By David Dayen, a lapsed blogger, now a freelance writer based in Los Angeles, CA. Follow him on Twitter @ddayen
This morning the Government Accountability Office released their second report on the Independent Foreclosure Reviews. Kudos to Rep. Maxine Waters for getting GAO to take another look. At the time that the OCC and the Fed scrapped the reviews and amended their consent orders, this review, expected to be critical of the process, was seen as a proximate cause.
Rep. Brad Miller, a North Carolina Democrat, told The Huffington Post that the report, which has not been released, was “critical” and that the Office of the Comptroller of the Currency, which administers the review, was aware of its findings. Miller said that that one problem the GAO was likely to highlight was an “unacceptably high” error rate of 11 percent in a sampling of bank loan files.
That indeed is in the report, though I wouldn’t call it a major feature. And there are only a few big bombshells or pieces of corroboratory evidence in the report. Furthermore, its narrow scope – GAO only looked at the regulators’ design and ovesight of the foreclosure reviews, rather than what the independent reviewers did, and in fact they used the bank consultant reviewers as primary sources – tends to give a very circumscribed picture of the reviews. You could even say that this report will help get the bank consultants off the hook by putting the blame on OCC and the Fed.
But the regulators were definitely part of the story here, and once you get through the government audit-ese, you can begin to see the picture of how they conspired to ensure the reviews would offer little to no value, and indeed attempt to exonerate the banks. The ensuing calamity only shows how the scheme worked too well, burying any evidence of borrower harm among an avalanche of deliberately cracked design.
Remember why these reviews came together in the first place. State Attorneys General and federal regulators were moving forward with a joint process in the wake of the robo-signing scandal, and feeling lots of grassroots pressure to bring cases against the banks. OCC’s move was a classic divide-and-conquer tactic. They split off with their consent orders, even though around that time, then-director John Walsh said before Congress that he believed there were no illegal foreclosures as a result of the servicing errors. So the thumbs were on the scale even then. The reviews were a political maneuver to reduce the leverage in the larger investigation, and protect the mega-banks, who all have OCC as their primary regulator, from these widespread claims of borrower harm.
Given that, we should expect that little or no thought went into designing an accurate method for determining harm. GAO verifies this in two ways. First, they confirm that nobody actually dealing with borrowers on the ground was consulted about how to best figure out the major categories for servicer errors. This is from the report:
In addition, regulators did not consult with community groups, such as national organizations representing housing counselors that have worked with individual borrowers on their loan modification and loss mitigation applications. These consultations might have provided input on challenges specific servicers and borrowers experienced with the range of loss mitigation and loan modification activities which could have assisted in identifying high-risk loan categories or program elements to consider. For example, our prior work surveying housing counselors found that while assisting borrowers with HAMP applications, counselors experienced challenges with servicers, including missing documentation, lengthy decision-making processes, and miscalculations of borrowers’ incomes.
A targeted process, that just took the biggest categories well-known to any housing counselor, and had the reviewers simply check those boxes, would have been efficient and broadly accurate. Sure, things would have gotten lost in the shuffle, but the overwhelming majority of abuse would have been apparent, and with far less effort. But nobody at the regulators asked the counselors.
Instead, they created what I’d call the “illusion of comprehensiveness.” In ensuring that no stone was left unturned, they devised a process laden with unturned stones. They set the consultants down a path that created a confusing, complex, kludgy system requiring up to 50 hours per loan file to review, which is simply ridiculous and unnecessary. From the report:
The number of test questions used by third-party consultants to conduct the file reviews varied. For example, one consultant told us that they had approximately 2,600 test questions with more than 4,000 discrete steps, while another consultant told us they had 16,000 test questions. Further, third-party consultants from whom we obtained information stated that their reviewers spent as many as 50 hours to complete a full file review, although review times varied depending on the issue and type of review.
What GAO leaves unsaid is that the consultants had every incentive to draw out the process and lengthen the reviews. $2 billion worth of reasons, in fact.
This led to the chaotic process that has been well-documented by Yves (who I’m sure will have plenty to say about this report). But too much of this reads like an apology for the bank consultants. GAO’s had a mandate to review how the regulators managed the consent orders, not how the consultants dealt with it. Therefore, we get a lot about the lack of communication to borrowers, the inadequacy of the initial planning, the failings of oversight over the reviewers, and so on. With the reviewers as sources, frequently the report gives off the vibe that the reviews were just too darn hard, that the regulators gave too much conflicting information. With the consultants as sources, the critical eye never gets turned in their direction in this report.
Nonetheless, GAO appears to know the scheme here. This conclusion is accurate IMO:
…the broad and expanding scope of the reviews and delays in defining key concepts could have been mitigated by more advanced planning from regulators, resulting in more efficient and effective reviews. Regulators may have missed opportunities to potentially narrow and refine the project scope—for example, through earlier definition of a harmed borrower or agreement on errors not resulting in remediation that may not have warranted additional review… conducting a planning process that involves all stakeholders provides an opportunity to examine preliminary information and pilot-test processes and procedures to help further define the scope of potential activities and hedge against the risk of future changes.
I’d say this is audit-speak for “the regulators screwed up royally.” And that was the whole point. OCC wanted these things to drown in data. It made it easier for the consultants to game the system. They may not have wanted the whole process to collapse, but as long as the evidence of borrower harm is ultimately relatively undefined, all the better.
You can read the report yourself – it’s at the bottom of this post. Just three more amazing things I flagged:
1) Like Treasury with HAMP – another program where the government regulator had different goals than they publicly stated – OCC and the Fed kept issuing new guidances and updates that would change the reviewers’ parameters (this, not the constant haggling between reviewers and their superiors, is used as the cause for delays in the IFRs). Regulators called it an “iterative” process, where they learned and refined as the reviews progressed. But it wasn’t so iterative that they got around to defining whether loan files missing documents counted as an error:
Guidance issued to consultants in March 2012 on how to determine whether a borrower suffered financial harm when key documents — such as documents that evidenced that foreclosure actions were taken or loan modification application documents — were missing indicated that consultants should treat those instances as errors on the part of servicers. However, this guidance noted that consultants should defer decisions on how to determine borrower remediation until regulators provided further guidance. In the absence of additional guidance, consultants indicated that they took a variety of approaches, including waiting to make remediation decisions, making preliminary considerations of whether the error might have caused financial harm, or working with their third-party law firm to review any applicable legal precedents. According to OCC staff, at the time of the agreements that led to the amended consent orders, they were planning to issue further guidance to consultants on remediation for borrowers with missing documents and they had informally provided additional direction about treatment of files with missing documents during their
regular meetings with consultants.
In the nearly two years of the review process, OCC and the Fed offered no guidance as to whether missing documents – really a basic servicer failing – counted as harm.
2) There’s a lot in here on sampling methodology, which the regulators clearly did not care about. Remember that there were basically two reviews, the review of files where the borrower submitted a complaint, and the “look-back” review, based on a sample of files. The regulators didn’t standardize the process, and so you ended up with differences in the expected population error rate (three consultants expected up to 10%, and eight of them assumed 0%, which completely changes how the sample gets viewed), the sample size per loan category, the categories themselves, and the review parameters. This effectively rendered the samples meaningless, at least as a comparative matter. When asked about this, OCC let loose with this gem:
OCC staff told us that their handbook on sampling methodologies—a reference their staff use for sampling and one that both regulators suggested consultants consider in designing their sampling approaches — generally assumes few or no errors will be found in a sample.
GAO explains why this is problematic. In their example, they show that the difference in expected population error rates changes the triggers that would lead to additional analysis of more loan files. Therefore, one consultant expecting a lower error rate would conduct additional analysis if one or more errors were found, but another would only conduct the analysis if five or more were found. Theoretically, they could be looking at the same sample and find the same amount of errors, but one consultant would do additional analysis and the other wouldn’t. It’s another way to game the system, filtering the sample methodology to ensure less rigorous analysis, for example. This is a great excuse by the Fed:
According to Federal Reserve staff, they anticipated that consultants would continue reviewing files until the sampling found no additional errors. However, unless a large majority of the population is examined for errors, reviewing files until sampling finds no more errors does not necessarily imply that all or most errors would be identified.
3) If the GAO’s analysis of how the payouts of the amended consent orders will work is at all accurate, we’re on the cusp of yet another train wreck. They’ve apparently learned absolutely nothing.
• They haven’t decided aspects of the reviews for SCRA violations and foreclosures on borrowers who weren’t in default
• They haven’t determined payment amounts for the different broad categories of borrowers
• They haven’t figured out what to do with leftover funds not claimed by borrowers
• They haven’t figured out the loss mitigation procedures for servicers, or criteria for what borrowers will be eligible
So here we go again.