Implementing CECL A Practical Approach. October 23, 2018

Transcription

1 Implementing A Practical Approach October 23, 2018

2 Implementing A Practical Approach 1 Today s presenters: Gary D Condit Assurance Director GBQ Partners LLC

3 Objectives 2 The primary objectives of today s presentation: Provide Illustrative examples of how to apply in a practical manner Provide suggestions on what data to assemble over the next several years to improve options for computations

4 What do I need to know when the regulators and auditors come knocking

5 The Methodology 4 There is no single method for measuring credit losses The FASB pronouncement is very clear in stating that it is does not specify a single method for measuring expected credit losses The NCUA also was a party to a joint letter with the Board of Governors from the FRB, the FDIC, and Office of the Comptroller of the Currency The joint letter validates and also clearly indicates that it recognizes that there is no single method under the standard Institutions should use judgement to develop estimation methods that are well documented, applied consistently over time, and faithfully estimate the collectability of the financial assets by applying the principles of the new standard The joint statement further states: The new accounting standard allows expected credit loss estimation approaches that build on the existing credit risk management systems and processes, as well as existing methods for estimating credit losses Institutions will need to consider how to adjust historical loss experience not only for current conditions, as is required under the existing incurred loss methodology, but also for reasonable and supportable forecasts that affect the expected collectability of financial assets.

6 OVERVIEW FOR BUILDING A VINTAGE POOL ALLOWANCE 5 In consideration of the requirements of the new standard and the initial views of the joint statement by the regulatory bodies we suggest that credit unions evaluate the historical and other data available. Since most credit unions have historically kept loss data and have access to loans disbursed by period we are going to go through a sample methodology today that consists of what I have divided into a 3 step approach to computing a loan allowance using vintage pools 1. Assemble the data 2. Use the data to forecast life of loan loss by vintage pool 3. Build the required allowance for each vintage pool

7 STEP 1 ASSEMBLE THE DATA 6 1. Determine how far back you have year end AIRES files, Loan trial balances or other reports that have the loan origination date, original loan amount, current balance amount and loan type. 2. For periods for which you have the data above obtain the loan charge off amounts(preferably net charge-offs) for each year and subtotal the charge offs for each year by the year the loan was originated 3. Create a table for each vintage year reflecting total loans, charge off and ending balance for each year

8 STEP 1 BUILD THE DATA VINTAGE POOL LIFE COMPLETE LOAN HISTORY Payments Net Charge-off Balance 2005 New 5,000, ,000 4,500, ,000 8,000 3,700, ,192,500 7,500 2,500, ,000 10,000 1,500, ,200,000 7, , ,000 3,000 0 Total 35,500 LOSSES OVER LIFE OF LOAN $35,500 (.71% OF ORIGINAL $5,000,000 LOANS MADE IN 2005)

9 STEP 1 BUILD THE DATA VINTAGE POOL LIFE COMPLETE LOAN HISTORY Payments Net Charge-off Balance 2006 New 10,000, ,500, ,500, ,789,500 10,500 6,700, ,191,000 9,000 4,500, ,985,000 15,000 2,500, ,183,000 17,000 1,300, ,000 10, , ,500 1, , ,000 1,000 0 Total 64,000 LOSSES OVER LIFE OF LOAN $64,000 (.64% OF ORIGINAL $10,000,000 LOANS MADE IN 2006)

10 STEP 1 CONTINUATION 9 Vintage New loans 5,000,000 10,000,000 11,000,000 9,000,000 8,000,000 Net C/O 35,500 64, ,000 70,000 80,000 Loss %.71%.64%.90%.77% 1.00%

11 STEP 1 BUILD THE DATA CONTINUE TO VINTAGE POOL LIFE PARTIAL LOAN HISTORY Payments Net Balance Charge-off 2011 New 12,000, ,500, ,000 9,000, ,000 7,000, ,000 4,500, ,000 3,000, ,000 2,000, ,000 1,000,000 Total 101,000 Est Add. 1,000 Although Pool is not complete it is pretty mature so I estimate that we will charge off another $1,000

12 STEP 1 BUILD THE DATA CONTINUE TO VINTAGE POOL LIFE PARTIAL LOAN HISTORY Payments Net Balance Charge-off 2012 New 10,000, ,000, ,000, ,000 15,000 8,000, ,982,000 18,000 6,000, ,983,000 17,000 4,000, ,000 17,000 3,000, ,000 16,000 2,000,000 Total 83,000 Est Add. 3,000 Although Pool is not complete it is 80% mature so I estimate that we will charge off another $3,000

13 STEP 1 BUILD THE DATA CONTINUE TO VINTAGE POOL LIFE PARTIAL LOAN HISTORY Payments Net Balance Charge-off 2013 New 9,000, , ,800, ,285,000 15,000 7,500, ,000 15,000 6,500, ,000 12,000 5,200, ,000 10,000 4,500,000 Total 52,000 Est Add. This pool is only 50% mature so I am going to exclude it from my historical base period and forecast the total pool loss using historical data

14 STEP 1 BUILD THE DATA CONTINUE TO VINTAGE POOL LIFE PARTIAL LOAN HISTORY Payments Net Balance Charge-off 2014 New 11,000, ,285, ,500, ,000 7,000 9,500, ,000 22,000 7,200, ,000 10,000 6,500,000 Total 39,000 Est Add. This pool not mature so I am going to exclude it from my historical base

15 STEP 1 BUILD THE DATA CONTINUE TO VINTAGE POOL LIFE PARTIAL LOAN HISTORY Payments Net Balance Charge-off 2015 New 14,000, , ,500, ,678,000 22,000 10,800, ,290,000 21,000 8,500,000 Total 43,000 Est Add. This pool not mature so I am going to exclude it from my historical base

16 STEP 1 BUILD THE DATA CONTINUE TO VINTAGE POOL LIFE PARTIAL LOAN HISTORY Payments Net Balance Charge-off 2016 New 16,000, ,2000, ,800, ,782,000 18,000 13,000,000 Total 18,000 Est Add. This pool not mature so I am going to exclude it from my historical base

17 STEP 1 BUILD THE DATA CONTINUE TO VINTAGE POOL LIFE PARTIAL LOAN HISTORY Payments Net Balance Charge-off 2017 New 18,000, ,000, ,000,000 Total 0 Est Add. This pool not mature so I am going to exclude it from my historical base

18 STEP 2 ESTIMATE LIFE OF POOL LOSSES 17 Vintage New loans 5,000,000 10,000,000 11,000,000 9,000,000 8,000,000 12,000,000 Net C/O 35,500 64, ,000 70,000 80, ,000 Loss %.71%.64%.90%.77% 1.00%.85% CUMM AVG..71%.68%.75%.76%.80%.81% Vintage New loans 10,000,000 9,000,000 11,000,000 14,000,000 16,000,000 18,000,000 Net C/O 86,000 73,900 90, , , ,800 Loss %.86%.82%.82%.82%.82%.82% CUMM AVG.82%.82%.82%.82%.82%.82%

19 Applying Historical Loss % to new loans 18 Vintage of Pool Pool Total Remaining Balance ActualLosses Forecasted New Loans Dec-17 Pool Loss Loss % From Inception Remaining ,000,000 - Historical 35, % 35, ,000,000 - Historical 64, % 64, ,000,000 - Historical 100, % 100, ,000,000 - Historical 70, % 70, ,000,000 - Historical 80, % 80, ,000,000 1,000,000 Historical 102, % 101,000 1, ,000,000 2,000,000 Historical 86, % 83,000 3, % avg ,000,000 4,500,000 Forecast 73, % 52,000 21, ,000,000 6,500,000 Forecast 90, % 39,000 51, ,000,000 8,500,000 Forecast 114, % 43,000 71, ,000,000 13,000,000 Forecast 131, % 18, , ,000,000 17,000,000 Forecast 147, % - 147,800 Computed Loan Balance 52,500, ,300.00

20 STEP 3 SUMMARIZE THE ALLOWANCE FOR LOAN LOSS 19 ALLOWANCE FOR LOAN LOSS VINTAGE SUMMARY DECEMBER 31, 2017 Vintage of Pool TOTAL Net 2012 Net 2013 Net 2014 Net 2015 Net 2016 Net 2017 LOSS CO CO CO CO CO CO ,000 (12,000) 90,000 (15,000) 75,000 (20,000) 55,000 (30,000) 25,000 (18,000) 7,000 (6,000) 1, ,000 86,000 (15,000) 71,000 (18,000) 53,000 (17,000) 36,000 (17,000) 19,000 (16,000) 3, ,900-73,900 (15,000) 58,900 (15,000) 43,900 (12,000) 31,900 (10,000) 21, ,300-90,300 (7,000) 83,300 (22,000) 61,300 (10,000) 51, , ,900 (22,000) 92,900 (21,000) 71, , ,400 (18,000) 113, , , ,300

21 Using other pooling criteria 20 In addition to just using total loan the computations can be broken in to categories such as: 1 Loan Types 2 Credit Scores 3 Loan officer 4 Indirect Dealer In addition external factors can be added to the historical data to determine if there is correlation between the credit union losses and such data.

22 STEP 2 ESTIMATE LIFE OF POOL LOSSES 21 Vintage New loans 5,000,000 10,000,000 11,000,000 9,000,000 8,000,000 12,000,000 Net C/O 35,500 64, ,000 70,000 80, ,000 Loss %.71%.64%.90%.77% 1.00%.85% CUMM AVG..71%.68%.75%.76%.80%.81% Avg Unemploymnt 5.5% 5.7% 8.3% 11.0% 9.4% 8.0% Avg Delinquent.5%.4%.8%.9%.7%.6% Vintage New loans 10,000,000 9,000,000 11,000,000 14,000,000 16,000,000 18,000,000 Net C/O 86,000 73,900 90, , , ,800 Loss %.86%.82%.82%.82%.82%.82% CUMM AVG.82%.82%.82%.82%.82%.82% Avg Unemploymnt 7.4% 6.8% 5.1% 4.9% 5.0% 4.6% Avg Delinquent.6%.5%.5%.4%.4%.3%

23 DATA BUILDING SUGGESTIONS 22 Retain year end ARIES file for every year For Loan charge off reports add the year of origination to the report so the charge off reports have charge off date, loan type, charge off amount and year of loan origination For Recoveries maintain date, loan type, recovery amount and year of loan origination If you go through a conversion be sure to archive AIRES of other loan detail from old system Begin doing computations now so you can update them each year between now and 2021 implementation date

24 Thank You!

25 PREDICTIVE REGRESSION ANALYSIS - SAMPLE 24 The objective of this predictive regression analysis is to first determine an estimate of the life of loan loss of a pool of new loans put on the books of the credit union. For purposes of this sample model the first step was to obtain the client's annual data from their call reports for a period from 1999 through Using the data GBQ built a summary of the annual activity of new loans, charge offs, recoveries and other loan data including delinquencies. The data was then used to determine average loan lives based on new loan, charge off and payment activity computed from the call report data. After the summary was prepared the next step was to determine if there were factors that correlated with the trailing losses incurred over the lives of the respective annual pool of loans. In order to determine this GBQ accumulated the following data to run against the total life of loan losses identified above. The data used was as follows: 1. Actual Delinquencies by year of origin 2. Ohio Unemployment Rates by year of origin 3. National Unemployment Rates by year of origin 4. National Real estate Delinquency rates This analysis was done for the entire period from 1999 through Using this data the factors above were run against the estimated life of loss estimates computed.

26 25 The factors above were run against the actual trailing losses to identify factors that correlated the best. These would be the factors with the lowest P-values in the regression. Based on this regression model all of the factors correlate well with the actual loss data. See highlighted P-values on regression The next step was to forecast the losses expected using the regression analysis predictive formula applied to 1999 through 2017 new loan volume. This is done on the "LOANS GROUPINGS" tab. See highlighted numbers at the bottom of the forecast Based on this analysis a summary of our predictive allowance for regular loans for the Credit Union is as follows:

27 26 Sample Credit Union Regression Data Downloard from Peer to Peer Call Reports 12/31/ /31/ /31/2011 Unsecured Credit Card $976,496 $967,432 $1,036,415 Unsecured $2,081,096 $2,040,384 $2,315,054 New Auto $2,791,986 $2,130,835 $1,978,122 Used Auto $13,371,478 $13,273,414 $13,189,351 1st Mortgage $17,111,600 $17,047,788 $18,659,835 Other Loans 1,376,045 1,338,375 1,401,273 Total 37,708,701 36,798,228 38,580,050 Control $37,708,701 $36,798,228 $38,580,050 Diff $ Allowance for Loan & Lease Losses $320,580 $247,258 $276,028 Credit Card C/O 4.65% 2.42% 2.82% Loans Granted YTD $20,502,484 $15,567,926 $18,959,743 Charge off % 0.44% 0.78% 0.59% Charge Offs Total $186,851 $311,818 $257,522 Recoveries $41,651 $22,496 $36,292 Net Charge Offs $145,200 $289,322 $221,230 Delinquencies $672,651 $706,740 $625,319 Regular 141, , ,295 Loan activity Summary Beginning Balance 27,768,034 36,732,205 35,830,796 New disbursements 20,502,484 15,567,926 18,959,743 less charge offs (141,444) (288,406) (228,295) Payments (11,396,869) (16,180,929) (17,018,609) Ending Balance 36,732,205 35,830,796 37,543,635 Control 36,732,205 35,830,796 37,543,635 Unemployment Rate 7.50% 11.25% 8.75% Average life Adjusted average life multiple Losses over average life 659, , ,470 Loss % to average loans 2.48% 1.76% 1.42% Loss % to New loans 3.22% 3.65% 2.71% Delinquent Loan % % % % Predicted loss to average loan Regression % % %

28 27

29 28