Sample Bank. ALM Model Validation for September 2018

Transcription

1 Sample Bank ALM Model Validation for September 2018

2 Table of Contents Executive Summary... 3 Regulatory Guidance on Model Risk Management... 5 The Scope of ProfitStars ALM Model Validation... 9 Review of Account Setup Review of the Monthly Update Process Historical Review Maturity & Repricing Review Historical Deposit Rate Correlation Analysis Review of Assumptions Review of General Model Setup Bank Regulatory Guidance on ALM Modeling

3 Executive Summary ProfitStars deems the institution s ALM model setup and assumptions to be reasonable subject to making the recommended changes discussed in the model validation report. Prior to making the changes, it is recommended that the proposed changes be thoroughly discussed with ALCO to understand both the rationale for the changes, as well as how the new assumptions impact the overall model results. The following recommendations are listed in order of their relevance and effect on accuracy of the model projections / calculations: High Significance Recommendations 1. Non-Accrual loan setups should be reviewed (pages 13-15). 2. Fair value assumptions should be reviewed to ensure the model is accurately valuing each product (pages 19-23). 3. The short-term treasury spot rate should be reviewed (page 28). 4. Pricing assumptions for deposits should be reviewed and compared to the results of the rate correlation and regression analysis performed (pages and 49-51). 5. An appropriate driver rate should be selected for decay setups (pages 43-47). Moderate Significance Recommendations 1. The recommended breakout within the deposit section should be considered. These recommendations will enhance the ability to model accurate cash flows and improve projected expense calculations (page 12). 2. Rate information and Maturity/Repricing/Balloon timing should be reviewed (pages 16-18). 3. Historical averages should be in balance each month to ensure proper ratio calculations in the model (pages 34-35). 4. Current portfolio rates should be reviewed for two accounts (pages 35-36). 3

4 Low Significance Recommendations 1. Special account codes should be reviewed (pages 24 and 26). 2. Income accrual basis settings should be reviewed and documented (pages 24-25). 3. Accounts set to use instrument level detail should be reviewed (pages 55-56). 4. The out of rate checking warning level in the model should be reviewed (page 58). 4

5 Regulatory Guidance on Model Risk Management Overview The validation of an Asset and Liability Management (ALM) model is a key part of the overall model risk management process. Since 2010, there have been three primary regulatory guidance letters that relate to overall model risk management and to model validations: 2010 Interagency Advisory on Interest Rate Risk (IRR) Management 2011 Supervisory Guidance on Model Risk Management 2012 Interagency IRR Management FAQ The above letters provide a solid foundation for the current regulatory framework for the overall assessment of model risk. These letters supplement the 2000 OCC bulletin on Risk Modeling (OCC ) and expand the scope of the requirements. The 2011 Supervisory Guidance on Model Risk Management (aka the Guidance Letter ) was published by the Board of Governors of the Federal Reserve System and the Office of the Comptroller of the Currency (OCC). While not all financial institutions are regulated by the Federal Reserve Board or the OCC, it is reasonable to expect that the other regulatory agencies adopt a similar stance with regards to the managing of model risk. The Guidance Letter states that a model consists of three elements which should be validated. The elements include: 1. Information input This element relates to the validity of the assumptions and data provided to any model. Model validations performed by ProfitStars primarily focus on the institution s assumptions to ensure that they are reasonable for ALM planning and stress testing. Although ProfitStars does not have perfect knowledge of an institution s specific ALM profile or objectives, we can provide guidance on the reasonableness of the assumptions used within the ALM model based on our experience and knowledge from talking to a variety of regulators and financial institutions. 2. Processing This element includes the method and equations used to transform the inputs into estimates. ProfitStars uses an independent third party to verify that its model calculates correctly and does not contain any material mathematical discrepancies. The model certification is completed on every other 5

6 model year s release and the latest one was performed by Angel Oak Advisory. A copy of the latest model certification can be found on the For Clients Portal. 3. Reporting This element translates the processed estimates into standardized information. The output results of reports are also verified during the model certification. If an institution creates customized reports, they need to verify the accuracy of the resulting outputs. It s also important to understand the context of any formulas being used, as well as time frames being reported on. This is critical in order to have useful data for making executive decisions. The Guidance Letter further advises that there are two primary risks associated with any model. The first major risk component is model errors. These types of errors include mathematical issues and design defects. As previously stated, ProfitStars uses an independent third party to test and ensure that the PROFITstar ALM model does not have any material issues such as mathematical or design flaws. A third model error relates to assumption quality. A proper model validation will examine the assumptions used and make recommendations for changes. The other major risk component is tied to incorrectly or inappropriately using a model. The PROFITstar ALM model is meant primarily for ALM and IRR analysis. When management uses the ALM model for other forms of analysis beyond what it is primarily designed for, it runs the risk of not getting meaningful or accurate results. Because of this, management must take care to control how the model is being used. Minimizing Model Risk In order to minimize model risk, the Guidance Letter recommends that institutions: 1. Establish limits on model use: In other words, models should be relied upon up to some limit. A proper validation can provide guidance on the accuracy of the models results and what other components need to be considered outside of the model. It is important to utilize both qualitative and quantitative data to get the most out of a model. 2. Monitor model performance: A best practice is to measure performance by conducting a back-testing of past model results based on initial estimates. A sensitivity analysis can be performed which isolates the key drivers that factor into the differences between actual and projected results for a given period. As noted in the 2012 Interagency IRR Management FAQ (aka the FAQ Letter ): 6

7 Management should conduct ongoing monitoring and outcomes analysis of model performance using the institution s results (Back-testing). Back-testing results can take on many different forms. Some of the most common items backtested in an ALM model include forecasted margin (volume/rate/mix factors), pricing estimates, prepayment estimates and decay estimates. Back-testing of all of these elements is not within the scope of a ProfitStars model validation. 3. Adjust or revise model: After a validation is conducted, there should be a number of recommendations that will result in changes to different parts of any ALM model. The ProfitStars validation will show the potential degree of importance for each of the various suggestions. 4. Supplement model results with other analysis: Performing additional analyses that supports the assumptions being used. Examples of this would be conducting a regression analysis to support pricing, prepayment or decay assumptions, based on historical consumer data. 5. Have a periodic Model Validation performed: The recommendation of the FAQ Letter is that financial institutions should conduct a periodic review at least annually but more frequently if warranted. It further states that if there are material changes made to a model, the changes warrant a more frequent validation. The Guidance Letter further states that: a) All model components are subject to validation: There needs to be a complete and thorough review in the validation. b) Validation work is subject to critical review by independent party(s), not just the line of business: While internal reviews are one aspect of the validation process, other parties should also be included in the validation process. Using a third party clearly provides an independent group not subject to the control of the model user(s). c) The group doing the validation must have the requisite knowledge, skills and expertise: It is acceptable to use parties such as an institution s internal audit area, as long as they have sufficient ALM knowledge. If the institution uses a third party, the party needs to have the skills and knowledge in ALM. An organization such as ProfitStars has employees doing validations with a strong background in ALM. d) The group doing the validation must have explicit authority to challenge users: Using a third party such as ProfitStars provides this form of objectivity and informed opinion needed. ProfitStars does not have a vested interest in 7

8 the financial institution following a particular balance sheet strategy. ProfitStars goal is to help the financial institution understand their risk-return tradeoff for any strategies that they look to undertake so that the institution can make an informed decision for themselves. 6. Have a validation report which is structured to include: a) That all the model aspects have been reviewed. b) Highlights of potential deficiencies. c) Whether adjustments to the model are warranted. d) A clear executive summary. e) A synopsis of the model that includes major limitations and key assumptions. The regulatory letters stress the importance of an independent party being involved in order to provide an unbiased opinion as to the current state of the ALM model and make appropriate recommendations. As summarized in the Guidance Letter, Independence is not an end in itself but rather helps ensure that incentives are aligned with the goals of the model validation The quality of the [model validation] process is judged by the manner in which models are subject to critical review. In this regards, an institution should look for independence in two forms: 1. Independence from the party responsible for the model updating and maintenance processes. ProfitStars does not maintain the models for its clients. An exception to this is our ALM service bureau clients. For these clients, we recommend that another party performs the model validation in order to avoid a potential conflict-of-interests. 2. Independence from providing recommendations that the third party could potentially benefit from. ProfitStars does not recommend (nor do we benefit from) one particular ALM or investment strategy over another. 8

9 The Scope of ProfitStars ALM Model Validation The ProfitStars ALM model validation is intended to address the key elements discussed above which pertain specifically to the validation of the ALM model inputs and assumptions. The following is a list of what is directly covered within the model validation document: Review chart of accounts setup. Evaluate chart of accounts to determine if there is appropriate breakout on fixed, variable, tiered and callable instruments. Review setup of non-accrual and participated loans to ensure best practice assumptions and procedures are observed. Review fair value discounting methodologies and modeling practices. Review the utilization and setup of key rates and yield curves. Examine application files to ensure all required downloadable fields and recommended fields are in use. Test results of model inputs & outputs based on the monthly update process. Run correlations & regressions of historical loan and deposit rates compared to market rates to determine possible pricing relationships. Assess if offering rates and portfolio rates appear reasonable and updated on a regular basis. Review reasonableness of the maturity and repricing cash flows. Review the estimated non-maturity deposit life assumptions utilized by the financial institution. Review the estimated prepayment assumptions utilized by the financial institution. Examine the use of key rate ties, balance sheet ties and other projection formulas available in the model setup. Verify use and setup of rate floors and ceilings within the model. Issue an Executive Summary that indicates the potential impact of key findings and recommendations. Issue a Report of Engagement which discusses the details of key findings and recommendations. Conduct a consultative call to review findings included in the written Report of Engagement. 9

10 The following list includes the items NOT addressed in the ProfitStars ALM model validation report. These items are still considered a critical but separate part of the overall ALM model risk management process: Earnings risk analysis Capital risk analysis Liquidity risk analysis Credit risk analysis Back-testing analysis ALM strategies analysis Review of policies and procedures relevant to the ALM/IRR process Review processes employed for model updates Review processes employed for ALCO/board/auditor/examiner preparation Reconcile ALM/IRR data to financial statements Ensure proposed recommendations are changed within the financial institution s model or a final review of the changes made (additional charge and follow-up time would apply if requested) 10

11 Review of Account Setup Balance Sheet Setup Account Detail The account detail within the balance sheet will determine how accurately the balance sheet will react during interest rate risk calculations and how the institution is able to forecast/budget. Our recommendation is to have enough detail to account for the unique characteristics and optionality of each product type. LOANS The model would benefit from additional breakout in the loan section of the balance sheet. For loan categories that have varying interest rates and terms, added breakout would allow the institution to more accurately project rates for new volume and ensure proper interest income calculations. Vehicle loans, personal loans and other miscellaneous types of loans have rates that vary based on the loan term and consumer credit scores. The used vehicle loan portfolio, for example, currently shows downloaded rates that range from as low as 1.74% to a high of 14.99%. Loan categories with such wide rate variations should be broken out by their term and risk rating to more accurately project offering rates and maturity schedules for new volume. Using a blended rate could misstate interest income when new volume is projected on loans that have this pricing structure. Additionally, the current setup means a single maturity interval is being used for new volume that may not accurately reflect the portfolio. Used vehicle loans are currently set to 36 months, which means all new volume in projections will use a 36 month amortization schedule. Separating these loans by terms will not only give the model a more accurate projection of future cash flows, it will allow more control when forecasting. For example, if there is anticipated future growth in 36 month used auto loans but fewer loans are expected to lock into 84 month contracts, the suggested breakout would give the model the flexibility to project these various scenarios. 11

12 Additional subtotaling in the loan section is also necessary. Subtotals would provide for better monthly and quarterly reporting within the model and also helps with balancing during the update process. INVESTMENTS The investment section would also benefit from extra breakout. Looking at the detail of the investment portfolio, mortgage backed securities appear to be both fixed and variable rate. However, the MBS security account in PROFITstar is set up as a fixed rate account. Variable rate investments should be separated in the chart of accounts to ensure the repriceability of these investments is modeled correctly. Callable investments should also be separated from non-callable investments. There is currently just one CD investment that is callable. If the institution is going to continue investing in callable instruments, these accounts should be separated out in the chart to ensure call features are modeled correctly. DEPOSITS Deposits are currently broken out by types and organized between demand accounts and time deposits. Regular certificates and IRA certificates are currently categorized by term, which follows our recommendation. Similar to recommendations made in the loan section, deposits should also be split out based on the rates they earn so that interest expense can be properly calculated. According to published rates, money market plus accounts have tiered rates based upon the current balance. Given that these funds have the ability to reprice, even a 5 bps difference in rate can have an impact on the earnings risk analysis. The balance should be used to separate these deposits so correct rates can be modeled. Additionally, several other non-maturing deposit accounts have minimum balance requirements to earn a dividend. This can be modeled as well by separating accounts that do not meet the minimum balance so that interest expense is not applied to those deposits. For liquidity purposes, CDs and IRA CDs should be divided out by dollar tiers of <$100k, >$100K and >$250K so that non-core funding dependence ratios can be determined. This will also give the model more granularity when doing projections if these CDs need to be projected differently based on the deposit size. 12

13 Non-accrual Loans When PROFITstar recognizes an instrument as a non-accrual loan via the Non-Accrual Flag loan field, it automatically codes the coupon as 0%. If Instrument Level Detail (ILD) is enabled for these accounts, all cash flows are placed in the first monthly time bucket. Not enabling ILD allows the user to manually adjust the non-accrual cash flows. It is recommended to not have ILD enabled for non-accrual loans so that cash flows can be adjusted based on their expected pay down. Consequently, projections for performing loans will be inaccurate if non-accruals are not split out from the rest of the portfolio and given their own account. We recommend separate accounts for non-accruals (either a single account for all loan types or multiple non-accrual accounts broken out by loan category). The ALM model does not currently separate non-accrual accounts from performing loans. Given the increase in non-accrual loans across the industry, it is recommended as a best practice that these loans are separated out in the model. According to the loan and lease analysis provided to the FDIC in June 2015, non-accruals make up 0.24% of the total loan portfolio. This is down from 0.32% reported in June 2018 and also indicates fewer non-accrual loans than the peer group, which reported a ratio of 0.94% in March It would still be beneficial to separate these loans to follow best practices in modeling these accounts and to ensure these loans are set up correctly should nonaccrual loan volume increase in the future. RATE SENSITIVITY Non-accrual loans should be set as rate bearing accounts with a 0% interest rate so that they are handled accurately in regards to earning asset calculations. A key rate tie would be used to prevent projected income from being calculated during rate shock calculations. This can be done in the Key Rate Ties tab that appears after setting the rate to Fixed. The recommended setup for this would be to tie the account to a standard rate, such as National Prime, and use a beta (multiply by) of This would prevent these loans from suddenly earning interest in a shocked scenario. 13

14 The following example shows a proper key rate tie setup for non-accrual loan accounts: CASH FLOWS & MATURITY SCHEDULE Another important aspect for non-accrual loans is the expected cash flow of these loans. Separating non-accruing loans from accruing loans on the balance sheet allows the institution to manually input the expected cash flow schedule (action plan) for nonaccruing loans in the Maturity & Repricing module. Due to the nature of non-accruing loans and the fact consumers are not holding true to contractual obligations, these loans do not likely follow contractual maturity terms to determine the cash flow schedule. Therefore, the cash flows generated by contractual information from the loan file are not likely accurate. Additionally, it is not assumed that these accounts will never be returned to regular accrual status or otherwise be charged off. The most accurate approach is to manually input when each non-accruing loan is expected to be recovered or charged off in the Maturity & Repricing module. I would generally expect a WAL of one year or less if the institution is aggressively trying to reduce this portfolio. As a best 14

15 practice, I recommend overriding the contractual cash flows in the maturity/repricing module with expected pay downs of non-accruals on a loan-by-loan basis. With current non-accruals using the contractual maturity schedule, the model is likely overstating the duration these loans are held on the balance sheet or the period of time they are held in a non-earning status. In order for these settings to be used for non-accruals, there would need to be a sort field in the loan file that distinguishes these loans from its regular loan portfolio. After reviewing the loan files, it does not appear as if a non-accrual code exists in the core system. This is covered in more detail in the Review of the Monthly Update Process. Loan Participations When PROFITstar recognizes an instrument as a participated loan, it should be set up to ignore the balances that are parted out and only model the portion of the loan that the institution retains. If the loan file does not contain the necessary fields to recognize participation loans, it is common to see clients create contra asset accounts in order to model the net loan balances. Contra accounts come from the general ledger and do not provide cash flow information. Consequently, projected loan cash flows and interest are inaccurate when using this modeling method. We recommend against using contra accounts to model loan participations. Although there is a participation contra account on the balance sheet, this account has never carried an active balance. As noted above, the use of a contra balance to model the sold amounts is not the recommended method of modeling participated sold loans. In order to accurately model the net cash flow, a participation code should be used in the loan file to determine the net amount. I recommend contacting our support department for more information on the proper modeling of these loans from an ALM perspective if participations are bought or sold in the future. 15

16 Rate Type/Maturity Type/Maturity Interval/Repricing Type/Repricing Interval The rate type options for rate information are: Fixed, Variable, or N/A for non-rate bearing accounts. These options determine if the account generates income or expense and whether the account is a fixed or variable rate instrument. The Maturity Interval, along with the Maturity Type, determines when new volumes will mature. The Repricing Interval, along with the Repricing Type, determines the frequency that the new volume of a variable account will reprice. The Maturity Interval is intended to reflect the contractual term of the instrument. This Maturity Interval is applied to any new volume that you project on the account. You can apply prepayments to that maturity schedule to assist in modeling an average life for the instrument. We recommend reviewing the chart of account settings at least annually. There are three fields that make up Maturity Timing. They are Maturity Type, Maturity Interval, and Balloon Interval. The Maturity Type options are: N/A: There is no maturity for the account. This option should be used for all nonrate-bearing accounts and accounts that do not have maturity associated with them (such as demand deposits). Number of Months: The maturity length for the account is expressed in number of months. Month of the Quarter: The account matures in a specified month in the quarter. The maturity interval value indicates which month (e.g., 2 = second month in the quarter). Month of the Year: This account matures in a specified month of the year. The maturity interval indicates which month (e.g., 1=Jan, 8 = August). This is used for accounts like Christmas Club accounts that mature once a year. There are three fields that make up Repricing Timing. They are Repricing Type, Repricing Interval and Initial Interval. The Repricing Type options are: N/A: There is no repricing for this account. This selection is for Fixed Rate accounts. Number of Months: The repricing for the account is expressed in number of months. Month of the Quarter: The account reprices in a specified month in the quarter. The repricing interval indicates which month (e.g., 3 = third month in the quarter). 16

17 Month of the Year: This account reprices in a specified month of the year. The repricing interval indicates which month (e.g., 1=Jan, 8 = August). For fixed-rate certificate accounts, the maturity and repricing intervals would match the term, in number of months, for each line item. Loans and investments require additional thought & analysis. While your current portfolio may be a good indicator of the term of your new volume, a detailed review of products booked in the most recent months is recommended. Balloon Intervals The balloon interval is used to determine the amortization stream for new volumes. There are three options: If the value in the Balloon Interval is zero, the account does not amortize. The entire balance matures based on the Maturity Interval. If the balloon term is equal to the Maturity Interval, an amortization schedule is calculated based on the Maturity interval (this account has NO balloon payment). If the Balloon Term is less than the Maturity Interval, a balloon payment will be made. The instrument will amortize based on the term specified by the Maturity Interval until the balloon interval is reached. At that point the remaining balance matures. A balloon interval of 1 indicates that the entire balance could mature at once. General guidelines for Balloon intervals are as follows: Loans should use the same balloon interval as maturity interval except for true balloon loans. For true balloon loans, use the stated balloon interval. Interest-only loans should have a balloon interval of 0. Amortizing investments (such as MBS and CMO) should be treated like loans. Non-amortizing investments should have a balloon interval of '0'. Time Deposits should have a balloon interval of '0'. Demand type accounts should have a balloon interval of '1'. All other accounts not fitting into one of the above classifications should have a balloon interval of '0'. 17

18 Comments on Rate Information: Account settings appear to be appropriate with the following exceptions: Acct # Account Title Current Rate Setting Recommendation 8 Mortgage Loans Fixed This account contains both fixed and variable rate loans. These accounts should be broken out into separate accounts so the variable rate loans are modeled correctly. Comments on Maturity/Balloon/Repricing Type & Intervals: Account settings appear to be appropriate with the following exceptions listed below and highlighted in red: Account # Account Title 6 Ag Loans R/E Farmland R/E Residential 31 Consumer Current Maturity Interval 6 Months 36 Months 12 Months 60 Months Current Balloon Interval 0 Months 36 Months 12 Months 60 Months Current Repricing Setup 1 Month Fixed Fixed Fixed Recommendation Although there are loans in this category that have 6 month maturity schedules, the average maturity of recent loans is over 20 months. The balloon interval of 0 also indicates that these loans do not amortize and only interest payments are due each period. This is fairly typical for these loans. However, if this is not the case this interval should be adjusted to match the maturity interval being used. The average loan originated in 2018 had a term of 72 months (6 years). The maturity and balloon intervals should be updated to reflect this. Loans originated in the last year were 5 year loans (on average). An interval of 60 would be more appropriate. Considering this loan category accounts for over 16% of total gross loans, this can have a substantial impact on earnings projections. The average term for the most recently originated loans in this group was 2.4 years. Therefore, a shorter interval of 24 to 36 months seems more reasonable. As mentioned earlier, this category may benefit from additional breakout due to the varying terms and rates within this loan group. I strongly encourage the institution to review the accounts noted above to ensure an accurate cash flow schedule is being calculated for new volume. Incorrect rate 18

19 information or maturity and repricing intervals can overstate/understate the amount of interest rate risk on the balance sheet. Fair Value Codes The correct Fair Value code needs to be selected for each balance sheet account in order for the Fair Value matrix to correctly calculate Fair Value of Portfolio Equity. There are a number of different ways to determine the value of each balance sheet account. Most non rate-bearing accounts are valued at their current balance. For rate-bearing accounts the valuation process involves calculating the present value of all future cash flows. Fair Value Treatment options include: N/A: These accounts are not included in the Fair Value calculation or the Fair Value reports. Book: The Historical, Current, and Projected Fair Value of this account are equal to the balance of the account. Quote: The Historical Fair Value of the account is manually input in the edit screens for the Historical Fair Value data item. Discount: The Current and Projected Fair Value of the account is calculated using discounting techniques (its Fair Value equals the sum of the discounted cash flows). This option is available only for rate-bearing Balance sheet accounts. FAS107/Discount: This option can be used to calculate the book value of Demand Deposit Accounts for FAS107 purposes and for Economic Value calculations, the accounts can use decay rates or core/non-core to discount their value. The general guidelines for setting up these codes are as follows: All non-rate bearing accounts except capital accounts are set to Book. All rate bearing accounts are set to Discount. Rate Bearing Securities, such as Treasuries, Agencies, Municipals, CMOs, MBSs, Asset Backed Securities and other like accounts where a market value can be obtained are set to Quote If a market value cannot be obtained use Discount. 19

20 Other accounts that add into Total Investments, such as Bank CDs, Mutual Funds, FHLB Stock, Federal Reserve Bank Stock, Corporate Bonds, Bankers Bank, Wescorp, BOLI, and other like accounts should use Book or Discount as appropriate. All capital accounts are set to N/A and are excluded from the calculation. Any Unrealized Gain/Loss Accounts on Investments are set to N/A and are excluded from the calculation. Any Valuation accounts such as premiums and discounts are set to N/A and are excluded from the calculation. Each time you make changes to the Maturity or Repricing Interval in your chart of accounts, you will need to review the Fair Value codes for each account. Refer to the guidelines explained above to determine the code that should be used. Accounts appear to have appropriate settings with the following exceptions: Account # Account Title Current FV Setting Recommendation 76 Fas115 (A/D) Quote None 105 CD Mkt Val Adj Book None INVESTMENTS Bonds, agencies and MBS accounts are set to use a quoted value. Although this follows our best practice recommendation, quoted market values appear to be the same as book values. Book value is not necessarily reflective of market value. MBS accounts are secured by pools of real estate loans that have scheduled cash flows. These securities will have varying degrees of risk and sensitivity to rate changes based upon the pool of mortgages that securitize the bond, which would impact how individual cash flows are discounted. This should give each security varying market values in both the flat and shocked rate scenarios that would be above or below the current book value. If interest rates were to increase, the value of the MBS would fall. This is especially true for long term bonds. Fair market values should be obtained for these investments in each rate scenario used in the model. The following table shows sample investments from the MBS portfolio with their most recent market values for various rate scenarios in relation to the book value: 20

21 CUSIP Book Value MV -200 MV -100 MV Flat MV +100 MV +200 MV MBU21 $146,103 $182,290 $170,052 $151,725 $141,451 $137,944 $137, ZP0 $135,479 $178,642 $162,338 $143,880 $136,739 $134,507 $133, YR4 $198,266 $268,525 $249,656 $217,250 $198,068 $191,673 $191,291 Total Value $479,848 $629,457 $582,046 $512,855 $476,258 $464,124 $462,300 Once the Enable Shocked Quotes option is turned on in PROFITstar, supplied market values can be entered into the history module. This is further discussed in the Review of General Model Setup. If quoted market values are not obtainable, the fair value treatment should be changed to discount by the treasury yield curve. LOANS Loan accounts are set up to discount by the treasury yield curve with a corresponding fair value spread. This follows our recommendations. The use of the treasury yield curve will account for the time value of money. Time value of money states that a dollar received today is worth more than a dollar received tomorrow. Although a 30 year mortgage loan provides greater earnings potential, the risk of not getting 100% of the principal back increases as the consumer has longer to pay the loan off. Discounting the loan portfolio by the treasury yield curve will value the short-term cash flows by a shortterm spot rate (less risky) while the long-term cash flows will be valued based upon long-term spot rates (more risky). Since the treasury yield curve is a risk free curve, a spread must be applied to account for the risk in pricing the loan portfolio. Spreads are currently being calculated by the PROFITstar model using the calculate spread option. Using this feature in PROFITstar will calculate the spread based upon an average life input for each loan portfolio. This method compares the spot price on the treasury yield curve (based on an average life input) to the account s current offering rate. Considering this process is automated, it offers an efficient way of applying spreads but this method does have limitations. For example, if a loan is priced above market (premium), then the spread that is calculated will be higher which can result in a lower fair market value of that loan account. Unfortunately, the PROFITstar method doesn t account for the premium that likely exists in pricing the loans above market. Ideally a loan spread would be determined by the institution so that a premium in pricing can be accounted for. Although this is more time consuming than the calculated option currently used, it gives a more accurate spread. Below is an example of 21

22 procedures that can be used for determining loan spreads which can be input into the chart of accounts manually: 1. Determine the new market rate for this item. If possible use a local market offering rate. If the local rate is not readily attainable use a regional or national rate. This rate should reflect the general credit quality of the asset, thus don t use an AAA auto rate for loans that are subprime. Also, use a rate with a similar maturity to the item. 2. The model shows the behavioral weighted average life for the chart of account line item. Using this maturity to determine the current yield curve spot rate. 3. The spread is the difference between the market rate and yield curve spot rate. 4. The following example illustrates this point, assuming all the mortgage loans are 30 yr. conforming mortgages. Account Real Estate Loans Source of Market Rate 30 Yr. FNMA Commitment Rate Market Rate WAL 3.314% 10 Years Treasury Yield Curve Spot Rate 1.96% (10 Yr. T-Note) Spread to Curve 1.354% *Market Rate Information as of March Document the results in a form similar to the above and key each spread into the chart of accounts within PROFITstar. DEPOSITS/BORROWINGS Deposit accounts are being discounted to a curve of Federal Home Loan Bank (FHLB) advance rates to value non-maturity and time deposit balances. This follows our recommendations as the FHLB rates are the closest corresponding market rates for deposits. Borrowing accounts that remain on the books for longer than one month are also set up to discount to a curve of rates, which follows our recommendations. Deposits should also have a cost associated with each product to account for the cost in servicing these products. Deposit accounts are currently applying a calculated spread, similar to loans, which does not follow our recommendations. Instead of a spread, a servicing cost should be used. Servicing costs can vary from one institution to another. At one point, there were servicing costs prescribed by the Office of Thrift and Supervision (OTS) that gave industry standard costs to use as a benchmark. The OTS servicing costs can be found in the OTS publication called Selected Asset and Liability Price Tables. Although the OTS 22

23 agency is no longer in existence, this is a common starting point used by many financial institutions until internal costs are developed. The servicing cost estimates from the former OTS study are: Deposit Type Servicing Cost Non-Interest DDA 2.57 Savings/Passbook 1.39 Transaction 1.80 Money Market 0.86 Certificate of Deposit 0.20 I recommend current spreads are replaced with applicable servicing costs for each deposit account. Although the OTS servicing costs mentioned here can be used as a proxy, it is recommended that servicing costs are derived from an internal cost analysis. Auto Update Actions Setting up the Auto Update Options makes the building of the maturity/repricing schedules fast and accurate. Before using the Auto Update option, you must first identify the appropriate action to be taken for each account on your balance sheet. The general guidelines for setting up these codes are as follows: All non-rate bearing accounts should be set to Balance in > Last Rate bearing accounts that are downloaded from an application file (Loan, CD, Investment) should be set to None Demand Accounts should be set to Balance in > Last when using Decay Rates, otherwise set to core/non-core. Accounts that you are inputting the cash flow should be set to None Overnight Accounts (Fed Funds) should be set to Balance in First Amortized Accounts that are not downloaded from an application file (Credit Cards, Student Loans) should be set to Amortize In process, suspense, clearing, settlement and unposted accounts should be set to Balance in First 23

24 Accounts have appropriate auto update settings. Loans and investments that come from the application file download are set to have an Auto Update Action of None. Additionally, non-maturing deposits are set to use the >Last option so that decay speeds can be applied correctly. Lastly, CDs are using a setting of None so that the downloaded contractual information is used to determine the maturity schedules. These settings follow our recommendations. Special Account Codes Special account codes are used to identify specific accounts or types of accounts. They are also used for such things as calculating ratios, cash flow reporting, and some calculations in Projections. All balance sheet special account settings appear appropriate with the following exceptions: Account # Account Title Current Code Recommendation 134 Sub Fixed Assets Fixed Assets None 139 Total Cash/Due Cash None 401 Total Borrowed Money None Total Borrowings Special account codes are used for certain ratio calculations within PROFITstar. In order for accurate ratio calculations, the recommendations should be made as suggested above. By having these subtotals set with a special code, the model is double counting these assets in several calculations. Taking this special account code off of these accounts will prevent this in the future. Additionally, the Total Borrowings Special account code needs to be set for the total borrowing account. Income Statement Setup Accrual Basis The accrual basis is used to calculate the monthly interest income or expense on an account. The basis is also used in calculating historical portfolio rates and ratios. While there are some industry standards, interest annualization conventions are at the 24

25 discretion of each institution for loans and deposits. Common industry conventions include: Account Type Accrual Basis All registered bonds 30/360 All conforming, 1st Lien RE products 30/360 Commercial RE 30/360 All other loans Days/365 All deposits Days/365 Overnight-type accounts Days/365 FHLB advances 30/360 Most accounts are using Days/365 for the accrual basis. Only two accounts are set to use 30/360 including the FHLB Stock account and the US Agency investments. Considering there are not set rules for how these should be applied, these settings should be reviewed to ensure the proper accrual basis is being used for each account. To review this, an income statement setup report can be run from the Report Wizard in PROFITstar. Either myself or customer service can assist with this if the institution needs help reviewing these settings. Federal and State Tax % There are three options for calculating federal and state income taxes in projections. The options are: None the system performs no tax calculation and makes the projected tax accounts editable. Effective Rate (the default method) the system multiplies the effective tax rate account found in Databank by the monthly taxable income number. Tax Table the system calculates income taxes on a year-to-date basis using the tax brackets and rate specified on the customizable tax table. In addition, the institution can identify if an account is taxable or tax exempt. This can be done by inputting a percentage on each detailed account that identifies the percent of a detail account s income or expense that is subject to State or Federal taxes. 25

26 Accounts with a State or Federal Tax value less than 100% result in an income statement adjustment when calculating the monthly State or Federal Taxable Income amount. The bank is currently forecasting a Federal Tax rate of 21% and a State tax rate of 10.84%. Based on the historical tax expense, and the recent tax reform, the effective rates appear appropriate. Special Account Codes Special account codes are used to identify specific accounts or types of accounts. They are also used for such things as calculating ratios, cash flow reporting, and some calculations in Projections. Income statement special account settings appear appropriate, with the following exception: Account # Account Title Current Code Recommendation 401 Total Borrowed Money None Total Borrowings The account noted in the table should be updated to use the recommended special account code. Databank Setup Shock Constraints Key rate accounts have a setting under Setup>Databank for Shock Constraints that enables the user to limit how shock scenarios impact each key rate. There are two options available for this setting: 26

27 Allow negative rates will not limit the movement of a market rate during a rate shock and allows the rate to go below 0%. Do not allow negative rates assumes the key rate will never fall below 0%. The setting of Do not allow negative rates is the recommended option for proper ALM modeling. Market rates would generally have a floor of 0% and this setting would prevent negative rates during interest rate shocks. Additionally, in Setup>Preferences there is an option to Allow negative Discount Rate due to Servicing Cost, and this is enabled by default. When applying a servicing cost on a deposit account, the servicing cost is subtracted from the discount rate. It is not likely that an institution would use a negative discount. Therefore, it is recommended that this option is disabled in the model. With the recent rise in interest rates, the talk of negative interest rates as a stimulus option has diminished. However, given the uncertainty that remains about the future of economic conditions, many financial institutions and examiners still want to stress the impact of a negative rate environment. If the financial institution would like to model the impact of a negative interest rate environment, the 2016a version of PROFITstar can accommodate this stress scenario. To model negative rates, the following steps must be done: 1. The shock constraint setting within Setup > Databank should be set to Allow negative rates for all applicable key rates. 2. To ensure an offering rate is allowed to go negative, the bank must set a floor on each account that is less than the projected offering rate. For example, if the financial institution would like to see the impact if the offering rate of cash held at the Federal Reserve Bank drops to -0.50%, a floor of at least -0.50% must be input on this account. Because not all product offering rates will go negative (i.e. loans), there is no global feature that can be turned on to model a negative rate environment. The model requires the user to manually input the negative floor rate on all accounts the bank wants to stress. On the following page is an example of how to set up an account to model the impact of negative rates: 27

28 Databank Setup (for Key Rate Tie and IRR Calculations) Key Rate Tie Setup (if applicable) Offering Rate Limits (in Projections tab of Chart of Accounts) Yield Curve Setup In Databank you may have multiple yield curves. For graphing purposes the key rate curve you choose needs to have the number of months until it matures placed in the account setup. Many clients are now discounting to a yield curve rather than discounting to an offering rate when calculating fair value. In order to determine the rate a particular balance would be discounted to, the PROFITstar Fair Value calculation matches up the balance in each maturity time frame (for each individual account) with its corresponding point on the yield curve. For example, based on the maturity/repricing schedule, the dollars scheduled to mature in the first maturity/repricing time frame are discounted to the one month rate on the yield curve. Both the Treasury and FHLB Yield Curves are being appropriately modeled. The Treasury Yield Curve is utilizing spot prices from a one month rate (Fed Funds) to a 30 year rate (30 Year T-Bond). Similarly, the FHLB advance curve is utilizing spot rates from 1 month to 20 years. This follows our recommendations for setting up these yield curves. 28

29 29

30 Review of the Monthly Update Process The purpose of this section is to assist in reviewing the process followed when loading client data into the model. This includes a review of the download files and the processing of the general ledger and application files. Download Setup Balance Sheet (PSGL.csv) All required fields are present and appropriately set up. Income Statement (PSGL.csv) All required fields are present and appropriately set up. The download is utilizing the MTD value for income and expense, which is the preferred method for importing these balances. Time Deposit File (PSCD.csv) The following fields should be added to the Time Deposit File: Missing Field Origination Date Impact of Missing Field Statistical items such as 'Number of New Accounts' and 'New Account Balance' cannot be determined without this field. Demand Deposit File (PSDD.csv) The following fields should be added to the Demand Deposit File: 30

31 Missing Field Origination Date Impact of Missing Field Statistical items such as 'Number of New Accounts' and 'New Account Balance' cannot be determined without this field. Loan File (PSLN.csv) The following fields should be added to the Loan File: Missing Field Non-Accrual Flag Interest Only Flag Original Interest Rate Origination Date Actual Maturity Spread to Key Rate P&I Payment Date Impact of Missing Field This field is used to identify any loans that are in a non-accrual status. This field will allow PROFITstar to automatically change the interest rate on these loans to 0.00% to prevent interest from being calculated. As mentioned earlier in the validation, non-accrual loans should be considered an earning asset and therefore should be ratebearing. This field will enhance projected income and interest rate risk calculations. It can also be used as a sort field to separate these accounts in the chart. This field will identify whether a loan record makes interest only payments. This field should also be used as an additional sort field to separate IO loans from P&I loans. Without this field PROFITstar would be unable to utilize Lifetime Ceiling/Floor Spreads. Statistical items such as 'Number of New Accounts' and 'New Account Balance' cannot be determined without this field. Since the Actual Maturity represents a principal only amount which includes the contractual maturity amount plus any prepayments received, this field can be used to help calculate the historic prepayment behavior of loan records. Without this field, alternative methods would have to be used to come up with prepayment history to use for future expected prepayment behavior. This field identifies the spread or margin above/below the rate index. This information is used to determine what rate each variable rate loan is eligible to reprice to. This field is critical given the low rate environment and the fact that a lot of loans are currently at or below their floor. In some cases, loans will not get above their floor even if market rates climb 100 bps. Not having this field can potentially cause an overstatement in interest income during interest rate risk analysis for earnings risk. This field is necessary for loans that are interest only for an initial time period, then make P&I payments for the remainder of the term. This field identifies the end of the interest only period, at which time the P&I period begins. It is necessary to provide correct cash flows for applicable IO records. 31

32 Missing Field P&I Payment Amount Impact of Missing Field This field identifies the principal and interest amount the institution will receive once the interest only period has ended. It is necessary to provide correct cash flows for applicable IO records. Investment File (investments.csv) The following fields should be added to the Investment File: Missing Field Origination Date Payment Amount Payment Frequency Rate Change Frequency Rate Tie Description Repricing Date Spread to Key Rate Impact of Missing Field This field identifies the date each security settled. Statistical items such as 'Number of New Accounts' and 'New Account Balance' cannot be determined without this field. This field is also necessary to determine an accurate average age for any amortizing security utilizing PSA prepayment assumptions. This field represents the next contractual principal and interest payment amount to be received and is required for amortizing securities. If this field is missing, PROFITstar will treat the security as a bullet instrument with principal due at maturity. This field identifies how often principal payments are received. If this field is missing, PROFITstar will utilize the next payment date field (if valid) to model the next principal and interest payment. All future payments will then default to 'monthly' payments. This field identifies how often each security is eligible to reprice. This field is less accurate than the Repricing Date as it will not necessarily model the contractual repricing terms. If this field is missing, PROFITstar will look to the Rate Tie Description field to determine future repriceability. This field is the third, and least accurate option of modeling the repricing characteristics of each variable rate security. Any alphanumeric character that appears in this field will translate to a 'monthly' repricing term. This field identifies the next date each variable rate security is eligible to reprice. This is the most accurate option of modeling when each security can reprice. If this field is missing or incorrect, that record will be treated as a fixed rate security. In conjunction with the rate index, this field will determine the future rate of each security. As with the rate index field, this field is required for accurate interest income calculations. If this information is missing, the security will not reprice correctly and effectively overstate or understate projected interest income. 32

33 Key Rate File (keyrates.prn) All required fields are present and appropriately set up. 33

34 Historical Review Databank Items Updated: The Databank stores a variety of financial and non-financial information that is used throughout the model. There are three major sections of items in Databank: 1) Check figures for balancing that include Total Assets, Net Income, and YTD Income, 2) Off Balance Sheet Statistics including Charge offs, Recoveries, and Delinquencies 3) Key Rates Check figures for total assets, net monthly income and YTD income have been consistently entered in the model for the last 10 years at the consolidated level. These check figures are useful to assist in the monthly balancing process. Other off-balance sheet items are also being tracked, including charge off dollars, recovery dollars and number of accounts that can be used by both PROFITstar and Profitability. I often see databank as one of the more underutilized areas of the ALM model, so it is good to see it effectively used as intended. Key rates provided by the monthly download via PROFITstar have also been populated consistently for the last 10 years. FHLB rates are entered manually each month as well. This follows our recommendations for best practices in ALM modeling. Balance Sheet Month End/Averages The month end balances are critical for financial reporting and projections. Out of balance issues may indicate problems with the download or incomplete data. Problems can occur in maturity/repricing and projections when history is out of balance. Average balances are used for both ratio and portfolio rate calculations. In reviewing the last 10 years of historical month-end and average balances, month-end balances are perfectly balanced to the GL and averages on only off in a few timeframes. Small variances of $1 can be attributed to rounding and ultimately these have no serious 34

35 impact on the results in the model. Averages are used in several ratio calculations, so ensuring that these figures are accurate is import to proper modeling. February 2012 was the most recent example of averages being out of balance and it was only for $47. This will not have a major impact on the model but these figures should be in balance in future months. Income Statement Income statement balances are used when completing many of the ratio calculations. For rate bearing accounts, the income or expense is used when calculating portfolio rates. In reviewing the last 10 years of historical income/expense values, all time periods are in balance at the consolidated level. Portfolio Rates The current month's portfolio rates are used in updating maturity/repricing. Missing portfolio rates in the history module could cause incorrect or missing portfolio rates in the maturity/repricing module, which in turn can cause incorrect calculations of interest income/expense in the projections module. There are two accounts with active balances that do not have a portfolio rate entered. These accounts are downloaded from the GL, so portfolio rates must be manually entered or calculated. The following accounts are missing portfolio rates: Account # Account Title Portfolio Rate Offering Rate 10 LOC-Visa Corp 0.000% 2.800% 12 LOC-Cash Advance 0.000% 0.001% 35

36 PROFITstar is able to calculate a portfolio rate on accounts that have interest income/expense and an average balance. Within the History>Rates section there is a Distribution option to Calculate Portfolio Rates. The following formula is used to apply a rate: (Interest Income/Expense * 1 / Basis For Individual Account) / Average Balance (Current Month) Offering Rates An offering rate is the current rate offered for new volume. The current month s offering rates are used in projections and Interest Rate Sensitivity Analysis. In the Interest Rate Sensitivity Analysis feature, the current month offering rate is used to calculate interest income and expense on rate bearing accounts. When no rate is entered as the current month offering rate, net interest income could be calculated incorrectly. The model does not appear to have offering rates. The last time offering rates were entered for any organization was in April 2018 and historic offering rate input is sporadic. The offering rates for the following accounts should be reviewed: Account # Account Title Current Portfolio Rate Current Offering Rate 6 Ag Loans 3.88% 3.25% 3.75% 17 R/E Frms 13.00% 12.00% 11.00% IMPACT ON IRR Average Rate on New Volume Not having offering rates in the current month will significantly impact IRR results. Although its interest bearing accounts are tied to key rates, which handles the projection of interest income and expenses, a lack of offering rates can impact the prepayment assumptions being used in the projection. Most of the loan portfolio is using the BondEdge mortgage prepayment assumptions, which utilize elastic prepayment speeds. These loans are also set to use the offering rate as the driver for determining which prepayments to apply. In not having a historical offering rate in the 36

37 model, PROFITstar assumes that shocked prepayment speeds should be used in the projection which will result in improper maturity forecasts. An example of this would be the R/E Arms-3 Yr category, which accounted for just over 7% of the total gross loans in December With an offering rate of 0% historically and a projected rate of 4.75%, the model is anticipating a very low prepayment amount on these loans because the model is reading this as a +475 bps shock. Therefore, PROFITstar uses the shocked prepayment speeds for these loans rather than the flat prepayment speeds it should use here. If the historical offering rate were changed to 4.75% OR if the driver rate for the prepayment speeds on this account were switched to something other than the offering rate, the flat rate prepayment speed would be appropriately modeled. 37

38 As shown here, this is likely understating the prepayments on these loans which in turn overstates the loans duration. This could have a major impact on both earnings risk and capital risk calculations. Offering rates should be updated appropriately and a prepayment driver other than the offering rate should be used to ensure proper IRR results. Even if regular projections or IRR analysis is not being conducted in PROFITstar, it is still important that offering rates are entered into the model. Historic rates can be used to analyze offering rate movement in relation to market rate changes. This type of data can then be used to help develop or review pricing assumptions used in the model. 38

39 Maturity & Repricing Review Maturity/Repricing Balancing Although maturity and repricing is static, this information is an integral part of your projections. The maturity and repricing data works in connection with your growth assumptions. This allows PROFITstar to accurately calculate future purchases as well as when variable rate accounts reprice. There were no balance or rate errors/warnings. Balancing maturity/repricing is a critical step in the monthly update process. Loan, investment and deposit cash flows are the backbone for proper projection and interest rate risk results. Maturity/Repricing Reasonableness The maturity and repricing schedules are used in the calculation of interest income and expense in the Projections module and as the starting point for both Fair Value and the Interest Rate Sensitivity Analysis results. The maturity schedule should represent the timing of the principal payments on the account. The repricing schedule should represent the timing of when the account can reprice for the first time. Contractual cash flows appear to be reasonable, with the aforementioned exceptions of callable investments and loans with prepayments. Making the recommended changes will result in more reasonable cash flows in the behavioral view by properly modeling these instruments. 39

40 Historical Deposit Rate Correlation Analysis A key component that financial institutions must address when attempting to effectively model deposits is the rate that will be earned from and paid to consumers, both in the current rate environment as well as future rate environments. Understanding how the institution s deposits have been priced historically provides valuable insight in determining the validity of the pricing assumptions being used in PROFITstar. As most balance sheets are liability sensitive with regards to repricing, modeling deposit accounts for purposes of interest rate risk assessment presents a challenge for most institutions as the sensitivity of the account offering rates to secondary market rates vary at different points in the business cycle. This is also magnified by the fact that most depository institutions maintain a large percentage of their natural deposit base in accounts that lack a specified maturity date. To properly assess the risk to future earnings or the risk to an institution s capital position, it is imperative that financial institutions periodically review and assess the validity of the pricing assumptions used to model deposits, as these assumptions are used to calculate the projected change in interest expense as well as to calculate the fair market value of the accounts. This section will provide insight into how deposits have been priced during the most recent business cycle. This analysis is structured to assess the rate change sensitivity of deposit offering rates in relation to changes in secondary market rates. In addition to assessing the magnitude of change, the analysis will assess the timing of the change by including various lags or delay periods. Components of the Study Correlation coefficients show directional movement between the historical changes in the offering rates and the historical changes in the selected market rates. In simple terms, a correlation of % means that the two data series are moving in the same direction 95.00% of the time. Higher correlation results indicate greater directional movement in the same direction. Alternatively, if the correlation observed is %, it means the two rates are moving in opposite directions 95.00% of the time. If the correlation is observed to be zero, it means that the data series are neither moving directly or inversely; no real pattern is evident. The correlation analysis will indicate the secondary market rate that each deposit has the 40

41 best correlation to but will not indicate the sensitivity to market rate changes (rising/falling rate environment). A regression analysis was performed to determine the sensitivity of each of the deposit rates to market rate changes. A regression is performed using the best fitting market rate and lag period against the offering rates to come up with the linear equation that characterizes the relationship between the rates. This linear equation produces a slope ( beta ) and y-intercept ( then add ) which can be incorporated into key rate tie (KRT) pricing assumptions directly into the PROFITstar model. Each linear regression produces an R-Square value. The R-Square value is used to measure the level of success of the regression in predicting the value of future offering rates. The higher the R-Square value, the better the ability of the market rate to predict changes in offering rates. An R-Square with a lower value corresponds to a lesser ability of the market rate. Low R-Square values can signify many things, including non-rate factors and pricing decisions that are made regardless of current market conditions. As a first step in assessing historical pricing decisions for deposit accounts, ProfitStars performed a correlation analysis of each of the selected market rates listed to the right and the offering rates for deposit accounts. Secondary Market Rates In the PROFITstar ALM model there were a total of 26 ratebearing deposit accounts eligible for this analysis. To achieve the most precision with the statistical analysis, each deposit account was examined on an individual basis for the period covering October 2008 to September 2018, or a period of 120 months. For accounts that are newer and didn t have data going back to October 2008, these accounts were reviewed for shorter periods of 72 and 96 months. Fed Funds 1 month Libor 3 month Libor 6 month Libor 3 month Treasury 6 month Treasury 2 year Treasury 5 year Treasury 10 year Treasury WSJ Prime The information provided in the resulting tables should be compared to the assumptions currently used in the PROFITstar model in the key rate ties section of the chart of accounts to determine the validity of the current pricing assumptions. 41

42 Non-Maturing Deposits Account Market Rate Correlation Multiply by Lag R-Square % (Beta) (Months) Value Regular Shares 6MLIB 98.34% % Club Accounts 6MLIB 98.16% % Money Mkt Shares 6MLIB 97.73% % IRA Share Accts 6MLIB 97.71% % Youth Accounts 6MLIB 98.16% % College Club 6MLIB 97.84% % In reviewing the non-maturing deposit rates, accounts correlated most closely to changes in the 6 Month LIBOR rate. Results showed a 2 month delay in reaction to changes in market rates, as indicated by the lag periods. This makes sense as financial institutions tend to wait to see how competitors react to rate changes while also waiting to gauge consumer demand. Beta coefficients that resulted from the study also follow industry trends. Regular Shares, Club Accounts and Youth Accounts had low betas when compared to Money Market Shares, IRA Share Accounts and College Club accounts. Normal share deposits tend to be less rate sensitive than higher yielding money market and IRA accounts. For many institutions there are key non-rate factors that drive a depositor s decision to maintain these accounts such as direct deposit, online bill payment, branch location etc. Therefore, aggressive pricing may not be needed to maintain these regular transaction accounts. Money market or IRA accounts, on the other hand, tend to be less sticky and depositors are generally more rate sensitive with these accounts. Higher beta results for these accounts make sense as these deposits likely price more aggressively to avoid losing funds to competitors. These results should be reviewed to determine whether the suggested betas and lags seem appropriate for use in the model. The correlation % returned on each account type is well above what I would consider to be a reasonable result for implementing key rate ties. Additionally, the beta and lag periods seem reasonable for use as they follow general logic and also remain within the range I typically see on non-maturing deposits. However, the period covered in this analysis was primarily a flat rate environment with little movement in market rates after With a period of rising interest rates looming, actual pricing strategies could vary from what the historically based results indicate. Internal experience and expertise with deposit pricing should generally be used in conjunction with statistical analysis when applying key rate ties to the model. 42

43 Time Deposits Account Market Rate Correlation Multiply by Lag R-Square % (Beta) (Months) Value 3 Month Certs 6MLIB 98.72% % 6 Month Certs 6MLIB 98.53% % 9 Month Certs 6MLIB 97.90% % 12 Month Certs 6MLIB 97.98% % 15 Month Certs 6MLIB 98.58% % 18 Month Certs 6MLIB 97.53% % 24 Month Certs 6MLIB 96.98% % 30 Month Certs 2YTreas 96.60% % 36 Month Certs 2YTreas 96.02% % 48 Month Certs 2YTreas 94.86% % 60 Month Certs 2YTreas 92.86% % 3 Month IRA 6MLIB 98.64% % 6 Month IRA 6MLIB 98.57% % 9 Month IRA 6MLIB 98.08% % 12 Month IRA 6MLIB 98.07% % 15 Month IRA 6MLIB 98.57% % 18 Month IRA 6MLIB 97.65% % 24 Month IRA 6MLIB 97.31% % 30 Month IRA 2YTreas 96.65% % 36 Month IRA 2YTreas 95.92% % 48 Month IRA 2YTreas 95.08% % 60 Month IRA 2YTreas 92.47% % 3 Month Certs 6MLIB 98.72% % 6 Month Certs 6MLIB 98.53% % CD and IRA certificates are strongly correlated to 6 Month LIBOR rates and the 2 Year Treasury rates. Results showed varying degrees of lag in reaction to changes in these rates, with shorter term certificates reacting more quickly than long term accounts. Beta coefficients vary from a low of to as high as This type of pricing behavior makes sense as short term CD rates tend to fluctuate more than long term rates and certificates are traditionally more sensitive to rate changes than non-maturing deposits. As with the non-maturing deposits, the correlation % returned on each certificate type is well above what I would consider to be a reasonable result for implementing key rate ties. Behavioral results for beta and lag periods also seem reasonable for time deposits. Even with these strong results, internal experience and expertise with deposit pricing should generally be used in conjunction with statistical analysis when applying key rate ties to the model. 43

44 Review of Assumptions Decay Rates The sustained volume of NMD inflows at historically low rates compounds the complexity of IRR management. Predicting future behavior of these depositors is a key component of IRR modeling. The combination of lengthening or shortening asset duration while managing liability stability directly affects IRR. - OCC Semiannual Risk Perspective (Fall 2018) The importance of modeling deposit behavior has become increasingly apparent. It has also become increasingly more challenging. Industry experts and examiners alike have taken note of the impact these assumptions have on interest rate risk and these assumptions have come under increased scrutiny. Unlike certificate deposits, nonmaturing deposit (NMDs) accounts do not have stated maturity terms that can be easily determined in an ALM model. Therefore, decay rates are used to give these deposits an implied life in the model by predicting deposit runoff. Decay rates are a way of stating that a certain portion of these deposits will leave the institution at some point in the future. Depositor behavior can change based on many circumstances and can be influenced by external factors such as market rate fluctuation, competition in the region, economic conditions etc. Not all deposit types behave the same, so choosing an accurate decay assumption for each deposit type is important. For instance, hot money sources such as money market accounts are generally expected to have a faster decay than more sticky money sources such as checking accounts. Typically decay speeds are expected to show some sensitivity to rate movements as deposit runoff may increase or slow down as rates change. However, some institutions can convincingly make the argument that market rate movements or market conditions do not impact their depositor behavior and a static speed can be used. Recent technological advancements have changed how consumers decide where and how they deposit their money. Institution that locks clients into a multitude of various add on products such as online banking, direct deposit, mobile banking and other options greatly reduces the influence rates have on a depositor s decision to open or close an account. It can also be argued that life events have a greater impact on deposit runoff. A job change that has caused a loyal customer to relocate, marriage/divorce, death etc. could all lead to a consumer s decision to close an account. Regardless of the theories surrounding depositor behavior, the fact still remains that depositors will spend their 44

45 money or close their accounts at some point in the future. ALM models need to account for this fact. The following illustration outlines data the OCC gathered from Q to Q The results of the survey indicate the annual decay rates utilized by more than 1500 banks. This will give the institution an idea of how others are modeling the anticipated runoff within the non-maturing deposit portfolio. Annual decay rates in the illustration table translate to the following average lives: WAL (yrs.) 25th Percentile Median 75th Percentile MMDA High Yield MMDA Now Savings Non-interest bearing Other Surge Deposits Since 2008, the industry has also experienced a surge in core deposits, which represent the extraordinary growth of the deposit base relative to previous historical levels. In essence, the percent of surge deposits can be calculated by taking the annual deposit dollar growth percent post-2008 minus the annual dollar growth percent pre-2008 to determine the presence of surge deposit dollars. 45

46 Core deposits surged during the recession and remain significant because of the low interest rate environment (see figure 22). Transaction accounts experienced a growth rate higher than in other periods of declining rates. Total deposits increased at a similar rate to past cycles, but customers migrated to transaction accounts thereby increasing the reliance on transaction accounts. These inflows from depositors looking for safety and liquidity could quickly reverse when rates rise. Accordingly, the OCC expects banks to carefully analyze the recent behavior of liabilities and to use this information to model alternative assumptions in interest rate models. - OCC Semiannual Risk Perspective (Fall 2015) Industry concern for surge deposits is statistically backed. As the above chart shows, there was a clear shift during the recession in which core deposits became a much more significant part of the balance sheet for many institutions. Surge deposits may not be easily calculated for newer deposit accounts and may not be present for all deposit types. For accounts where a surge percent is determined, this balance should be treated differently from the balance of core deposits. The threat of these surge deposits being non-sticky and leaving at some more immediate point in the future is an area of concern, both from a liquidity and a fair value perspective. Therefore, modeling the potential impact of surge deposit runoff is vital to proper risk analysis and should be included in the institutions assumptions for deposits. 46

47 The current decay estimates and the resulting WAL calculation used on demand deposit products are outlined below: Acct # Account Title Decay Method Driver Rate WAL 173 Demand Deposits Monthly Rate Fed Funds TOTALLY FREE CKG Monthly Rate Fed Funds FREE BUSINESS CHK Monthly Rate Fed Funds COMM BUS CKG Monthly Rate Fed Funds Checking W/Int Monthly Rate FED FUNDS Citof Paragould Monthly Rate FED FUNDS Prime Plus 55 Monthly Rate FED FUNDS Attny Aolta Chk Monthly Rate FED FUNDS The Works Chking Monthly Rate FED FUNDS PLUS CHK/INT Monthly Rate FED FUNDS VIP CHK/INT Monthly Rate FED FUNDS LOW MIN CHK/INT Monthly Rate FED FUNDS WALL STREET C/I Monthly Rate FED FUNDS BUSINESS INT CKG Monthly Rate FED FUNDS HIGH INTEREST CK Monthly Rate FED FUNDS NO MIN CHK/INT Monthly Rate FED FUNDS ECONOMY CHK/INT Monthly Rate FED FUNDS HEALTH SAVINGS Monthly Rate FED FUNDS Super Now Accts Monthly Rate FED FUNDS Regular Savings Monthly Rate FED FUNDS Savings No Int Monthly Rate FED FUNDS School Savings Monthly Rate FED FUNDS Corning Bank MM Monthly Rate FED FUNDS Money Market Monthly Rate FED FUNDS HiFi Account Monthly Rate FED FUNDS Working Money Monthly Rate FED FUNDS NW BRANCH DEPOSI Monthly Rate FED FUNDS - *Accounts with a WAL of - do not have a current balance. Based on the resulting deposit lives, these assumptions seem reasonable. However, it should be noted that these decay speeds are static speeds that do not vary with changes in market rates. This can best be explained by non-rate factors (such as branch convenience, direct deposit ties, auto-bill pay ties etc.) playing a larger role than rate factors in a consumer s decision to keep their account where it is, particularly on accounts with inherently lower sensitivity to rates, like regular checking products. When compared to the OCC survey results, hot money sources (i.e. money markets) are slightly lower than the median average life results that is between the 25th 47

48 percentile and the Median. The Now and Savings portfolios have an annual rate that is higher than the median which translates to an average life that is between the Median and the 75th percentile. It is important to understand that the OCC s survey results are not benchmarks. Rather they are a guide to determine how reasonable the current assumptions are. I recommend a more comprehensive study should be conducted to determine if these assumptions are reasonable. This type of study can be completed internally or through a third party and could be as simple as a closed account analysis. The PROFITstar model has the ability to track closed account data when the proper fields from the deposit file are available. This data can be viewed in the Historic Balance Sheet within PROFITstar or reports can be generated to show this statistical information with the average life of closed accounts from month to month. This information can be used to come up with statistically backed internal decay speeds for each product type. Prepayment Speeds Prepayments are the difference between the contractual payments that are to be paid monthly and the actual payments received. Prepayments can be tracked by taking the first month maturity from the previous month's maturity schedules in PROFITstar and what was actually received in payments for the month. The actual payments for the month are: Current Month's Balance - Previous Month's Balance - New Volume for the month. Prepayment assumptions affect liquidity risk, the amount of new volume needed to achieve monthly projected volumes, and interest rate risk. Downloading your loans into maturity/repricing captures the contractual maturities. However actual principal cash flows received may be more than what is contracted to be received. Additional cash flow from loans and investments means additional liquidity that if not planned for could result in a loss of potential revenues. The additional liquidity could also mean missing goals for growth in individual accounts or overall institution growth. The additional principal cash flows also mean that more volume is available to reprice. This means unanticipated interest rate risk. 48

49 As rates change, so do prepayment patterns. Typically there is an inverse relationship between prepayment speeds and the movement of interest rates. When interest rates go up prepayment speeds slow down and when interest rates go down prepayment speeds increase. This process can be modeled in PROFITstar using the prepayment elasticity factors. Typically there are six rate scenarios modeled: 3 Rising Rates and 3 Falling Rates. However, the institution can modify these scenarios as they see fit. As you project changes in offering rates, or as rate shocks are applied in an IRSA or Fair Value calculation, the prepayment speeds will fluctuate based on your elasticity settings. This allows you to further customize the prepayment effects in various rate scenarios. Loan accounts are all using some form of prepayment, with the exception of the Construction and Secured Business loans. Many of the loans are using the PROFITstar model to calculate prepayments based upon 12 months of historical principal payment amounts received from the core system. This follows our recommendations and using internally driven prepayments is preferred. However, using the auto calculated prepayment speeds results in the application of a static prepayment. This means the loans will not prepay at a faster rate when interest rates fall nor will loan prepayments slow down as interest rates rise, as is typical for most consumer loans. This could potentially overstate or understate the loan maturities in projections. The real estate loan portfolio is using BondEdge mortgage prepayment speeds. This follows our recommendations for prepayment options when internal data is not available. These prepayment assumptions are based upon pools of mortgage backed securities gathered from Interactive Data's BondEdge Fixed Income Analytics. Pools include both Conventional and GNMA securities for different terms, rates and origination dates to account for the seasonality and vintages of each loan record. Additionally, these prepayments offer elastic assumptions. This means that a different prepayment is assumed when there is a shift in market rates, rather than using a static speed that does not change with rate fluctuation. Student loans are using prepayment assumptions provided from an internal source. As with other assumptions in the model, there should be strong documentation of how these speeds were derived. 49

50 Lastly, the offering rate is being used as the driver rate for all prepayments. This should be changed to a more appropriate market rate as prepayments are generally reliant upon external rate factors, rather than the accounts own offering rate. The following table shows the current prepayment setups, along with current contractual and prepaid Weighted Average Lives: Acct # Account Title Prepay Method Auto Contractual Prepayment (Months) WAL WAL 5 Personal Loans Auto Calculate Years 1.6 Years 6 OD Protection Auto Calculate Years 0.1 Years 8 New Vehicle Lns Auto Calculate Years 2.0 Years 9 Used Vehicle Lns Auto Calculate Years 1.8 Years 10 Othr Secured Lns Auto Calculate Years 2.1 Years 11 Indirect Lending Auto Calculate Years 0.0 Years 12 Indir Used Veh Auto Calculate Years 0.0 Years 13 Mobile Home Lns Auto Calculate Years 2.7 Years 14 Share Secure Auto Calculate Years 1.5 Years 17 Fed Student Lns Constant Prepay N/A 5.2 Years 3.5 Years 18 Priv Student Lns Constant Prepay N/A 4.5 Years 3.2 Years 19 Land Loan Auto Calculate Years 5.2 Years 20 Equity Plus Auto Calculate Years 2.0 Years 21 Equity Plus 2 Auto Calculate Years 2.1 Years 22 HELOC Auto Calculate Years 1.3 Years 23 Int Only HELOC Auto Calculate Years 3.0 Years 24 F/R 5 Year Mtg Auto Calculate Years 1.6 Years 25 F/R 10 Year Mtg BondEdge (Conv 10) N/A 3.5 Years 1.9 Years 26 F/R 15 Year Mtg BondEdge (Conv 15) N/A 6.5 Years 2.3 Years 27 VAR HE-Qtrly Auto Calculate Years 3.3 Years 28 ARM 3 Yr Q BondEdge (Conv 10) N/A 13.5 Years 3.0 Years 29 ARM 5 Yr Q BondEdge (Conv 10) N/A 15.9 Years 3.2 Years 30 ARM 7 Yr Q BondEdge (Conv 10) N/A 15.6 Years 3.4 Years 32 F/R 15 Yr CUSO BondEdge (Conv 15) N/A 5.2 Years 2.3 Years 33 F/R 30 Yr CUSO BondEdge (Conv 30) N/A 9.7 Years 2.9 Years 34 V/R 1st Mtg CUSO BondEdge (Conv 10) N/A 10.1 Years 3.3 Years 38 F/R BLS LNS BondEdge (Conv 10) N/A 7.3 Years 3.4 Years 39 V/R BLS LNS BondEdge (Conv 10) N/A 9.8 Years 3.3 Years 40 Business Loans BondEdge (Conv 10) N/A 8.1 Years 3.2 Years 50

51 Key Rate Ties Key Rate Ties provide a method of telling the model how to calculate projected offering rates based on a key rate account in Databank. Consider using the Key Rate Tie feature on your loans, investments, and deposits (savings). The benefits of this option are faster projections since fewer account offering rates have to be projected and more accurate rate shocks since the account offering rate is determined by the Key Rate tie. Key rate ties are being used on all of rate bearing accounts. LOANS Loans are assuming a full and immediate shock of interest rates. These accounts are using the key rate ties to project interest rates in the model, but they do not limit price movement in shocked environments. This is a common setup for loan products. DEPOSITS Non-maturing deposits are using varying betas and lags, with betas of either 0.5 or 1.0 and lag periods of 3 or 6 months. Additionally, some of the lags used are immediate while others assume an average lag that spreads the rate change over time. Certificate and borrowing accounts have key rate ties that assume immediate shocks of interest rates. Similar to the loan ties, these key rate ties are used for projecting rates rather than as a way to model behavioral price changes in varying market environments. These settings follow our recommendations and I would consider these assumptions to be on the conservative end of what I typically see used in other ALM models. Strong documentation for how these pricing assumptions were derived should be maintained. The following table shows the current key rate tie setup: Account # Account Title Key Rate Beta Lag 4 Paric Purchased FED FUNDS Ag Loans NAT PRIME LOAN Tax Free Loans NAT PRIME LOAN R/E MTG DEPT NAT PRIME LOAN R/E Arms-6Mo NAT PRIME LOAN R/E Arms-1Yr NAT PRIME LOAN R/E Arms-3Yr Fix NAT PRIME LOAN R/E Arms-3 Yr NAT PRIME LOAN

52 Account # Account Title Key Rate Beta Lag 13 R/E Arms-5Yr Fix NAT PRIME LOAN R/E Arms-5 yr NAT PRIME LOAN R/E Arms-7Yr Fix NAT PRIME LOAN R/E Arms-Jr Mtg NAT PRIME LOAN R/E-Frms NAT PRIME LOAN Real Estate-Fix NAT PRIME LOAN R/E-FRM Pool #1 NAT PRIME LOAN R/E Const 1 to 4 NAT PRIME LOAN R/E CONS NON 1-4 NAT PRIME LOAN R/E Farmland NAT PRIME LOAN R/E Residential NAT PRIME LOAN R/E HOME EQUITY NAT PRIME LOAN R/E Resid Mult NAT PRIME LOAN R/E Commercial NAT PRIME LOAN R/E Comm Variabl NAT PRIME LOAN Commercial Var NAT PRIME LOAN Student Loans NAT PRIME LOAN Installment Loans NAT PRIME LOAN Consumer NAT PRIME LOAN Credit Card Lns NAT PRIME LOAN Other Consumer NAT PRIME LOAN Commercial NAT PRIME LOAN NWA LOANS NAT PRIME LOAN US Treas (AFS) FED FUNDS US Agen (AFS) FED FUNDS US Agen (HTM) NAT PRIME LOAN Muni (HTM) NAT PRIME LOAN FED TAXABLE MUNI FED FUNDS % Muni (AFS) FED FUNDS TARP Investment NAT PRIME LOAN Fed Resrv Stock BASE FHLB Stock FED FUNDS Fed Funds Sold FED FUNDS CD w/ Other Banks FED FUNDS MBSs FED FUNDS Money Market FED FUNDS Christmas Club NAT PRIME TRAN Checking W/Int NAT PRIME TRAN Cityof Paragould NAT PRIME TRAN Prime Plus 55 NAT PRIME TRAN Attny Aolta Chk NAT PRIME TRAN The Works Chk NAT PRIME TRAN PLUS CHK/INT NAT PRIME TRAN VIP CHK/INT NAT PRIME TRAN LOW MIN CHK/INT NAT PRIME TRAN

53 Account # Account Title Key Rate Beta Lag 193 WALL STREET C/I NAT PRIME TRAN BUSINESS INT CKG NAT PRIME TRAN HIGH INTEREST CKG NAT PRIME TRAN NO MIN CHK/INT NAT PRIME TRAN ECONOMY CHK/INT NAT PRIME TRAN HEALTH SAVINGS NAT PRIME TRAN Super Now Accts NAT PRIME TRAN Regular Savings NAT PRIME TRAN School Savings NAT PRIME TRAN Money Market NAT PRIME TRAN HiFi Account NAT PRIME TRAN Working Money NAT PRIME TRAN NW BRANCH DEPOSI NAT PRIME CD Time Deposits NAT PRIME CD FHLB O/N BORROW FED FUNDS FHLB SHORT BORRO FED FUNDS NW BORROWINGS FED FUNDS FHLB # FED FUNDS FHLB # FED FUNDS FHLB # BASE FHLB # BASE FHLB # FED FUNDS FHLB # FED FUNDS FHLB # FED FUNDS FHLB # NAT PRIME - LOAN 1 0 Rate Ceilings & Floors Are Rate Ceilings & Floors set up in the chart of accounts to download? On adjustable rate loans, floors and ceilings can be added to reduce the risk to the consumer and the institution. Lifetime limits specify the maximum and minimum rates over the lifetime of the instrument. In addition, the change in rate may be limited each time the instrument reprices (periodic floor and ceiling). If the download file includes ceilings and floors, this information for each loan can be incorporated in the repricing schedule for the current volume. If this information is not available you still have the option of manually entering the information into your maturity/repricing schedule. 53

54 The loan extract file contains absolute lifetime ceilings and floors. In order for these ceilings and floors to be applied to projected income and interest rate risk calculations, each variable rate account must have a setting within the chart of accounts set to download. This option is activated on variable rate loans, which allows the ceilings and floors downloading from the loan file to be used in PROFITstar. This increases the precision of interest rate shocks for variable rate loans and follows our recommendation. Rate Ceilings & Floors Setup on New Volume In addition to the option to download floors and ceilings on existing balances, you can also set floors and ceilings on new volume. Within the earning asset portfolio, the bank has input new volume ceilings and floors on a few cash equivalent and consumer loan accounts. Based on the loan file, most of the existing variable rate loan portfolio has contractual floors. I recommend the bank begin inputting ceilings and floors for new loan volume based on input from your loan department. The bank has input new volume ceilings and floors on all applicable deposit accounts. Given the volatility within this portfolio, I recommend the bank periodic review the ceilings and floors to ensure they represent the absolute ceiling and floor on each deposit product. Callable Instrument Option Callable instruments are difficult to model because their maturity date changes under different rate scenarios. Checking this option activates a number of options, which will help you model callable instruments more accurately. 54

55 If the callable feature is in use, fields for each of the seven scenarios in IRSA/FV will be active. The options are Call, Maturity or Calculate. If Call is selected, the instrument will mature at the call date specified. If Maturity is selected, the instrument will mature at the maturity date. If Calculate is selected, the user will determine during the download process the rate at which the instrument will be called. This section is not currently applicable to the bank s balance sheet since the bank currently does not have any callable securities. Projection Formulas Projection formulas can be used to calculate either the projected volume of an account or the offering rate. Formulas will not be available for accounts that already have a key rate tie, balance sheet or income statement tie, optimized or are already systemcalculated. When formulas are active the projected values of the related accounts are not editable. As previously noted, the Treasury and FHLB yield curves are using databank projection formulas to correctly model each curve. These two formulas were reviewed earlier and no changes are recommended. There are no other projection formulas being used at this time. Instrument Level Detail Instrument level detail can be utilized in several areas of the model (Maturity/Repricing, Projections, IRSA, and Fair Value). For variable-rate accounts, this provides more precise income and expense results, by improving the accuracy of repricing calculations for detail records that have rate ceilings and floors. In addition, differences may be observed when instrument detail is used in conjunction with accounts utilizing ProfitStars Loan Prepayment Estimates or PSA prepayments, which factor in the average age of the instruments. 55

56 While instrument level detail (ILD) is useful for adding precision to variable rate limits and prepayments, keep in mind that it does add time to the monthly download process. Therefore, it is recommended that ILD is only enabled on variable rate products that have valid ceilings/floors downloading and fixed rate products that are utilizing BondEdge prepayment assumptions. ILD is turned on for all time deposit accounts. These are fixed rate accounts that do not use floor and ceiling information. I would suggest turning ILD off on these accounts. Month-end Optimization When the Month-End Optimization option is used, the Projected Month-End balance, Average Balance and Income/Expense balance are calculated by the system based on the user-defined parameters. This option is only available for detail accounts. Month-end optimization allows PROFITstar to balance the forecast based upon the accounts with this setting active. This setting is currently activated on the following accounts: Account # Account Title 101 Fed Funds Sold 385 Fed Funds Purch These two accounts are balancing accounts. If these accounts are actively used then any projected balances associated with these accounts need to be shifted to other accounts on the balance sheet to properly allocate the balancing position. As a best practice, it is recommended to set up a generic asset balancing account and liability balancing account and use these for balancing purposes. 56

57 Maturity/Repricing Timing These options allow the user complete control over what day of the month accounts mature and reprice as well as what day new purchases are made. The Maturity Day is used to determine when an instrument matures, and affects both an account s current balance and new volume. The default value is Mid-month. No interest is accrued on the day the instrument matures. Changing from the default value may make your averages not balance. The Repricing Day is used to determine when a variable-rate account will reprice. The default value is Mid-month. The Repricing Day setting affects an account s current balance, as well as its new volume. Changing this setting will impact the income or expense generated for an account in the months when repricing occurs. Interest accrues at the new rate on the day the account reprices. The New Purchases Day setting controls when new volumes are added during the month. The default value is Mid-month. Interest begins accruing on the Purchase Day. Loans, investments and deposit accounts are set to mature, reprice and purchase new volume in the middle of the month. This follows our recommendations as most maturities and purchases do not typically occur on a specific day of the month. However, if there are products that reprice on a specific day of the month, this setting can be changed to reflect the appropriate repricing day. Several FHLB borrowing accounts are set to mature on the first day of each month. 57

58 Review of General Model Setup Error Tolerance Levels These values determine your threshold for showing balance or rate errors in the model. These values are used in history and maturity/repricing to determine if you are out of balance. The following error tolerance levels are currently set: I would suggest reducing the rate check warning. Given the current low rate environment, even a small variance can cause significant change in earnings. Identify out of balance for average balances option When this option is checked, the system will display a # sign next to the date if the Calculated Average Assets and Calculated Average Liabilities & Capital, in that time frame, are not in balance. The balance sheet averages are important because they are used in the calculation of ratios and portfolio rates. This option is not activated in the model. Average balances are used for several ratio and rate calculations in the model. Being out of balance could lead to inaccurate ratios or rates. Therefore, I recommend this feature gets turned on in the model setup. 58