Modeling a bank’s loan portfolio
I’m at a decent FIG focused Pe fund. All my banking experience was in asset management or payment processing.
I have just been asked by the MD to model out the loan portfolio of a target bank (that we are looking to acquire) and stress test the portfolio. I’ve never done This before and Don’t have a junior to consult. Can someone whose invested in banks: 1) please walk me through how can I determine loan growth rates for the different types of banks, and 2)how should I go about stress testing the portfolio it to see how many NPLs will come about / losses / what provision for risk losses should we be underwriting?
Bank operates in retail, credit card, mortgages and commercial segments. I found some approaches that project total loans outstanding in the country as % of its GDP and multiply that by bank’s market share. But that seems stupid and too simplistic in this approach. Any ideas?? Thanks so much in advance
Anyone smart?
Ignore
Would be very helpful if you could walk me through both sections if you had all the information that you desired. Many thanks
Anyone more knowledgable?
??Anyone?
I only have some experience with the consumer side. For at least that portion of the business you will need to do some work on the losses and prepays (hopefully this is done and you have accounted for movements in rates and that impact on prepays).
You should cut up the portfolio using some of the ways below 1. Loan type - mortgage, student, personal, payday etc. 2. Credit tier - think bands of FICO score or ability to repay (income level) 3. Term - 3yr is less risky for default than 5yr but 3yr will have a higher prepay rate
Existing portfolios should be treated as one cohort performing in-line with historical levels. I would model out the go forward portfolio based on monthly vintages. No banks/lenders are originating significant consumer paper right now - based on that, I wouldn't assume any origination growth for the next few years. Reason being if you assume significant origination growth, despite draconian losses/prepay assumptions, it can still drive top line growth.
Provision should be a buffer to the expected loss rate. Example is if you expected a CDR of 4% you likely need to have a ratio of capital against that rate which will depend on the size of the bank etc. There is legislation on this which I am blanking on.
Feel free to DM me if you any other questions.
(edited for the provision piece)
As an aside, if you are at a FIG focused PE fund, look at your firms materials on similar types of deals. I would find it hart to believe they haven't looked at a bank or spec fin company before now. Check the info tab of the excel, maybe the Asso/VP who worked on that deal is still around? I have found that looking through old materials often helps me find answers to questions....maybe I have been lucky
At the start of the the pandemic, we were asked by our MD to do stress tests and here’s what we did. Ideally we would start by benchmarking the net loans (as there are already provisions accounting for previous NPAs).
Now you can assume that new NPAs would require 100% provisioning (again, the depends on the regulations on provisioning; my geography requires 100% provisioning) and you can deduct the equal amount of provisioning from the core capital. You can do it for 50bps of increase in NPAs and calculate it to a point where you reach the minimum required core capital.
Another way it could be done is you assume stress scenarios where you assume that corporate book will undergo 10% increase in NPAs, and you can calculate core capital at various scenarios of 50%,60%...100% level of provisioning and highlight the point where it breaches the regulatory minimum core capital.
A bit new to FIG, but seems you are knowledgable. I am trying to learn more here. Can you please confirm if provisioning reduces the net income hence the tangible common equity number, and an increase in NPA increases the risk weighted assets as well as reduces the net loans (thus the interest income)?
Also - how did you estimate the 10% increase in NPA? I understand you can look how historically NPA assets have trended and apply a growth rate, but since this time it's different, I am curious what underlying calculations / thought process drove your assumption
Apologies for getting back late to you. Let me answer your question point by point.
Yes, provisioning reduces net income as any new provisions created in the quarter flow through income statement. The line item is usually called "Impairment expense" and often lies below operating expenses and above PBT. Yes the tangible equity number is also affected as the income flows into equity
An increase in NPA reduces net loans indirectly as the latter is calculated as Gross Loans (including NPAs) less provisions, and NPAs often have to have 100% provisioning (varies from regulator to regulator, but I have read 100% is pretty standard). On RWAs, I am not exactly sure as I have never dug too deep into its calculation, and have just used it at face value
I stated 10% number as an example, but a good way to estimate can be the following: for a bank focused on corporate lending, you can look at default rates within IG or HY Bond indexes and they can serve as a good proxy (this too has its drawback, but is pretty good proxy in my opinion). For retail portfolio, you can look at the split between secured loans (Housing loans, loans to finance motor vehicles, etc) vs unsecured loans (Credit card loans, personal loans, etc). On secured loans, you can look at repossession numbers (I think housing repossession stats should be available), which can serve as a good proxy, whereas credit card default numbers are pretty common knowledge.
The stress test can be performed to varying degrees of complexities based on the data available, and using various scenarios, but naturally I would advise against it. Ping me if you want further colour.
https://www.federalreserve.gov/publications/files/2020-dfast-results-20…
Pg 23 (PDF pg 31) has a table with aggregated projected loan losses for the banks that participated in the most recent Fed stress test.
Understand you guys probably aren’t buying PNC or US Bank but it makes for a good proxy and allows you to apply different loss rates to different categories of your target company loans. So that could be scenario 1 and account for the “this time is different” aspect. I still think it’s instructive to look at ‘08 crisis losses especially on a relative basis. It gives you an insight into the underwriting culture of the bank (if they were willing to put garbage on their balance sheet bank then do you have a reason to believe that’s changed? Etc.)
As for growth, not sure how long you can wait but SMID cap banks are starting to report earnings. I would just assume industry average (krx index is a decent proxy) 2Q loan growth continues for the next year or 6 quarters. After that GDP loan growth seems reasonable unless this is a small bank that had been growing at a healthy clip.
In the dfast results there’s also discussions on Pre-provision net revenue. Once you model that, as another poster mentioned above I’d just apply 1:1 losses:provisions (although CECL accounting may mess with this, so you might have to look at what ratio a few banks have been reporting) to get you to a pretax income (or loss) number. This will obviously flow into your equity in the burn down analysis so you’ll be able to see where capital levels shake out.
Cumque at voluptate illo nihil. Tempore consequatur rem perferendis sit tempore.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...