Sandrew on Finance

Comments on finance and capital markets by a practitioner.

Bot-proof Contact:
[Title of this blog, no spaces] [at] Google's public email service (hint: rhymes with shemale)

Importance of LSAT Scores & UGPA in Law School Admissions at Selected Top Law Schools

[Ed. note: I found some more data that indicate that the selection bias of the LawSchoolNumbers.com data used to produce the graphs below is very significantSome acceptance rates are grossly overstated. I am attempting a fix.]

Every year around this time, thousands of people applying to law school stress about their prospects of being admitted to their top choice of school.  As many applicants are aware, the Law School Admission Test (LSAT) plays a key role in the admissions process at many law schools, particularly those in the coveted “Top 14” (T14) rankings, a fact often bemoaned by applicants and admissions officers alike. 

Another important number for an applicant is his or her UGPA, or “uniform” grade point average.  I put uniform in scare quotes because GPAs are made uniform in only a perfunctory sense by being put on the same scale (e.g. A- is awarded a 3.67 grade point), whereas no effort is made to adjust grades across schools or courses (a difficult exercise, I grant).

As a prospective applicant myself, I found myself caught up in this stress.  I wanted someone to demystify for me, in cold numbers, just how important the LSAT is to one’s chances of being admitted.  But rather than rely on one of the many books written on the subject of law school admissions “secrets,” and wanting some relatively current data, I took it upon myself to do my own research.  If nothing else, the research process was cathartic for me.  I hope you also find it useful. 

Note on Methodology: I did not prepare a regression, which would be an obvious way to answer the question: How much to these variables matter?  The reason I didn’t run a regression is twofold: (1) For most schools in my sample, the relationship between scores and acceptance rates are non-linear, and I don’t have the tools at hand to efficiently run such a regression.  What’s more, the data sets, while large in terms of numbers of applications, aren’t quite large enough to give a clear and confident picture of acceptance rates at all points on the curve.  For example, very few applicants score in the 178-180 range, and as such, the calculated acceptance rates shown below are based on the few data that are available.  (2) I’ve prepared the below figures from the perspective of an applicant, who might be asking: Given my LSAT score (and UGPA), what are my chances at getting in to School X?  Given this objective, I figured it would be more useful to provide the data simply (i.e. graphically) rather than statistically, which would be much more prone to misinterpretation.

Figure 1

Observations on LSAT Scores

  • We all know that the LSAT matters a great deal to law school admissions officers, but how much?  It appears that most top schools (Yale and Harvard less so) have 3 buckets of LSAT scores (Yes, No, and Maybe; alas, Yale and Harvard drop the Yes bucket).  To a first approximation, the extent of the relationship between one’s LSAT score and the acceptance decision at a particular school shows up in the width of the Maybe bucket (as measured in LSAT points).  For example, look at Columbia: for all applicants earning over a 175 on the LSAT (irrespective of UGPA), the likelihood of acceptance is very high (90% or more; the Yes bucket), and among applicants earning less than a 167, there were no accepted applicants.  So the “Maybe” range for Columbia is 168 to 174, a range of 6 points on the LSAT.  Compare this to NYU, where the Maybe range is half as wide as Columbia’s (though somewhat more forgiving to applicants) at 168 to 171.
  • Interestingly, at the top-of-the-top law schools (Harvard and Yale, and I suspect also Stanford), while LSATs certainly matter, there are no shoe-ins (no Yes bucket).  This is evident in the longer and shallower slope of the curves illustrated for these schools.   For other Top 14 schools, the curves are very steep, perhaps an indication that feedback effects—between and among applicants, admissions officers, and school ranking purveyors—are more at play among the schools ranked 4-14 than at the super-elite Top 3 schools. 

Figures 2 & 3

Observations on UGPA

  • I also tried to tease-out the relationship between UGPA and acceptance rates.  While it certainly exists, the relationship is far less strong than that of LSAT scores and acceptance rates. 
  • Figure 2 is best viewed in relationship to Figure 1.  You’ll note that most school’s curves have shifted to the left, when filtered for high-UGPA applicants.  The extent of this shift gives some indication of the importance of UGPA.
  • For one law school, Georgetown, I’ve hashed out acceptance rate-LSAT curves for each of the 5 quintiles among the sample applicants in Figure 3.  As one would expect, lower-UGPA applicants generally require higher LSAT scores to improve their chances of acceptance.  Or, seen a different way, one could say that for some LSAT scores, the UGPA matters a great deal at GULC.  For example, with an LSAT of 168, you’re a virtual lock (90%) to be accepted if your UGPA is in the first quintile of applicants, whereas you’re a long shot (about 10 to 1 against) from the 4th or 5th quintile.

Final Thoughts

I hope this helps shed light on the What are my chances? questions that many law school applicants have.  Clearly LSATs are important to law schools, but these illustrations give some sense of just how important they are, and for whom that extra point on the LSAT matters most.  Lastly, I’d like to acknowledge that I’ve been silent on the normative question about whether law schools ought to put as much weight on the LSAT as they appear to, on which I have no comment at this time.

Comments (View)

Sense & Securitizability: Theater Review of Monetizing Emma

Past portents of the far future can seem oracular when they hit or campy when they miss.  But accurate or not, they can have lasting appeal.  Imagining the near future, however, is a riskier gambit—flirting between sage social commentary and irrelevancy—and has a much shorter shelf life.  For Felipe Ossa’s pre-crisis script for “Monetizing Emma,” set in the 2013 world of structured finance, the risk pays off, ripening nicely in the six years since it was penned.

Let me get this out of the way: some of the acting and stage directing are a bit amateurish, and some casting decisions are distracting.   That said, Nitya Vidyasagar is excellent in the lead role, and both James Arden and Janice Mann turn in capable performances as  co-villains.  That’s all I can reliably say of the production itself, and so I’ll instead focus my attention on the play’s biggest draw, the script.

Mr. Ossa’s delightful premise imagines financial innovation run amok, where investment bankers have concocted a new security backed by shares of teenagers’ future lifetime earnings.  Smarmy bankers (played by Arden and Mann) compete to identify the best “assets”—overachieving teenagers short on means but long on potential—and to sign them to a contract that will cede a portion of their future earnings to a securitization vehicle, in exchange for which they receive a handsome lump sum that more than finances their college tuition.  Emma Dorfman (Vidyasagar) is one such overachiever, a top-of-her-class 15-year-old that the bankers see as a potential asset for their teen portfolio—that is, provided Emma can kick her Jane Austen obsession before she becomes (gasp!) an English major.

Mr. Ossa backs up his clever premise with witty and efficient writing, generously peppered with sneaky jokes calling on an eclectic range of reference material, not just structured finance geekery.  His send-up  of angsty Austen fandom is deft, and pop cultural and political references are slyly dropped into the dialog.   So sneaky are the jokes, for example, that I only heard one audible guffaw (my own) at a slick and unexpected reference to a West African dictator.  I found myself wishing this were the extent of the humor, however, as repeated attempts at slap-stick fell flat.  And while the cast is slim at only six characters, the roles of Emma’s two bratty high-school tormentors—responsible for the least compelling moments of comic relief—could be pruned from the script without a loss.

Perhaps owing to when it was written, and to its credit, the play avoids overtly satirizing the perverse incentive structures of the securitization model.  To do so would have been tedious and overreaching.  Instead the commentary on finance and financiers is subtle yet effective, leaving room for a well crafted and entertaining plot.  At its heart, “Monetizing Emma” is more caper than satire, but that’s no knock.   Watching the swindles committed by too-clever-by-half bankers backfire can be just as much fun to watch on stage as it is in real life.

Comments (View)

Reverse Engineering the ABACUS Timeline

[Hat Tips: Many thanks to @alea_, @crookery_, and @interfluidity.]

If only I had the offering circular for the Goldman-structured synthetic sub-prime mortgage CDO ABACUS 2007-AC1, which ostensibly closed on April 26, 2007, I might be able to fill in some of the gaps in what’s been publicly reported/leaked about the deal. Alas, I don’t, and so I’m left to speculate. I’ll try to sum up what I view to be the most likely timeline of events for the much ballyhooed deal and related bilateral transactions.

[Note: A lot of this is speculation. Unless a specific day and date is provided, I’m guessing as to what happened and when.]

December 2006: Paulson & Co. approaches Goldman Sachs expressing interest in purchasing protection (i.e. betting against) a basket of a BBB-rated, subprime-backed mortgage bonds that closed around the second half of 2006 (i.e. “06-2”).

NB: Goldman calls this a “reverse inquiry”—“reverse” because it the idea was hatched by one of its clients (in this case Paulson), rather than by some clever Goldmanite (e.g. Fabrice Tourre). Also note that the fact that trade germinated from an idea the short side (i.e. Paulson) rather than the long side (i.e. ACA and IKB) is moot to this nomenclature; reverse inquiry simply means it was not Goldman’s idea.

Late December/Early January 2007: Goldman calls potential investors (among them, German bank IKB)—probably from a list of past investors in previous ABACUS-branded deals—and asks if they would have interest in taking a long position in a 06-2 BBB subprime mortgage bond-backed synthetic CDO. Goldman gets enough interested parties (none of them yet committed, obviously) to start structuring a deal. But there’s a hitch. The investors want to know how the portfolio (which is to be static) will be selected; they want a reputable CDO manager calling the shots, which in a way is an outsourcing of their due diligence on the underlying portfolio.

First week of January 2007: Goldman approaches ACA to propose they serve as the “Portfolio Selection Agent” for the proposed CDO. ACA jumps at the opportunity—this is a service they can charge for, after all—and probably considers it a blessing that Goldman came to them, and apparently only them.

January 8, 2007: Goldman introduces ACA to Paulson at a meeting in Paulson’s offices among Goldman’s Fabrice Tourre and representatives of ACA and Paulson. At of this writing, we don’t yet know what, in these initial conversations, Goldman told ACA about Paulson’s involvement. Paulson and Goldman both allege that they were transparent with ACA, and that ACA knew that Paulson intended to go short the proposed structure. The SEC alleges differently—specifically, that Goldman (not Paulson) misled ACA into believing that Paulson was to be long the transaction. ACA has been silent, probably because they no longer exist—incidentally, for reasons not unrelated to the sorts of actions they took in connection with the ABACUS deal in question.

January 9, 2007: Goldman sends ACA the first proposed portfolio, concocted by Paulson, in an email with the subject line “Paulson Portfolio.” This initial portfolio consists of 123 subprime RMBS.

NB: According to one of Goldman’s defense letters to the SEC, the entire space of subprime RMBS deals that fit the parameters for the ABACUS deal totals only 293 bonds.  This may become important in the SEC suit.  If Goldman can show that even if ACA had selected securities for the portfolio in a vacuum without input from Paulson, then the portfolio would have gone bust anyway and the investor losses would have been the same, then Goldman can invoke the complex “no-harm/no-foul” defense.

January 9, 2007 to February 26, 2007: Paulson and ACA debate the contents of the portfolio. The parties agree on the final portfolio of 90 RMBS securities on February 26.

February 26, 2007: Goldman launches a road show to pitch the proposed deal to potential investors, including a snapshot of the expected capital structure and ratings. The pitch book shows a $2B total notional amount, and an optimistic Goldman seems hopeful at the time that they can drum up enough interest to fund $700M worth of notes.

NB: Since the portfolio wasn’t agreed upon prior to this date and the deal terms not yet finalized, Goldman not could have gotten one of the ratings agencies to commit to a rating by this date. For reasons discussed below, the terms that appear in the pitch book are not the same as those of the final deal.

March-April 2007: Goldman finalizes the terms of ABACUS 2007-AC1. After shopping the proposed deal to IKB and other investors, Goldman discovers that investor interest in the $700B of funded notes it’s been pitching has waned. This is probably due in no small part to the RMBS market beginning to show substantial weakness by this point, falling from prices in the high 90s at the time Goldman initially phoned potential investors back in December-January, to the low 80s by mid-April.

Goldman tries to spin this as good news—an opportunity for more attractive yields—but alas, the pitched deal is substantially undersubscribed. When the dust settles, only one of the original potential investors, IKB, is still interested. They commit to $150M. But once again, there’s a hitch. IKB wants ACA, the ostensible portfolio selection agent, to have some skin in the game. ACA, who you’ll remember stands to earn fee income from IKB for this service, agrees to pony-up $42M for some to-be-AAA-rated paper.

April 26, 2007: ABACUS 2007-AC1 closes. While we don’t yet know the exact terms of the notes at closing, I’ve found evidence in Bloomberg of only two classes: $50M of Class A-1 Notes (paying LIBOR+85bps) and $142M of Class A-2 Notes (paying LIBOR+110bps), both originally rated AAA by S&P (and a month later by Moody’s too). Based on the alleged facts from the SEC complaint and corroborated by Goldman’s defense letters, IKB held all $50M of the A-1 Notes and $100M of the A-2 Notes, with ACA picking up the remaining $42M of A-2 Notes. Concurrent with the deal closing, in which Goldman purchases structured protection exactly mirroring the terms of the issued notes (via a single-tranche bespoke basket default swap) from the ABACUS 2007-AC1 special purpose entity (a legal entity registered in the Cayman Islands), Goldman sells this protection in a back-to-back transaction to Paulson. It’s unclear whether Goldman keeps a spread between these two trades as consideration for helping broker and structure the deal. It’s also unclear whether Goldman earned its reported $15M fee from Paulson at the time of this closing or at the time the supersenior trades closed (see below).

NB: I don’t know much more about the exact terms of these notes or the ultimate size of the reference portfolio underlying them. As @alea_ has helpfully pointed out to me, it’s not essential that they bear the same total portfolio notional as the subsequent supersenior trades (see below). And so since I’m already speculating… One possible structure might be thus:

May 2007: Paulson wants more protection (i.e. a bigger short position) than the $192M Goldman managed to close via ABACUS. Based on the terms of the pitch book, it appears Paulson was prepared to short a $1.8B (10%-100%) tranche of the $2.0B reference portfolio. Paulson asks Goldman to see what it can do to arrange a big supersenior bet on top of the already-closed ABACUS deal. They’re looking for a protection seller for a 45%-100% supersenior tranche up to a notional of $2B.

NB: Perhaps Goldman and Paulson had an agreement (written or otherwise) to solicit investors and protection sellers up to certain target amount that Paulson had in mind. And perhaps Goldman’s fees were tied to hitting this target.

Later that May 2007: Goldman’s Tourre goes back to ACA to see if they’re interested in writing supersenior protection. ACA’s parent company, ACA Capital, just so happens to be in the business of providing financial guarantee insurance (i.e. writing protection) on structured products. ACA agrees and the parties negotiate a 50bps premium for protection on the supersenior tranche on $1.8B of total notional reference the same ABACUS portfolio. But there’s another hitch. Goldman and ACA can’t agree on the terms of the collateral arrangement.

NB: I don’t know if Goldman solicited anyone else to write the supersenior protection, but I can think of one notorious name who got themselves up to their eyebrows in the stuff. In any event, it would have made sense to ask ACA first, since they’d already done the due diligence and wagered $42M of their own money on a more subordinate tranche.

Still May 2007: Goldman calls ABN, looking to find someone willing to work out a solution to the impasse between Goldman and ACA Capital. ABN agrees to intermediate the Goldman-ACA Capital deal by effectively guaranteeing ACA Capital and agreeing to post collateral on its behalf to Goldman, all in exchange for 17bps. Goldman’s risk managers are pleased, but now Tourre has to go back to Paulson with mixed news. He managed to find a protection seller, but only for a $1.8B 50%-100% tranche and for a rather high total premium (for supersenior deals) of 67bps. Paulson, perhaps recognizing that the RMBS market is on the verge of a significant downturn (as it turned out, it was), seizes the deal on largely these terms. But Paulson knows it still has some goodwill with Goldman. So it tries to leverage its relationship with Tourre to convince Goldman to write protection on 45%-100%, as originally contemplated. To get it done, perhaps Paulson agrees to pay some incremental premium to Goldman for retaining the slim 45%-50% supersenior tranchelet. Tourre works his magic with the risk managers and closes the deal.

May 31, 2007: Both supersenior deals close. Shortly thereafter, the RMBS market tanks hard, kicking off the financial crisis that would culminate over a year later with the collapse of Lehman Brothers.  But that’s another story.

Comments (View)

Waste Not

Baruch at Ultimi Barbarorum tried his hand at motivational speaking this weekend, imploring the econoblogosphere to get its act together on financial reform.   I applaud the effort, but as pep talks go, this one was a bit of downer:

With [the economy recovering and memories of the darkest days of the financial crisis receding] I fear some of the spice has gone out of the commentary… [T]he econoblogosphere is moving back to where it used to be, which is to cater to a niche… with a circumscribed influence.

… Political money and public ignorance has corrupted civic decency in the US to such an extent that doing the right thing for the Republic appears quite impossible.

All these things should make econo-bloggers all quiver with righteous anger, which we would communicate to our readers who would then rise up en masse against the legislators captured by Big Banking, and some form of actual reform might even take place.

…To be sure, there are brave souls who carry on the fight… But the rest of us seem to have stopped caring so much. The minutiae of practical policy is much less amusing than making lots of money in financial markets that really, truly appear to be on the mend.

…At its best, the econo-blogosphere can be the last haven of truly independent, non-captured, and crucially, informed, commentary able to affect policy and opinion makers positively. It used to do just that. It may not help in the end but let someone at least try.

Get your game back on, people. Get some fire in the belly again. A crisis is a terrible thing to waste, and it looks like we are on the verge of wasting ours.

It’s a powerful message—don’t waste our crisis!—and it’s been echoing around the econoblogosphere for months.  But I see things playing out differently.

We haven’t wasted the crisis because there was nothing to waste.   To be sure, the public is angry.  But it’s folly to have hoped that the public’s anger ever could have been channeled into grassroots support for meaningful reform.  The anger is too diffuse.  Worse, to attempt to corral this anger in service of reform is to risk having it turned against you by professional anger-wranglers employed by anti-reform opponents.  It’s too easy for these opponents to convince the ill-informed that government is the problem, and to tar you with the same broad brush in the process.  Like a loose fire hose, public anger is only useful if it can be controlled—otherwise it is unwieldy and potentially dangerous.

Some believe that it is the duty of We Finance Bloggers to inform the public—to walk them down the garden path with reasoned argument for which specific reforms  we need and why.  It is not.  The cynical reason is that meaningful reform is impenetrable to the public.  Voters don’t know the meaning of resolution authority—to take one example—and they never will.   Try as we might, we cannot shame lawmakers for kowtowing to the interests of those who profit from deregulation.  The less cynical reason is that it is not our job to shape minds in the public sphere.  Leave that job to the professionals—i.e. politicians, pundits, and PR flacks.  (Besides, our reach is too niche.)  Rather, it is our job to debate the merits of specific reforms, in the hopes that the right people are listening.

Which brings me to the good news.  From where I stand, and contrary to Baruch’s call-to-arms, the last two months have been the richest yet for wonkish policy discussion within this little econoblogosphere we inhabit.  Policy recommendations are getting more specific, debate is happening, and the intended audience is indeed taking notice.  Yes, there are some unpleasantries.  Fissures are forming in the ranks of those previously aligned on regulatory reform, as the merits of specific reforms are judged differently by the varied players in the technocracy.  And yes, it’s taken quite some time for the powers that be to get the ball rolling.  But that’s exactly what we should hope for and expect.

So while I’m not naïve to the realities of sausage-law-making, I’m sanguine that great things are happening in the ‘sphere and that we may yet see its collective fingerprint on a meaningful reform bill.

Keep it up.

Comments (View)

Those Greek Swaps

Heidi Moore has a great write-up of the Goldman-Greece swap story.  Do read it.  But don’t let her convince you that there’s nothing in this story to lend credible ammunition to the squid-eating Matt Taibbis of the world.  Moore lets Goldman and the rest of the banks off too easy.

But the swaps weren’t illegal, and Goldman was giving its client — Greece — what it asked for. The bank followed all the rules to do it. True, the swaps weren’t disclosed, but you can blame the Maastricht rules for that: No country has to disclose these swaps, and that’s partly why they’re so popular with politicians and also why you never hear about them in the popular media.

Since when is “following the rules” the standard of ethics against which we’re to judge bankers?  All manner of regulatory arbitrage schemes concocted and pitched by bankers were (and in many cases still are) legal, but that does not absolve them of blame.  And that’s exactly what these swaps are: an arbitrage of disclosure.   Greece borrowed money from Goldman in a clever manner that skirted the rules about disclosure.  Just like banks used structured credit products to create high-yield (i.e. risky) investments that were complicated enough that they could dupe rating agencies and regulators into thinking they were safe.

These schemes share a common thread: they facilitated risk-taking by obfuscating it.  Yes, we should blame the gamekeepers for designing too-flimsy rules or for poorly enforcing them.  But we should also blame the poachers for finding and profiting from the gaps in those rules.

Comments (View)

On the Impossibility of Measuring Model Risk

This week The Economist poked a little fun at the quants:

JPMorgan Chase holds $3 billion of “model-uncertainty reserves” to cover mishaps caused by quants who have been too clever by half. If you can make provisions for bad loans, why not bad maths too?

And in response to this revelation, Francine McKenna wondered how the auditors could have signed-off on the models:

If you need $3 billion of “model reserves” how [does] PwC attest to [the] models underlying valuations, estimates and reserves?

It’s worth noting that these model-uncertainty reserves not only comply with GAAP, but are mandated by it.   So in response to Ms. McKenna’s concern, there is in fact a “GAAP for that.”

FAS 157 Par. C16: This Statement clarifies that the measurements should be adjusted for risk, that is, the amount market participants would demand because of the risk (uncertainty) inherent in a particular valuation technique used to measure fair value (such as a pricing model) and/or the risk inherent in the inputs to the valuation technique (a risk premium notion). Accordingly, a measurement (for example, a “mark-to-model” measurement) that does not include an adjustment for risk would not represent a fair value measurement if market participants would include one in pricing the related asset or liability. [Emphasis mine.]

OK, so now we understand why banks have to measure model risk, but how do you do it?  Well, if you’re being honest, you don’t.  Model risk is impossible to measure.  Here’s why.

Pricing Models as Interpolation

First, it’s important to understand what a pricing model is and why they are used.  Pricing models are used for two purposes: valuation (that is, to come up with fair values for instruments that do not have directly observed prices—e.g. OTC derivatives) and risk management (that is, to measure the sensitivities of instruments to particular risks for the purpose of managing an overall book).  Let’s put aside for now the risk management purpose and focus on the valuation.

The majority of OTC positions are not “marked-to-model” in any meaningful sense of that term.   Yes, there are pricing models used to value them, but they’re not the scary kind of marks that skeptics rightly call “mark-to-make-believe.”   Most of the time, the pricing model is simply a fancy (and sometimes expensive) tool to interpolate between observed market prices.

Let’s say I have an interest rate swap.  I can observe the market prices (rates of various maturities) and as long as my swap is within the range of my observations, then my pricing model is calibrated to market.  The only modeling I’ve done is to build a rate curve based on observed inputs and used this curve to discount the contractual cash flows of the swap.  This is simply a robust way to interpolate the value of my swap from observed quotes on similar instruments (i.e. other swaps).  Now this is obviously a very simple example, but this model-as-interpolation view can also be said of more complicated, but traded, instruments like synthetic index CDOs.

What’s this have to do with model risk? When models are calibrated to observed market prices, and hence where the model is used an interpolation tool, the model risk is (pretty much) already captured by the model.  This is true even if the model is “wrong”.  If the model calibrates to market, it already reflects the market’s view of the model risk—at least with respect to the observed instruments to which it’s calibrated.  I should add that even if you’re interpolating between observed prices, you might have residual model risk—how much residual risk (which could be significant) depends on the granularity of observed data, among other things.

True Mark-to-Model Positions and Why Model Risk is Immeasurable

But wait.  If most positions are marked to prices interpolated between observed quotes, what about the rest?  Here’s where we get into the true mark-to-model issues, and where model risk is most prevalent.  Thankfully, these are easy enough to identify on a balance sheet.  They are anything noted as a “level 3” fair value measure—i.e. instruments where the value significantly depends on the model itself and on the unobservable inputs or parameters thereto.  Think of a CDO-squared or a bespoke synthetic CDO.

I promised I’d get to the point about the impossibility of measuring model risk, so here it is:

  1. Model risk is the risk that you’re using the wrong model.
  2. The space of possible models is infinite.  That is, there are an infinite number of models to choose from, including those not yet discovered.
  3. No one knows what the right model is.  If you knew which model was the right one, you’d already be using it.  Even if most market participants agree on a model today, they might discover a better model tomorrow, or simply decide that no model is sufficient to assess the risks (this has happened).
  4. Judgments about the amount of model risk are necessarily qualitative.  The best I could hope for would be to say that this model feels more certain than that one.
  5. Model risk is recursive.  Even if I could quantify the level of model risk, what model would I use to measure the impact of that model risk on fair value?  Where are the models of model risk?  Even if they existed, those model risk models have model risk, no?

That $3B Model-Uncertainty Reserve

If model risk is immeasurable, where did JPMorgan’s $3B come from and what does it mean?  As to where it came from, I don’t know the specifics, but I suspect they’ve either: (a) shocked the unobservable model inputs by some arbitrary amount and taken the worst of the lot or (b) run some “shadow models” (i.e. run the same positions through multiple known models) and taken the worst of the lot.  Either way, the result is arbitrary.  So as to what it means: not much.  At best, it gives us some insight into the subjective judgments of JPMorgan management with respect to the quality of their models.  So yeah, not much at all.

Comments (View)

Deposit Insurance and Moral Hazard

Over the weekend, I listened to Russ Roberts’ EconTalk interview with Charles Calomiris.  There’s far too much in this interview to unpack in a single post, so for now I only want to address Calomiris’ take on deposit insurance.  He’s against it, and pins deposit insurance and other risk subsidies as the primary cause of many recent financial crises.  I’m for it.

Calomiris’ broad claim is that government policies distort individual incentives to assess risk, and that these perverse incentives cause financial crises.  He looks at past financial crises—which he defines as domestic banking losses exceeding 10% of GDP—and finds that while there were only four such crises globally from 1874 to 1913, there were 140 of them from 1978 to the present.  So what happened? According to Calomiris, what caused the frequency of crises to spike was that governments began implementing policies to guarantee (or otherwise subsidize) certain types of risk-taking.  His thinking is the familiar cry of moral hazard: when governments bail-out banks and their depositors, individuals are more willing to take risks knowing they’ll be able to share losses with their fellow taxpayers.  The risks then build up in the banking sector and cause financial crises.

So which government policies are to blame, specifically?  “Most important” of these government policies, says Calomiris, is deposit insurance.  Deposit insurance, he argues, distorts depositors’ incentives to keep tabs on bank risk; and these perverse incentives are a necessary, if not sufficient, condition for financial crises.

Sandrew here.  I grant that it’s not crazy to cry moral hazard in the face of deposit insurance, but I’m not yet convinced it has much of an effect.  For one, the frequentist evidence feels a bit light.  But more damning is that it’s hard for me to imagine a significant population of depositors who are capable of assessing bank tail risks, even if so incented.  Lose your savings in a bank failure?  Tough cookies, Ethel, you knew the risks.  What, can’t you read a balance sheet?  On this issue, it’s fair to call me a paternalist.

Calomiris dismisses the view that risk-bearing depositors mightn’t be so great at assessing risk.  But interestingly, he later double-backs on himself when it comes to bank ownership.  Here he is on fragmented ownership:

We’ve designed these [banking] institutions, by regulation, to be relatively immune to corporate governance—there isn’t sufficient number of stockholders with large, concentrated shares that can really discipline management…. What we need is concentrated ownership… particularly [by] sophisticated institutional investors.

It’s important to recognize here that fragmented owners are still owners.  They have exactly the same economic incentives, in aggregate, as concentrated owners—the incentives are simply dispersed.  And fragmented owners still have the voting rights to kick-out management.  What Calomiris is pointing out is that small, unsophisticated investors either don’t know or don’t care enough to act in their own self-interest with respect to corporate governance.  And yet he puts endless faith in Ethel the Depositor to discipline bankers with respect to murky and difficult-to-comprehend tail risks.  Strange.

Deposit insurance makes sense to me.  The majority of depositors lack the wherewithal and/or information required to make informed decisions about the riskiness of their deposits.   So they pool their resources (taxes) and hire the FDIC to look after it on their behalf.  Since they wouldn’t be very good disciplinarians in the status quo ante, the moral hazard cost of this insurance strikes me as negligible.  A small price to pay for an end to depositor-driven bank runs, if you ask me.

Comments (View)

To Dos

I haven’t written anything here in a long while.  Sorry about that.  Thanks to some encouraging words from Mike at Rortybomb, I think I’ll hammer out a couple posts soon. In the meantime, I’m going to use this space to make note of the things I want to write about:

  1. On the Connection (or Lack Thereof) between TBTF and Systemic Risk
  2. On Moral Hazard and Deposit Insurance, Briefly
  3. My Own Personal Financial Innovation: Convertible Student Loans
  4. Hedging Is Evil: Part II
Comments (View)

Regulatory Capital Equivalence Principle

Caution: Though brief, this post will likely be impenetrable to anyone who doesn’t have a working understanding of regulatory capital requirements, securitization, and how rating agencies fit into both.

Here’s an idea for Basel III.  Next time we design new capital requirement rules, let’s try to put in the following fail-safe principle to curtail the regulatory arbitrage.

Regulatory Capital Equivalence Principle: If the sum of the capital charges on a group of assets come to X, then the capital charge on a securitization backed by those assets must also equal X.

Let’s break this down with an example.  Say I’m a bank and I have a $100 of loans on my books.  Suppose the risk-weighted capital charge on those loans is $8.  Now let’s say I securitize those loans into three tranches via a CDO thus:

Tranche A: $60 of AAA-rated CDO securities

Tranche B: $30 of BBB-rated CDO securities

Tranche C: $10 of unrated equity in the CDO

Let’s say I find a buyer for 50% of each tranche and retain the rest.  On balance, I’ve sold off exactly half the risk in this basket of loans and retained the rest.  So my capital charge on what I’ve kept should be half of the original capital charge [$8 x 1/2 = $4].  Simple.

Unfortunately, this isn’t how things work under Basel II. (If I’m getting this wrong, someone please correct me.)  Instead, regulators look at my new CDO securities and the ratings thereof to drive my capital charges.  Notice how most (60%) of the securities issued by the CDO are rated AAA.  AAA-rated securities backed by corporate credits get a pretty low risk weight under Basel II.  So my capital charge on the $30 of AAA-rated securities retained is, say, $0.50.  Now the other bits that I retained (the BBB and unrated securities) have a much higher risk weight, so for thsoe I have to hold $2 in capital.  That brings my total post-securitization capital charge under Basel II to $2.50.

But wait a minute.  My original capital charge on the loans was $8.  The CDO securitization chopped up the loans and sold-off exactly half the risk to other parties; the other half I retained.  Since I own the same proportion of each tranche (50% of each), I didn’t alter the overall credit risk characteristics of what I retained, except to reduce the size of the exposure from $100 to $50.  So I should have exactly half the capital charge ($4).  But under Basel II’s standardized approach, I get to reduce my capital charge to only $2.50.  In other words, regulatory arbitrage freed-up $1.50 in capital.

I’m not sure how prevalant this type of regulatory arbitrage is today, but I would hope a wise regulator would find a way to plug this gap.  It strikes me that one simple way to close-down this arbitrage opportunity would be to require that the total risk weight across all tranches of a securitization must equal the risk weight for the underlying assets backing the deal.  I recognize that it might be difficult to opperationalize, as you’d have to effectively allocate that risk weight across the different tranches and communicate that risk weight to all holders of the newly issued securities.  Perhaps that could be accomplished by tying a risk weight to each tranche of a structured product at issuance, similarly to how ratings are doled out.  Of course there are other challenges too, such as how to handle the changing risks weights as time elapses and the risks evolve.  But I don’t see those challeneges as fatal flaws in the general principle that capital charges should be substance-based and blind to the legal form in which the risks are conveyed from one party to another.

Comments (View)

Zombie Securitization Market (and other bad metaphors)

Securitization refuses to die.  Bloomberg is reporting that Morgan Stanley is planning to breathe new life into a recently downgraded CLO originally packaged by Goldman Sachs in 2007.  Some key details are lacking in the Bloomberg piece, but here’s what I can piece together from the article and some other digging.  I’ll do my best to put it in plain English so you can see how (I think) this works.

Quick Primer on Securitization

In 2007, Goldman securitized a basket of $502 million of secured corporate loans.   What I mean by “securized” is that Goldman sold the loans to a special-purpose legal entity called Greywolf CLO I, which financed this purchase by issuing $502 million in securities backed by those loans.  So basically, Goldman sold the loans to a special legal entity that it created (for a fee), which sold interests in those loans as an investment product to a bunch of investors (probably mostly institutional investors like pension funds and maybe some hedge funds).

Now, the securities issued by Greywolf are not all samey-samey (or “pari passu” in the jagon).  One (or more) batch of securities (“tranches” in the jargon) were designated to be first in line to absorb any losses on the loans (“subordinated” in the jargon) and hence more-risky.  This of course is by design.  Because of the presence of these more-risky subordinated securities, the rest of the securities are somewhat protected against those first losses (“senior” in the jargon), and hence are less risky.

How do we know how risky the senior and subordinated tranches are?  This is where the rating agency comes in.  For a fee, the agency will take a look at the portfolio of loans and the relative sizes of the tranches (that is, how much first loss protection is absorbed by the junior tranches), do some fancy math, and presto!  They have their ratings.  The senior tranche gets rated super-safe (“AAA”) and the subordinated stuff gets rated not-as-safe.  Now again, this is all by design.  Goldman, like the rating agency, has it’s own fancy mathematical models to figure out how big the subordinated tranches have to be in order to just barely get the top AAA rating for the senior tranche.  What’s more, they probably repeated the process on the subordinated tranches—slicing them up so that some bits are kinda-safe (“investment grade”) and the ugliest bits are absolute garbage (appropriately called “junk,” or in the more colorful language of this crisis “toxic waste”).  That part’s not so important right now.  What you need to know is that Goldman (for a fee) chopped-up a portfolio of corporate loans into more-safe and less-safe bits, and sold off those bits to investors in the form of rated (again for a fee) securities.  That’s securitization.

[NB: More specifically, it’s a CLO or Collateralized Loan Obligation, a kind of loan securitization.]

What Morgan Did

So far, we’ve only been addressing Goldman’s original securitization of the loans, which took place in 2007.  Fast forward to 2009.  In June of this year, the rating agency that had originally rated the senior tranche AAA back in 2007, reassessed this rating and downgraded the senior tranche to a lower rating (A3, which is still in the investment grade category, but much less desireable than AAA). Times have changed, and the securities that the agency ostensibly thought were super-safe in 2007 are now considered only sorta-safe.

Investors love AAA securities like Garfield loves lasagna.  They can’t get enough.  But the securities originally sold in 2007 are no longer AAA.  The lasagna went bad.  Enter Morgan Stanley.  Morgan sees in these Greywolf securities an opportunity to do something very similar to what Goldman did with the basket loans in the first place.  They can buy-up the spoiled senior securities originally issued by Greywolf, sell them to a special-purpose entity, and sell-off new tranches of securities backed by the once-AAA (now A3) Greywolf senior securities.  Just like what Goldman had done in 2007, they can do some math to find the magic level of subordination required to mint a good chunk of brand-new AAA securities out of a bigger chunk of now-lesser-rated old Greywolf securities.  That is, they take some of the spoiled lasagna, scrape off the bad layer and sell the rest of the layers as a good-as-new (if smaller) piece of pasta.  Oh, and they sell the bad layer too.  This is re-securitization.

Greywolf Resecuritization

Why Do I Care?

The Bloomberg article gives off the impression that Morgan’s “first of its kind” transaction to repackage an underperforming CLO security is something innovative.  In my opinion, it’s not.  Resecuritizing CLOs (or pieces thereof) is basically a copy-cat of the outstanding practice of resecuritizing other types of securitized products, such as mortgage securitizations (e.g. REMICs).  And resecuritization in general is basically a re-hash of the idea that gave rise to securitizing the original loans/mortgages/etc. in the first place.  There’s nothing groundbreaking here—at least not from what I read in the Bloomberg article.

What’s interesting to me about this story are two things.  For one, it tells me that investors still have an appetite for newly minted securitized products.  I had thought investors no longer trusted securitization.  The conventional wisdom for at least the past two years seemed to have been that the models used to structure the deals were broken and/or based on flawed assumptions, that the underlying assets were worse than advertised, and that the ratings agencies were either complicit in deception or simply incompetent.  I had thought that securitization—as a process by which risky loans could be classed-up into AAA bits, less-than-AAA bits, and dreck—was dead and burried; that we had seen the last such deal to be birthed and were simply waiting for the inevitable extinction of the species as deals were unwound or died of natural causes.  But this deal tells me there is some activity still going on and that investors are still willing to buy newly AAA-stamped products.  Is the trust back?

For two, the article is just an interesting story of Wall Street one-upsmanship.  Goldman created the structure in 2007, and Morgan subsequently pinched the restructuring job from under their noses when an opportunity arose after the June downgrade.  Morgan doesn’t have any special sauce here.  Anyone capable of doing a CLO deal could have done it.  And who knows; maybe others were trying?  But Morgan moved fastest.  I wonder if Goldman isn’t a bit embarrassed.

Comments (View)