The worst of both worlds
Originally published 1/10/2011 © 2021 Olson Research Associates, Inc.
After years of being in the A/L modeling business I’ve heard both sides of the “out-sourced” versus “in-house” modeling debate. Either side could make quite an exhaustive list of pros and cons for each, sort of a good news/bad news list of characteristics. In my opinion you can simply boil it down to these “best” and “worst” characteristics:
Out-sourced modeling - “Best”
The model is updated regularly with standardized data. The reports generated are consistent each quarter so comparisons to past quarters are possible (and very useful). The model is run by experts in the field who understand the modeling inputs and can explain the modeling results. Being updated regularly insures good data quality and integrity.
Out-sourced modeling - “Worst”
The model may be somewhat inflexible. The range of variables that can change may be limited. You typically can’t change the chart of accounts. Because the report set is standardized it is also difficult to change or customize.
In-house modeling - “Best”
In contrast, an in-house model can be very flexible. Given the appropriate amount of time and training an in-house model can accept a wide variety of inputs, and create customizable reports and graphs for further analysis.
In-house modeling - “Worst”
Because the model is operated in-house it must be maintained and updated by bank personnel. Given that their plate is already full with financial, regulatory, and auditing requirements, updating the model often lags behind – sometimes only being updated every six months or so. And because they’re not model experts they may leave certain critical assumptions at the default setting (or worse at a setting that made sense two-years ago, but not today).
While this is not even close to being an exhaustive list, it does capture the most popular arguments for and against out-sourced or in-house modeling.
Getting the benefits from an in-house model requires a substantial dollar investment, on-staff expertise, and a considerable amount of time.
Here’s the problem – most of the banks that choose an in-house model don’t invest the appropriate amount in all three (dollars, expertise, and time). When that happens, many banks that choose an in-house model instead get:
An inflexible model - no one at the bank knows how (or has forgotten how) to change the chart-of-accounts or run a custom report.
A model that isn’t updated regularly – given the other financial, regulatory, and auditing requirements, the data gets updated very infrequently.
In other words many banks that run in-house models get the worst of both worlds.