After watching its stock tank 50% this year while denying that it needed capital, Bank of America's management has begun to acknowledge reality.
The bank raised $5 billion by selling preferred stock and options to Warren Buffettdiluting common shareholders in the process. And now, as previously promised, it has sold half its stake in China Construction Bank for $8 billion.
These moves are good news for the bank's employees and shareholders, as well as for the U.S. taxpayer, which will be on the hook if Bank of America's management flies the company into a mountain.
But many analysts believe that Bank of America will need to raise a lot more capital before it gets back on sound footing.
These analysts believe that Bank of America is still overstating the value of some of the assets on its balance sheet. When the company is finally forced to recognize the real values of these assets, this theory goes, the bank will once again have to fill a major capital hole.
Last week, we described the general concerns of one analyst, who is focused on a specific portion of Bank of America's humongous balance sheet: The company's portfolio of residential mortgages and home equity loans.
Below, we put some numbers on this possible exposure.  Based on the analysis below, in this one asset category alone, Bank of America could be under-reserved by tens of billions of dollars. And that doesn't include its ongoing litigation exposure.
As we described last week, the analyst we spoke to is concerned that the performance of Bank of America's whole loans (mortgages and home equity loans) will ultimately mirror the performance of a national pool of "securitized" mortgage loans that were made during the housing bubble.
(The analyst is not a "short-seller." But he wants to preserve his anonymity so Bank of America and others won't be mad at him.)
Bank analysts have much more detail on the performance of the industry-wide securitized loan pool than they do on the individual banks' whole loans. And the analyst thinks it is fair to use the performance of the securitized loan portfolio as a proxy for the banks' whole loan portfolios.
The embedded losses in the national securitized loan portfolio are much higher than the losses Bank of America and other big U.S. banks have reported thus far on their whole loan portfolios. And the analyst believes that the banks are using the leeway given them by U.S. accounting rules to make the whole loan portfolios look better than they actually are. (Presenting a rosy view of loans allows the banks to avoid taking write-offs and, thus, avoid having to raise additional capital and further diluting their shareholders.)
Specifically, a recent analysis of the industry-wide portfolio of securitized loans by Amherst Securities breaks them down as follows:
TOTAL SECURITIZED LOANS: 4.6 million loans worth $1.2 trillion
This is made up of:
  • ALWAYS PERFORMING LOANS: 2.2 million loans worth $606 billion, 51% of total principal
  • NON-PERFORMING LOANS (in default): 1.4 million loans worth $370 billion, 31% of total principal
  • "RE-PERFORMING" LOANS (loans that were in default that are now performing, at least temporarily): ~900,000 loans worth $204 billion, 17% of total principal
So, in other words, of all the securitized loans outstanding, 49% (~$600 billion) are troubled, of which 31% (~$370 billion) are in default.
And there are two other points to keep in mind about the securitized loan pool:
  • The "recovery rate" on Non-Performing non-prime loans is only 36% (this is the portion of the original money owed that the lender gets back after foreclosure)
  • A net 2% of the "Re-Performing loans" become "non-performing" again each month (based on the May-June rate).
Assuming the industry-wide whole loan pool looks similar to the securitized poolwhich the analyst I've talked to thinks is a fair assumption, given that they were both originated late in the housing bubble when home prices were highwe should eventually expect to see similar performance on the banks' whole loan portfolios.
Bali House
Does Bank of America's mortgage loan portfolio look like this?...
Image: Elite Houses
How much exposure does Bank of America have to residential real-estate loans? How is Bank of America saying these loans are performing? How much has the bank reserved for possible loan losses? Is it possible there are tens of billions of dollars of "embedded" losses hidden on the balance sheet?
These are the questions Bank of America analysts have to ask, even though Bank of America is not providing enough information to determine definitive answers.
As of June 30, per Bank of America's financial statements, here's what Bank of America's residential whole loan portfolio looked like:
  • Total residential loans: $413 billion (Composed primarily of $265 billion of first and second mortgages and $132 billion of home-equity loansthe latter of which are generally junior to the first mortgages and, thanks to plummeting house prices, may no longer have any actual "home equity" backing them up).
  • Total non-performing loans (per Bank of America): $19 billion, or 5% of the portfolio
  • Total provisions for loan losses: $21 billion, or 5% of the portfolio
In other words, Bank of America has classified 5% of its loan portfolio as "non-performing" and reserved $21 billion to cover the expected losses from these and other loans that go bad.
housecollapse tbi
Or this?
So how does that compare to the performance of the securitized loan portfolio described above?
It looks downright fantastic!
In the securitized portfolio, 31% of the loans are "non-performing," versus only 5% of Bank of America's. And another 17% of the securitized loans are "re-performing," many of which will slip back into non-performing.
*UPDATE: A sharp reader points out that the definition of "non-performing" is different for the securitized loans versus the bank loans, so this is comparing apples and oranges. (My mistake--I'm sorry.). The comparable level of "non-performing" loans on an apples-to-apples basis would be 15% or 20% for Bank of America (depending on whether you focus on first mortgages only--20%--or include home-equity loans). Some other factors do mitigate this and make BOFA's situation worse, so the general conclusion is the same.  See the addendum below for all the arcane details.
What scares the analyst I spoke to is his belief that much of Bank of America's loan portfolio may actually be just as bad if not worse than the securitized portfolio, despite what Bank of America is telling everyone.
In other words, the analyst thinks that 35% or more of Bank of America's loans might end up going into default, versus the 5% 15%-20% the bank says are in default today.
So how much would Bank of America have to take in losses if the analyst is right? (Or, put differently, how much would Bank of America have to increase its loan-loss provisions by to account for the likely performance of these loans?)
This analysis is very complex, even with the data available, and it involves several different types and classes of loans (first mortgages, second mortgages, home-equity loans, accruals, non-accruals, etc.). To keep things relatively simple, I'm going to take a very broad-brush approach. Doing this involves some technical inaccuracies, but it gets us to basically the same place.
Assuming Bank of America's loans mirror those in the securitized portfolio, here's a look at what the numbers might look like:
  • Total residential loans: $413 billion
  • Total non-performing: 31%, or $128 billion (vs. ~$19 billion currently, using the bank definition of "non-performing"--$60 billion or so, if you use the securitization methodology )
  • Total re-performing: 17%, or $70 billion
In other words, if the securitized pool proves a reasonable proxy for Bank of America's loans, about 35% Bank of America's $413 billion of residential real-estate loans, or ~$145 billion, might eventually be in troublethe non-performing percentage, plus a portion of the "re-performing." 
(Of course, this "proxy" concept is a rough analysis. But without having detailed performance data on Bank of America's loans like we have on the securitized loans, it's impossible to get a clear picture of the situation.)
The next question is what sort of "recovery" Bank of America might get from these loansand, therefore, what its loan-loss provisions should be.
As you'll recall, the "recovery rate" for non-performing loans in the securitized poolthe loans that go to foreclosureis a dismal 36%.  Loans that are permanently modified with a principal reduction, meanwhile (according to my analyst's estimate) might be expected to have a new carrying value of about 70% of the original loan amount.
If Bank of America's resolution of its potentially troubled loans were accomplished via foreclosure or principal writedowns, and its recovery rates mirrored those in the securitized loan pool, Bank of America might end up losing about 50% of the value of these loans.
So, what is Bank of America's exposure under this scenario?
A lot more than is currently reserved.
Possibly many tens of billions of dollars more.
It's possible that the analyst I've spoken with is wrong and that Bank of America's whole loans are vastly superior to the loans in the securitized pooland, therefore, that its loan loss provisions are conservative.
It's also possible that the housing market and economy will soon start to recover in earnest and that lots of non-performing and "re-performing" loans will quickly become fully performing again.
But it's also possible that the housing market and economy will continue to deteriorate, in which case the performance of the securitized loan pooland Bank of America's loansmight get even worse.
And it's possible that the analyst's logic is sound and that Bank of America will ultimately have to face the reality that the losses embedded in its whole loan portfolio are VASTLY higher than it has currently admitted and that it needs a lot more capital to offset them.
And this is only Bank of America's residential loans were talking aboutwe haven't even gotten to the commercial real-estate loans, consumer credit loans, European exposure, derivatives, and other exposures on the company's $2.2 trillion balance sheet. Or the potentially enormous liabilities associated with the mortgage-underwriting behavior of Bank of America's subsidiary, Countrywide, for which the company seems to be hit with a new lawsuit every other day.
So it's no wonder that some analysts are persuaded that Bank of America needs to raise more capital.
Here are some more specifics on the status of Bank of America's mortgage portolio:
The bank has a total of $60 billion of loans that are "troubled"--15% of its overall $413 billion portfolio. This includes loans that are 30-89 days past due ($12 billion) and loans that are 90+ days past due ($49 billion). These troubled loans are composed of both first mortgages ($56 billion) and home-equity loans ($4 billion). Of the total first mortgages on the bank's balance sheet, nearly 20% are past due. Of the total home-equity loans, only 3% are past due.
The delinquent percentage on the HELOCs is surprisingly low. The analyst we spoke to speculates that this is because homeowners are using the HELOCs as a source of ready cash, and they don't want to lose access to it. So they are staying up to date on the HELOCs even as they go delinquent on their first mortgages. This makes the analyst concerned that the HELOCs are a much bigger risk to Bank of America than the low delinquency percentage would suggest. Many of them are likely junior to first mortgages on houses whose value has plummeted, so they may not be backed up by any equity in the houses.
Overall, 15% of Bank of America's loans are in some form of delinquency, which would be categorized as "non-performing" in the securitized pool of loans discussed above. (Under the bank's definition of non-performing, only 5% are non-performing.)  On top of this percentage, however, the bank also has the HELOCs, which the analyst believes are far flimsier than they appear. And the analyst is also concerned that we don't have a clear picture of the rate at which Bank of America's loans move from one bucket to another (in other words, loans that were delinquent becoming temporarily "re-performing").