PROBABILISTIC ESTIMATES IN STRATEGIC PLANNING

One of the problems with static rate shocks or forecasts in bank risk management is that they do not give managers information about how likely a scenario might be to occur. What is the probability of rates rising 100bp in the next year, versus rising 200bp? Banks expecting rates to rise 200bp, may be shocked to learn that there is a very real scenario that says while short-term rates rise, the 2-year area (where many banks are the most sensitive), may actually decline. As we have said before, it is important for a bank to have a rate view in order to better manage risk and increase performance. Probabilistic forecasting takes things one step further and assigns probabilities to A/L rate shocks or forecasting. This method provides a single, integrated set of risk estimates banks can manage towards. If you consider a bank's rate view as the "most probable" case, then an important part of contingency planning is for management to understand the probability of the next most likely case. This exercise serves to focus management attention on more important scenarios. A bank will always have loss numbers should rates rise an immediate 300bp, but is that scenario worth planning for? Should the bank take a lower yield now if a rate scenario has less than 5% odds of occurring? After banks get back from their strategic planning sessions and start to model new initiatives, run the scenario through the ALM model with various scenarios to better understand the strategy. Our primary rate view, which we place a 60% probability of occurring, says that rates are 50bp greater in the 2Y area one year from today. Additionally, we place a 25% probability of a 100bp drop in the 2Y rate and a 15% probability on a 150bp increase in that rate. Notice that our statement outlines what portion of the curve we are talking about (we do the same for Fed Funds and the 10Y area) and puts a time horizon on the analysis. Each new strategy can be run separately through the model and a scenario run given each rate view. By multiplying estimated earnings and affect on the economic value of equity from our ALM model with each of these set scenarios and averaging them together (essentially arriving at a weighted average), we can now choose the strategy that gives the best return. Rolling out a new mortgage product, may outperform all other products if rates drop, but because our primary view is rates up and flat, it would not offer the return of instituting a new cash management sweep account. As a result, we can table the new mortgage product for a year when we think rates will fall. The same analysis can be had for a new branch, waiving foreign ATM fees, becoming an SBA preferred lender or any number of strategic initiatives. Better still, the act of modeling volume, pricing, revenue and expenses, gives management a better understanding of the strategy. The downside is that using probabilities of occurrence inserts some subjectivity, but, it is better to focus management on high probability events, than to treat all the same, or to focus on the wrong ones. In other words, it is better to be approximate and probably right, than precise and probably wrong.