Linus Wilson claims Treasury Accepts a Lowball Price for TARP Warrants
I read the paper quickly and it seems Linus is making some pretty wild and laughable volatility assumptions.
The warrants have 9.6 years left to expiration. The paper says: “The low end volatility is the annualized historic volatility. This is obtained from calculating the daily instantaneous returns from January 1, 2009, to May 8, 2009….”
A little over 4 months of data, drawn from a period of unusually high stressed market conditions to evaluate a 9.6 years warrant?
As an update from the initial Linus Wilson report, you may be interested in knowing that our analysis shows that, with the exception of Old National Bancorp, bank warrant repurchases have been made at close to “fair value” prices.
Pluris Valuation Advisors has just completed a study valuing the warrants of all 265 public banks participating in TARP using data accumulated from several years worth of TARP transactions.
A copy of the study is available here: http://www.plurisvaluation.com/site/liquistat.html#2.
You can also download a white paper on valuing warrants here: http://www.plurisvaluation.com/site/pressroom/whitepapers.html.
Espen Robak, President
Addendum: the haircuts depend on the time left, the vol, and (most important) the moneyness.
The main reason why illiquid (untraded) warrants are worth less than their Black-Scholes values is their illiquidity. And the TARP warrants are illiquid even if they’re registered for resale. The registration does not create an active market for these instruments, or any market at all. So these are “tradable” only in private transactions no matter what. The B-S model was never designed for valuing warrants; it applies to continuously traded options on continuously traded shares only (otherwise, the hedging “delta” assumptions don’t hold). Non-tradable illiquid warrants generally sell in private transactions at 20-80% haircuts from the B-S formula value. See http://www.plurisvaluation.com/site/LiquiStat-WP.pdf
Linus:
THX for your comments. I still think your low estimate as high given that using the recent historical record results in an upwards bias due to the unusual circumstances seen lately.
The 37.10% vol drops to 34% over 10 years and 30% over 15 years still including the recent period, and ex that recent period, you would be challenged to find any 10 year period over the mid to high 20s.
I agree your estimate gives the Treasury the benefit of the doubt and that a market derived price would be much better.
I don’t use the conventional model as I don’t think it works for long dated options/warrants first because it assumes the bank will survive to the term with 100% probability, something that is unrealistic, and second in the real world, arbitrage dictates the use of a “risk-free” rate where you can borrow and lend so swap rates are what I would use not treasuries.
Regarding the Bloomberg article, not very clear what the method/assumptions are except for volatility: “On May 11, the day the U.S. announced the sale, the stock’s option-implied volatility, derived from market prices of stock options that are traded daily, was 61 percent, according to data compiled by Bloomberg.”
I suspect the writer is using very short-term options to evaluate volatility and that doesn’t make any sense.
There was a typo that was repeated in the descriptions of the tables 2, 3, and 4. That typo has been corrected now at http://ssrn.com/abstract=1404069. Nevertheless, my paper correctly said on page 6 of the first and current draft,
“The low-end estimate of volatility was 37.10 percent. It was derived from daily closing stock prices and quarterly dividends paid from January 1, 2002, to May 8, 2009.”
Thus, there was over 7 years of historic prices behind that estimate. I have updated the description of the tables so there will no longer be any confusion about the time interval used to generate the 37.10 percent number.
I never used the historic volatility from January 1, 2009, to May 8, 2009. If I had done so, I would have found that the historic volatility over that period was 87.74 percent. That is higher than the high-end estimate of volatility that I used of 72.89 percent in the paper. I think my estimates gave the U.S. Treasury the benefit of the doubt. For example, my estimates are also much lower than Bloomberg’s at http://www.bloomberg.com/apps/news?pid=newsarchive&sid=ae2fQFMrDer4
Ultimately, I would like private investors decide what the warrants are worth. It is private investors who determine what the fair market price is. I don’t think the TARP warrants should be an exception.
Sandrew:
I misspoke, the bank exercise rights are for the preferreds, not the warrants.
Linus:
I think we have had an unusual period of high volatility for a while and I don’t expect that to last 10 years or at least very long shot that it will. But basically agree that a market price or public auction would be better than behind closed doors negotiation to set the price. Also the warrants can hardly be said to have been issued at a fair price, at least 0 isn’t a fair price for me.
“the banks have the exercise rights”
Huh?
Dear Alea,
I will look for the typo. I calculated daily instantaneous returns from January 1, 2008 to May 8, 2009 for the high estimate. For the low estimate, I calculated daily instantaneous returns from January 1, 2002 to May 8, 2009. If you don’t like my historic volatility estimates, you could always choose another time period. The methodology for calculating historic returns is explained in more detail in my paper “The Goldman Sachs Warrants” at http://ssrn.com/abstract=1400995. Perhaps you could calculate a different time period that would give you a lower historic volatility estimate. I am always open to suggestions about how to obtain better estimates.
CapVandal:
“I’m not sure that Black Scholes gives reasonable answers for long term options”
Pretty unlikely to give the right answer, specially since the seller (bank) has the exercise right.
I’m not sure that Black Scholes gives reasonable answers for long term options. Elizabeth Warren has already claimed that the preferred purchases were significantly below “market value.” Getting paid back at par seems like it would be a good outcome, unless they are revising their estimate of the value of these.
More importantly, the point of the entire exercise was to stabilize banks — not extract market like prices when the capital markets weren’t willing to put anything into banks.
I’ve seen this line of reasoning more than once:
“Miller, the North Carolina congressman, said the Treasury should have insisted on terms for taxpayers similar to those Warren Buffett secured for Berkshire Hathaway Inc. shareholders when he invested $5 billion in Goldman Sachs in September.”
This fails to recognize that the Treasury gets 1/3 of future profits in perpetuity from the recipients. Not to mention the knock on effect of the “improvement” in the entire economy based on the capital support. The treasury gets roughly 20% of incremental GDP.
The usual suspects are certain that when the banks talk they are lying and when they are quiet they are stealing. Overall, the net spreads on the preferreds should give the government a decent spread plus profits from the warrants. I suppose if you count the expected losses on C, AiG, Lehman and the FDIC, the total losses will be a couple percent of GDP — consistent with historical, post war bank failures in mature economies.
It might be a typo. Still there are many other dubious assumptions. Treasury didn’t pay a fair price for the warrant, they were extorted for free under duress, remember not all banks wanted or needed to be in this thing.
More importantly the banks have the exercise rights which is not bullish for the warrant price. Ideally, a public auction would have been a good idea to get a price although I doubt it would be much different than what Treasury got so far, maybe even less. The banks are eager to get out, that they would pay more than a rational investor is entirely possible.
I too skimmed it, perhaps a bit closer than you. What you’ve quoted (from the footnote to Table 2, 3 and 4) looks like a typo.
From the body:
“The low-end estimate of volatility was 37.10 percent. It was derived from daily closing stock prices and quarterly dividends paid from January 1, 2002, to May 8, 2009.”
That’s a 7.4-year look-back–2 years shy of the term of the warrant, but I don’t see why they need match. Plus, a hvol with a 9.6-year window might look even higher than 37%, given what I recall to be a period of high volatility in 2000-2002.
In any event, a long-term vol of 37% doesn’t strike me as high, given where the short-term calls were trading.
This doesn’t taste like snake-oil to me, but the conclusion is pretty weak anyway. The government accepted a bid that might be low by a few hundred thousand bucks. Big whoop.