Liquidity Forecasting Christian Pichlmeier, MUFG 1 Disclaimer The - - PowerPoint PPT Presentation

Liquidity Forecasting

Christian Pichlmeier, MUFG

1

Disclaimer

The views expressed in this presentation are mine and not necessarily the views of MUFG Union Bank.

2

Traditional Forecasting process

3

Forecasting Inputs/Outputs

4

Funding portfolio: Fed Funds, CP/CD, FHLB, BRCD, Repo

Cost of Funds % Targeted Allocation (as determined by funding strategy) Target Funding Gap (Assets, Deposits)

Base Scenario: § Assumes Business as

• usual. Continue to issue

unsecured funding. § Hold a certain % of FHLB capacity open. Adverse Scenario: § Access to unsecured funding sources may be curtailed especially assuming rating downgrade impact. § BRCD issuance increasing. Credit Rating assumption

Balance Funding Gap and Funding Strategy

Funding Strategy: available funding sources and associated costs, concentration risk, TLAC Funding Gap: projected loan growth, required buffer assets, RWA

5

Limitations

6

Reliance on line of business forecasts Reliance on management developed funding strategies Reliance on Financial Reporting to calculate the funding gap User errors through manual entries Changes in business strategy, financial markets and business environment, regulatory standards or tax laws

Build-out of Cash Flow Forecasting

7

Components of a CF Forecast

8

Contractual Maturities Intercompany Transactions New Business Funding Renewals Customer Options Other potential events that may impact liquidity Reasonable assumptions about future behavior of assets, liabilities and off- balance sheet exposures Contractual Cash Flow New Business Behavioral Assumptions 1 2 3 Include sufficient detail to reflect the capital structure, risk profile, complexity, currency exposure, activities of the company Legal Entity / Business Line / Currency

Considerations to develop CF Forecasting

• How reliable is the business forecasting process?
• How granular is the forecast? Does the business provide a breakdown

by the required EPS categories (new business, behavioral assumptions, contractual)?

• How do we bridge between O/N and the first business line input

(commonly 1 month)?

• How timely is the business forecast provided and how often is it

updated?

9

CF Forecast Framework Development

10

Reliability of Forecast Granularity

• f the

Forecast Tenor Buckets Frequency

• f Updates

CF Forecast Starting Point

11

Reliability of Forecast Granularity

• f the

Forecast Tenor Buckets Frequency

• f Updates

Business tends to overestimate Loan and Deposit Growth Business Breakout and Treasury view don’t align EPS requires short term and long term CF projections up to 1 year Short Term projections to be updated daily, LT projections monthly

Case Study: Deposit Forecast - Granularity

Line of Business Segmentation (Industry Type) Product Type Wholesale Banking Commercial Banking Checking Accounts Saving Accounts Trust Checking Accounts Saving Accounts Financial Institutions Checking Accounts Saving Accounts Government Deposits Checking Accounts Consumer Banking Retail Deposits Checking Accounts Saving Accounts Promotional Deposits Saving Accounts CDs

12

Case Study: Deposit Forecast – Historical Analysis (Trend)

13

Threshold Threshold 2015 2016 2017 Total 2015 2016 2017 Total 0.02 0.01 0.29 0.68 >.6 0.24 0.00 0.71 0.78 >.6

Daily Ending Balance Regression line Monthly Ending Balance Regression line

Trend Analysis

The trend analysis visually demonstrates if there is evidence that ending balances follow along a trend line. R2 is calculated and compared with a threshold (in this case, 0.6). If a trend is identified by surpassing the threshold, the results are used as part of the forecasting process for long-term cash flow predictions. Monthly ending balances over a 3 year horizon

Case Study: Deposit Forecast – Historical Analysis (Seasonality)

14

Threshold Mean f% a% Mean f% a% Mean f% a% Mean f% a% Mean f% a% Mean f% a% (0.05) $

57% 50%

1.18 $

43% 54%

(0.13) $

60% 51%

(0.41) $

60% 52%

(0.04) $

58% 50%

0.12 $

41% 50%

%>60

Day of the Week (Represented best in seasonality) Mean, Frequency(%) of total, Amount(%) of total

Seasonality Analysis Total Monday Tuesday Wednesday Thursday Friday

Threshold Mean f% a% Mean f% a% Mean f% a% Mean f% a% Mean f% a% Mean f% a% Mean f% a% 14.38 $

98% 100.0%

15.53 $

91% 97%

(5.58) $

79% 77%

(12.69) $

100% 100%

(5.69) $

83% 90%

(6.73) $

100% 100%

0.12 $

41% 50%

%>60 Total

Week of the month (Represented best in date of the month) Mean, Frequency(%) of total, Amount(%) of total

Week 1 Week 2 Week 3 Week 4 Week 5 Week 6*

The seasonality analysis attempts to find out whether a pattern exists either in terms of days of the week (1) or week of the month (2) by analyzing frequency and amounts of cash flows over a 3 year horizon. f% = frequency of daily cash flows in the same direction as the mean e.g. for Week 2 data (contains for a total of 3 years every 2nd week of the month based on 5 business days totaling approx. 180 data points), 91% of these cash flows move in the same direction a% = amount of daily cash flows in the same direction as the mean e.g. following example of Week 2 above, of the 91% of those cash flows moving in the same direction equal to 97% of all amounts, meaning those cash flows moving in the opposite direction equal only 3% of the absolute cash flow movement.

Week of the month Day of the Week

1 2 1 2

*The frequency of observations is shown for each bin which is determined by the standard deviation of the cash flow movements.

*

We are looking for bell curves which are either to the left or to the right of zero which would indicate seasonality. If observations are spread around the 0 we might not be able to identify seasonality.

*

Case Study: Deposit Forecast – Final Concept

15

§ The results of the trend/seasonality analysis is translated into cash flows leveraging the equation of the trend analysis and overlaying it with potential seasonal patterns. § The forecasted balances based on the quantitative model (dark blue line) are compared to the business forecast (purple line) and provide a check in regards to the validity of the more subjective business forecast. § In the shown example below, the trend analysis contributes only a very small portion to the forecasted cash flow while the seasonality overlay impacts the overall cash flow the most.