How much of a sucker is the United States? When it comes to repatriation, so far we've been such an easy mark that corporations pay real money to bank on it.
The idea of a tax holiday on overseas earnings has reared its slobbery head again, this time in connection with funding infrastructure projects. There is more to say about this scam elsewhere. This is an article about the game theory involved: the strategy a company employs in keeping its cash overseas or at home, depending on how soon it thinks the U.S. government is going to give it another freebie on its taxes. Fair warning: I am not an economist, I am a mathematician. However, Paul Krugman says that toy models give pretty good insight anyway, and I'll be keeping this simple. If anyone would like to run a similar analysis with more knowledge of how corporate debt and asset management works, I would be all ears.
Here's the abstracted scenario: a corporation has income it can designate as being earned in the U.S., or abroad. If it is designated as being earned in the U.S., it is available for investment and other usage, after paying U.S. corporate income tax. If it is designated as being earned abroad, eventually it expects to bring the money back to the U.S. during a tax holiday. Until then, to get any use out of the cash, the corporation must borrow against it, by issuing corporate bonds. This costs it money, which would eventually erode the value of the cash it holds overseas, and hence the corporation will only pursue this option if it expects the amount saved during the tax holiday to be such that, after borrowing costs and future taxation, the amount repatriated will be greater than the amount it would have held had it simply declared the income as domestic earnings and been taxed on it immediately.
This simple game neglects everything fancy: all costs associated with holding money overseas, differences in interest rates that can be earned on holdings in various countries, other ways to employ overseas cash, et cetera. To model the corporation's borrowing cost on its assets, I will assume that it borrows against its overseas cash at roughly the same debt/asset ratio that holds for the entire company (an assumption deleterious to the argument since it has more reason to borrow against its overseas cash).
The current largest overseas account holder in America is Apple Corporation. Apple has about a 27% debt/asset ratio. It can borrow long-term at about 4%. Thus, at a ballpark, it pays roughly 1% of its assets per year for the privilege of keeping its cash overseas -- even less if we acknowledge that some of this debt is not due to borrowing against overseas cash, and that its shorter-term debt is at even lower rates.
The current statutory federal U.S. corporate tax rate is 35%. Call it 28% after tax loopholes in the standard tax code. Suppose that a tax holiday is offered so that Apple pays 10%. How much is an overseas dollar worth to Apple once brought back if it must pay 1% per year to keep it overseas, and then pays a 10% tax rate on it, if it expects that this holiday will happen x years in the future? One dollar becomes 0.9*(0.99)^x. Meanwhile, this dollar would have been worth 0.72 if brought back immediately, with no interest paid on loans to be able to use it.
Thus, keeping the cash overseas and waiting for a holiday is beneficial to Apple if it thinks that tax holidays happen frequently enough that 0.9*(0.99)^x > 0.72. Solving, we find that Apple will hoard cash overseas awaiting a tax holiday if it believes that x < 22.2. This time horizon is even longer if we use Apple’s lower-cost shorter term debt.
When was the last tax holiday instituted? 2004. When can the next one be made? Well, if it gets baked into the next budget, 2018. 14 < 22, so if that's the case, Apple won that staring contest, and the U.S. lost out on 18% of Apple’s income in taxes over the intervening period.
So how do we turn the math on its head? The nation’s goal is to convincingly broadcast a sufficiently long time horizon to the next tax holiday that corporate taxpayers believe the cost of regularly doing business in this way is more than the benefit. We stiffen our spines enough that corporations judging the tenor of U.S. politics believe that repatriation holidays come around once in a generation at best, 25 years or more between arrivals, or, better yet, "basically never." Having an actual period of 25 years pass since the last tax holiday would probably go a long way in that direction. (Much more drastically, we could also jettison the system entirely, and consider a less gameable system like sales factor apportionment, for which the location of corporate assets is irrelevant. The complications involved in such a move, however, are beyond my ability to analyze skillfully.)
What if Apple decided, after a tax holiday failed due to public outcry despite Republican control of government, that a tax holiday had become an abnormal idea, and was unlikely under any foreseeable political regime? In this simple model, paying any amount to keep the money overseas would be fruitless: Apple would repatriate everything at current tax rates. The government loses no revenue by never blinking, under which circumstance Apple never blinking would basically give itself an avoidable, permanent asset tax. In other words, if we would just have some spine we would win this game instantly and completely. But calling it "spine" implies that the incoming Congress' and President’s actual goal is to optimize the result of the game for the nation as a whole, rather than, say, to enrich a small number of wealthy individuals.