Securitization: Don't Make Transparency the New Opaque
(Apologies: “opacity” doesn’t have the same ring. If you prefer, this is the KISS article, for Keep It Simple Stupid.) If the Federal Deposit Insurance Corp. (FDIC), currently digesting comments to its proposed securitization safe harbor rules, and the Securities and Exchange Commission (SEC), now posting comments on its proposed revisions to REG AB, are to be believed, transparency is the critical ingredient to “restarting” asset securitization markets. Transparency is promoted in a couple of guises: simpler deals (fewer tranches etc.) and loan level information at issue and thereafter. The call for loan level information strikes me as exceptionally misleading. It makes the regulators look like activists, when in fact they are either ignorant of the market they would regulate or they are cynically making an empty display of regulatory zeal. Loan level performance data is already available on the great majority of non-agency deals. In fact, the degree of disclosure provided on non-agency MBS was envied by Ginnie Mae/GSE analysts and investors who understood that the credit characteristics of underlying loans (LTV, loan purpose, credit score, etc.) influenced the prepayment experience, and hence the interest rate sensitivity, of the bonds. In August 2002, staff from the SEC, the GSE regulator (then OFHEO, now Federal Housing Finance Agency (FHFA)) and Treasury Department conducted a study of MBS disclosures to determent whether enhancements were desirable. The results of that study were published January 2003 in “Enhancing Disclosure in the Mortgage-Backed Securities Market”. The study concluded that “the significant evolution of disclosure standards in the offer and sale of MBS - whether of GSEs, Ginnie Mae or private-label MBS - in the past has been nearly entirely market driven. Furthermore regarding loan size, geographic distribution, property type, occupancy, LTV, credit scores, degree of documentation and loan purpose, private issuers provided detailed statistical information on pool loans while the GSEs did not. Missing from all MBS offering disclosures was information on debt-to-income ratios of underlying borrowers.” However, following deal closing, servicers transmitted very detailed information on each loan at issue which includes debt-to-income ratios and all terms necessary to calculate scheduled P&I. Monthly updates track loan balances, delinquency status, coupon changes and payment recasts (if applicable), and so on. About the only items missing are updated credit scores or appraisals, names and addresses. (It is possible to map loan data to public credit histories stripped of names and addresses to deal data by zipcode and original lien amount. Analysts who do this can estimate with considerable accuracy changes in credit scores, how related non-mortgage debt is being serviced, whether second-lien mortgages exist and who holds them, and so on. Using home price data at the zip code level - a variety of vendors can provide this - it is also possible to estimate current LTVs. HousingWire coverage of research reports is replete with examples of analysts’ findings using such techniques with loan level data. I participated in such analysis as far back as the early 90s, but I’ve seen the work since the late 80s.) Needless to say, Ginnie and the GSEs responded to the 2002 findings (and the market that called for the study) by expanding the items included in at-issue and monthly pool updates. Now, however, the regulators have apparently forgotten that private-label MBS were the template for improved disclosure in government guaranteed and sponsored MBS. If you read the regulators, policy geeks and bloggers, MBS investors have forgotten as well. It’s hearsay, but apparently they are telling the regulators they would never have made such bad investments if they’d known what was inside. I believe this is called passing the buck or shifting the blame. If they had read the prospectus and looked at the loan level data for other deals by the same issuer, or similar issuers (they needed to do that to make basic relative value decisions), they would have known what was inside the MBS. If they couldn’t make sense of the prospectus, they should have realized they were in over their heads. If they didn’t subscribe to the data themselves, they could ask the dealer (who had a small army analyzing this to support sales and trading) - and if the dealer stonewalled, they should know to step away. If a particular issuer wasn’t providing this data, they should know to step away or lower the bid significantly. Only if the loans turned out to be other than as represented in the prospectus, prospectus supplement and term sheet, do they have a complaint about inadequate loan level data. In fact, they and the SEC have a lawsuit. The fact is, everyone but the regulators knows that no one reads the prospectus. And I know from experience how few investors demanded loan level analysis of subprime issuers before they bought into the dream. The dream of the fattest yields for a given rating available in the bond market. Well, except for CDOs made of MBS, but that is a different story. (CDO were black boxes. I say buyer beware. This market blew itself up before - as recently as the late ‘90s - and apparently too few investors learned any lesson from the debacle.) Conspiracists to the Rescue It’s not bad enough that regulators and their cronies have concealed the existence of loan level data. There is a band of righteous crusaders in the weeds of this debate calling for these “new” disclosures to be made on a daily basis. I wouldn’t be bothering you, patient readers, about this, if I were sure no one is listening. Alas, proponents of this scheme attract listeners because they are convinced Wall Street firms broke down the chinese walls, acquired up-to-the minute daily performance data from their conduit and subprime lender subsidiaries, and used it to short the market with credit default swaps (CDS). That’s right, they are playing the Wall-Street-bet-against-its-own-securities card. It’s the kind of argument many people, incensed as they now are with Wall Street, don’t stop to question. So let us now, if only in the spirit of devil’s advocacy, dispute it. The Drummer Boy Leading the charge is Richard Field, Managing Director of TYI, LLC, a boutique portfolio strategy firm (Bloomberg calls him a day trader). I first noticed Richard via his comment on the FDICs proposed securitization rule but he’s gotten the attention of reporters at Bloomberg, CNBC, the New York Times, the Chicago Tribune, and Fortune Magazine, among others. The Financial Times has run an Insight piece from Field, and he garnered a speaker slot at ABS East 2009. (Links to all are available on the TYI website.) Last week, my favorite banking industry analyst, Chris Whalen brought Field on as a guest writing about “Covered Bonds and the Reform of Structured Finance” in the April 12, 2010 Institutional Risk Analyst (IRA). (Unfortunately it’s rolled into archive, but the IRA archives are well worth the annual subscription fee.) The occasion for writing about covered bonds was the recent covered bond legislation introduced in the House, but the legislation simply provides a fresh segue to Field’s thesis that daily data is necessary to level the playing field between investors and the big Wall Street banks. His comments in IRA are addressed to the SEC, which has not specifically addressed the timing of the “new” data disclosure. Field explains why it should: “In the ABS market, a select few have access to loan-level performance information on a daily basis. Firms such as Goldman Sachs and Morgan Stanley have subsidiaries involved in originating, billing and collecting loans backing ABS. They receive “fresh” loan-level performance data on a daily basis that they can use for trading days, even weeks, before most other market participants receive the information.” By contrast, other participants have to wait to receive the “stale” data distributed monthly or, he asserts, less frequently. “It has been reported” says Field, that players like Goldman Sachs and Morgan Stanley who had the data on a daily basis “recognized that risk was mis-priced, stopped buying new securities by late 2006 and in fact went further and shorted the subprime market.” Yet another $1.75trn in non-agency mortgage backed securities, home equity loan backed securities and CDO were issued between Goldman and Morgan stepped back and the market collapsed in 2007. Think about it. Even if it is true that these banks used their originators like stethoscopes, they were only ahead of the rest of the market by maybe a month (reporting is done as of the 25th of the month). In other words, whatever Goldman and Morgan might have learned from insider channels at the end of 2006, investors saw in the first month or two of 2007 when the loan level data was processed. As a matter of fact, what everyone saw -- coming through every orifice, in loan level data, research reports and conferences, remittance reports -- drove spreads wider and prices down (which everyone investor would see in month-end dealer marks). Falling prices pushed the collateral calls and rising haircuts on repos, killed the Bear Stearns funds, forced the raters to begin -- late, as was their wont -- mass downgrades in second quarter 2007. The accumulating troubles in subprime were media events as well -- the first subprime mortgage banking companies started dying in 2006. Wasn’t that information? As a great poet of American popular music observed, it is not necessary to have a weatherman on staff to know which direction the wind is blowing. Inside Information Through a Pinhole Those big dealer’s purported stethoscopes were pretty dinky too. Let’s take the case of Goldman, which owned a small subprime lender, Senderra. (I have to admit that, having heard in my first days on Wall Street that Goldman, above and beyond the usual drug test, required a lie detector test of prospective employees, I have been unenthusiastic about the firm, regardless whether it was true or not. And Goldman does not publish MBS research, though I believe they have “desk” analysts who could talk to investors with questions. Why this did not discourage investors from buying their MBS issues, I do not know. Given my wall of indifference to daily doings at Goldman, I was ignorant of the firm’s acquisition of Senderra and only learned of it when Richard Field kindly provided me with one of the Wall Street Journal articles on which he bases his claims.) Anyway, Goldman invested in Senderra at startup, in late 2005 and bought it outright in early 2007. In an April 13, 2007 article in the Charlotte Business Journal, staff writer Will Boye quotes Tony Plath, finance professor at UNC Charlotte, who opined that Goldman, like others, was bottom fishing in hopes that the subprime market would come back. (Read on -- better scoop on Goldman’s motives below.) The news on subprime was already very bad and very public. Writes Boye, “The U.S. subprime-mortgage market has fallen on hard times, with rising loan defaults leading to problems for some lenders. New Century Financial Corp., one of the largest subprime lenders, has filed for bankruptcy. Locally, Wells Fargo Home Mortgage, the mortgage unit of Wells Fargo & Co. [stock WFC][/stock], is eliminating 250 jobs at its Fort Mill facility because it is cutting back on its subprime lending.” Wouldn’t an investor think that, if rising loan defaults could sink and shrink lenders, they could mess up your portfolio real bad? I mean, wasn’t the media busily providing the daily updates already? But I’m digressing. I was going to show you how narrow a look into the world of subprime Senderra might have afforded Goldman. Inside Mortgage Finance estimated total 2007 subprime origination at $192.5bn and The Mortgage Lender Implodometer indicates that Senderra averaged about $30.8m originations a month in 2007. That works out to about 0.19% of 2007 subprime originations. Analyzing monthly loan level data across the subprime universe would provide a vastly more detailed and comprehensive picture of subprime’s unraveling, and would pinpoint performance by geographical location, loan product, layered risk combinations. In particular, it would put specific issuers, vintages and bonds under the microscope. Extrapolating from the experience of one small operation in the Carolinas would not indicate which bonds to buy credit default swaps on (the technique dealers, hedge funds and others used to short the subprime market). Even if you used the ABX index to short the market, it would still be far more effective to track the underlying deals to determine which contracts to buy. More to the point, having an originator in pocket would not tell bond investors which bonds to sell or how soft the bid would be for specific bonds when they offered them for sale. For that, they could look at monthly performance on a loan level basis (including the rates at which loans go from one delinquency bucket to a worse one, how long foreclosure takes by state, how long liquidation takes by zip code, what kind of loss severities are being experienced, etc.). For that they would have to compare their bonds to other bonds that had recently traded. Misrepresenting the Press I can sympathize if the media misunderstands how MBS are constructed, perform, trade, are tracked, analyzed, etc., etc. Even the big names in financial reporting are under deadlines, have editors who want it slicked up and dumbed down and were clueless about the complexities of mortgage and other asset-backed securities until the crisis struck. If they had been able to do a better job, I wouldn’t have neglected consulting opportunities and started writing columns. But I am not sympathetic with people who treat items in the press as the truth. Even the best is “infotainment” and needs to be read carefully. I am even less sympathetic with folks who, in the service of their agenda, rewrite what they find in the press and present it as evidence. This is something I believe Richard Field does. In his response to the FDIC’s ANPR and his March 30, 2010 response to the Bank of England’s request for comments regarding its Consultative Paper on Extending Eligible Collateral in the Discount Window and Information Transparency for Asset-backed Securitizations, Field states: “On January 21, 2010, the Wall Street Journal discussed Goldman Sachs’ acquisition of a subprime mortgage lender. Goldman invested in the subprime lender when it was launched in 2005 and bought the firm in 2007. According to the article, ‘mortgage experts say the acquisition likely gave Goldman a clearer view of the market as other parts of the company made bets on home loans.’ These bets generated nearly $4bn in profits for Goldman.” First of all, the article wasn’t about the acquisition, it was about layoffs at Senderra and its impact on Ft. Mill, South Carolina, where Senderra and other subprime mortgage units were located. The stuff about Goldman is buried half-way down. Second, the 2007 acquisition came late in the unravelling of the subprime market, so I would argue scoop reaching Goldman from Senderra is stinking old news and, had they been real experts, the WSJ’s sources would have known that. (I don’t blame the WSJ reporter, his editor expected precisely such expert color.) Third, it appears Bloomberg’s reporters penetrated considerably deeper into Goldman’s motives for buying Senderra outright. Citing a November 19, 2008 Bloomberg article, Implode-o-meter notes that Goldman converted its acquisition into an FHA shop to refinance their portfolio of subprime loans. If so, that was a boon for investors! He Said WHAT? Here’s another of Field’s assertions that had me stumped. In his comment on the FDIC’s ANPR, he repeats a quote he found in Total Securitization, taken from a Global ABS Researcher Panel Discussion on June 3, 2009. Apparently a J.P. Morgan [stock JPM][/stock] vice president said “in an investor survey just carried out by the bank asking what would bring them back to the market, 60% said a greater level of deal information was their number one requirement.” This was offered after another panelist was said to observe: “We don’t have all the information we need. The loan-level data is not available. We can’t rely on the rating agencies, so we need data that allows us to make our own educated forecasts. The more information we get ... the better.” I could not believe any of the wonderful analysts I know at J.P. Morgan could have said such a thing, as I am a great admirer of their analysis of loan level deal data. So I emailed the quote to them. They were mystified. They said, we don’t even use the title of vice president. It took a day to track it down to a gentleman in the UK who said the dates matched the European ABS conference, it was his survey, which anyone on the team might have quoted at the meeting. And he explained, “This is being taken totally out of context if applied to the US - in Europe we don’t have any loan level data reporting and have quarterly reporting in the majority of jurisdictions.” (The emphasis is his.) Enhance the Monthly Data, Spare Us the Daily Afflicted as I was with insatiable data greed throughout my career as an analyst, I could never object to enhancing loan level data, by expanding coverage to, for instance, provide more detail on the range of potential loss mitigation and modification actions a servicer might take, standardizing the formats and definitions, putting it in easily accessed public locations and so forth. And please make it mandatory. For the investors sake and because I am a completist. But doing so won’t change the facts -- that diligent investors will use it, read the prospectus, familiarize themselves with the originators underwriting criteria and business model, and the un-diligent will not. This will make it marginally easier to look at a single deal, but it will not change the fact that an institutional investor -- a professional investor -- does not own just one, but many, perhaps hundreds of these MBS. Moreover, professional investors do not assess a bond in a vacuum, but relative to many similar bonds. This means data bases, and data bases either require full time staff to build and feed them, or the services of vendors who, at a minimum, aggregate all the deal data or build reporting systems and analytic user interfaces for the data. So cheaper it will not be (though the SEC thinks it will be). But please, no daily data. Having poured over monthly reports and diced and sliced monthly data, the thought exhausts me. It’s not just that I am lazy. I’m not too lazy to read a prospectus. But I know in my gut it is extravagantly impractical. It’s also intellectually unappealing. Monthly data is fully intuitive. It reflects the loan accounting cycle. Mortgage payments are made 30 days in arrears. (By convention a month is considered to be 30 days long.) That means the January payment is due on February 1. Most lenders allow a 15-day grace period. From the 16th day to the end of the month is a penalty period. Borrowers are charged a penalty fee if they pay during this period, but the loan is not counted as delinquent until the 1st day of the next month, when the next payment is due. This event is reported to the credit bureau and may incur yet more penalty fees. (I’m skipping the technical difference between the Office of Thrift Supervision (OTS) and the Mortgage Bankers Association (MBA) method of determining delinquency. It matters, but not for this discussion.) Subsequent delinquencies are reported at 30-day intervals. Watching the Paint Dry The milestones on the delinquency path are 30, 60 and 90 days. Current thinking is that if a loan reaches 60 days, it has a convincing likelihood of default (and if strategic default becomes a greater fad, 30 days might raise the alarm), but in the old days 90 or more days was as far out as credit analysts toted them up. Watching loans move from current to late to delinquent on a daily basis will be like watching paint dry. Necessary if you have are managing servicing interventions designed to keep borrowers current and cure them if they aren’t, but a monumental analytic problem and time drain for someone trying to manage a portfolio that is also subject to a host of market forces. Simply put, senior, triple-A bond holders won’t do it. They’ll say, just show me the 30, 60, 90 buckets please, I don’t want all this paper on my desk. The investors who might care are the ones holding unrated or residual pieces and rated credit tranches taking losses or about to. But it will still be like watching paint dry, waiting for loans 90+ to roll to foreclosure, from there to modification, short sale or REO, waiting for the REO to be liquidated. I do not believe the market will price differential delinquency or daily changes in the number of loans 90+ of in foreclosure. Too much can change in either direction. Its the discrete events that people will study: how many loans went into foreclosure later to be modified, how many loans were modified only to re-default (and of course, how many loans going 30-days delinquent ultimately default, etc.). Happily, these are the kinds of events that are presently being analyzed under microscopes in dealer research departments, at the big money management firms, by hedge funds and private equity firms. Using monthly loan level data.