The Importance of Loan Loss Reserves

Last post I said:

I'm going to have to make a post comparing this [Exchange Bank of Santa Rosa's] reserving with some of the companies Fairfax owns credit default swaps against. The allowance for loan losses doesn't even cover the 90+ day delinquencies for many of these companies, even after some very large loan loss provisions recently.

Well today, I made the comparison using two other household names in Fairfax's CDS portfolio: Countrywide Financial and Washington Mutual. Before going into that though, I want to comment that it was very difficult to find these numbers in their earnings report, and 30+ delinquency numbers could only be found by using FDIC Call Thrift Data. And for Countrywide and WaMu, getting a true understanding of their financial conditions is very complex.
(Numbers in 000's)

First to reiterate, is Exchange Bank of Santa Rosa's at the end of Q3:

Exchange Bank of Santa Rosa
(Numbers in 000's)

Gross loan portfolio: 1,173,194
Allowance for loan losses: 20,930
30-89 days delinquent: 1,327
90+ days delinquent: 7,890

30 day + delinquencies / gross loans = .78%
90 day + delinquencies / gross loans = .67%
Allowance for losses / gross loans = 1.78%


Now let's look at Countrywide Financial. First, a few things to note about Countrywide. One is that Countrywide is seperated into a mortgage originating and a banking operation. The originating business doesn't provision for losses, because they expect to sell their loans. For purposes of this exercise, we will only look at their banking operation because that involves the loans they plan on holding on their balance sheet. But it should also be remembered that Countrywide's originating business pumps out loans at very high multiples of equity. For example, in the latest quarter they produced 90 billion of loans. Annualize the rate and compare it to 15 billion in total equity, (traditional banking included) and you start to see what I am talking about. So at any given point, Countrywide is also holding a large amount of loans that it does not consider part of its portfolio and so doesn't reserve for. And because in the current credit turmoil, we saw them transfer 12 billion of loans to their banking portfolio this quarter in order to avoid selling them at current market prices. But anyways, I will be using only the banking operation numbers today.

Another thing to note is that since Countrywide does not provide 30 day delinquencies, I had to look them up using Call Thrift Data. For one, the most recently available report is from the 2nd Quarter, so the numbers are slightly lagging. Second, the numbers slightly differ between what the company reports, likely because of other subsidiaries. So do not accept these as precise, but very rough estimates. If anything though, it is underreporting delinquency numbers.

Countrywide Banking
(Numbers in millions)

Gross loan portfolio: 80,430
Allowance for loan losses: 1,127
90 days + delinquent at end Q3: 1,433
90 days+ delinquent at end Q2: 941

30-89 days delinquent at end Q2: 1,790

So we see that 90+ delinquencies shot up between the 2nd and 3rd quarter, which means 30-89 day delinquencies probably went up in Q3 too. But if we just assume the same 30-89 day numbers for Q3, we get:

30+ day delinquencies / gross loans = 4.00%
90+ day delinquencies / gross loans = 1.78%
Allowance for losses / gross loans = 1.40%

Now a few other things (I told you this was going to be complex). The gross loans number of 80.4 billion is inflated because Countrywide transferred 12 billion in loans over during the quarter from their originating business. But most of these loans were made recently, so they would either not be delinquent yet or they went delinquent during the third quarter, for which the figures aren't available (remember we had to use Q2). So these numbers are slightly inflated in that respect. Second, Countrywide holds 45 billion in Adjustable Rate Mortgages (ARMs), with many of those being Pay-Option ARMs. These are much more difficult to go delinquent on because any payment that can not be made is just added to the principal balance. But either way, we see that Countrywide's reserving is still inadequate.

With all that cleared up, it should be much easier to explain Washington Mutual. The same issues mentioned above apply here, with respect to originating vs banking business and the 3 month lag in some of the numbers.

Washington Mutual Banking
(Numbers in millions)

Gross loans = 227,348
Allowance for loan losses = 1,889
Non accrual loans at end of Q3= 4,577
Non accrual loans at end of Q2 = 3,443

90+ days delinquent and accruing at end of Q2 = 245
30-89 days delinquent at end of Q2 = 2,910

Using the same Q2 numbers going into Q3, we get:

30+ days delinquencies / gross loans = 4.58%
90+ days delinquencies / gross loans = 3.40%
Allowance for losses / gross loans = .83%

Again, Washington Mutual also has 138 billion in ARMs, with 56 billion of them Pay-Option ARMs, which do not normally go delinquent until much later on. Calculated Risk recently provided this chart with regards to recent deteriorating in ARM performance.
We see the recent trend has been much worse. So to summarize the point of this entire post, I'd say look at two percentage figures:
1. Allowance for loan losses to 30 day delinquencies, and
2. Allowance for loan losses to 90 day delinquencies.

Basically, the percentage which your allowances cover delinquent loans. The higher the better, and remember this is giving a lot of benefit to the doubt to CFC and WM. (lagging delinquencies, ARMs)

EXSR
1. 228%
2. 266%

CFC
1. 35%
2. 79%

WM
1. 18%
2. 24%

The significance of all this is that an under-reserved bank has no cushion for declining credit performance and will have to provision more for losses in the future, meaning their current earnings figures and equity are likely inflated and unreliable. Conversely, a well-reserved bank has a cushion for any adverse credit trends and the income figures are much more stable and protected.