Applied to the 25 largest US banks in 2023, and breaking down PBV, ROE, Tier 1 capital ratios, deposit growth, interest spreads and dividend yields into above and below median values, with green (red) indicating cheap (expensive).

JP Morgan Chase is hands down a higher quality bank than Citi, on almost every dimension, but Citi is a better investment, right now, than JPM Chase. I have owned JPM Chase for a while, and I see not reason to sell, but Citi is my buy now.

There is a link between high (low) price to book ratios and high (low) returns on equity, but that link has a lot of noise, and it has become even noisier during the 2023 crisis and investors scramble to adjust.

Pricing is about mismatches. If you are looking for a under priced bank, you want that is trading at a cheap price, with none of the reasons for that cheapness (such as high risk, low ROE, sensitive deposits).

One reason for those difference is the divergence in returns on equity across banks. While 50% of US banks had ROE between 9.24% and 13.75 in May 2023, there are outliers in both directions.

For some value investors, the contrarian take is that banks are now cheap. An alternative view is that banking, as a business, has been disrupted, and its capacity to generate excess rates has faded. I am with the latter group.

Looking across US banks in May 2023, while the median price to book ratio for banks has dropped below one, there is variation in price to book ratios across banks, with a few trading at premiums and some trading at significant discount.

Looking at price to book ratios for all US banks, in the aggregate, you can see that banks have never recovered the trust that was lost in the 2008 crisis. Even as bank ROE has risen, price to book ratios have stayed stuck around one.

Applying this model on Citi Bank in May 2023, I get a value of $69/share (stock price = $43) with a story of a stodgy, low growth bank with a ROE that lags its cost of equity forever. If you disagree with my story, make it your own in this spreadsheet:

To price a bank, the pricing metrics that you use have to be equity-based (PE, PBV). Price to book works better at banks than at other firms because of mark-to-market rules and the link between book equity & regulatoy capital.

After 2008, I have little faith in either assumption (sensible people or regulatory oversight). I created a bank-specific version of free cash flow to equity, where reinvestment into regulatory capital takes the place of net cap ex and working capital.

While you have a choice between valuing equity and valuing the entire business with most firms, with banks, you can only value equity. Debt and deposits to a bank are raw material not a source of capital.

With equity valuation, estimating free cash flows to equity, i.e., cash flows left over after reinvestment and debt payments, for a bank is complicated by the difficulties in estimating cap ex, working capital and debt cash flows.

Not surprisingly, analysts valuing banks have fallen back on using dividends as free cash flows to equity, but with a dividend discount model, you are implicitly assuming that banks are run by sensible people & that the regulatory framework works.

Following up on my good banks/bad banks post, I look at banks through investor eyes, where price takes center stage. A good bank at too high a price is a worse investment than a less well-run bank at a bargain basement price.

The question of which ERP is the best one is not a theoretical argument, since a good measure of ERP should provide better predictions of actual returns in future periods. (You are looking for positive correlation, the higher the better, with returns).

Earnings and dividend yields offer shortcuts to estimate these implied premiums, but the former has had extended periods of over & under shooting and the latter has been undercut by the shift to buybacks.

The implied ERP is a reflection of the ebbs and flows of markets over time, and its rises and falls over time underlie the history of US stocks. Booms are associated with lower ERPs and busts with revisions upwards. We started 2023 with the ERP at 5.94%!

The implied ERP moves with risk premiums in other markets, for the most part, rising and falling together. The periods (late 1990s, 2002-2007) where they diverge have historically preceded major market corrections.

I offer an alternative, where I estimate a forward-looking premium, based upon equity market levels today and expected cash flows, and backing out an internal rate of return. That expected return, with the risk free rate netted out, is the implied ERP.

As an aside, the much touted small cap premium, used by some investors as the basis for picking stocks and appraisers to adjusts costs of equity upward for small companies, has been missing in action since 1981. In 2023, it is more fiction than fact.

You can ask investors or market observers what they expect to earn as an equity risk premium, but talk is cheap, and these survey premiums tend to be static and more reflective of the past than good predictors of the future.

The standard approach to estimating ERP is to look at history, estimating what you would have earned on stocks, relative to risk free investments (treasuries). Not only are historical premiums backward-looking, but vary depending on estimation choices.

I wrote my first survey paper on equity risk premiums (ERP) in October 2008, and have updated it every year since. My 16th update for 2023 is now available for download here: It is long & boring. So, the summary in linked tweets...

The equity risk premium matters because its level determines the prices at which equities trade, and every argument about whether markets are over or under priced can be translate into one about whether the current ERP is too high or too low.