By Abigail Caplovitz Field | March 13, 2012
updated for style and to add Six Sigma
On my first read through of the consent agreements among the bailed-out bankers (B.O.Bs), the Feds and the States I saw much as had been promised. One thing I hadn’t seen coming, however, was that the B.O.Bs would now be allowed to systematically overcharge borrowers and steal their homes. Seriously. Who cares about $1 million or $5 million penalties if horrible damage can be inflicted without punishment?
To see what I’m talking about, you need to look at Exhibit E-1. (It’s in all the consent agreements; here’s Chase’s.) Exhibit E-1 is a 14 page table titled “Servicing Standards Quarterly Compliance Metrics.” That is, it’s a table that details what, precisely, will be monitored to make sure that the B.O.Bs are meeting the very pretty servicing standards detailed in Exhibit A (again part of all the agreements.)
Note: You may want to print out table E-1 while reading this, or at least keep it open in another browser window; these metrics are shocking enough you won’t want to take my word for it, you’ll want to verify I’m citing the text correctly.
Now, the table doesn’t come right out and say, we, the federal and state governments of the United States of America do hereby bless the institutionalization of servicer abuse, but it should. To understand why, you need to keep your eye on how the table’s columns are defined. For most issues, the critical columns are C “Loan Level Tolerance for Error” and D “Threshold Error Rate.” Later I’ll talk about the problems in Column F, the “Test Questions.”
When Error Isn’t Error
Loan Level Tolerance for Error, Column C, is defined as:
“Loan Level Tolerance for Error: This represents a threshold beyond which the variance between the actual outcome and the expected outcome on a single test case is deemed reportable” (bold mine; see footnote 1 page E-1-14)
Get that? Any error up to the threshold doesn’t count; it’s not reportable error. Now see footnote 2, which defines “Threshold Error Rate” (Column D):
“Threshold Error Rate: For each metric or outcome tested if the total number of reportable errors as a percentage of the total number of cases tested exceeds this limit then the Servicer will be determined to have failed that metric for the reported period.”
Only “reportable” errors count, and only if enough of those are reported can get a servicer in trouble under the settlement.
Let’s check out how these definitions interact in a few of the metrics. Let’s start with the metric called “incorrect loan mod denial.” (Top of page E-1-2.) The point of this metric is to protect you from being rejected for a loan modification you actually qualify for.
According to Column C, the loan level error tolerance for income errors is 5%. As a practical matter, here’s what that means: Imagine your household income is $80,000. Imagine that at $80k the bank’s formula says you get a modification and thus you can keep your house. But the bank doesn’t use $80k in its math; it uses $77,000. So the computer rejects you, and you lose your home to foreclosure. Does law enforcement care about the bankers wrecking-your-life error? No, because the $3,000 error, while enough to deny you the mod, isn’t 5% of your income. So the error was too small to count.
With metrics like these, what incentive do the B.O.Bs have to eliminate errors, or at least fundamentally minimize them? Doesn’t law enforcement understand that people whose homes depend on the banks’ math need much, much better?
But wait, it gets worse, because of Column D–the Threshold Error Rate. It’s not enough for the B.O.B.s to make such a big mistake that it’s “reportable” when they rejected you for a mod and foreclosed wrongly. They have to make five reportable errors in every hundred files reviewed before they get in trouble! Since the B.O.B.s are dealing with millions of people seeking mods, that’s a lot of A-OK big mistakes–50,000 for every million mod applications. Sweet.
Contrast that standard of not caring until a 5% error happens at least 5% of the time with the approach of Six Sigma. Six Sigma is the business process Motorola invented because Motorola wanted to manufacture defect-free products. Six Sigma aims for 3.4 errors in a million runs. Imagine–only 3.4 wrongly rejected loan mods instead of 50,000. If Motorola can do that, why can’t BofA? Why can’t JPMorgan, Wells Fargo, Citigroup or Ally? Imagine everyone’s joy if the B.O.B’s manufactured defect-free products! Instead, when a Six Sigma Black Belt tells Chase that its credit card debt collection processes are broken Chase fires her rather than try to solve the problem.
Ok, now that we understand that Columns C and D combine to let a large number of life-changing errors not trigger enforcement action, let’s see how Columns C and D let the bankers systematically overcharge people.
Systematically Ripping You Off Is Okay
Skip ahead to page E-1-6. The first metric is “Fees adhere to guidance,” and applies to people who are 60 days or more delinquent. The point of the metric is to make sure the B.O.B.s aren’t larding on junk fees to people who are already struggling with their payments. Column C says it’s not reportable error unless “Amounts [are] over stated by the greater of $50 or 3% of the Total Default Related Fees Collected.”
That means it’s okay to charge everyone $49.99 too much, or even more, when 3% is more than $50. Column D says 5%, meaning, the B.O.B’s won’t get in trouble if they charge 4 people in every 100 an unlimited amount too much. Given the millions of people who will be 60 days or more delinquent during the life of this deal, that’s quite a license to steal.
If the B.O.B.s want to refer you to foreclosure, they can steal even bigger amounts from you; check out page E-1-4, “Pre-foreclosure Initiation-Accuracy of Account Information.” Column C says that when your account is referred to foreclosure the “Amounts [can be] over stated by the greater of $99 or 1% of the Total balance”. So if you owe $200,000 the B.O.B.s can say you owe $202,000 and it’s not reportable error. And again, thanks to Column D, B.O.Bs can overcharge four in every hundred defaulted borrowers lots more by ‘mistake’ and not get in trouble. In sum, the banks could give themselves a 1% gratuity when starting every foreclosure, bump it up to whatever they felt like 4% of the time, and not get in trouble. Nice.
But it gets worse; that’s just fees and balances for delinquent folk. Let’s return to page E-1-6, and look at the second metric, which applies to everyone with a mortgage: “Adherence to customer payment processing.” According to Column C, it’s not reportable error for the B.O.Bs to tell their computers that you paid less than you did, if “Amounts [are] understated by the greater $50.00 or 3% of the scheduled payment.”
Since most people don’t pay more than what they owe each month, posting less than you paid would seem to make you delinquent when you’re not. How can that be ok? What are the consequences? The servicing standards say the banks have to take your payment if you’re within $50, (See page A-5 at 3.a) but if your mortgage payment is $2000/month, 3% is $60. What if you start facing fees? What if you were trying to bring your account current and the bank screws up the data entry and starts foreclosing? Why isn’t that potentially devastating error reportable?
And again, it gets worse because of Column D. Again, reportable error has to happen 5% of the time to matter. There’s more than 50 million mortgages in the country. 5% of 50 million is 2.5 million. In a single year the banks can tell their computers that 2.5 million people paid so much less than they in fact paid that it’s reportable error, and still the bankers won’t get in trouble.
Most plainly, the bankers can tell 2.5 million people:
“Hey, you didn’t make your payment this month, your check’s short and we’re putting it in the no man’s land of a “suspense account” triggering delinquency and fees, even though you really did pay in full and have the canceled check to prove it. And guess what? No one but you cares; law enforcement won’t even consider dinging us for it.
I’m struggling with the same level of disbelief I had when I first learned that banks were systematically committing forgery.
Watch Out for Test Question Clarifications–Okaying Home Theft
Even metrics that look tough superficially turn out to be cruelly weak. For example, take the very first metric in the table, page E-1-1, “Foreclosure sale in error”. If it happens, that means the B.O.Bs sold your house when they weren’t supposed to. On first glance, things look good: no loan level error is tolerated (Column C is N/A). Column D looks tough, but only if you don’t think much about it: only a 1% error is tolerated.
When 1 million homes are foreclosed, that’s 10,000 sold wrongfully.
In 2011 banks foreclosed on nearly 2 million homes according to BusinessWeek (stat on p. 2 of story), so if that metric were in place last year, nearly 20,000 homes could’ve been effectively stolen from people and the B.O.Bs wouldn’t get in trouble. [Update: CoreLogic puts the 2011 completed foreclosure at 861,128, or 8,611 allowably sold wrongfully; it notes that since Sept. 2008 there’s been 3.3 million completed foreclosures, which would be 33,000 allowably sold wrongfully in that time. My point is still the same: it’s way too many.]
But that 1% isn’t the really big flaw in this metric. The biggest problems with this metric are hidden in the “Test Questions,” which are Column F.
Focus on the parenthetical qualifications that start with question 2:
1. Did the foreclosing party have legal standing to foreclose?
2. Was the borrower in an active trial period plan (unless the servicer took appropriate steps to postpone sale)?
Surprise! It’s not reportable error if the B.O.Bs sold your house during an active trial mod, if they tried to stop the sale from happening.
3. Was the borrower offered a loan modification fewer than 14 days before the foreclosure sale date (unless the borrower declined the offer or the servicer took appropriate steps to postpone the sale)?
Again, it’s not reportable error if the B.O.Bs sold your house while you were evaluating or responding to their mod offer, if they tried to stop the sale.
4. Was the borrower not in default (unless the default is cured to the satisfaction of the Servicer or investor within 10 days before the foreclosure sale date and the Servicer took appropriate steps to postpone sale)?
Wow–it’s not reportable error to sell your house even though you weren’t in default, so long as you foolishly cured the default too close the sale date and the B.O.Bs tried to stop the sale of your home.
5. Was the borrower protected from foreclosure by Bankruptcy (unless Servicer had notice of such protection fewer than 10 days before the foreclosure sale date and Servicer took appropriate steps to postpone sale)?
Again, you can have the law on your side–you’re protected by the bankruptcy court–but the B.O.Bs can sell your house anyway if you dawdled in declaring bankruptcy and the bank tried to stop the sale. I wonder what a bankruptcy judge would make of that provision?
See, in four of the five questions the B.O.Bs have found a way to make yes mean no: Yes, we violated the bankruptcy stay; No, it doesn’t count toward the 1% error rate. As a result, 1% of the foreclosure sales checked by the monitor isn’t the real threshold for getting bankers in trouble. It’s 1% plus all the wrongful sales that this settlement says are ok anyway.
Too Big To Be Competent
I hate the term Too Big To Fail because it’s a loaded premise presented as fact. But looking at the weasel parentheticals, maybe we should start asking if the banks as too big to be competent. I mean, why do the banks need a ‘hey, we tried but didn’t have enough time to stop the sale’ exemption? If the B.O.Bs want their lawyer or trustee to call off a foreclosure sale, all they need is two things: a) to contact their agent and b) have a competent agent.
What does “took appropriate steps to stop the sale” mean, anyway? Does it mean that someone at the bank left a message or two with foreclosure counsel? If the B.O.B.s made a real effort to stop the sale but their agents did it anyway, why isn’t that the B.O.B’s fault for having incompetent agents? Doesn’t giving the B.O.B. a pass remove any incentive to have competent (and thus more expensive) agents?
Wrongfully selling someone’s home should be a strict liability issue. Strict liability is, well, strict: no one cares what you were trying to do, what your intentions were, what you did or didn’t do. Did the harm happen? Then you’re responsible.
Before you give me any, hey, let’s be reasonable here, a business needs to operate and we’re so big some mistakes will happen, remember what we are talking about: homes; property rights; land records; fundamental fairness. How can the B.O.Bs be held to any standard other than strict liability when it comes to wrongfully selling a home?
Note, I’ve only dissected a handful of key metrics; remember, there’s 14 pages of these, each with their Columns C, D and F.
So here’s the real question: given these metrics, this institutionalization and legitimation of servicer abuse, how did all our meaningful law enforcers do this deal?