The TARP, or Troubled Asset Relief Program: remember that bailout, the very first in an increasingly ugly family of bastard children the new recession left at our doorsteps? The Sunlight Foundation, a nonprofit that “uses the revolutionary power of the Internet to make information about Congress and the federal government more meaningfully accessible to citizens,” has unearthed new details about the program’s birth and entry into the world. It’s reporting that—gird yourself for a moment—the language that set the price under which the federal Treasury would get warrants, exchangeable for common stock in bank companies, changed. And that might have cost taxpayers big time.

Warrants allow the government, at some point in the future, to buy stock in a company at a price that’s set in advance. The idea was that the value of these banks would go up over time; the government could buy stock years from now at the pre-set lower price, sell them back into the market eventually (or right away), and pocket the difference, making the initial investment worthwhile.

What happened here was that Treasury originally said that this pre-set price, often called the strike price, would be calculated based on the price of the stock over the course of 20 days before A) the government’s investment was made. Then, that changed, so that the price would be derived from the price of the stock over the course of 20 days before B) the company’s application to participate in TARP was approved by Treasury.

Why does this matter? Two reasons: First, B) the date of approval is always earlier than A) the actual date of the investment. And second, bank stocks have been generally (and definitely were over the autumn, if you didn’t notice) falling.

See this 8-K filing (a “current event” report on major items of shareholder interest), by major institutional banker State Street Corporation to show how taxpayers could have gotten burned. It said it got Treasury approval to participate in the Capital Purchase Program on Oct. 13. But Treasury says, in this report, that the transaction was only completed on Oct. 28—more than two weeks later. In the meantime, the strike price of the warrants increased from $44.25 if calculated via the original method to $53.80 when calculated by the new method, a 17.8 percent difference. (See our calculations of the stock price in the different periods using Google Finance data here: These match the Sunlight Foundation’s calculations.) In other words, we earned the right to buy stock at the old, inflated prices, rather than the new, lower prices, as was originally intended.

With all banks, the government earned the right to buy 15 percent of the preferred stock investment in common stock. State Street’s issuance of $2 billion in preferred stock means $300 million in warrants for common stock—except that those warrants become worthless until the stock price hits $53.80. With the earlier language, those warrants would have been valuable as soon as the stock price hit $44.25. The warrants are exercisable for 10 years—but what if State Street never climbs above $44.25 in that period? Then the warrants will not have been used in either situation. What if they go above $44.25 but never reach $53.80? That’s where the foregone value is: The warrants are now worthless until the price reaches $53.80. Say we had used the old method and the price settled at $50, but the government feared the price would soon go down again. Then the government could’ve trigged the warrants, bought at $44.25, immediately sold at $50, and realized an appreciation of 13 percent on the original $300 million investment: or a total return of $39 million to the taxpayers, versus none under the new method.