To Warren Buffet and his business partner Charlie Munger, an amazing company is often worth a high share price. Unfortunately for ordinary investors, in the banking industry, figuring out when it's worth ponying up for an expensive bank can be really, really hard.

Don't worry, though -- we're here to help.

Some big-picture thoughts on bank valuation
Before we dive in, let's take a minute to talk generally about valuations of bank stocks. Most analysts turn to the price-to-book-value ratio as the gold standard of bank valuation. A bank with a P/B less than one is considered pretty cheap, and a bank with a P/B above two is generally considered expensive.

What drives that valuation, though? Warren Buffett looks to a bank's return on assets. In an interview on CNBC back in March 2013, he explained:

Yes, well a bank that ... earns 1.3% or 1.4% on assets is going to end up selling above tangible book value. If it's earning six-tenths of a percent, or five-tenths of a percent on assets, it's not going to sell below. Book value is not key to evaluating banks. Earnings are key to evaluating banks, and you earn on assets.

Now, it translates to book value, because it -- to some extent, because you're required to hold a certain amount of tangible equity, compared to the assets you have. But you've got banks like Wells Fargo and USB, that earn very high returns on assets, and they sell at a good price to tangible book.

If we accept that ROA is a driver of a bank's valuation, we still need to find a way to predict how a bank's ROA will change in the future. An investment is, after all, a bet that a company's stock will go up in the future; today's price is simply our entry point.

Looking for a correlation
According to data from the FDIC's Quarterly Banking Profile, the industry averaged a return on assets of 1.02% in the third quarter of this year. Using the same FDIC data, I searched for a correlation between ROA and other, more predictive bank metrics. There are many factors that contribute to a bank's ROA, of course. Our objective is to find the one or two metrics that drive performance in an outsized way.

At the end of the analysis, loan loss provisions stood out as the biggest influence on ROA. This makes sense, as the provision for loan losses is the most variable -- and sometimes largest -- of a bank's expenses.

The loan loss provision correlates at nearly 85%. The correlation is quite obvious when graphed:

Banks lose money when loans go bad. When loans go bad, the loan loss provision spikes as banks put aside more money to cover those losses. Therefore a bank that is putting aside a lot of money into the loan loss provision is highly likely to be struggling in general. That means a lower ROA.

This answer seems blindingly obvious: The best banks for the long term -- the banks that deserve to trade at a premium -- are the banks that make good loans.

The $1 million question: How do you know which banks are making good loans?
Investors do not have the luxury of visiting banks and reviewing loan files. We simply cannot know whether a bank's current underwriting and credit culture are strong or weak. Our only hope is to look at a bank's historical performance during downturns. 

Fortunately for us, history repeats itself, and it's a reasonable bet that banks that have historically weathered past recessions well will also be the banks that outperform in future downturns.

It's at this point in the analysis where fellow Fool and bank-stock guru John Maxfield would chime in about the efficiency ratio -- and he'd be 100% correct. The theory goes something like this:

  1. More efficient banks can turn a profit on less revenue because expenses are held in check.
  2. Thanks to that efficiency, these banks do not need to chase higher-yielding assets to overcome excess expenses and meet profitability goals.
  3. Those higher-yielding assets are invariably riskier than other lower-yielding assets -- that's just the nature of asset pricing.
  4. When the econcomic cycle turns from boom to bust, the efficient banks avoid the losses associated with those riskier assets because they didn't need them in the first place.
  5. Therefore, the best banks are able to maintain those strong efficiency ratios during downturns just as well as in boom times because their loan loss provisions do not spike when those risky loans go bad. Inefficient banks will have the opposite result, suffering outsized spikes in their efficiency ratios and loan losses while those losses are absorbed.
In short, the best banks will have the most consistent efficiency ratios. The ratio won't spike every few years as the credit cycle changes over time.

We're rounding third base, heading for home plate
Let's quickly summarize our findings so far before moving on. First, we know the banks that deserve a premium valuation are those with better-than-average returns on assets (hat tip to Warren Buffett). We also know that over time, the primary driver of ROA is a bank's loan loss provision. Put another way, the banks that consistently make good loans are the banks that will produce reliable and above-average returns on assets.

And finally, we now hypothesize that looking at a bank's historical efficiency ratio during past downturns is a good proxy for understanding whether that bank makes quality loans. With quality loans, the logic flows full-circle back to ROA and a high-quality bank stock worth a premium.

Boring banks, it turns out, are the best banks
Warren Buffett has been a vocal shareholder and proponent of Wells Fargo (NYSE:WFC) and US Bancorp (NYSE:USB). We'll examine these two banks, assuming they'll be strong performers. I'll also use Bank of America (NYSE:BAC) and regional bank Fifth Third Bancorp (NASDAQ:FITB) to round out this experiment.

Using data from S&P Capital IQ going back to 1996, we can compare the consistency of these banks' efficiency ratios relative to each other and to the industry. We would expect that the banks with the most consistent results -- in the form of fewer standard deviations -- would command a higher premium.

BankStandard Deviations 1996-2013Price-to-Book Ratio
Fifth Third Bancorp 7.28 1.2
Bank of America 10.72 0.8
US Bancorp 3.93 2.1
Wells Fargo 2.53 1.7
Large bank industry average 3.78 1.8

P/B data from Yahoo! Finance. "Large banks" are defined as banks with total assets in excess of $10 billion. Average P/B value for large banks was estimated using a sample of 65 large banks and data from S&P Capital IQ.

The results turned out pretty close to our expectations. Both Wells Fargo and US Bancorp have stable efficiency ratios during this period, while Bank of America and Fifth Third have far more volatile results. B of A and Fifth Third have also seen their stock prices lag both the broader markets and the banks Buffett favors in terms of performance and price-to-book value.

What this means -- and what it doesn't
This is admittedly a tiny sample size. Further work beyond the scope of this article would be needed to fully flesh out an airtight conclusion. However, we're definitely on to something here. These relationships are real, and we have shown that these key metrics drive both strong bank performance and premium valuations.

How do you invest in bank stocks the Warren Buffett way? You find banks with strong returns on assets and a history of maintaining efficient operations through multiple credit cycles. These banks have exhibited that they have the operations, management, and credit culture to win over the long term.

Jay Jenkins has no position in any stocks mentioned. The Motley Fool recommends Bank of America and Wells Fargo. The Motley Fool owns shares of Bank of America, Fifth Third Bancorp, and Wells Fargo. Try any of our Foolish newsletter services free for 30 days. We Fools may not all hold the same opinions, but we all believe that considering a diverse range of insights makes us better investors. The Motley Fool has a disclosure policy.