Byrne's Eye View

When the first history of the subprime mess is written, it will relegate C-BASS to a footnote, but eventually, we'll see it as iconic -- the WebVan of 2007. The brief history is this: the company was set up to help MGIC and Radian securitize some of their assets. C-BASS did about as well as you'd expect for a while, because the loan-and-mortgage slicing-and-dicing business was doing very well, indeed. And then a few structured products fell apart, and there really wasn't a market for C-BASS's assets (which put it in roughly the same position as a bank that has accidentally foreclosed on every single borrower -- in a new business they didn't understand). C-BASS managed to scuttle a merger between its parents, since neither side could figure out which was more to blame for the disaster.

And now, the latest: in February, C-BASS bid way, way above the market price for Fieldstone, a company in the mortgage origination business. Now, in November, Fieldstone is bankrupt.

In just a few years, C-BASS has ruined two parent companies and one subsidiary, costing the financial markets many times its own net worth in lower asset values. I'm impressed

Suppose a gang of Radicals for Capitalism takes control of the US in a bloodless coup. To maintain their popularity (and their libertarian street cred), they agree to only pass laws that correct fallacious actions in others. And even then, all they really do is enforce textbook economic thinking. For example, this junta (whose membership probably overlaps a lot with the Junto) wouldn't be able to declare abortion permanently legal or permanently criminal, but they would be able to prevent a politician who considered abortion wrong in speeches to consider it proper in practice.

A few specific proposals they might endorse:

• A common economists' bugaboo is the ignorance of opportunity costs. Someone earning $10 an hour, but willing to waste an hour to save $5, is 1) $5 poorer than he could be, and 2) $5 poorer in such a way that society is $10 poorer, because it's out an hour of his labor even though he's ahead $5 in savings. To avoid this, reverse the compensation equation: instead of paying someone $10 per hour of work, pay them $240 per day of existence, and charge them $10 per hour of non-work. Suddenly one hour to save $5 is a $5 net loss for the time-waster, too.


• To prevent extravagant government promises, the government would be forced to sell fractional shares of future tax revenues, and fractional shares of future welfare benefits (auctioning off 1% of 2015's income tax revenue and 1% of 2015's social security benefits, for example), and to avoid higher expenditures or lower taxes unless assets exceeded liabilities.


• Robin Hanson would probably insist that pundits preface pontifications with some kind of win-loss record showing how accurate their predictions were compared to 1) expert opinion, 2) popular opinion, and 3) the actual outcome, weighted by how much the pundit bet on each outcome.

I'm almost afraid to add this to my wishlist.

Especially since I've heard rumors that there's a scanned copy floating around online somewhere.

The WSJ reports possibly imaginary rumors that Warren Buffett is buying or reinsuring bond insurers. I have no idea if it's true or not -- if there were a prediction market, I'd be selling this at 10%, buying at 1%, and watching with interest in between.

But what's annoying about this is that the article's argument is, roughly, 1) Warren Buffett buys things nobody else is willing to buy, and 2) nobody else is willing to buy bond insurers. I can think of a few situations over the last few years when nobody -- including Buffett -- was buying. Airlines, Internet Incubators, Enron, Worldcom, Tyco, steel, homebuilders, subprime lenders, etc. And in none of those situations did Buffett end up buying so much as a single share.

This kind of article is a cheap shot -- or a free call. Nobody is going to remember someone for being the 99th person to say "Buffett is buying!" and be wrong, but almost any financial journalist would love to be the one in a hundred who says that and happens to be right.

Back when efficient markets were a little more voguish, one theory was that all returns were compensation for risk: that if you wanted to get rich, all you had to do was accept higher beta, and if you wanted stability, you'd do well to forget about getting paid for it. That theory bumped into the inconvenient fact that people who just buy volatile stocks don't do especially well.

Let's say an investor wants to mimic the performance of the S&P 500, using one of three baskets of stocks. Basket "A" is a group that roughly approximates the S&P. Basket "B" correlates with the S&P, but is only half as volatile. Basket "C" correlates, but is twice as volatile. Now, our lucky investor can choose all kinds of permutations that will mimic the S&P. 100% of assets in "A", 200% of equity in "B" paid for with debt, or 50% cash and 50% "C" would all do it. The only real difference is that the second two choices have to be rebalanced to maintain the ratio (if, for example, the S&P drops 50%, portfolio "C" will be 67% cash and 33% S&P, and will thus be 67% as volatile instead of 100%).

The only implication for investors is that a basket of stocks that mimics the S&P is likely to be less liquid (because people aren't constantly trading to rebalance). This, of course, is completely wrong, since S&P stocks are substantially more liquid. So much for that theory, too.

A sub-implication is that low-volatility stocks should have more 'momentum', since as they go up, an investor would have to add to the portfolio to keep the proper equity/assets balance. This implication is also belied by the data, which tend to show (in my experience) that trend-driven stocks are either a) companies with some amazing, possibly imaginary new product, or b) natural resource companies formed to exploit some amazing, possibly fictional mineral deposit.

I think I'll leave the quanting to quants and stick with value investing.

Every few years, someone will publish a new study showing that a very simple, automatic investment strategy outperforms the indicies (and, by extension, most institutions). Usually it's something like "Buy the bottom decile of price/book value stocks, and short the top quartile." These strategies have been eerily successful over the last fifty years: apparently accountants following simple depreciation schedules can actually outperform your average Harvard MBA.[1]

But every automated strategy relies on taking a great number of different items (e.g. mortgages, equities), and treating them as more or less identical. It's true that, all else being equal, a large enough pool of any product is going to look just like the product as a whole. But all else isn't equal, because by treating securities that way, investors make themselves suckers -- CDOs and index funds are the buyers of first resort for junk from more analysis-oriented portfolios, so they end up disproportionately short the good stuff and long the dreck.

But that's something you can take advantage of with -- you guessed it! -- an automated strategy. Here's what I would suggest: at the beginning of the year, divide stocks into deciles based on P/E, P/cash flow, P/book, market cap, etc. Then, within each decile, short the top percentile and buy the bottom percentile. So instead of buying the bottom 10% of stocks and shorting the top 10% of stocks by P/B, you'd buy the bottom 1%, short the bottom 9-10%, buy the bottom 10-11%, etc.[2] Assuming the relationship between deciles holds between percentiles, the best outcome is outperformance, and the worst is a hedge against the failure of these know-nothing strategies.

[1] Read that whole talk. It's great.

[2] When this stops working, you can trade the mill-iles. But I'm pretty sure there aren't enough stocks to take this any further.

Bloomberg says SIVs may be over, forever. That because asset-backed paper was backed by bad assets this year, that kind of capital structure will never, ever be used again.

I'm guessing this is what people said about high-yield bonds financing LBOs in about 1991.

I just got some pretty standard spam: someone gave me bad feedback on eBay (which I haven't used since I tried to sell my Social Security benefits a few years ago), and of course they want me to go to a website that looks like -- but isn't -- ebay.com, so I can resolve it.

The nice part is the footer of the email: it's a link to eBay's guide to detecting spoofed emails. The link, of course, goes to a .ru domain that tries to steal my eBay password. These phishers have a sense of humor

When I hear a typical fund manager describe his strategy, it goes something like this: "First, we figure out a basic thesis, like 'Interest rates are too low!' or 'Demographic changes in California will raise the value of west coast real estate!' and then we try to get exposure to that by shorting overvalued banks or buying cheap REITS and construction companies." This sounds like typical investment patter -- the kind of thing you might hear from a Soros-level trader or just another hack -- but it's absolutely, terribly misguided.

It's wrong because it isn't a strategy -- it's two strategies. One is very broad Zeitgeist Arbitrage: speculating on how wildly our broad assumptions differ from the facts. But another is Nano Arbitrage: taking advantage of discrepancies between the prices of individual securities (like two banks with different P/E ratios, or two homebuilders with different backlog/revenue ratios). It's strange to claim that both levels of arbitrage are necessary in the same strategy, especially since they require such different skills.

If a manager is really good at figuring out major trends, like which way oil, interest rates, or real estate will go in the next twelve months, why not make a straight bet on them and ignore the other details? What is a smart macroeconomist doing puzzling over 10-Qs? Similarly, if an investor's real talent is in finding disrepancies in valuations, why kill their Sharpe ratio by mixing in macro bets when all it takes is some short-selling and leverage to trade exclusively based on relative value?

Macro investors who are genuinely good at what they do should spend their time coming up with a macro thesis, and invest in it by looking for high-delta opportunities. Usually, that will mean buying directly into whatever variable they're trading (using interet rate derivatives to speculate on interest rates, instead of doing so by proxy with financial companies). An ostensibly macro fund shouldn't have a portfolio like this.[1] There's a promiscuous mixing of skills and strategies here, and the only way I can explain it is that it's a way to trade returns for conventionality -- which has always been a losing trade.

[1] I'm pretty sure I know what they're doing, here: if you want, you can decompose an oil company into a set of call options on oil, where the marginal cost is the strike price and the cost of the option is the fixed cost of producing oil. But that takes one Excel spreadsheet too many -- if these companies are actually undervalued oil assets, why not buy the stock, short the oil, pocket the difference, and forget about the macro thesis?

When a supermodel starts playing the currencies market specifically to bet against the US dollar, it sounds to me like the crash has gone too far.

Never thought I'd see a speculative attack on Intrade as it happens, but here it is:

As of now, Ron Paul is third on Intrade's Republican nomination market, ahead of John McCain and Fred Thompson. But looking at the chart shows a conspicuous number of rapid purchases, just like every other speculative attack. Give it a little while, and this should collapse back to normal. After all, the only new Paul news is bad news.

Alex Kates writes that Facebook applications suck because the only really popular ones are 1) awful, and 2) viral.

We use 'viral' too often to remember that it's not necessarily a good thing. A 'viral' video is, obviously, a video that people see and want to forward to their friends, who see it and want to forward it, etc. But any virus will slough off features that aren't necessary to getting passed on -- which is what has happened to viral Facebook apps. They aren't 'viral' anything -- they're just pure viruses. The most common I've seen are 'vampire', 'zombie,' and 'werewolf' applications, with which a user 1) adds the application, and 2) tries to get other users to add it, too. It's okay that these things exist, even if they're bothersome to ignore, but I'm going to request some honesty:

Someone needs to create a Facebook AIDS application. Add it, and you're Facebook-AIDS positive! Try to infect all of your friends!

A while ago, I wrote about the dangers of 'recursive' markets: imagine a betting market that quantifies someone's trustworthiness. Now imagine a dispute that gets resolved by comparing someone's trustworthiness numbers. And now imagine how the results of that dispute affect the market. Suddenly, having a low trust-metric lowers one's trust metric, and vice-versa, to the point that noisy markets or a large bid-ask spread can turn someone into a presumed liar just because past market moves correlate with future changes in market fundamentals.

It happened. There's a rumor that a major presidential candidate has done something sexually scandalous (Dennis Mangan wonders what it could possibly be). James Miller suggests that it's Obama, because his estimated odds of winning are lower than one would expect. In other words, a partial reputation futures market is being used to determine a candidate's actual reputation: Obama must be doing poorly in real life because he's doing poorly in the market -- thus he should do worse (and thus, and thus, etc.).

Prediction markets are strange: they make market inefficiencies more efficient.

Dilbert author Scott Adams has suggested a giant externality-generating conspiracy in the course of trying to defend his blogging hobby as a business.

A few years ago I tried an experiment where I put the entire text of my book, "God's Debris," on the Internet for free, after sales of the hard copy and its sequel, "The Religion War" slowed. My hope was that the people who liked the free e-book would buy the sequel. According to my fan mail, people loved the free book. I know they loved it because they emailed to ask when the sequel would also be available for free. For readers of my non-Dilbert books, I inadvertently set the market value for my work at zero. Oops.

So I've been watching with great interest as the band "Radiohead" pursues its experiment with pay-what-you-want downloads on the Internet. In the near term, the goodwill has inspired lots of people to pay. But I suspect many of them are placing a bet that paying a few bucks now will inspire all of their favorite bands to offer similar deals. That's when the market value of music will approach zero.

Is that really what he suspects? That there's a massive, uncoordinated conspiracy to temporarily overpay for a product to radically increase production, causing a glut that cuts prices further? Discussion questions:

1. Is this the most parsimonious explanation?


2. Is it plausible that this exists as a conspiracy?


3. Is it plausible that this exists as a random phenomenon?


4. If this kind of thing works, why not try it with, say, oil? Or, if you want to be all economical about it: why doesn't this work with any product that, like recorded music, has a very high fixed cost and a very low marginal cost? Does this explain overpaying for semiconductors in the late 90's to ensure lots of below-capacity foundries in the 2000's?


5. And anyway, where does he get his data? The only sales figures I've heard for the new Radiohead album are imaginary, so what makes him think it's not like any other easily-downloaded album?

Radiohead is not revolutionary: they're just admitting that the Internet has changed the music industry; Scott Adams is not insightful: he's come up with an implausible hypothesis to explain imaginary facts. Stick to drawing comics, monkey-brain!