Contrary to popular belief, I believe both Citigroup (NYSE:C) and JPMorgan (NYSE:JPM) offer a larger margin of safety within the large-cap bank space than US Bancorp (NYSE:USB). * Based on my analysis, USB should generate a low-teens return for long-term investors, a worse return profile than both C and JPM.* While USB's return profile is similar to Wells Fargo (NYSE:WFC), I believe that USB is a riskier investment because my analysis assumes that the firm can generate a sustainable long-term Return on Tier 1 Common Equity of over 20%, which is higher than I have assumed for any other bank. If this assumption proves too aggressive, the expected return on the stock can fall dramatically as the appropriate Price-to-Tier-1-Common (P/T1C) multiple would be much lower than I have estimated. You will find more detail about this in my Sensitivity Analysis section.

US Bancorp is my 4th choice among large-cap US banks. My #1 pick is Citigroup, my #2 pick is JP Morgan, and my #3 pick is Wells Fargo. I've written about each of these banks in my previous articles here:

- Own Citi For A 20% IRR With Limited Downside - Most Compelling Large-Cap Bank
- Own JPMorgan For The Second Best Risk/Reward Profile In The Banking Sector
- Wells Fargo: 3rd Best Large-Cap Bank Stock Offering Mid-Teen Returns And Limited Downside

This is the 4th article in what I intend to make a five-part series analyzing the risk/reward profile of the five largest banks--Bank of America (NYSE:BAC), Citigroup, JP Morgan, Wells Fargo, and US Bancorp. I've applied a consistent methodology to analyze the expected return and the potential downside risks of each of these banks. This particular article will focus on US Bancorp and is broken up into the following sections:

- Results
- Methodology
- Assumptions
- Sensitivity Analysis
- Bottom Line

**Results: Expect mid-to-low teens, but given the high multiple, additional upside is limited. **

In the table below, I show my risk/reward ratio and the Price-to-Tier-1-Common Equity Ratio (P/T1C), stock price, & IRR (Internal Rate of Return) for both my base case and my downside scenario across various holding periods. * So for example, if your holding period is 3 years, I expect the stock to be trading at a 2.6x P/T1C ratio, which implies a 3yr target price of $41 and an IRR of 14% including dividends in my base case.* In my downside scenario, I would expect the stock to trade at a 1.9x P/T1C ratio, which implies a 3yr target price of $30 or about a 3% annualized return.

While not shown here, I've completed a similar analysis for the other four large-cap banks, Bank of America, Citigroup, JP Morgan and Wells Fargo. My analysis shows that Citi and JP Morgan offered a better risk/reward profile. I've provided links to those articles above.

**Methodology: A modified return on capital approach to estimating long-term value.**

For those of you who have read my Citi, JP Morgan and/or Wells Fargo write-ups, much of this methodology section will be repetitive, and you can skip down to the Assumptions section. If you're new to my approach, hopefully you'll find this section useful (and maybe even interesting).

Here's the basic concept; I've built a reasonably simple model, which forecasts Tier 1 Common Equity Per Share (T1C/shr). I arrive at my T1C/shr estimate by projecting a Return on Tier 1 Common Equity (ROT1C), which helps me estimate net income. I then make a few more assumptions around DTA utilization, dividends, buybacks and share count growth to project year-end T1C/shr. My forecasts are shown below:

I then estimate the stock's value using a range of Price-to-Tier-1-Common Equity Ratios (P/T1C). By multiplying my T1C/shr estimate by a P/T1C ratio, I can generate an estimated price for the stock. Since I don't know what the **right** P/T1C ratio is, I've layered on a probability distribution for each P/T1C ratio across different time periods. One general assumption I have made is that over time, the base case P/T1C ratio will migrate towards the appropriate P/T1C based on my long-term ROT1C assumption. This is discussed in more detail in the Assumptions section, but * for US Bancorp, I believe the long-term P/T1C ratio should approach 2.7x (vs. 2.4x today)*. Below, I show my probability distribution assumptions for each period.

To calculate my base-case estimated P/T1C for each period, I multiply my probability assumption by each P/T1C ratio in the range and sum up the results. This creates a weighted average P/T1C for each period, which is my base case assumption. Below, I show a table which walks through this math for the 2014 period. Note that I use my 2013 year-end T1C/shr estimate of $15 to calculate my 2-year holding period target price of $36.

Once I have my estimated base-case price, I can add the cumulative dividends I expect to receive over that period ($1.60) to the price and calculate an IRR. As my calculation essentially implies that all dividends are paid at the end of the period, it understates the true IRR. Below, I show the IRR calculation for my base-case assumption in 2014 (2-year holding period).

*IRR = [(Price _{t+n} + Cumulative Dividends) / Price_{t+0}] ^ (1/n) - 1*

- Price
_{t+n}= $36 - Cumulative Dividends = $1.60
- Price
_{t+0}= $30 - n (holding period) = 2 years

**IRR = 13% in my base case scenario w/ a 2-year holding period**

To calculate my downside scenario, I again use my probability distribution graph. With this graph, I can estimate the bottom 10% P/T1C ratio. For US Bancorp in 2014, my downside scenario estimates a P/T1C ratio of 1.8x. Said another way, 10% of the time, I expect my P/T1C to be equal to or worse than 1.8x in 2014 (after a 2-year holding period). This analysis is somewhat similar to a VaR calculation in that it doesn't estimate the max downside, but the expected loss at a given probability. Below, I attempt to show where that downside scenario falls on my 2014 probability distribution graph for US Bancorp.

Now on to my actual forecast model. To forecast Tier 1 Common Equity growth, my model essentially starts with 2011 year-end Tier 1 Common Equity and then grows it each year as follows:

**Beginning of Period [BOP] T1C**

- Add net income
- Add disallowed Deferred Tax Asset [DTA] utilization
- Subtract dividends
- Subtract buybacks

**End of Period [EOP] T1C**

Additionally, I also forecast the average number of diluted shares in each period. I assume that shares have a natural growth to them of 1.5% from the firm issuing shares to employees, but that this growth is offset by buybacks. I discuss the assumptions embedded in my forecast model in more detail in the next section.

**Assumptions: Shave between 6-11% from Analyst Estimates over the next 3 years.**

My analysis relies on many, many assumptions, and while I've tried to be conservative in making these assumptions, some readers will disagree with my inputs. For this reason, I've made my model available to anyone who wants to do their own analysis and stress the model or look at how less conservative assumptions affect the outcome (model available for download here). Below, I walk through the biggest drivers of my model.

**Net Income**

To forecast Net Income in 2012-2014, I start with the street's GAAP EPS estimates from Thomson Reuters of $2.75, $3.00, and $3.37, respectively. I then layer on additional losses of $0.4b, $0.6b, and $0.4b, respectively. My additional losses help compensate for my view that The Street tends to be overly optimistic in its earnings projections. For 2015-2020, I forecast a Return of Tier 1 Common Equity (ROT1C) for each year to estimate net income. **For US Bancorp, I believe the firm's long-term sustainable ROT1C in 21%, and I use that for each year.**

To arrive at my 21% long-term sustainable ROT1C, I use history as my guide and perform a modified DuPont Analysis. I look at the last 10yrs of data unmodified (i.e. I leave the 2008-2009 financials losses in without making any adjustments). This assumption essentially assumes that a traditional credit cycle tends to include a crash like we saw in 2008, a reasonably conservative assumption, but not insane considering we have Europe hanging over our heads. My modified DuPont Equation is as follows:

*ROT1C = Profit Margin x Asset Turnover / Regulatory Capital*

- Profit Margin = Net Income / Revenue (pre-provisions)
- Asset Turnover = Revenue (Pre-provisions) / RWA
^{1} - Regulatory Capital = Tier 1 Common Equity / RWA
^{1}

^{1} RWA = Risk-weighted Assets as defined under Basel 1

Note: Tier 1 Common is calculated under Basel 1 for all periods

In the table below, I show the historical average, one standard deviation up and down, and median for the full history and the last four quarters for each component of my DuPont analysis. The full data series is available as part of the model and was constructed using both CapIQ & Bloomberg data.

Using this table as my historical guide, I then make estimates for what each of these components can be going forward, taking into account the many, many regulatory changes that are being implemented across the industry. The biggest change I'm forecasting is that regulatory capital will need to increase almost 15% above the 10yr historical average. This is a smaller increase than I expect for BAC, C, and JPM, but reflects my view that under Basel III, USB will only need to hold in excess of 8.5% T1C ratio as a less systematically important bank than its money center peers. Below, I show my estimates and my forecast for long-term sustainable ROT1C of 21% based on these estimates.

**Long-term Appropriate P/T1C Ratio**

Next, using the dividend discount model and making a few adjustments, I can back into the appropriate P/T1C, given an assumed payout ratio, discount rate, and ROT1C. The math works as follows:

*Dividend Discount Model: Price = Dividend _{t+1} / (r_{equity} - g)*

- where g = ROT1C x (1 - PayOut)
- where Dividend
_{t+1}= Dividend_{t+0}x (1 + g) - where Dividend
_{t+0}= T1C x ROT1C x PayOut

P/T1C = ROT1C x PayOut x (1 + g) / (r_{equity} - g)

Making the following assumptions:

- PayOut Ratio = 85% (includes both dividends & buybacks)
- Discount rate = 10% (my standard hurdle rate)
- ROT1C = 21% as calculated above

* I arrive at a long-term sustainable P/T1C ratio of 2.7x plugging in the assumptions above.* In my analysis, I forecast that by 2016, US Bancorp should trade right around that level in my base case (the stock currently trades around 2.4x P/T1C).

**Dividends**

I have assumed that the firm pays $0.75/shr of dividends during 2012 and continues to increase its dividend payout each year targeting about a 30-35% dividend payout ratio. My dividend assumptions and payout ratios are shown below.

**Stock Buybacks**

I have assumed the firm repurchases $1b of shares during 2012 and continues to step up its buyback program in each subsequent year, eventually targeting a total payout ratio, including both dividends and share repurchases, of nearly 85% by 2020. My buyback assumptions and buyback payout assumptions are shown below.

**Disallowed Deferred Tax Asset [DTA] Utilization**

US Bancorp doesn't have any disallowed DTAs, because unlike Citi and Bank of America, which accumulated large NOLs during the financial crisis, US Bancorp was relatively successful in navigating the crisis. So while my model has the ability to deal with disallowed DTAs, it doesn't come into play for US Bancorp.

**Probability Distributions for P/T1C Ratios**

As I discussed above, I've made a series of assumptions around the likelihood of each P/T1C ratio being the correct future ratio. I generally assumed that the weighted average P/T1C would migrate toward the appropriate sustainable long-term P/T1C of 2.7x that I calculated using the dividend discount model above. This, of course, may not happen, but feel free to change my probability distributions as you see fit.

**Sensitivity Analysis --A 21% estimated ROT1C means less of a margin of safety, and thus, more risk to the stock. **

**Lower Long-term Sustainable ROT1C Estimate to 17% (vs 21% Currently)**

In the table below, I show my updated return profile if I cap my sustainable long-term ROT1C at 17% (vs 21% previously). This 17% ROT1C estimate is still higher than I have forecast for any of the other large cap banks. As I show below:

- C - 10.25%
- BAC - 10.5%
- JPM - 12.0%
- WFC - 15.0%

I have updated by P/T1C distribution tables by shifting my probabilities 0.75x to the left. This essentially targets a long-term P/T1C of 2.0x, which is roughly what you'd expect the multiple to be if the ROT1C drops to 17%. My average base-case IRR over all time periods falls to 1% (vs 13% previously).

**3x Larger Additional Losses in 2012-2014**

In the table below, I show how my returns are affected if I triple my additional losses for each of the next three years. In this scenario, I'm forecasting additional losses in 2012-2014 of $1.2b, $1.8b, and $1.2b, respectively (vs $0.4b, $0.6b, and $0.4b previously). Again, my loss estimates are deducted from The Street consensus estimates. My average base-case IRR over all time periods falls to 10% (vs 13% previously).

**Bottom 5% Expected Returns**

In the table below, I update my return analysis looking at the bottom 5% of expected returns (my previous downside scenario was the bottom 10%). This provides some perspective on what kind of returns investors should expect if things really get ugly. The key take-away is that with a long enough time horizon, 4yrs+, even in a very difficult environment you should be able to generate a very modest return and avoid a loss of principal, based on my model.

**Upside--Stock Trades At Top 10% Of My Probability Distribution**

While I'm much more concerned with downside risk than I am with upside potential, it is still helpful to have an understanding of the return possibilities in a very bullish upside scenario. Using the same methodology I used to forecast the potential outcome in the bottom 10% scenario, I can alternatively look at the top 10% scenario. The table below looks at this outlier scenario to the upside. Under this very bullish scenario, the stock could generate IRRs in the low-to-mid-twenties over a 3 year period.

**Bottom Line -- USB's price appears to already reflect the firm's strong growth potential and does not offer a sufficient margin of safety to warrant an investment.**

I continue to prefer the risk/reward profile of both C & JPM over USB. In my view, the stock's valuation does not provide a sufficient margin of safety, particularly if I've over-estimated the firm's long-term sustainable return on capital.