Communicator Archives: March 2002

Choose another issue

What's Wrong with My Model?

As we have discussed in previous issues, the current regulatory buzzword is model validation. Regulatory agencies need to know that the bank's model presents accurate analytical information. Sometimes during the model validation process or when examining reports, results are not as previously predicted. The next question is usually, "What's wrong with my Model"?

If you use A/L Benchmarks, the answer to this question is quite simple. Nothing is wrong with the model. However, correcting the issue is a little more complex and requires a close examination of the inputs.

The primary data source for the A/L Benchmarks model is the bank's FDIC call report because the data is assumed to be accurate and consistently reported by all banks. To this data, A/L Benchmarks uses standard assumptions based upon industry norms, mathematical relationships, accounting principals and deductive logic. But what if the standardized assumptions are not appropriate for your bank?

A/L Benchmarks allows inputs beyond the call report in the form of Managerial Input to further customize the bank's call report data. Absent these Managerial Inputs, the A/L Benchmarks model will rely entirely upon the call report data and standardized assumptions. If you find during model validation or examination of the reports that the model did not report your risk position accurately, the data going into the model is inappropriate.

How can you make sure that the assumptions are accurate? First, be sure that you know what assumptions are being used. Your Executive Report package has a section entitled, "Model Methodologies and Assumptions", which provides you with the model assumptions and shows the supplemental data you provided for the analysis. Second, in the event that the standard assumptions are not appropriate for your bank, you can supply Supplemental Data to more closely reflect your institution. The most common inputs are listed below.

Historical Assumptions:

• Loan cash flow characteristics
• Loan maturity and repricing data
• Securities download from portfolio accounting system
• Securities cash flow characteristics
• Core deposit decay assumptions
• Deposits maturity and repricing data
• FHLB advance characteristics

Forecast Scenario Assumptions:

• Market and internal rate forecast
• Provision expense & charge offs
• Loan growth assumptions
• Loan pricing spreads
• Deposit growth assumptions
• Deposit pricing
• Security growth assumptions

Perhaps your bank's characteristics do not fit neatly into these forms. Your model is not a black box. Olson Research Associates, Inc. deals with clients with assets from 15 million to 9 billion and each bank is unique. Over time, nearly 60% of our clients have begun to provide supplemental data and/or an investment download. An analyst is always available to help you sort out the most appropriate modeling technique for your bank.

"Accurate and timely identification and measurement of interest rate risk are necessary for proper risk management and control."^* The modeling process works to achieve accurate measurement of risk but this measurement hinges entirely on the quality of the data inputs. Timeliness is in complete control of the bank...the sooner you submit your data, the quicker you have your analysis.

Model validation is likely to be an examiner topic for some time. As of the fourth quarter 2001, there is a new section in your Executive Reports dedicated strictly to helping you prepare to review or validate your A/L Benchmarks model. It serves as a guideline to validate the accuracy of the inputs to the model. If you can't validate the model, we need to work together to improve the inputs to the model.

    *Joint Agency Policy Statement on Interest Rate Risk

▲Top

"Fed's Olson Doubts Case for More FDIC Coverage"
American Banker (03/13/02) P. 4; Garver, Rob

Former community banker and new Federal Reserve Board Governor Mark W. Olson^* says there is no evidence to support an increase in the $100,000 deposit insurance coverage limit. Olson insists all banks are still receiving deposits above the limit, and widespread deposit losses to non-banks is not evident. Though many community bankers are in support of legislation presently in the House Financial Services and Senate Banking committees, Olson believes banks should consider the stricter regulations that would result from a boost in coverage. Nevertheless, Federal Deposit Insurance Corp. Chairman Don Powell has pledged to continue to fight for a higher limit, at least for periodic increases based on inflation. Meanwhile, Olson warns that bankers should be on the lookout for potential difficulties in retaining deposits as the economy recovers and investors move back to the stock market. Moreover, risk management systems and loan officers will be tested as weakened credit quality becomes apparent. Finally, Olson says new loan officers will encounter the consequences of an economic decline for the first time, adding to the responsibilities of senior managers.

   ^*Governor Mark W. Olson is not affiliated with ORA

▲Top

We would also like to thank our clients for all of the referrals. Keep them coming!

Online managerial input is almost here! In an effort to speed up the delivery of your A/L Benchmarks reports, our technology team has been preparing our website to accept an online version of the Managerial Input forms. Paperless submission of data is just a click away! This online version should be ready for client use during the 2nd quarter of 2002.

▲Top

Let's look at an example. Bank A has a balance of 10,000 in small certificates of deposit earning an average rate of 3%. During a 200 basis point shock analysis, in the rates-down environment, the average yield would drop to 1%. Contractually, you must still pay the bearer principal +3% while the certificates are only "worth" principal +1% in the market. The value of the liability is inflated, which will ultimately reduce the value of equity (A = L + E).

▲Top