatever developments the economy might bring.
After all, there's a big difference between being ready for a spate of mortgage refinancings and getting an unexpected big prepayment check at a time of low interest rates, when it is difficult to reinvest profitably.
It seems to me that passing up asset-liability models because you didn't understand what made them work would be like refusing to switch from gas lights to electricity because you couldn't figure out what was inside the wires. Mr. Nadler is a contributing editor of the American Banker and professor of finance at the Rutgers University Graduate School of Management.