Neil Barofsky (New York University)

Neil Barofsky's new book on his experience as the government watchdog for the bank bailouts of late 2008 is getting a fair bit of press, both for its accusations that the White House did too much to protect its political allies in the financial sector and too little to help everyday homeowners. But another accusation is getting much less attention: He thinks the Treasury Department doesn't know how to count.

Currently, the Treasury Department says (pdf) that the Troubled Assets Relief Program (TARP), the bailout program for which Barofsky served as inspector general, will cost taxpayers a total of $43.32 billion, when gains from it and related asset buys are taken into account, or $59.75 billion when only TARP gains are included. But Barofsky alleges in a recent Bloomberg column that these figures include revenue gains that it shouldn't include, and excludes losses that it should include.

Specifically, the $43.32 billion figure includes gains the Treasury department made from stock of the American Investment Group (AIG) that was bought with non-TARP money. If it wasn't bought with TARP money, Barofsky reasons, it shouldn't be included as a TARP gain. Without considering these other shares, the higher $59.75 billion estimate is correct. Matthew Anderson, a spokesperson at the Treasury department, argued in an e-mail that including the other shares is appropriate, noting a previous report from Barofsky himself stating (pdf) that "All of these infusions to AIG are linked inextricably." In an e-mail, Barofsky calls this characterization of his past views "laughable." The bottom line, he says, is that Treasury is "including Fed gains as if they were part of a Treasury investment, when they were not."

Barofsky also faults Treasury for not taking into account lost revenue due to changes in the tax treatment of AIG, General Motors, and Citigroup following their bailouts. Usually, companies are allowed to "carry forward" past losses for up to twenty years in the future to reduce their tax burdens. So if a company loses $5 billion one year and gains $7 billion the next, it can carry forward the $5 billion and only pay taxes on the $2 billion difference.

However, section 382 of the tax code limits the ability of companies to carry forward losses from companies they bought. The idea is to prevent companies from buying smaller, loss-heavy companies just to reduce their tax burden. According to a paper (pdf) by Harvard Law School's J. Mark Ramseyer and Indiana University's Eric B. Rasmusen, this provision would have meant that if private buyers acquired AIG, General Motors, and Citigroup, those buyers could have carried forward none of the bought-out firms' losses, as firms are limited to carrying forward an amount equal to the total stock value of the company purchased times the interest rate charged on tax exempt bonds. "I believe the technical term on Wall Street for what Citi was worth then is 'bupkis,'" USC's Ed Kleinbard, who was chief of staff for the Joint Committee on Taxation at the time, tells me, so the provision could well have wiped out all tax carry forwards for the three firms. To prevent this, Bush's Treasury Secretary Hank Paulson exempted the government from that provision when it bought shares of AIG, GM, and Citigroup.

As a consequence, the three companies got to carry over a combined $90.38 billion in losses, according to Ramseyer and Rasmusen. Since all three firms are big enough to pay the top corporate tax rate — which is a 35 percent flat tax — that works out to $31.633 billion in lost revenue, over the next decade or more. But some, like Kleinbard, defend the move, saying that it doesn't make sense to apply a provision meant to prevent firms from lowering their tax loads to the government itself. "The rationale was, 'Don't be stupid, that's not what the rule's about,'" he says.

What's more, it could be the case that, absent that exemption, it would be harder for the government to sell shares of the companies in the future, as their inability to deduct past losses would have made them less valuable. If the shares sold for enough less, the new revenue could be swamped, and not exempting the firms could have actually cost taxpayers money. Barofsky dismisses this idea, noting that with the exemption, "The windfall goes not just to treasury, but also, in proportional amounts, to all of the other shareholders. Therefore, it would seem that the benefit would go to the many under the current scheme, while without a tax break, all the benefit would've gone exclusively to the taxpayers."

That's something of an evasion. Even if interests other than the taxpayer's made money off the tax exemption, that doesn't answer the question of whether the government would have made more money without it. There, the evidence is mixed, and thus the Treasury's decision to leave the exemption out of its calculations seems fair. Where Barofsky's case is much stronger is on the $16.43 billion in non-TARP AIG gains. That really is a case where Treasury is counting money that wasn't part of TARP as a gain for the program, at least in one of its estimates. Regardless, whether TARP cost $59.75 billion, as the government's estimate without non-TARP gains has it, or $91.38 billion, as Barofsky's math would have it, the total cost came in a whole lot less than the $356 billion that the Congressional Budget Office estimated the program would cost in March 2009. It cost real money, but a fraction of what we thought it would.