IMF: Re-Remics are being used to resecuritize senior private-label mortgage-backed security (MBS) tranches that have been downgraded from their initial AAA levels. In a typical Re-Remic, a downgraded tranche is subdivided into a new AAA-rated senior tranche and a lower-rated mezzanine tranche (see figure). About $25 billion were issued during the first half of 2009, mostly against MBSs backed by prime mortgages. Given that most of the AAA privatelabel MBS tranches issued between 2005 and 2007 have been downgraded, the potential for this market to grow is substantial. However, although these transactions are playing a useful role in dealing with the overhang of legacy assets, they are partly driven by rating/regulatory arbitrage.

Re-Remic issuance is being driven by a number of factors, including the need to maintain the AAA ratings that many investors require to hold these securities. Maintaining AAA status canresult in substantial capital requirement reductions.For example, the new Basel II risk weight on a BB-rated tranche is 350 percent under the standardized approach, whereas it is 40 percent on an AAA-rated resecuritization. Also, for banks and insurers, big rating downgrades can trigger “other-than-temporary-impairments,” which have to be recognized immediately through the income statement. These consequences can be avoided by replacing the downgraded securities with new AAA-rated Re-Remics. In the figure, the new AAA-rated senior tranche comprises 70 percent of the structure, with a mezzanine tranche that absorbs the first 30 percent of losses. Additional credit enhancement is provided by an option for the new senior tranche to be resubdivided into two “exchange classes” should it lose its AAA rating. Also, there is a hedge fund demand for the mezzanine tranches as a means to take a leveraged credit bet. The holder of the senior tranche that was downgraded to BB could then hold the new AAA tranche, and sell the mezzanine tranche to an investor desiring distressed securities. Hence, only 30 percent of the original holding is sold at distress prices, and the risk-weighted par value of the holding goes from 350 to 28 percent (70 percent of 40 percent). Even if the bank were to retain the mezzanine tranche, the riskweighted par value could still be less than the original 350 percent.
For example, for single security-backed Re-Remics, the default probability-based rating methodologies used by DBRS, Fitch, and S&P will typically pass the underlying bond’s rating through to the new mezzanine tranche (emphasis mine). Hence, in the example transaction, the total riskweighted par value would decline from 350 to 223 percent (70 percent of 40 percent on the AAA-rated tranche plus 30 percent of 650 percent on the BB-rated tranche).¹ In this regard, it is notable that Moody’s has been virtually shut out of the Re-Remic rating business, possibly because it rates on the basis of expected loss, which is tougher on mezzanine tranches than the default probability basis (Fender and Kiff, 2005), and thus issuers prefer not to have Moody’s rate their potential securitization.

Although Re-Remics and similar repackaging transactions are playing useful roles in dealing with the legacy asset overhang, they also serve to illustrate the vulnerability of ratings-based regulations to gaming and shopping. Also, these new securities remain exposed to further downgrades if economic and housing market conditions worsen. However, the information underpinning these securitizations and the methodologies applied to their ratings are likely more robust than before and thus pricing is likely to reflect risks more appropriately.

Source: Restarting Securitization Markets: Policy Proposals and Pitfalls, IMF

¹ The new risk weights would be even lower if they were calculated with the securitization exposure weights (20 and 350 percent, respectively, on the AAA and BB tranches), rather than the resecuritization exposure weights (40 and 650 percent). The Basel Committee has defined a resecuritization as a securitization where “at least one of the underlying exposures is a securitization exposure” (BCBS, 2009), but some market participants are hopeful that single-security repacks may not be considered resecuritizations (Mayer Brown, 2009).

Richard Thaler is quoted in the latest issue of The Economist saying the following about the EMH:

The [Efficient Market Hypothesis] has two parts. The “no-free-lunch part and the price-is-right part, and if anything the first part has been strengthened as we have learned that some investment strategies are riskier than they look and it really is difficult to beat the market.” The idea that the market price is the right price, however, has been badly dented.

If I understand the paper Efficient market hypothesis and forecasting by Timmermann and Granger correctly, then the hypothesis does not have two parts but there are actually two completely different versions. The first one talks about informational efficiency, which is about forecasting, and the second one is built on the notion that the prices should reflect intrinsic values.

The first version basically says that when you trade on the basis of publicly available data you can only expect to earn the risk-adjusted return. This does not rule out fat tails, stochastic volatility or even bubbles. More technically:

.

E[Q_t+1R_t+1 | Ω_t] = 0

.
where Q_t+1 is a discount factor (containing investors preferences/risk aversion) and Ω_t is a given information set. As you can see, we are only talking about the conditional expectation and not about the conditional variance, higher moments, or the shape of the return distribution. The message is that just because their has been a market crash that doesn't mean the EMH is wrong. [Actually, is there a reason why the intrinsic value of companies should not drop by half?]

The other question is what investors actually do with the information set at any give point of time. Was the information set we had in 2007 already screaming that house prices will crash and investors didn't make use of this information or did the extent of the problem only gradually build up in the information set?

The only thing I know for sure is that the more I think about this the less I think I know what I'm talking about. The good news is that I don't think I'm alone!

Today I was going through some past editions of The Journal of Portfolio Management and stumbled across a special edition in memoriam of Fischer Black (1997).

In Remembering Fischer Black Jack Treynor writes:

It's easy to forget that over Fischer's career finance changed:

• From a verbal to a mathematical discipline.
• From accounting-centered to economics-centered.
• From suppressing uncertainty to giving it a central role.

Thirty years ago, nobody would have defined finance as the economics of uncertainty. Risk was a cop-out--for explaining why the future had departed from deterministic forecasts. Before the revolution, we regarded randomly fluctuating markets as evidence of irrationality of market prices. Now we view departures from random fluctuations as "anomalies."

In Fischer Black: Some Personal Memories Jonathan Ingersoll writes:

I first met Fischer Black in the summer of 1975 on the day I presented a paper to the faculity of the Graduate School of Business at the University of Chicago--an intimitating place to give a seminar. The people there are always prepared and ready to question your assumptions, your methodology, and your findings.

The paper was a theoretical and empirical analysis of dual-purpose funds using the Black-Scholes option pricing model. Both Myron and Fischer were then on the faculty at Chicago and in the audience. Fischer asked me how I might incorporate stochastic volatility into the model I'd developed.
It was a question for which I was not prepared. The Black-Scholes model did not have stochastic volatility. I had no idea how it might be incorporated into a theoretical pricing model. And the Black-Scholes model itself was still just a fledgling. It had been tested by the authors on OTC put and call data, but not yet empirically applied to any other type of claims. I'm sorry to say, Fischer, I had no answer for you that day in 1975. [...] Since the crash of 1987 and the subsequent growth in implied volatility smiles, the study of stochastic volatility has become something of a cottage industry among both academics and street practitioners who follow derivatives. But Fischer was there long before other were even aware that it was an issue.

This leaves me with two questions:

1. How would developed economies look like today in case no financial innovation would have been made over the last 50 years.
2. Could there be a severe economic crisis without any of those financial innovations (i.e. with consumers still levering up and believing that house prices will ever increase, greedy managers maximising short-term profits, etc.)

It's amazing that--given the current crisis--nobody thought it would be worth to come up with a general defense of financial innovations so far. Just to give an example: Most quants knew about the shortcomings of a given theoretical model. But given those limitations, they at least had a certain pricing tool and started to offer companies who did not want to take a directional risk the possibility to hedge* almost any exposure at more or less reasonable prices.

* ad counterparty risk: No, I don't think that in the larger picture AIG or LEH play such a big role. And lessons learned.

Bill Luby writes (Barron's):

The 30-day VIX time horizon primarily captures what I call "event volatility" that is associated with scheduled events such as important economic data releases, earnings reports, as well as various scenarios associated with high-profile geopolitical and other events which are likely to cast a shadow over the next month.

By comparison, the 93-day time horizon of VXV guarantees it will encompass an entire earnings cycle, a full quarter of the economic-data-release cycle and two FOMC meetings. The VXV is more focused on long-term systemic threats, measuring something akin to "structural volatility" in the markets.

VIX vs. VXV

Investors who trade based on volatility signals usually look for signs that implied volatility is overextended to the upside or downside. This is traditionally done by comparing implied volatility measures such as the VIX to recent historical volatility or to previous implied volatility levels for the same implied volatility measure.

Since the launch of VXV in November 2007, investors have been able to use a third approach: comparing two different volatility indices with different fixed-time horizons to determine the extent to which short-term volatility expectations in the form of the VIX are aligned with longer-term volatility expectations in the form of the VXV.

related items:
VIX:VXV Ratio Sell/Short Signal, Bill Luby
Who is Selling Wholesale Vol and Why?, Zerohedge

Lusardi, Tufano and TNS Global asked 1000 U.S. residents the following question:

You purchase an appliance which costs $1,000. To pay for this appliance, you are given the following two options:

a) Pay 12 monthly installments of $100 each;
b) Borrow at a 20% annual interest rate and pay back $1,200 a year from now.

Which is the more advantageous offer?

Result:

Here are all the results from the survey. Unbelievable. The questioning was done over the phone which could distort the results a bit but it's nevertheless a disaster.

It's amazing how many crappy stories you find these days on the EMH. Guys, go and read

Efficient market hypothesis and forecasting, Allan Timmermann, Clive W.J. Granger, International Journal of Forecasting 20 (2004) 15--27

and help Clive Granger stop spinning like a freaking pulsar in his grave.

Does anybody know why Wolfram|Alpha is calculating overlapping returns?

Two good slides from Axel Leijonhufvud's presentation at the 37th Economics conference organized by the Austrian National Bank:

A fallacy of composition

• "Do not put all eggs in same basket" is a good rule for a bank
• If everyone made safer by diversification, surely the economy is then safer?
• WRONG
• With (almost) everyone operating in (almost) all markets, the connectivity of the network has qualitatively changed

Herd behaviour

• Hard to opt out of process... also for those who realize risk is increasing
• Loan officer who does not lend, risk manager who does not play along, banker whose branch is not "doing enough business", hedge fund operating with less leverage than the competition... not likely to last

As you can imagine, Leijonhufvud and many other panelists mentioned the high level of leverage in the system. Though most academics know that one has to differentiate between direct leverage (loans by banks and prime brokers) and indirekt/embedded leverage (derivatives), they somehow always forget to keep in mind that the risk involved with leverage depends on the risk of the levered asset and that a stand alone leverage ratio doesn't provide any useful input. During my time as a hedge fund analyst I monitored market neutral funds who where 20 times levered but half as risky as hedge funds who where only 2 times levered but had all their money in options. I really hope regulators keep this in mind.

related items:
IFRS vs US-GAAP: European Banks Leverage Overstated, Alea

Bloomberg writes:

[Carry trades] produced average annual returns of 21 percent in the 1980s with no down years, the best of four commonly used currency strategies, according to ABN Amro Holding NV indexes.

In the 1990s, carry-trade investors suffered three down years, including a 54 percent slide in 1992, ABN Amro data compiled by Bloomberg show.

From 2000 to 2005, the trade was again on top with average gains of 16 percent. Then it dropped three years in a row in 2006-08, the longest streak since 1976-78, for an annualized average loss of 16.5 percent through Feb. 28. <> Last month, the carry trade roared back.

Let's go and pay a HF manger 2/20 for a trade your grandma could put on and will kill you once every ten years.

related items:
Biggest Money in Currencies is Made Selling Options, Bloomberg

In a recent Night Talk on Bloomberg interview (via MR) Robert Engle said:

"[The ARCH model] is the key ingredient for short-run risk forecasting for financial markets. <> We know some things about financial markets, which is if they are really volatile today, they are likely to be pretty risky tomorrow. And that kind of information allows you to update every day of what the risk is of a particular position."

For Non-Quants: Imagine you have an estimate of the daily volatility. Could be the squared daily return or a volatility measure based on intraday-data or ... Now for calculating tomorrow's risk, you could just take an n-day average of your daily volatility. Something a little bit more sophisticated would be a weighted average where you'd put more weight on recent data points (today, yesterday) and less weight on past data points. With an ARCH model, you actually estimate those weights, i.e. "optimize" them by looking at past data.

Methinks the problem with ARCH models is that you know what to expect (like when choosing between a 5-day or 10-day moving average). A.) Is your volatility forecast just based on the last couple of days, it will be highly erratic. It will do fine when there are successive days of either only large moves or only small moves (volatility clustering) but deliver many wrong signals in case big moves are more or less evenly dispersed between average and small moves. B.) When your volatility forecast is based on too much data it will be slow to react to sharp increases or decreases in daily volatility but on the other hand don't deliver a wrong signal in case there is only a short-lived correction or inaction.

I ran an ARCH(5) regression and an ARCH(10) regression on the last 10 years of S&P 500 returns. According to standard information criteria, the optimal lag-length is actually 5. But information criteria didn't change significantly when changing the lag-length and I can imagine that they depend more on the data window. The plots cover the last 200 trading days. The estimated (daily) standard deviation from the ARCH models was multiplied by sqrt(265) to annualize it:




Anybody with practical experience around?

R: A Language and Environment for Statistical Computing, R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria.

PS: You actually won't believe the coefficients for the ARCH(5) model (cross checked with a different package):