Thursday, October 22, 2009

Bank Loss Reserves: Not an easy challenge

With most of the big banks having reported 3Q earnings I wanted to take a look at how we're progressing with loss reserves. Unfortunately, I did not find good news.

Here's what I did. I took a look at Wells Fargo, J.P. Morgan and Bank of America. All three reported a decent breakdown of their loan exposures by type. (Click on each name above for their earnings presentation). I then took the loan loss estimates used by the Fed in the stress test and multiplied each bank's exposures by the loss estimates. Note that the Fed basically had four loss estimates. They had a "Baseline" and a "Adverse" scenario, then they had a high and a low estimate within each of those.

Then I compared that loss estimate with the combined loan loss reserve and current write-downs the bank has taken. This analysis is far from perfect. Each bank reports things a little differently, so I had to make some judgements. Plus who really knows whether the Fed's SCAP estimates for loan losses are right. Still, the Fed's guess is as good as any, and the exercise should be illustrative of how close we are to the end of loan loss provisioning.

Alright, so on the left is estimated losses in all four scenarios, the bottom two blue bars being the "Baseline" and the top two being the "Adverse." Then on the right is losses realized/provisioned to date.

First, Wells Fargo.

Ugh. Almost enough to cover the range of "Baseline" losses, but we should keep in mind, the baseline assumed unemployment would average 8.8% in 2010. Its currently 9.8%. The adverse assumed 2010 unemployment of 10.3%. Given that unemployment is almost assured to go higher before it falls, I'd be shocked if the number weren't closer to 10.3% than 8.8%. Not that loan losses couldn't possibly outperform the unemployment rate if you will, but I'd consider the red area more likely than the blue area.

Moving on to Bank of America.

About the same, although the mix of loans makes the distribution of blue and red a little different. Regardless, a long way to go here.

Now J.P. Morgan.

A little better. Jamie Dimon has more than the Baseline SCAP loss estimate covered. Still, you'd think there are more losses coming here.

So comparing the three, I ran the losses already accounted as a percentage of the Baseline Low and Adverse High (in other words, the lowest and highest loss estimates in the SCAP) for each bank. Below is that chart.

Ugh again. At least with J.P. Morgan I can squint my eyes real hard and suppose that they might be getting close, especially given J.P.'s relatively small commercial loan book. But looking at Wells Fargo and Bank of America, these are still banks that are going to struggle to keep up with losses as far as I can see.

Now consider this. Bank of America 2014 bonds are +210, Wells Fargo +150, J.P. Morgan +140. Shouldn't Wells and BofA be a lot closer? Shouldn't there be a bigger gap between Wells and JPM?


In Debt We Trust said...

I know you don't like to talk about equities but there's something I believe you should consider. The Vix has fallen dramatically w/in the past few weeks and there has been a massive number of put buyers flooding in.

B/c of the bonds-equity inverse relationship, this seems to imply a further magnitude of risk taking out there by investors . . . heedless of reasoned analysis like your article presents.

Rich said...

Wells has a HUGE loan writedown that came from marking down the WB portfolio to absurdly low levels when it was acquired. This doesn't show in the loan loss reserves, but was instead a part of the acquisition accounting.

Also the net interest margin at Wells is higher than any of the other banks, so they can earn their way out of the problem.

k1 said...

Great analysis, it's good to see some facts (or reputable estimates) on this matter. Rich mentions an issue I was just thinking about, that the analysis might be missing the time factor. Put another way, you might want to project the rate of accrual of loss reserves.

As I understand it, one benefit to the banks of the steep yield curve and relief from mark-to-market is that it allows the banks to earn their way out from under the huge overhang of potential losses.

So the rate at which the various TBTF banks are reserving current earnings against future losses should tell us how long it will take a given bank to fully write off their worst-case scenario, assuming earnings remain constant.

We can discuss whether this projected timeframe represents the bank's estimate for how long it will take those losses to appear, or the bank's reality for how long it will take them to accrue enough reserves.

Anonymous said...

I know this idea is completely ridiculous, but how about the American banks plow all the profits back into their capital accounts, use that money to write down assets, rinse and repeat for a few years, then we can, perhaps, have a real banking system again?

Accrued Interest said...


Fair point. If I'm reading the numbers correctly, the loan amounts are net of writedowns already taken. Which makes the analysis less than perfect. Ideally you'd want gross loans on the left and the merger writedown included on the right.

But your point really just makes BAC look all the worse vs. JPM and WFC, because the later two made large mergers for a big chunk of their ugly loans.

hooligan said...

good article, i would like to expand the argument away from the need to do complex analysis and draw a different parallel. Let's face it the ludicrous situation where the government/fed/regulators of banks with worse balance sheets than the majority of those they lend money too are able to charge 5% borrowing costs whilst borrowing at zero from the Fed. We have all been focussing on the escalation in u/e 6 to 16%, the default rate average across all loans escalating to 10% whilst missing what crime is being perpetrated on the 75 % of people who have (not of 80/90/ or 120% FHA sponsored mortgages) LTV's of 60% 50% 40% ... Let me put it another way. Banks have 10% Tier 1 capital (leave aside how they got it or whether its real or whether tangible common equity is a better measure), this means they have an equivalent loan to value ratio of 90%. These banks receive money at zero from the Fed, they take this money with this crap balance sheet and lend it individuals and companies with 50% loan to value at 5% (whether through mortgages or loans). I am not talking about the fails and defaults that make the headlines. I am talking about the majority of company and individual balance sheets that are sound and have only 50% loan to value of assets...even after market sell-offs. Here is the rub, who on earth says that our Government should sponsor a structure where those who are highly leveraged (banks) should receive a 5% cheaper funding rate than those who are lowly leveraged (companies and individuals). This is sickening and even worse, the longer it persists, the smaller will become the pool of companies and individuals with lower loan to leverage ratios. It has always failed, it will always fail and why on earth are we still supporting a failed (bank based) model? There should be an instution that can lend money at lower than bank borrowing rates because the security is better..why don't we have a system that charges interest according to loan to value, so that those with high ratios (banks) pay more than those with low ratios (majority of companies and individuals).

GS751 said...

Really interesting way of looking at things.

Raphael said...

I was browsing through Barclays investor presentation and dug out a few numbers. I thought they were interesting because they paint a picture quite similar to what I've heard about many other of the large banks. From page 4, you can see that most activities have posted gigantic...