Today in the Tax Policy Colloquium, we discussed the above paper, available here, albeit with technical limitations on the session.
Of the two authors, only Nancy Staudt was able to attend. By the way, she was our second-ever colloquium guest, back in January 1996, when she presented her paper, Taxing Housework. (Our first guest ever was Louis Kaplow discussing state and local taxes; in week 3 we had Bill Andrews discussing corporate taxation and the new view. My co-convenor at the time was David Bradford.)
Anyway, Nancy, unfortunately but understandably, had to leave early. She barely got in yesterday from Los Angeles, switching to a flight that wasn't canceled by yesterday's East Coast storm, and I gather managed to escape this evening just ahead of tonight's East Coast storm. But this required her leaving our session early. I could certainly understand the problem; twice in the last few years I've been trapped in Los Angeles (mid-semester) for 48 extra hours due to an East Coast storm.
To fill the gap, Victor Fleischer participated (from San Diego) by Skype. I really don't like doing this because the technology still isn't so great, unless you have a higher capital investment site than NYU can offer. Vic apparently heard less than half of what was being said on our end, and inevitably we had lag, Internet freezes, etcetera. But it was good to have him participate even virtually, and this enabled us keep the session going under some simulacrum of quasi-normality for the full time.
Anyway, the topic of the paper is a type of deal that has gotten some attention in the tax press, known as a "supercharged IPO." As discussed in the paper, these deals have two main attributes, and the relationship between the two was a main topic of interest. The first attribute is that, in certain initial public offerings (IPOs) in which a given start-up company is taken public, the parties deliberately arrange a taxable, rather than a tax-free transaction. The second attribute is that, in a very few deals but these being the ones that were studied in the paper, the parties agree to a "tax receivables agreement" (TRA). Under a TRA, the buyer agrees to make certain payments to the seller, in effect as deferred installment sale payments, the amount of which depends on the tax savings realized by the buyer, post- transaction, from tax attributes acquired in the course of the deal (and enhanced by the fact that the deal was deliberately made taxable).
OK, let's give an example, before turning to the fact, which came out during the session, that this was not generally the universal or even typical pattern in a "supercharged IPO." Suppose the following. A highly successful start-up business, conducted as a partnership, has assets with a value of $1 billion and a tax basis of zero. Suppose that all of the assets would yield capital gain, taxable (during the pre-2013 years covered by the paper's data analysis) at a 15 percent rate. In other words, no depreciation recapture or "hot assets" (to use the operative tax lingo) that would be taxed at the ordinary income rate. Suppose, moreover, that if the assets were newly acquired, they would get 10-year straight line depreciation. Thus, if acquired for $1 billion, they would yield depreciation and amortization deductions of $100 million per year for 10 years. The tax savings would be $35 million per year for 10 years at the 35 corporate tax rate, assuming that the taxpayer always has enough taxable income for the year to claim all of the available deductions at that rate.
OK, suppose the taxpayers could sell the asset to themselves for $1 billion, for tax purposes. They'd pay $150 million of tax on the $1 billion capital gain. Saving $35 million per year from the cost recovery would yield them the equivalent of a 10-year annuity (again, assuming certainty of realizing the full value). These tax savings would have a present value of $270 million if one uses a 5% discount rate.
Therefore, the self-sale, if permissible would save $120 million of tax in present value. While you can't do that, using an IPO to do it arguably applies the following. In a competitive market, buyers who would have paid $1 billion just for the assets should also pay $270 million for the annuity, so the total sales price, in a taxable deal, should be $1.27 billion rather than just $1 billion. (To keep the arithmetic simple, I am ignoring the fact that this changes all the relevant dollar amounts, and thus the true amount might settle a bit further north.)
OK, so if this is the right scenario to be thinking about - and knowledgeable practitioners in the room suggested that perhaps it actually is not - then Conclusion 1 is that, obviously, the parties should do a taxable rather than a tax-free deal. No need for a fight between sellers who'd want to avoid tax and buyers who'd want to maximize their deductions, since the right way to approach the problem is as one of collective tax minimization. Once you've made the "pie" as large as possible, at the expense of the fisc, there's plenty of time to divide the loot between the two of you so that you're both better off. E.g., at a $1.27 billion price, the buyers get fair value while the sellers are compensated for their $150M tax liability by a $270M increase in their sale price. (Yes, I realize still that I am not fixing the numbers to include the capital gain on the extra sale price, etc.)
Let's call this the underlying "tax arbitrage" here (offsetting 15% & 35% rates create a net tax saving despite the adverse timing of having income before deductions). But that just concerns doing a taxable rather than a tax-free deal. (Practitioners suggested, however, that more realistic scenarios include (a) sellers are going to realize gain anyway, by selling the stock they acquire within a few months even if it's a tax-free deal, so why not get the basis step-up, and (b) a separate set of considerations that guide purely corporate transactions.) Even if we accept the scenario, however, what we haven't explained is why the parties, at least in a small set of cases, might include a TRA.
A typical TRA might provide the following. For each of the next 10 years, Buyer shall make a payment to Seller that equals 85% of the tax savings enjoyed by reason of the tax benefits. So under my facts, each year 85% of the $35 million tax benefit from the cost recovery deductions (i.e., $29.75 million) would be paid to the Seller.
OK, again for arithmetical simplicity, let's make it a 100% TRA, so the annual payments are $35 million. (This is frowned upon in practice, of course, because it would wholly eliminate the Buyer's incentive to actually use the tax benefits.) With a 100% TRA, the mechanism for paying $1.27B to the Seller might be $1B up front, plus $35M per year for 10 years. The question is, why do this? $1.27B in present value can be paid whether you use the TRA or not. And even if you want some deferred payment, why should it depend on the tax benefits. Use or non-use of the TRA should be a matter of complete indifference to the parties - indeed, whether or not they have done a taxable deal, unless there is more to the story.
Here are the 6 theories we discussed that might explain the use of TRAs in practice:
(1) Buyer myopia - Just about everyone who actually knows about these deals insist that the form reflects buyers' failing to value the tax benefits appropriately - and at the limit, valuing them at zero. "That's what the bankers tell everyone," the tax lawyers in the deals will explain to you if you ask. This seems decidedly odd as it implies a market failure that seemingly could be exploited by savvy arbitrageurs (or simply higher bidders) who understand the value of the tax benefits. But if we assumed it were true, it would make use of the TRA an efficient mechanism between the parties. You assign a given asset to the party that actually places a higher valuation on it. As we will see, while this theory may remain hard to accept, arguably all the other theories fare even worse.
(2) Careless buyer doesn't read the fine print - To put this in the strongest possible form, although in practice it may overlap with Theory 1, suppose the Buyers are willing to pay $1.27 billion as that is the value of the business assets plus the tax assets, it fails to read the fine print at page 496 of the transaction documents. Hence, they fail to realize that they are paying twice for the same tax assets: once up front and a second time through the TRA. This is an even harder theory to credit - TRAs are apparently highlighted not smuggled in - but it starts to look more like Theory 1 if we posit that the Buyers don't so much overlook the TRA as value it at zero (in the extreme case) due to their mysterious myopia about the value of tax assets. And in the intermediate case it's just the way they prefer to pay the overall sale price, since they have a lower estimate of the tax assets (even if not zero) than the Seller.
(3) TRA spares the parties the trouble of having to value the tax assets - Rather than fight about how much they're worth, why not just assign them to the Seller through the TRA. But the question here is, why are they harder to value than everything else? You have to figure out how much the company is worth, and once you're projecting annual pre-tax and after-tax earnings you're pretty much there so far as valuing the tax assets is concerned. So this theory is less than wholly persuasive.
(4) Buyers don't like the risk associated with the tax assets - Will the IRS allow them all? Will there be sufficient taxable income to use them in full right away? But if we view this purely as a matter of risk aversion, it seems peculiar in this context. Typically what's happening in an IPO is that the entrepeneurs who bear a concentrated business risk are selling it into the general marketplace in order to diversify. The buyers already were diversified and remain so (indeed, they may become a hair more diversified by reason of doing this). So why would the buyers be more averse to the tax risk here than the Seller?
(5) Lemons problem from asymmetric information - OK, now it might seem that we are finally getting somewhere. This was my favorite theory going in. Suppose the Seller knows more than the buyers about the true value of the tax assets. Is it overstated? Are they subject to successful IRS challenge? With asymmetric information, they have the used car problem. Even if the tax assets truly are worth what they seem, how can they prove this to the buyers? The very fact that they are selling creates a bit of natural suspicion that this is what motivates them. Now, the lemons problem clearly can be a big problem in selling stock to less-informed third parties. But why is it distinctively associated with the tax assets? E.g., asset basis is negotiated in the deal and may be hard for the IRS to challenge. Future profits that would permit the tax benefits to be used in full is already an issue on asymmetric information grounds. How much worse do the tax assets make it? Plus, apparently in the typical TRA the buyers' payment to the Seller is based on the tax benefits claimed, and no refund is subsequently due from the Seller if the IRS disallows tax benefits on audit. In sum, therefore, it is hard to really get this theory off the ground.
(6) Upselling by the lawyers - They want to be able to charge for arranging a TRA, so they tell the parties what a great idea it is. Only, it's apparently the bankers who push for these things, and they aren't increasing their own compensation by doing this unless this makes the deals higher-priced overall or easier to close.
So even though I don't like Theory 1, I am pushed back towards accepting it by the universal testimony of the players plus the weakness of the alternative theories.
If one accepts all this, what should one think of TRAs? They look pretty innocuous, serving merely as devices for permitting efficient pricing and costing the IRS zero on any deal that would have been made with the same overall price terms in any event. (Tax attributes aren't traded from lower-valuing to higher-valuing parties - it's just the after-tax benefit that they shift around.) There would be an SEC / capital markets / consumer protection for banning them if we believed, as under Theory 2, that they are a device for duping investors into overpaying. But that is hard to credit given how prominently they're disclosed.
So about the best one can do, if one wants to argue against them, is to claim that they are associated with the ability to execute tax arbitrage deals (like my $1B / $270M example above) that otherwise would founder due to differential valuations by the parties of the tax benefits. In other words, in this scenario one would ban TRAs in order to discourage tax arbitrage deals by impeding efficient pricing of the deals. (Note, however, that you can also potentially use a TRA in a tax-free deal.) But this one as well is pretty hard to credit. To what extent do we believe that impeding efficient pricing would actually impede "bad" deals while leaving "good" ones (e.g., those where people just want to diversify) unharmed?
I conclude both that there is no particular reason to go after TRAs - instead, address the tax arbitrage directly if one is concerned about it - and that they are simply aren't that big a deal from the policymakers' standpoint. But still interesting to discuss at the session.