The Regulatory Focus on Interest Rate Risk: What to Expect and How to Comply Conference Call will begin at 10:00am CT, lines open at 10:50am CT Audio: 855-749-4750 Access Code: 920 722 897 # You can also listen to the conference call audio using your computer speakers. To maximize the webinar viewing area, please select View Full Screen from the menu or press Alt-Enter on your keyboard. Presented by: Jeffrey F. Caughron, Managing Director & COO Ryan W. Hayhurst, Managing Director jcaughron@gobaker.com ryan@gobaker.com
Rate Environment: Fed Funds, 2yr T Note, and 10yr T Note 10yr Yield Driven Primarily by 1. Less Treasury Issuance 2. Strong International Demand 3. Low Inflation / Deflation Threat 2yr Yield Driven Primarily by Expectations of Fed Tightening 2
3
Effective Governance Processes Directors need a basic understanding of IRR commensurate with the institution s activities o o Policy Approval including risk limits Document participation in board and ALCO minutes Management should maintain an effective IRR measurement system which provides meaningful data to inform the directorate of exposure levels o Policies should require regular IRR measurement and meaningful risk limits. Role of ALCO is to coordinate balance sheet strategies, manage liquidity, and monitor IRR exposures. o o ALCO needs representation from major operational functions (e.g. lending, deposit gathering, investing.) One function is to help develop and review the key assumptions used in analyzing exposures Independent Review of Processes o o o Does IRR management process function according to policy guidelines? Does IRR measurement system reasonably estimate exposures based on reasonable assumptions? Overall assessment of risk management procedures (model validation, back testing, etc) Risk Mitigation Strategies: o o o Natural Hedge (adjust cash flows and/or relative durations of assets and liabilities) Complex Hedging (off balance sheet derivatives need thorough understanding of risks) Contingency Funding Plan (liquidity risk management) 4
Effective Governance Processes: Establishing Policy Limits Limits should not be so low as to frequently require exception approval or refinement, and they should not be set so high as to allow for an unacceptable level of IRR. Thresholds and limits should reflect current market conditions and balance sheet posture. For example, what was an achievable margin ten years ago may not be realistic in the current environment. Limits should reflect the unique characteristics of the bank, it s business model, market area, etc. 5
OCC Range of Practices Memo: Risk Limits 6
OCC Range of Practices Memo: Earnings at Risk 7
IRRM Clients Net Interest Change: Unchanged Rate Scenario & +300bps Scenario The average IRRM client can expect slight margin compression in the next 12mos if market rates remain unchanged If rates rise (parallel) 300bps in the next 12mos, the Net Interest Changes is a positive 2.02%. 8
OCC Range of Practices Memo: Economic Value of Equity 9
IRRM Clients Economic Value of Equity (EVE): +300bps Rate Scenario For a +300bps rate shock, the average IRRM client can expect a 8.35% depreciation in the (theoretical/hypothetical) fair value of equity. One standard deviation from the mean gives us a range of 27.49% to 10.79%. 10
Developing Key Assumptions Well supported, institution specific assumptions are necessary for good IRR analysis Management should give particular attention to non maturity deposit price sensitivity (or betas) and decay rates, the reasonableness of asset prepayment assumptions, and key driver rates customer behavior may not reflect past behavior when market rates change in the future. o o Sensitivity Testing of Key Assumptions (e.g. NMD Stress Test) Deposit Migration Simulation The IRRM model uses a Maturity Distribution Schedule instead of decay rates to determine NMD average lives. In IRRM, the driver rate can be any rate, but is most often assumed to be Fed Funds. 11
Common Weaknesses Found During Review of Assumptions Use of peer averages without consideration of bank specific factors Lack of differentiation between rising and falling rate scenarios Over simplification of balance sheet categories leading to potentially faulty analysis Lack of qualitative adjustment factors to historic data (e.g. not considering a higher run off factor for surge deposits) Inadequate sensitivity testing to evaluate how results would change in response to changes in critical assumptions Source: FDIC Supervisory Insights from Winter 2014 12
NMD as a % of Total Funding Has Surged With Low Rates 13
Questions NMD Balances: How long lived and how stable? Will our deposits stick around for a long time? How to determine historical average life: o Open Close Study Based on historical behavior, what s the average life of our core deposits? How to determine balance stability & potential surge balances: o Surge Balance Analysis Based on historical behavior, how sticky are our depositors and how much of a surge in NMD did we experience over the past five years flow versus our peers? Regulator Concern: The future may be different 14
Questions about NMD Pricing: How reactive is our pricing and how quickly do we react? How aggressively will we need to price our deposits in reaction to a Fed tightening? How to determine shift sensitivities (pricing betas): o Historical Price Analysis: Sensitivities Based on observed historical behavior, what s our normal pricing reaction to changes in market rates? How to determine time lags: o Historical Pricing Analysis: Time Lags Based on observed historical behavior, how quickly do we react to changes in market rates? Regulator Concern: The future may be different 15
OCC Range of Practices Memo: Repricing Betas 16
Estimating NMD Betas Using Call Report Data 17
Estimating NMD Betas Using Monthly Institution Supplied Data A Non Maturity Deposit (NMD) sensitivity (beta) analysis will provide empirical evidence to support your assumptions. Focus on distinct periods of rising (2003 07) and falling (2000 03 & 2007 13) rates. Results should not be used rigidly since the next period of rising rates may not look like the last. Instead, use actual historical data as a starting point for determining appropriate sensitivities. 18
Example Open Close Study to Estimate NMD Average Lives Tracks account open date and closed date to determine an actual average life for accounts (MMDA, NOW, Savings, etc.) Query core processor for open accounts and any recently closed accounts Results will often demonstrate very long average lives (8 12yrs) 19
Open Close Study A Non Maturity Deposit (NMD) Open Close Study can be used to provide empirical evidence to support your assumptions. For accounts that were closed in a given time period, obtain the date opened and average balance (YTD or LTD). This data can be used to calculate the true average lives of closed accounts. Results should not be used rigidly since factors such as demographics may cause deposits to behave differently in the future. 20
Example NMD Analysis to Estimate Decay MMDA Deposit Balance April 13 April 12 April 11 April 10 April 09 April 08 April 07 April 06 Accounts Open on: April 06-32% 34,330,089 37,883,043 39,941,999 38,102,344 37,429,422 37,473,074 42,247,033 51,034,496 April 07 6,173,696 10,059,312 8,553,936 8,362,733 10,114,226 15,758,102 18,427,352 April 08 117,990,973 251,474,388 217,561,115 223,751,787 224,880,925 247,100,676 April 09 303,373,221 503,775,635 455,500,795 492,001,738 540,982,260 April 10-50% 533,256,592 972,829,857 926,990,838 1,078,107,545 April 11 720,101,243 1,392,390,905 1,371,170,024 April 12 685,804,487 1,645,297,130 April 13 120,717,123 MMDA Number of Accounts April 13 April 12 April 11 April 10 April 09 April 08 April 07 April 06 Accounts Open on: April 06-78% 507 560 613 673 833 1,553 1,888 2,285 April 07 70 79 86 98 125 181 184 April 08 1,152 1,721 1,833 2,100 2,574 2,616 April 09 2,981 4,077 4,524 5,159 6,567 April 10-53% 5,536 8,299 9,520 11,700 April 11 6,883 10,930 13,200 April 12 6,855 12,771 April 13 1,250 21
NMD Analysis: Potential Surge Balances 22
Cost effective versus local markets or other wholesale funding alternatives. Flexible maturities: 1 mo to 30yrs. Can match funds according to your asset/liability needs. A single certificate per maturity. No collateral is required. No early withdrawal feature. Only upon death or adjudication of incompetence declared by court of law. Excellent resource for Contingency Funding Plans Through Baker Group arrangement with Financial Northeastern Companies 23
Estimating Loan Prepayments Prepayment estimates should be differentiated by product For most accounts, some level of prepayments is better than none Generally speaking, most prepayments should decline as rates rise and increase as rates decline Mortgage prepayments could use consensus estimates (Bloomberg Median or YieldBook) Some core processors may be able to calculate historical prepayments, but many cannot 24
Sensitivity Testing: Non Maturity Deposits Whatever Baseline Assumptions You Use, Stress Test Them Institutions should incorporate stressed assumptions for non maturity deposits in IRR models FDIC Three Ways to Stress NMD Assumptions (Sensitivity Tests) 1. Ratchet up pricing betas (shift sensitivities) and reduce time lags in order to mimic an aggressively competitive environment for NMD 2. Reduce Average Life (and Duration) Assumptions in order to assess the EVE impact of lower duration liabilities 3. Simulate a migration of NMD balances into more rate sensitive funding (time deposits or wholesale funding) considered to be the most realistic depiction of what may happen in the next rate cycle Demographics Access to information 25
Case Study NMD Migration: Balance Sheet Before vs After Before After The simulation shows a 20% decline in Non Maturing Deposits, replaced by Fed Funds (Borrowed Funds Category) 26
Case Study NMD Migration: Earnings Simulation Before 27
Case Study NMD Migration: Earnings Simulation After As migration takes place, interest expense rises more rapidly and earnings feel the impact 28
NMD Migration Simulation Results: Eight Sample Banks from TX, KS, ND, CO, OK, KY NP = Non-Parallel Rate Scenario (+400/+100bps) NMD Migration Stress Test: 20% Bank Total Assets NMD/TDM 12mo NIC NP 12mo ROA NP EVE NP 1 70mm 37.2% 7.72 0.06 7.62 2 850mm 61.0% 10.45 0.75 8.12 3 80mm 55.0% 4.47 1.02 20.48 4 120mm 51.2% 4.51 1.52 0.66 5 200mm 57.2% 1.18 0.71 6.57 6 325mm 44.3% 7.96 1.22 19.28 7 100mm 44.7% 4.16 1.29 19.85 8 550mm 42.5% 5.75 0.91 2.39 Average 287mm 49.1% 4.65 0.92 3.60 29
Developing an In House Independent Review Eleven step guide to developing an in-house independent ALCO review Banks are expected to monitor the effectiveness of their key internal controls either as a part of the internal audit process or by means of an appropriate independent review, and managing IRR is no exception. 30
The Iron Triangle of Regulatory Compliance Back-Test: Can be done internally or supplied by vendor Independent Review of Processes: Must assess the adequacy of all ALCO processes Board Education? A Robust Model? Stress Testing? Assumptions Reviews? Policy Limits? ALCO Minutes? Model Validation: Can be supplied by vendor the independent review should involve assessing the institution s measurement system of IRR, including the reasonableness of assumptions, the process used in determining assumptions, and the back testing of assumptions and results. FFIEC 31
32
Model Validation Validation: o Third Party o Periodically Updated o Documentation on file Institutions that use vendorsupplied models are not required to test the mechanics and mathematics of the measurement model. However the vendor should provide documentation showing a credible independent third party has performed such a function. FFIEC 33
What to Expect During an IRR Exam Expectations contained primarily in two documents: o 1996 Joint Agency Policy Statement on Interest Rate Risk o 2010 Interagency Advisory Effective IRR requires o Informed Directors o Capable Management o Appropriate Internal Resources Bankers should be prepared to discuss the results of their IRR measurement system and be able to describe key assumptions and assumptions development It isn t enough to simply produce reports management must demonstrate understanding Well documented board and ALCO minutes will help examiners understand the bank s risk management practices, etc. 34
Common Examination Findings Insufficient evidence of board and senior management discussion of IRR Policy limits unrealistic or uninformed Use of default (not bank specific) model assumptions Inadequate documentation or support of assumptions Lack of (assumption) sensitivity testing Lack of 300bp or 400bp interest rate shock scenarios Lack of nonparallel yield curve shift scenarios Lack of backtesting or backtesting over an insufficient period of time (3 mo. Vs. 12 mo.) Need for regular independent reviews Independent reviews do not cover all required areas Lack of independent review expertise Outdated or missing Model Validation Source: FDIC 35
Interest Rate Risk & ALCO Checklist A pre exam checklist for management review Director Education Ensure directors have a basic understanding of IRR and the bank s ALCO processes Provide directors with access to educational resources on interest rate risk Board minutes should reflect director participation in IRR discussions Regular (Quarterly) ALCO Meetings to Review & Discuss Reports ALCO minutes should reflect a demonstration of sound processes that quantify risk to earnings & capital Regular (Quarterly) Standard & Non Standard Stress Tests 100bp, +100bp, +200bp, +300bp, +400bp, Non Parallel (e.g. +400bp/+100bp Bear Flattener) Ramped Rate Shifts & Immediate Rate Shocks 12 & 24 Month Horizons Earnings at Risk & Economic Value of Equity Annual Validation Obtain most recent Validation Letter for Model (validates the math of the model) Back Test your reports over a 12 month period (validates the results) Independent Review of ALCO Process (validates the process) Annual Assumptions Review Use Back Test to determine if assumptions are generally reasonable Periodically perform analysis to ensure assumptions reflect institution s profile and activities (e.g. Loan Prepayments, NMD Sensitivities, Open Close Study, Decay Analysis, Surge Balances, etc.) Annual Sensitivity Testing (aka, assumptions stress test e.g. increase NMD betas, shorten NMD average lives by 50%, run a migration simulation from NMD to CD s, etc.) Annual Review of Investment & ALCO Policies 36
Resources from The Baker Group LP www.gobaker.com Resources from FDIC www.fdic.gov 37