Core Deposit Modeling 2011

Core Deposit Modeling:
Hot Topics and Current Best Practices
May 3 2011, 10:30 AM
presented by Bank Risk Advisors: Fred Poorman Jr., CFA, Managing Principal
questions or comments? email: fpoorman@bankriskadvisors.com
More information is available at www.bankriskadvisors.com
1

overview per conference brochure
Join this fast paced session to explore key issues including:
 rate sensitivities in a rising rate environment;
 valuation of core deposits;
 sensitivity analysis - regulators and best practices;
 liquidity concerns;
 core deposit studies and behavioral inputs;
 core deposits best theoretical approach
 conclusion & summary
 Please ask questions!
 We will note AL model examples as applicable.
2

what is a core deposit?
 Accounting: FASB definition, GAAP Fair Value disclosures
 “the definition will include deposits that do not have a contractual
maturity that management believes are a stable source of funds.”
 Quantitative approach to core/stable
 Look at balances/flows over time
 Regulatory comment at Midwest regional April 2011; “Surely you don’t
believe that all your recent deposit growth is true stable, core”
 Product specific behaviors
 Qualitative approach to core/stable deposits
 Bank Risk Advisors (Matz) has an 11 factor-approach to defining
stability of deposits for Liquidity modeling, including stress-testing
 Core Deposits are usually divided into core/noncore
 This binary approach is a good first step, however the reality is likely
more complex and is best viewed on a continuum that is product-
and/or market environment determined.
4

different core deposit perspectives
 Accounting: FASB definition, GAAP Fair Value disclosures
 “the definition will include deposits that do not have a contractual
maturity that management believes are a stable source of funds.”
 ALCO: balance sheet composition, pricing, key assumption
 ALM modeling: multi-scenario liquidity, income & value
 Equity markets: equity analysts, price multiples
 Liquidity: forecasting & stress testing
 Marketing: market share, pricing & branding/positioning
 Profitability: FTP & Liquidity FTP
 Please see Leonard Matz presentation next session
 Regulatory: impact on CAMELS
5

a core deposit modeling term clarified:
rate sensitivity or “beta”
 sensitivity is the relationship between one (independent)
variable and another (dependent) variable. for example, the
relationship between 3 month LIBOR and a bank money
market rate. if the independent 3ML increases 1% and the
bank MM increases 0.50%, the sensitivity is 50%.
 this is also called the rate beta. In linear regression analysis β (beta)
represents a standardized partial slope coefficient from the following
equation: 𝛾 = 𝛼 + 𝛽 ×
 For example MM rate (𝛾) = 𝛼 + 𝛽 × + ∈, where
 Y is the estimated MM rate
 a is a constant, intercept, or floor
 B is the slope or partial sensitivity (0.5 in above)
 X is 3ML
 e is the error term, which is usually ignored except for stats folks
 Important for modeling income sensitivity!
6

linear regression “beta” example: y = 3.0 + 0.5x;
50% sensitivity (Anscombe’s quartet from Wikipedia)
7

another core deposit modeling term clarified:
deposit decay
 Decay and retention
 Decay is also referred to as runoff.
 complement of retention, use degradation models
 Retention is how much you retain or keep.
 complement of decay, use survival models
 Example: 10% of Account balances go away.
 10% decay.
 90% retention.
 Generally speaking, these are modeled using “survivability”
and/or “degradation models”
 there are well-known approaches to statistical modeling of this data
from non-financial sectors. these tend to be more robust than the
typical decay modeling approaches of many banks and consultants.
 Important for modeling value sensitivity!
8

retention (survival) analysis
example from JMP software
9

yet another core deposit modeling term clarified:
correlation
 Correlation (r)
 Formally, a correlation coefficient is defined between the two random
variables (x and y, here). Let sx and xy denote the standard deviations
of x and y. Let sxy denote the covariance of x and y. The correlation
coefficient between x and y, denoted sometimes rxy, is defined by:
rxy = sxy / sxsy source:economics.com
 The tendency of two random variables to move together:
 1 is perfectly correlated
– Could be same data
– Fed Funds and 1 month LIBOR in the long run is close.
 0 is uncorrelated
 -1 is perfectly negatively correlated
 Correlation does not imply causation.
 "Empirically observed covariation is a necessary but not sufficient
condition for causality.“ Tufte
10

same example, correlation = .816, R2 = .67
(Anscombe’s quartet from Wikipedia)
11

rate sensitivities in a rising rate environment,
the way we were!
12

rate sensitivities in a rising rate environment,
where are we going? forward?
13

rate sensitivities (CDs) in a rising rate
environment
14
Preliminary Deposit Sensitivities
Floor Sensitivity
Fed move Product Index Start End Alpha Beta R squared
1.75 1 yr CD 1 yr Swap May-99 Dec-00 - 0.90 0.67
4.25 1 yr CD 1 yr Swap May-04 Jun-06 0.10 0.85 0.99
- 1 yr CD 1 yr Swap Jun-09 Mar-11 0.65 1.06 0.28
4.25 1 yr Jumbo CD 1 yr Swap May-04 Jun-06 0.11 0.85 0.98
Weekly national average deposit rates from bankrate.com
Weekly market rates from Bloomberg L.P.
Statistical measures from SAS Stats, JMP, Bloomberg L.P.
See forthcoming White Paper/BALM Article for disclaimer, methodology, results

rate sensitivities (MMDAs) in a rising rate
environment
15
Floor Sensitivity
1.75 MM Jumbo 3 M LIBOR May-99 Dec-00
4.25 MM Jumbo 3 M LIBOR May-04 Jun-06 0.47 0.79 0.99
1.75 MM 10k tier 3 M LIBOR May-99 Dec-00 - 0.65 0.49
4.25 MM 10k tier 3 M LIBOR May-04 Jun-06 0.80 0.51 0.97
missing data points

rate sensitivities (NOW/Int. Checking) in a rising
rate environment
16
Floor Sensitivity
1.75 NOW/Int Check 3 M LIBOR May-99 Dec-00 - 0.35 0.71
4.25 NOW/Int Check 3 M LIBOR May-04 Jun-06 - 0.23 0.55
4.25 NOW/Int Check 3 M LIBOR May-04 Jun-06 0.33 0.15 0.55
4.25 NOW/Int Check Multi-factor May-04 Jun-06 0.79

rate sensitivity issues
 Does past apply?
 Market & channel changes
 FDIC rate caps for “problem” institutions
 Regulatory fiat of no growth for “problem” institutions
 Is this time different vs. mean reversion
 Quantitative & qualitative approaches
 Do local, regional, or national markets apply?
 Impact of deposit concentration limits for 2-4 large banks
 Internet and brokered CDs
 Do you have early withdrawal penalties for CDs?
 Are they valuation-, or make whole-, based?
 Are they enforced?
 Option (step up, rate change) CDs considerations?
 ALLY Bank doesn’t for advertised products and we (U.S. taxpayers)
own it!
17

valuation of core deposits per GAAP
 In little noticed actions in September and November 2009,
FASB issued the following:
1. The value of the core deposit liability would be determined using
a present value of the average core deposit amount discounted by
the difference between the alternative funds rate and the all-in-
cost-to-service rate over the implied maturity.
2. The core deposit liability amount that would be subject to the
remeasurement would be determined as an average amount over the
implied maturity time period, which would result in the consideration of
future deposits. Considering and valuing future deposits would result in
an intangible asset being reflected in the valuation.
3. The Board agreed that core deposits would qualify for remeasurement
changes to be recognized in other comprehensive income. The
balance sheet presentation of core deposits would be subject to
previous presentation decisions.
 Last point means you can offset AFS Investment value volatility.
Rarely implemented.
18

valuation of core deposits per GAAP
 present value of the average core deposit amount
 This is what the industry does.
 For example, using the Fiserv AL model
 discounted by the difference between the alternative funds
rate and the all-in-cost-to-service rate
 Difference means Funding rate – (Servicing cost – Fees)
 Brokered CD rate = 2.00%
 Servicing cost = 1.50%
 Fees = 0.50%
 Difference = 2.00% - (1.50% - 0.50%) =1%
 Is this standard practice? Please play the raise your hand game 
 over the implied maturity
 over can be interpreted to mean across the period, implying periodic
cash flows
19

over the implied maturity?
 Implied:
 “involved, indicated, or suggested without being directly or explicitly
stated” source: dictionary.com
 “In financial mathematics, the implied volatility of an option contract is
the volatility implied by the market price of the option based on an
option pricing model. In other words, it is the volatility that, when used in
a particular pricing model, yields a theoretical value for the option equal
to the current market price of that option.” source: wikipedia.com
 An analogue would be that given the market price, discount rate, you
could solve for:
 a single bullet maturity
 multiple maturities, given periodic cash flows
 Maturity: “a financial term indicating the final date for
payment of principal and interest” source: wikipedia.com
20

what is “over the implied maturity?”
 Most practitioners use an estimated:
 periodic cash flow (sometimes called decay rate)
 final maturity (some only use decay, OTS tables)
 Behavioral studies are one approach we suggest (more
later)
 Regulatory alert: buyer beware!
 One ALM brokerage/consultant uses the complement of estimated rate
sensitivity?
 .1 sensitivity = 10 years, .5 sensitivity = 2 years, etc.
 Another ALM brokerage/consultant uses very wide ranges
 NOW accounts range from 1.2 to 15.8 to 39.9 years?
 Uses complement of standard deviation of average life?
 From established consultants, cash flow is typically less and
average lives are more than implied by market prices
21

valuation notes
 SNL definition includes retail CDs
 need to adjust for mix and pricing; may be problematic
 Both assisted and unassisted transactions include situations
when total assets, total deposits, and core deposits were
acquired at a discount
 Deposit rates are above market
 Mix is unfavorable with little true core
 Institutions are sold for less than book value
 Failed, loss share institutions
24

sensitivity analysis: regulatory perspective
 Interagency Advisory on Interest Rate Risk Management
(OCC 2010-1a):
 assess the sensitivity of the institution to changes in market rates and
important assumptions underlying the metrics used
 stress testing, which includes both scenario and sensitivity analysis, is
an integral component of IRR management
 stress testing should include a sensitivity analysis to help determine
which assumptions have the most influence on model output.
 sensitivity analysis can be used to determine the conditions under
which key business assumptions and model parameters break down
 institutions should ensure the reasonableness of asset prepayments,
non-maturity deposit price sensitivity and decay rates
 sensitivity testing of key assumptions that exert the greatest impact on
measurement results. When actual experience differs significantly from
past assumptions and expectations, institutions should use a range of
assumptions to appropriately reflect this uncertainty.
25

sensitivity analysis:
industry and better practices
 Interagency Advisory on Interest Rate Risk Management
(OCC 2010-1a) notes many better practices
 Example: large bank presents sensitivity analysis and
earnings impact of key assumptions monthly
 Recommendation for community banks:
 ALCO should review and approve key assumptions at least annually
 When significant change occurs, for example:
– Core Deposit study
– New pricing strategy
 Sensitivity analysis assists with an intuitive understanding of the affect
of key assumptions
 MMDA example
 Sensitivities of 25%, 50%, 85%
 CD example
 Sensitivities of 75%, 85%, 100%
26

liquidity concerns
 normal “business as usual” scenarios
 Base case rate cash flow assumptions by product
 May have scenario-specific flows based on difference from market rate
 Analogous to prepays
 liquidity scenario analysis & stress testing
 Define base congruent with IRR analyses
 Core Deposit flows vary by:
 scenario type (credit, fraud, systemic risk, etc.)
 stress level (mild, moderate, severe, extreme)
 Liquidity stress tests demonstrate another value of core deposits
 Liquidity Funds Transfer Pricing (Liquidity FTP)
 Required of large banks per Interagency Policy Statement on Funding
and Liquidity Risk Management (OCC 2010-27a)
 Better practice for all banks
 Remember to consider in pricing and valuation
27

core deposit studies and behavioral inputs:
regulatory and accounting compliance (1)
 Data requirements
 Request account level detail for 10 years
 Realistically need for an entire rate cycle to observe rate & balance
behaviors
 Data aggregation
 Product mapping & account type consistency over time frame
 Consider ALM model set up
 Draw curves for balance behaviors (we use JMP/SAS)
 Remember Anscombe’s quartet, this is a great reason to draw curves
 Life distribution (Excel example next page)
 Fit life by econometric factor x,y,z
 Degradation (decay)
 Survival (retention)
 Parametric
 Non-parametric
28

core deposit studies and behavioral inputs :
29

30

 Rates paid by account type
 Realistically need for an entire rate cycle to observe rate & balance
behaviors
 Product mapping & type consistency over time frame
 Consider ALM model set up
 Analyze rates vs. market rates & economic factors
 Single factor LIBOR or Fed Funds great starting point
 80/20 principle at work here
 Easy to understand, explain, and model
 Initial and lag effects (Arnold & Hawkins 1999, Poorman & Hawkins
2001)
 Step-wise multi-factor model
 May have greater explanatory power (R2)
 May be difficult to understand, explain, and model
 Initial and lag effects
31

core deposit studies and behavioral inputs:
better practices
 Integrate rate and balance behaviors at product level
 Need entire rate cycle & complete data set
 Need well-defined behavioral equations (“Valuing Core Deposits”,
Sheehan, 2004)
 Validation issues (out of sample & back testing)
 Use customer data to understand depositor behaviors
 Issue: Customer A has 5 products, how does the customer manage his
portfolio?
 Customer moves money between products; looks like savings goes
down but actually moving money to MMDA or Promo CDs.
 Data and analytical complexity.
 May need to integrate with CRM system & customer demographics
 balance migration between accounts
 balance changes within accounts
 new & closed accounts
 Drives pricing decisions, need management/organizational buy in
32

is there a “best” theoretical approach to core
deposits?
 Jarrow model is most often cited:
 The arbitrage-free valuation and hedging of demand deposits and credit
card loans”, Robert A. Jarrow, Donald R. van Deventer, 1999
 Using a market-segmentation argument, core deposits are equated with
an exotic interest-rate swap and valued via an arbitrage-free valuation
methodology. In this model, balances “change randomly based on both
the level and average of past market rates” (rates=Treasury rates).
 They note valuation is firm-specific
 Valuation of deposit is:
33

is there a “best” theoretical approach to core
deposits?
 Short answer is “No”.
 Longer answer is that the usually cited, theoretically correct
academic articles with closed form solutions may result in
nonsensical durations and values
 However, there have recently (last five years) been Core Deposit
articles that make sense from both a theoretical and practical
perspective. There are primarily from Europe.
 My favorite title is: “Optimal Deposit Pricing: There is no ‘One-Size-
Fits-All’’ Valuation Approach”, Blochlinger and Zurcher, 2010
 “The resulting Nash equilibria agree with the empirical finance
literature on deposit pricing: size matters.”
 “We show that there is no ‘one-size-fits-all’ approach, that is, the
valuation for deposit accounts must be bank-specific.”
 Three different bank strategies are noted for pricing & valuation.
Pricing, sensitivities, and valuation may also be categorized based
on strategy delineation.
34

conclusion & summary
 consider deposit rate sensitivities (betas) from prior cycles
 is this time different? why and how will this affect pricing?
 will provide paper on website in near future
 FASB has spoken on core deposit valuation
 definition: management believes are a stable source of funds
 prepare to justify your perspective to regulators and others
 market valuation
 note valuation trends, need to reverse engineer for benchmarking
 behavioral studies by qualified consultants provide:
 inputs for modeling and valuation, insights for pricing
 are metrics reasonable & statistically appropriate and valid
 academic models and papers on Core Deposits
 back in vogue, at least in Europe
 no one size fits all approach
35

About Bank Risk Advisors
 About Bank Risk Advisors
 Provides market, liquidity, and credit risk measurement and validation
solutions to banks and credit unions
 Founded in 2010 from the integration of four niche bank consulting
firms. Predecessor firms date back to 1994.
 People: Consultants average 20+ years experience. Backgrounds
include academia, audit, banking, brokerage, consulting, regulatory.
 Core Deposit Consulting Services team
 Leonard Matz, Principal, Liquidity Risk Consulting Services
 Mike Arnold, PhD, Sr. Consultant, Principal ALCO Partners
 Fred Poorman Jr., CFA, Managing Principal
 Howard Stern, PhD, Sr. Consultant, Principal Stern Consulting
 Extras on new website and client SharePoint site will include:
 ALM modeling benchmarks, updated quarterly
 White papers, published articles, presentations, links to books, stuff
36

Core Deposit Modeling 2011

More Related Content

What's hot (20)

Viewers also liked (16)

Similar to Core Deposit Modeling 2011 (20)

More from Bank Risk Advisors (6)

Core Deposit Modeling 2011