Commercial Bank Valuation Questions & Answers

Question 1

How are public comps and precedent transactions different for a bank?

Answer 1

See question #8 in the High-Level section – the mechanics are the same, but the screening criteria (Total Assets or Deposits) and metrics and multiples (ROE, ROA, P / E, P / BV…) are different.

One other difference is that you must be very strict with your selection criteria – for example, you should not include all bulge bracket banks in a set of public comps because Deutsche Bank, Credit Suisse, and UBS are European and because GS and MS are less diversified than JPM, Citi, Wells Fargo, and BoA.

Question 2

You mentioned Return on Equity as an important metric. What’s the difference between Return on Equity, Return on Common Equity, and Return on Tangible Common Equity, and which one should we use?

Answer 2

• Return on Equity = Net Income to All / Total Shareholders’ Equity
• Return on Common Equity = Net Income to Common / Common Shareholders’ Equity
• Return on Tangible Common Equity = Net Income to Common / Tangible Common Equity

None of these is “better” than the others – they’re just measuring different things. Many analysts prefer the latter 2 because Preferred Stock and Noncontrolling Interests aren’t part of a bank’s core business operations.

Question 3

With normal companies, a revenue or EBITDA multiple might be closely linked to the company’s revenue growth, EBITDA growth, or its margins. Which metrics and multiples are closely correlated for banks?

Answer 3

For banks, Return on Equity and P / BV are closely linked and banks with higher ROEs tend to have higher P / BV multiples as well.

If a bank is returning an extra high amount on its Shareholders’ Equity, you’d expect that the market would value it at a premium to that Equity. Higher ROE = Better returns for shareholders = higher stock price and market cap for the bank.

EPS Growth and P / E are sometimes correlated as well, though it is generally a weaker correlation than ROE and P / BV.

Question 4

What’s the normal range for P / E, P / BV, and P / TBV multiples for banks?

Answer 4

The correct answer here is, “It depends on the bank, the region, the business model, the size, and so on.”

For large, US-based commercial banks, P / E multiples in the 5-15x range are common; P / BV multiples are usually around 1x, and P / TBV multiples are closer to 2x depending on the Goodwill and Intangible Assets of the bank.

These are very market-dependent so hedge your answer as much as possible – but at the same time, realize that having a P / E multiple of 100x or a P / BV multiple of 50x would be extremely weird no matter what bank or region you’re looking at.

Also note that by definition, P / TBV must be greater than or equal to P / BV because Tangible Book Value is always less than or equal to Book Value.

Question 5

Do we care more about Book Value-based multiples or Earnings-based multiples when analyzing banks?

Answer 5

Book Value-based multiples are more reliable because of the one-time charges that show up in EPS; also, ROE and P / BV are highly correlated whereas P / E doesn’t correlate as strongly with EPS growth.

Question 6

What’s the flaw with both Earnings multiples and Book Value-based multiples for commercial banks?

Answer 6

The flaw with both of these multiples is that management has a lot of discretion with the Provision for Credit Losses (affects EPS) and the Allowance for Loan Losses (affects Book Value).
For example, they could report an artificially higher or lower Provision for Credit Losses to lower or boost earnings, which would in turn increase or decrease the P / E multiple.

Question 7

Why can we not use a DCF – even a Levered DCF – to value a bank?

Answer 7

A normal Unlevered DCF would never work because it excludes Net Interest Income, which can be 50%+ of a bank’s revenue.

But even a Levered DCF would not work well because Changes in Working Capital can be massive for a bank, and CapEx is tiny and does not represent re-investment in its business.

For a normal company, CapEx represents reinvestment in its business, but for a bank “reinvestment in business” means hiring more people – so you would need to find training and hiring expenses and then capitalize and amortize them based on the “useful lives” of employees.

Good luck finding that information in bank filings.

Question 8

Walk me through a Dividend Discount Model.

Answer 8

In a Dividend Discount Model, you start by making assumptions for ROA or ROE, the target Tier 1 or Tier 1 Common Ratio, and the Risk-Weighted Asset growth each year.

Then, you project the bank’s Net Income based on its Shareholders’ Equity and the ROE assumption, or its Total Assets and the ROA number; you project RWA based on your initial set of assumptions.
You then check to see what the Tier 1 or Tier 1 Common would be WITH the Net Income from the period you’re looking at added in.

Next, you issue Dividends such that Tier 1 Capital + Net Income – Dividends is equal to the minimum Tier 1 or Tier 1 Common required (e.g. if Tier 1 or Tier 1 Common is currently $100, Net Income is $10, and you need at least $105 of Tier 1 or Tier 1 Common in this period, you could issue $5 worth of Dividends).

Then, you discount all these Dividends based on Cost of Equity and add them up, calculate and discount the Terminal Value based on P / E or P / BV, and add it to the present value of the Dividends.

Question 9

How do you calculate the discount rate differently in a DDM compared to a normal DCF?

Answer 9

First, you use Cost of Equity rather than WACC because you’re calculating Equity Value rather than Enterprise Value.

Also, in the Cost of Equity calculation you do not un-lever and re-lever Beta because similar banks have similar capital structures and because banks cannot exist on an “un-levered” basis.

Question 10

How is a Dividend Discount Model for a normal company different from a DDM for a bank?

Answer 10

For a normal company, you don’t need to start with ROE or ROA and RWA and work backwards to calculate Dividends based on the required regulatory capital.

Instead, you can simply project Revenue down to Net Income as you normally would in a DCF, assume a simple payout ratio, discount and add up the Dividends, and then calculate Terminal Value based on P / E.

Question 11

Should we use Return on Assets or Return on Equity to drive a DDM?

Answer 11

Either one works, but ROE is more common. Warren Buffett has argued that ROA is the better measure of value for banks because its Total Assets – not its Shareholders’ Equity – drive Net Income, but many analysts prefer ROE because it’s closely linked to P / BV multiples.

Question 12

What are the flaws with using a DDM to value a bank?

Answer 12

Just like a normal DCF, it’s hyper-sensitive to assumptions and how you calculate the Terminal Value; often you don’t have enough information to make accurate predictions for Dividends issued in future years.

A DDM may not work well if the bank does not issue Dividends; also, other returns of capital such as Stock Repurchases or Stock-Based Compensation are not captured by the DDM, which is a problem for many banks that have extensive stock repurchase or share/option compensation programs in place.

Question 13

What makes the biggest difference in a DDM: the Payout Ratio, the Discount Rate, the Net Income growth rate, or the Terminal Value?

Answer 13

Just like a normal DCF, the Terminal Value typically makes the biggest difference because it usually represents over 50% of the total value. After that, the Discount Rate, followed by the other 2 criteria has the biggest impact.

As always, hedge your answer by saying, “It depends on the specific bank, but usually…”

Question 14

How do you calculate the Terminal Value in a DDM?

Answer 14

The same way you calculate the Terminal Value in a normal DCF: the multiples method or the Gordon Growth method.

The only difference is that you use P / E, P / BV, or P / TBV for the multiples method, and for the Gordon Growth method you use Final Year Dividends * (1 + Terminal Net Income Growth) / (Cost of Equity – Terminal Net Income Growth).

Question 15

Could you use a Dividend Discount Model to value a bank that does not pay Dividends?

Answer 15

Yes, but you have to assume that it starts paying Dividends at some point in the future or else the value would be $0.

Question 16

How is a Residual Income Model different from a Dividend Discount Model?

Answer 16

The setup is very similar and you still “work backwards” to calculate Dividends based on the target Tier 1 or Tier 1 Common Ratio.

The difference is that instead of summing the present value of the Dividends, you sum the present value of the Residual Income (also known as Excess Returns) instead.

Residual Income is simply ROE * Shareholders’ Equity – Cost of Equity * Shareholders’ Equity – basically, how much actual Net Income exceeds your Net Income expectation.

Then, you add the present value of these Excess Returns to the current Book Value of the bank and that’s the Equity Value.

The intuition: “Since this is a bank, let’s assume that its current Book Value is its Equity Value. But if it generates higher returns than we expect in the future, let’s discount those returns and add them to the Equity Value as well – because a bank that generates higher-than-expected returns should be worth more than its Book Value.”

Question 17

What are the advantages and disadvantages of a Residual Income Model compared to a DDM?

Answer 17

The advantage is that the Residual Income Model is grounded in the bank’s current Balance Sheet rather than assumptions 5-10 years into the future; the disadvantage is that often it doesn’t tell you much beyond the obvious – that Book Value and Equity Value are close for banks.

Question 18

Explain why you often do not see a Terminal Value calculation in the Residual Income Model.

Answer 18

Remember the formula for Residual Income: ROE * Shareholders’ Equity – Cost of Equity * Shareholders’ Equity.

Often, you assume that ROE = Cost of Equity in the long-term in Residual Income Models, so there is no Terminal Value.

If you did not make that assumption, you would calculate Terminal Value with Residual Income in Year After Final Year / (Cost of Equity – Terminal Net Income Growth), discount it by the Cost of Equity, and add it to the present value of the Residual Income each year and the current Book Value of the bank.

Question 19

Which Return metric – Equity, Common Equity, Assets, Tangible Common Equity, and so on – should you use to drive a Residual Income Model?

Answer 19

You should use Return on Common Equity because that corresponds to the Common Equity Value you’re calculating.

Return on Equity includes Preferred Stock, which you don’t want, and Return on Tangible Common Equity will give you numbers that are much different from Common Equity because it excludes Intangibles.

Question 20

Can you use a formula to link ROE and P / BV?

Answer 20

The most common formula is: P / BV = (ROE – Net Income Growth) / (Cost of Equity – Net Income Growth).

That assumes that the Payout Ratio for Dividends and the Net Income Growth are both constant numbers; if that’s not true, you would need to separate the formula into multiple stages.