Who really pays a digital services tax, and who would pay the tariff against it

On Friday 26 June 2026, President Trump posted on Truth SocialTruth SocialThe social-media platform President Trump uses to post public statements. that any country imposing a digital services tax on American technology firms would face a 100% tarifftariffA tax a government charges on imported goods, paid to the importing country's customs authority by the company bringing the goods in. on its goods, a duty that 'will supersede Trade Deals made with the Country, whether implemented, signed, or not.' Those tariffs, he wrote, would be imposed immediately. It was, almost word for word, the threat that pushed Canada to abandon its own digital tax in June 2025, one day before the first cheque was due.

The stated logic is protective. A digital services tax, or DST, is a charge a government levies on the local revenue a big platform earns inside its borders. Think of it as a 2% or 3% toll on what firms like Google, Amazon, Meta and Apple book in a given country, designed to catch companies that sell heavily somewhere without leaving much taxable profit there. Washington's position is that these taxes single out American champions, and that a tariff is a fair shield.

That framing rests on a hidden assumption: that the people hurt by a digital tax are US tech shareholders. If that were true, a retaliatory tariff would at least be aimed at the right injury. So we asked one narrow question and chased the data until it answered.

Who actually ends up paying a digital services tax? And if it is mostly local sellers and shoppers, who would bear a 100% tariff imposed to defend US firms?

The short answer, which the rest of this piece walks through step by step: on the one channel where we can measure prices to the penny, the foreigners the tariff claims to protect Americans from are largely taxing themselves. And a tariff to defend US tech, if it is ever actually collected, would be a tax on Americans. Both halves of that sentence come with caveats, and we will be explicit about them.

The cleanest evidence on who pays a digital tax comes from a 2025 study with an unglamorous but exact method. In 'Navigating the Amazon: The Incidence of Digital Service Taxes,' economists Dominika Langenmayr and Rohit Reddy Muddasani tracked prices on Amazon's marketplace before and after several countries switched on their digital taxes. It is a natural experimentnatural experimentA real-world event that switches a policy on at a known date, letting researchers compare before and after as if it were a controlled test.: the policy flips on at a known date, so you can compare the same products just before and just after.

Here is why that setting is so useful. When a country turns on a digital tax, Amazon does something it announced openly: it adds a separate fee to the bill of the third-party sellers who use its marketplace, naming the tax as the reason. (A third-party sellerthird-party sellerAn independent merchant who sells goods on a marketplace like Amazon rather than the marketplace selling the item itself. is an independent merchant who lists goods on Amazon rather than Amazon selling the item itself.) The researchers then watched what those sellers did next: swallow the fee, or pass it to shoppers? Because millions of near-identical products carry timestamped prices, you can measure the answer to the penny.

What they found: after the tax went live, the consumer price net of VATnet of VATA price with value-added sales tax stripped out, so you are comparing the underlying cost rather than the checkout total. (that is, with sales tax stripped out so you are comparing the underlying cost, not the checkout total) rose in France, Spain and the UK. In Italy the response was weak and not clearly distinguishable from no change at all. The sellers, in other words, did not eat the fee. They moved it down the line to shoppers.

Rising prices alone do not tell you the size of the shift. So the researchers compared the extra amount buyers paid against the tax the government actually collected. That ratio is the pass-throughpass-throughThe share of a tax that a company shifts onto its customers through higher prices rather than absorbing itself.. A pass-through of 1.0 means buyers cover the tax exactly. Below 1.0, the platform and its sellers absorb part of it. Above 1.0 is the striking case economists call over-shiftingover-shiftingWhen a company raises prices by more than the tax it owes, so customers pay more than the government collects.: buyers pay more than the state takes in.

Across France, Spain and the UK the marketplace pass-through landed roughly between 1.1 and 2.7. In plain terms, for every euro or pound the government collected, buyers paid somewhere between about 1.10 and 2.70. A useful analogy: imagine a corner shop told it owes a new 10-cent levy on a tin of beans, and it quietly rounds the shelf price up by 20 cents, using the tax as cover to take a little extra. The customer pays the levy and then some; the shop is no worse off and may be slightly better off.

Now connect that to who bears the tax, because over-shifting does not by itself prove the platform pays nothing. The chain is this. The tax is billed to the platform on its local revenue. The platform passes the named fee to its third-party sellers. Those sellers, the study shows, raise net-of-VAT prices by as much as or more than the fee. So the money is recovered from sellers and shoppers, and sometimes more than recovered. That leaves the platform's shareholders a small slice of the burden, and in the over-shifting cases possibly none. The tax meant to hit Silicon Valley is, on this channel, paid by people in Lyon and Leeds.

The advertising channel points the same way, though it is not in this study. When the first DSTs took effect, Google publicly added itemized DST surcharges to advertiser invoices, for example a 2% fee on UK ads, telling advertisers the tax was the reason. A surcharge named on the bill is pass-through in its plainest form: the advertiser, often a local business, pays it. That is consistent with the marketplace finding, but it is a separate fact and we flag it as needing its own confirmation rather than folding it into the headline number.

Canada offers a second angle, this time on legal remittancelegal remittanceWho is formally required to file and hand the tax over to the government, which is not the same as who bears the economic cost., which is who is billed and hands the money over, as opposed to who ultimately bears it. Canada's DST was sold politically as a charge on US tech giants. When the Canada Revenue Agency tallied what had been collected before the tax was killed, the split was telling.

Of roughly C$647 million collected, about C$148 million, near 23 percent, was remitted by US-based firms. The other 77 percent came from everyone else: Canadian and other non-US filers. And remember, this is only the legal split, who wrote the cheque. The marketplace incidence evidence from Step 2 says the economic burden slides even further off the platforms, because the firms that did remit recover much of it from their local customers. A tax framed as falling on American companies was, even on its own paperwork, mostly remitted by others, and in real economic terms more so.

Canada then folded under the tariff threat. It announced in June 2025, one day before the first payment was due, that it would scrap the tax, and the repeal received royal assentroyal assentThe final formal approval that turns a bill passed by a Commonwealth parliament, such as Canada's, into law. on 26 March 2026. The CRA is now refunding the roughly C$647 million it had collected. Hold that fact; it is also the strongest argument against this article's own thesis, and we come back to it.

A tariff is the mirror image of a DST when it comes to who is billed. It is collected at the US border, from the American company importing the goods. A 100% tariff on French goods is not a charge Paris pays. It is paid by the US importer, who generally passes it to American buyers, exactly the pass-through mechanism we just watched work in reverse.

Make it concrete. A 100% tariff on French goods lands first on importers and buyers of French wine and champagne, luxury leather and cosmetics, and aircraft and aircraft parts, three of the larger US import lines from France, along with some pharmaceuticals. Double the border price and either the importer absorbs it or, more likely, the shelf price climbs and American consumers pay. And because the EU has a record of matching US tariffs with its own, US exporters, from bourbon distillers to farm and machinery firms, would face retaliation in Europe. The injury the policy claims to answer is largely borne abroad by foreigners; the cost of the answer is largely borne at home.

We deliberately do not put a single ratio on that mismatch. An earlier draft compared France's annual DST take against total US imports from France, but those are different currencies, different years, and a tax-revenue figure against a gross-trade-flow figure, not comparable quantities. An honest like-for-like number, DST actually borne by US firms versus tariff cost actually borne by US importers, would require data we could not verify for this piece. So we make the direction claim, not a magnitude claim.

Here is the objection that does the most damage to our own thesis. If the tariff is never collected because the threat alone makes countries back down, then 'it would tax Americans' is conditional, and the realized cost to Americans may be close to zero. That is not hypothetical. It is what happened in Canada, where the threat killed the tax before a single tariff dollar was charged. The glossary entry for deterrencedeterrenceUsing the threat of a penalty to stop someone from doing something, where the goal is to change behavior rather than to collect the penalty. is doing real work here: the goal of a threat is to change behavior, not to collect the penalty.

But the scoreboard cuts both ways. The threat killed Canada's tax and capped France's attempt to double its rate. It did not remove France's existing 3% tax, did not touch the UK's 2% DST, now at record receipts near 800 million pounds, and did not stop the European Commission from floating a bloc-wide 3% digital tax in late May 2026. So the deterrent is partial: it moves the rate on larger partners and can kill a tax not yet in force, but it has not reversed the trend. The fair conclusion is narrow. If the tariff stays a threat that works, it is cheap for Americans and our cost framing does not bite. If a country holds firm, as the UK and the EU so far have, and the tariff is actually imposed, then the cost falls on American importers and buyers, to defend US firms from a tax those firms largely do not bear.

It is worth steelmanning the other side, because the incidence finding is not the same as saying DSTs are pointless. The standard case for them is that the largest platforms book enormous local sales while reporting little taxable profit in those same countries, by routing profit elsewhere. A DST is a blunt attempt to tax that local activity directly. Some pass-through is expected for any tax on a business, just as a corporate tax or a fuel tax partly reaches consumers; the fact that buyers bear much of a DST does not, on its own, make it a bad tax, any more than it makes VAT a bad tax.

What the evidence does undercut is the specific political story on both sides: the story that a DST is a clean hit on foreign tech shareholders, and the mirror story that a tariff 'defending' those shareholders is costless to the home country. Both claims fail the incidence test. The OECD's Pillar One was meant to replace this messy unilateral fight with a shared rule on where the biggest firms pay tax, but it remains stalled, which is why the tariff-versus-DST standoff keeps recurring.

So the finding is not that one side is righteous. It is that the same overlooked mechanic, pass-through, governs both the tax and the tariff, and both political slogans ignore it. A tax sold as hitting foreigners is largely paid by locals; a tariff sold as protecting Americans would, if charged, be largely paid by Americans.