Michael Knoll has a cool new paper estimating the revenue effects of changing the tax treatment of carried interest. Depending on various assumptions, he comes up with about $3 Billion a year (see paper at p. 12; if character is changed to ordinary income, then additional tax collected would amount to between 2.4 and 3.4 billion). I'd previously done a very rough back of the envelope calculation to come up with $4-6 billion a year.
Among Knoll's key assumptions:
1) $200 Billion a year invested in private equity. Is this too low? This is a high estimate by historical standards, but low given recent trends. On the other hand, the recent credit market shakeup may slow fundraising for a while.
2) Scope of the change. Knoll only looks at private equity, but at a minimum the change would also apply to hedge funds. Hedge funds that make long-term investments might also be affected. It's also possible that real estate, timber, and oil and gas partnerships might also be affected, depending on how the politics play out in DC.
3) Volatility. Knoll uses a Black-Scholes model to value the carry and estimates volatility of the average fund at 20%. No idea if this is right - average volatility for a portfolio company would be much, much higher, but of course most PE funds have 10 or more companies in the portfolio. I suspect we could find some historical return data that would help here. Unfortunately, academics are at a distinct comparative disadvantage here, as the best data sources are proprietary and expensive.
I'm also not sure what the Black-Scholes model gets you here, i.e., why it's better to use option methodology rather than just looking at historical returns for the sector as a whole.
Knoll also discusses some possible re-structuring of the carried interest that might take place, and how that might affect revenue. He notes, for example, that incentive fees might be paid by portfolio companies instead of the funds themselves, which -- if the portfolio company has a high effective tax rate -- generates a valuable tax deduction. On the other hand, this changes the economics of the deal in ways that the LPs might not like (by measuring carry on a company-by-company basis rather than an aggregate basis), and it's not clear how many portfolio companies have a high effective tax rate. (Recall that most portfolio companies take on a lot of debt in connection with the buyout.)
It will be interesting to compare Knoll's paper with government's revenue estimate and methodology.
TrackBack URL for this entry:
Links to weblogs that reference Knoll on Carried Interest:
1. Posted by Matt on August 21, 2007 @ 7:55 | Permalink
Is there really an argument for taxing carried interest at "receipt" (i.e. at the time of investment)? That struck me as simply bizarre; it's like taxing someone on their expected lifetime income when they're offered a job. If you're treating carry as compensation for services, you should tax it as ordinary income when it is really received, i.e. when the GP gets the check.
I think the idea of taxing when the carry is granted was just an excuse to use Black-Scholes equations and make this paper olook more complicated than it really is. I could've probably written it in one sentence, something to the effect of "Divide 35 by 15 and multiply that by the amount of private equity carry that gets paid each year."
2. Posted by Vic on August 21, 2007 @ 11:39 | Permalink
There is an argument for taxing carry on the date of grant, although I don't think it's ultimately persuasive. Taxing carry on the date of grant is, of course, administratively challenging, especially since such profit-sharing rights are not always vested. And you are right that taxing unrealized human capital is inconsistent with our usual reluctance to taxing endowment rather than income. So I think the better way to tax carry is on receipt of cash on the back end, or perhaps using a modified accrual method.
I'm not sure what the Black-Scholes methodology accomplishes here; presumably Professor Knoll didn't think projecting returns from the historical data was as useful as a model, or perhaps there is a theoretical reason that I can't figure out that makes the Black Scholes methodology more sound.
3. Posted by Dan on August 22, 2007 @ 11:04 | Permalink
Anyone know of any other estimations out there besides this one?
4. Posted by Manuel on September 19, 2007 @ 9:53 | Permalink
I tried to calculate the results of Knoll in his paper, but i can't find the same numbers. For the Black&Scholes Model he uses S=100, but which asset price P he uses?
Thank you very much