A Major Problem with the Pro-Border Adjustment Tax (BAT) Arguments

On Episode 35 of the Lara-Murphy Show, I walked through a Scott Sumner blog post about the GOP tax proposal, which includes a “border adjustment tax” (BAT). This would tax imports at 20% but give a tax rebate to domestic firms for their exports. Supporters of the plan such as Martin Feldstein argue that the US dollar would appreciate 25% and completely cancel out any negative effects on international trade flows, while netting the US government about $120 billion in net tax revenue (because imports exceed exports). However, I think there is a major problem with Feldstein (and the other BAT supporters’) argument–the dollar wouldn’t fully appreciate, and so yes indeed the new 20% tax on imports would make them more expensive for American consumers.

How the Dollar Appreciation Logic Works, If We Start With Balanced Trade

My argument in a nutshell is that the argument for the harmlessness of the BAT on international trade works, if we started from a position of balanced trade–i.e. where US imports equaled US exports. I spent the whole podcast of Episode 35 trying to help listeners understand the mechanism, but let’s walk through it again here to refresh our memories.

For convenience, assume the US just trades with Europe, and that the USD and euro originally trade at par with each other–that is, $100 USD gets you 100 euros on the foreign exchange market. Further suppose that originally, the US exports $3 trillion worth of goods to Europe, while it imports $3 trillion worth of goods from Europe. There is also no tax at all levied by the US government originally.

Now the US government changes things. It imposes a 20% tariff on imports into the US. That is, for every $100 that American consumers want to spend on foreign imports, the authorities in Washington take $20 for themselves and only let $80 leave the country and go to the European producers.

However, on the other hand, the US government also enacts a 20% subsidy on exports out of the US. That is, for every $100 in revenue that US companies receive from foreign buyers, the authorities in Washington send those producers an additional $20.

Now what happens to the volume of imports and exports in this situation? At first you might think imports would be crushed while exports would soar. However, experts like Martin Feldstein keep telling policymakers and the public that such intuition is totally wrong. What happens in this scenario that the USD appreciates 25% against the euro, so that prices to US consumers and producers remain unchanged, and (in this particular scenario) there is no fiscal impact on the US government.

To see why this is so, let me first get you to realize that the two actions would cause an appreciation of the USD. First, consider the import side. When the new 20% tax on imports is first introduced, all of a sudden European goods are more expensive to Americans; in order for the Americans to get the same net dollar amount to buy their imports, they have to fork over 20% more. Thus the American consumer demand for European goods would drop (from the perspective of the Europeans who are only seeing after-tax revenues). Compared to the original equilibrium, the moment the new import tax is levied, Americans would be less eager to import European goods. Therefore their desire to convert USD into euros would diminish, meaning (other things equal) that the dollar-price of a euro would fall. In other words, the dollar would strengthen (or appreciate) against the euro.

Now think about the export side. Because US exporters are getting an extra payment from Washington for every unit they sell to foreigners, the supply of US goods (from the perspective of Europeans) would increase, meaning the official sticker price quoted to European consumers (at first) would drop. Because European consumers now thought American goods were cheaper than before, they would want to buy more of them than before. So they would take more of their euros to the foreign exchange market and try to buy more dollars, with which to buy more US products. Again, the result would be to push up the value of the dollar vis-a-vis the euro.

Now for the purposes of this blog post, just grant me for the sake of argument that those two forces working together end up making the dollar appreciate 25% total against the euro. That is, when the dust settles, $100 USD now buys you 125 euros in the foreign exchange market.

In this case, Martin Feldstein claims there will be no significant impacts from the new import tax / export subsidy scheme. He writes:

[I]f the border tax adjustment is adopted, the dollar will increase by 25% relative to other currencies. A 25% rise in the dollar lowers the cost of imports by 20% (just enough to offset the increase in import prices caused by the 20% tax), while raising the cost of US exports to foreign buyers (just enough to offset the implied 20% subsidy).

I think this is mostly right–but not exactly right, as we’ll see–so let me walk through our particular example so you can see what Feldstein has in mind. Specifically, we are going to show that if the USD rises 25% against the euro, then it (mostly) washes out the impact of our 20% import tax and 20% export subsidy.

Before, when an American consumer spent $100 on a European product, it would exchange for 100 euros and that’s what the European producer pocketed.

Now, the American still spends $100 out of pocket. But the authorities in Washington skim $20 off the top (that’s the import tax). So only $80 flow out of the country into the hands of the European exporter. But when that exporter wants to convert his receipts into euros, he finds that on the foreign exchange market, $80 USD now fetches (80 x 1.25) = 100 euros. So, a typical US consumer can still spend $100 out of pocket, and receive the same amount of goods that a European producer is willing to ship for 100 euros. This is exactly the original situation, so there’s no reason for American consumers or European exporters to change their behavior because of the 20% import tax.

A similar condition prevails for American exporters. Originally, if they charged $100 for a batch of exports, then their European customers would pay 100 euros for them. Now, in the new situation, the European consumer spends 100 euros out of pocket (as before) for the goods in question. But on the foreign exchange market this only fetches $80. However, the US producer gets not only the $80 in revenue from the European consumer, but the subsidy from Washington, which works out to ($80 x 20%) = $16, so that his total revenue is $96 on this sale.

With the particular mechanism I’ve adopted for calculating subsidy payments, we see it’s not an exact wash. But the problem isn’t about the economics so much as it’s about using percentages on moving targets. If we had Washington send the US exporters not $16, but instead $20 (which is 20% of the original $100 sticker price for those exports), then it would be a perfect wash.

Already we see a potential problem with the standard treatment of these issues; to say you are “subsidizing exports 20%” means two different things, when the baseline level of exports (quoted in US dollars) changes because of the subsidy.

However, let’s put aside that quibble. As our example showed, if the dollar rises 25% against the euro, then Americans still want to import the same amount of European goods as before, and US producers want to (just about) export the same amount of American goods as before. So this means the volume of imports and exports should stay the same. Furthermore, the US government breaks even on the whole affair; for every chunk of imports where it raises $20 in import tax, it has to send $20 to US exporters for their chunk of exports. (Again, this isn’t quite right–a straightforward application of our original rule actually meant there would only be a $16 export subsidy on the mirror side of the $20 import tax. So even here the details are important, but let’s wave away that complication and assume the authorities tweak the subsidy rules so that they end up giving the “neutral” $20 payment to American exporters.)

I am sorry that already this post has become complicated, but hey that’s the situation we’re in. What I wanted to show you, the reader, is that if we start from an initial equilibrium of balanced trade–where the US had $3 trillion of imports and $3 trillion of exports–then imposing a 20% tax on imports coupled with a ~20% subsidy to exports would largely have no impact. It would make the dollar appreciate 25% against foreign currencies, so that the penalty on imports and the bonus on exports would largely cancel out. Furthermore, the US government would collect the same amount in tariff revenue that it paid out in export subsidies, so that there would be no fiscal impact on the Treasury’s finances.

This is the baseline result that guys like Martin Feldstein have in mind, when they assure people that the Republican proposal for a border adjustment tax (BAT) wouldn’t make imports more expensive and wouldn’t constitute a protectionist barrier to trade.

What Happens If We Exempt a Quarter of the Exporters From the Subsidy?

Now let’s tweak things. Suppose we start with our original setup: The US has balanced trade of $3 trillion in exports of goods, matched with $3 trillion of imports from foreigners, and originally there are no taxes. In this original setup, the USD trades at par with the euro; i.e. $100 USD fetches you 100 euros in the foreign exchange markets.

As before, the US government now introduces a 20% import tax, coupled with a 20% export subsidy. However, there’s one slight twist: Instead of applying the subsidy to all exporters, instead the new rule is designed to only apply to three-fourths of the exporters. That is to say, a quarter of the American companies who export goods to Europe read the new legislation and see that they are not eligible for the export subsidy.

Now what happens? Well, the dollar will still appreciate. The forces pushing it up from the import side are identical to our first analysis. The forces from the export side are also similar, except that they are only operating at 75% strength.

When the dust settles, the USD will have strengthened, but not as much as before. Perhaps $100 USD now fetches you 122 euros, not the 125 we concluded in our first analysis.

OK, so what? What is the big deal if the USD appreciates (say) 22% but not the full 25%?

Well the problem is that such an outcome would wreck our previous results about the harmlessness of the new measures. If the USD doesn’t fully appreciate, then American consumers will indeed perceive European goods to be more expensive than they were before, and European exporters won’t net as much after-tax revenue (even accounting for the stronger dollar) from sales to the US. Other things equal, Americans will import fewer goods than before.

On the flip side, the American producers eligible for the subsidy will want to export more than before (because the dollar hasn’t strengthened enough to fully offset the subsidy). However, the non-eligible exporters get crushed (because the dollar strengthened while they received no offsetting subsidy). Furthermore, because Europeans are now selling a lower total volume of goods into the US, they have less income with which to buy American exports. So US exporters would perceive that foreign demand for their goods dropped, even setting aside the issue of the exchange rate. All in all, Americans would export fewer goods than before.

Thus, if the USD didn’t fully appreciate 25%, then the new scheme would reduce American trade with Europe. It would on net act as a “protectionist” measure, rearranging production patterns in an inefficient way such that Americans and Europeans ended up with a lower standard of living per capita. (There could be individuals who benefited from the new arrangement, but their gains would be more than offset by the losses of those who were hurt.)

Last thing: The US government would probably reap net revenues from this tweaked arrangement. Even though the volume of imports and exports would both drop, because of the exemption of one-quarter of American producers from the export subsidy, in general it would be the case that the Treasury collected more in tax revenues from the new 20% import tax, than it had to pay out in compensation for the new 20% export subsidy. Thus, the Treasury would collect net revenues from the new scheme.

This is intuitive: If the Treasury is gaining revenue from the new scheme, then clearly it must be affecting the “real economy.” If the Treasury could somehow raise revenue without changing the pattern of international trade and the options available to American consumers and producers, then that would be a magical free lunch money pump. (This is a point I got from reading Alan Reynolds’ critique of Feldstein.) So it makes sense, to reiterate, that if we tweak the arrangement such that the US government collects more in import taxes than it pays out in export subsidies, that overall this acts as a “protectionist” measure that distorts trade flows and has negative economic consequences.

One Last Tweak: Call the Exempted Exports “Stocks and Bonds”

Now we have one more tweak. Go back to our original equilibrium, where the US imports $3 trillion worth of goods from foreigners, while Americans sell $3 trillion worth of stuff to foreigners. There is originally no tax.

Once again, the US government intervenes by imposing a 20% tax on imports and a 20% subsidy for exporters, and once again it exempts a quarter of the American operations that sell things to foreigners. However, this time I’ll give you the extra information that the quarter of US exports that are exempt are commonly called “stocks and bonds.”

In other words, rather than exempting American exports of software or jet aircraft, instead the US government says, “If what you ‘sell’ to a German citizen is $100 worth of stock in a US corporation, or $100 of Treasury bonds, then you don’t get an export subsidy. That doesn’t count when we say American exporters get a 20% subsidy from us, for everything you sell to foreigners.”

I submit that the analysis in this situation should go through as in my second stage above. Before, I didn’t get specific and tell you what the US exports were; they could have been cars, wheat, natural gas, etc. If now I label them “stocks and bonds” that shouldn’t have any immediate effect on the analysis that we gave.

Thus, I think we should conclude that if the original $3 trillion in US exports including roughly a quarter of “stocks and bonds,” while the other three-quarters were “goods” in the traditional meaning, and if the US government said its export subsidy only applied to the “goods” but not the “stocks and bonds,” then we know what happens: The dollar strengthens, but not the full 25%. On net, the US government earns net revenue, because there are more imports getting taxed than there are export goods getting subsidized. And the whole scheme acts as a protectionist measure that reduces overall trade flows.

Martin Feldstein himself agrees with me partly, when he writes:

But if the border tax adjustment would not improve the US trade balance, why are congressional Republicans eager to enact it? The real reason that it would boost tax revenue substantially, without increasing the burden on US consumers or producers. Currently, US imports and exports are 15% and 12% of GDP, respectively. Given the difference of 3% of GDP, the 20% import tax and 20% export subsidy raises a net 0.6% of GDP, now equal to $120 billion a year.

The border tax adjustment therefore pays for about two-thirds of the $190 billion cost of the corporate tax cut [another component of the GOP tax plan–RPM], and an even larger share when the lower corporate rate’s favorable effect on growth is taken into account. And, because there is no change in prices paid by American consumers or received by American exporters, that tax is borne by foreign producers, who, owing to the dollar’s appreciation, receive less in their own currencies for their exports to the US.

So Feldstein views the net tax receipts (due to the US’ trade deficit) as a good thing, a source of apparently painless money for the Treasury.

Yet how can this be? Surely if foreign exporters are bearing a new tax, they will respond by selling fewer goods to Americans.

Feldstein thinks he has isolated the impact of the scheme and “proven” that it can’t possibly hurt Americans, by showing (in his mind) that the dollar would appreciate the full 25% to offset the effects of the scheme on Americans.

Yet as I hope I’ve demonstrated above, his argument is faulty. Yes, if we started out in an initial equilibrium where imports equaled exports and then imposed the 20% import tax / export subsidy (with due allowance made for the percentage-applied-to-a-moving-baseline subtlety), then the dollar would appreciate 25% and basically cancel the whole thing out. Furthermore, the US Treasury wouldn’t gain or lose money.

However, if we started from a balanced equilibrium and then only applied the export subsidy to three-quarters of exporters, surely the outcome can’t be the same. Surely the dollar wouldn’t fully appreciate, meaning imports and exports would fall, and the Treasury would end up reaping net receipts because it was taxing a greater volume of imports than the exports it was subsidizing.

I submit that this is equivalent to the real world, where annualized imports are about $2.9 trillion and annualized exports are about $2.3 trillion. The roughly $600 billion gap is “the trade deficit” and is accounted for by foreigners accumulating (on net) roughly $600 billion more of US assets than Americans acquire of foreign assets. If you agree with me that literally exempting a quarter of merchandise exporters would mess up Feldstein’s analysis, then surely the fact that one quarter of our current “exports” are actually assets (which don’t receive a subsidy) should similarly mess up Feldstein’s analysis.

Where Did Feldstein Go Wrong?

Martin Feldstein is telling us that the GOP’s proposed BAT is a good idea because the trade deficit implies it will raise large amounts of revenue from foreigners with no impact on Americans. As my step by step analysis demonstrated, this can’t be right. In order for the US government to raise large amounts of revenue from foreigners, the scheme must be acting as a protectionist measure.

The specific flaw in Feldstein’s case is his assumption that the dollar would rise a full 25%; we’ve seen that it won’t. Feldstein also gives a non sequitur when he writes:

Although it looks like [the GOP plan’s BAT] would reduce imports and increase exports, that will not happen. As every economics student learns, the trade balance depends on the difference between domestic saving and domestic investment. Because the border tax adjustment does not change saving and investment, it wouldn’t change imports and exports. Instead, the changes in taxes on imports and exports would lead to a rise in the value of the dollar that offsets the direct impact of the border tax changes.

Yes, it’s true that the trade deficit (or more accurately the current account deficit) equals the difference between domestic investment and domestic saving. If there’s more total investment in the US than Americans are saving, the gap must be supplied by foreigners who (on net) are investing more in American assets than vice versa.

But Feldstein is assuming his conclusion. If in fact the dollar doesn’t fully appreciate, and this ends up hurting both imports and exports, and the US government effectively imposes a net tax hike on foreigners, then surely this might reduce foreigners’ willingness to invest in US assets. Feldstein is basically arguing in a circle: He’s assuming the BAT would have no impact on foreigners’ behavior, and then from that concluding that the BAT would be innocuous.


Martin Feldstein is an authority on the theory of taxation and international trade; he co-authored an excellent treatment (albeit with some typos) of these matters with Paul Krugman.

However, I think he (and some other major economists) are making a basic mistake when they reassure free traders that the Republican tax proposal won’t act as an implicitly protectionist scheme. They are relying on a textbook result that doesn’t seem to apply to our current situation.

I realize these matters are extremely difficult; they would have to be, if I’m accusing giants like Feldstein (and Krugman too, who basically agrees with the baseline analysis) of making such a simple mistake. But intuitively, Feldstein is arguing that there’s a free lunch: that the Treasury can raise more than $100 billion per year from foreigners, without making imports more expensive to American consumers. That can’t be right. I think I’ve shown in this post where Feldstein specifically goes wrong in his chain of reasoning.

Overall the GOP tax plan might make sense, but policymakers and the public should go into this with eyes open.