Daniel sent us this one — and it starts with something that actually happened last week. On July seventh, Ireland made it a criminal offense to buy a widget made in an Israeli settlement in the West Bank. The trade volume here is under ten million euros a year. That's a rounding error. But the mechanism they used to do it — a bureaucratic system for distinguishing Jewish communities from non-Jewish ones in the same territory — raises a much bigger, weirder question. If you look at every trade relationship between every UN member state, subdivided into imports, exports, goods, and services, do you ever actually find a zero? Or is there always some tiny trickle of trade, somewhere?
Daniel's pushing past the Irish bill itself to ask something genuinely interesting about the statistical distribution of global trade. What's the yardstick for negligible trade? Does every country have its own de minimis level, or is there some global threshold? And the academic point underneath it all — minus embargoes and the absence of diplomatic relations, does every country trade with every other country to some extent, or are there economic relationships that basically don't exist?
Which is exactly the kind of question that sounds simple and turns out to be a data collection nightmare. So let's start with what Ireland actually did, because the mechanism matters for the trade data question.
The Occupied Territories Bill passed on July seventh. It criminalizes the purchase of goods produced in Israeli settlements in the West Bank and East Jerusalem. And the way it works is fascinating from a trade law perspective — it relies on a bureaucratic mechanism to distinguish between Jewish communities and non-Jewish communities in the same territory. Goods from Jewish settlements are banned. Goods from Palestinian communities in the same geographic area are not.
Which is why Daniel called it bigoted legislation, and he's not wrong. Ireland hasn't extended anything like this to any other conflict zone — not Crimea, not Western Sahara, not northern Cyprus, not anywhere. It's a targeted mechanism for one specific set of communities in one specific territory.
During the debate, there were voices arguing the bill didn't go far enough. The ban should extend to services, not just goods. If you're going to boycott settlements, why stop at physical products? Why not include consulting, software, legal services, architectural work?
That's where the trade data question gets interesting. Because you can't really do services in the same way. The bureaucratic mechanism for distinguishing settlement goods works through physical origin labeling — where was this product made, what's the postal code of the factory. Services don't have origin labels. You can't look at a consulting contract and say "this advice was generated in a settlement.
And that goods versus services distinction is the crack in the sidewalk that leads to Daniel's bigger question. If services are so hard to trace, how do we even know what trade exists between any two countries? The Irish bill debate exposed something fundamental about how we measure international commerce — we're really good at counting containers on ships and really bad at counting everything else.
That's the Irish bill and the question it raises. But to answer whether every country trades with every other, we need to understand how trade data actually works — and where it breaks down.
Let's start with the main source. The UN COMTRADE database is the big one — it's where researchers go to look at bilateral trade flows between countries. You've also got the IMF's Direction of Trade Statistics, and then every country has its own national customs agency reporting data upward. The system works through something called harmonized system codes — every product that crosses a border gets a six-digit code, and customs agencies record the value, the origin, the destination.
Already you can see the problem. That system is designed for physical goods crossing borders with paperwork. A shipping container full of coffee beans? A software license sold online from a company in one country to a customer in another?
And even for goods, there are reporting thresholds. Many countries don't report trade below certain values. If you're a small country with a small customs agency and limited resources, you're not going to spend hours documenting a five hundred dollar shipment of handicrafts. You'll just let it through.
Right out of the gate, "zero trade" doesn't mean zero trade. It means zero trade above the reporting threshold that someone bothered to document.
This is where the COMTRADE data gets revealing. Researchers have found that in any given year, roughly thirty to forty percent of all possible country pairs show zero recorded trade. That's a huge number. If you take all one hundred and ninety-three UN member states and look at every possible bilateral pair, nearly four in ten of those relationships show nothing in the official data.
That's goods, right? What about services?
Services are a disaster. The WTO estimates that services trade data coverage is about sixty percent of what it should be. And that's being generous. When you try to drill down to specific bilateral services flows between smaller economies, the data often just doesn't exist. The WTO's own services trade database has massive gaps.
Daniel's specific example — how many services did Denmark import from Costa Rica last year? What's the answer?
We don't know. There is no public data for that bilateral services relationship. For goods, we can actually find numbers — Denmark imported about two point three million dollars in goods from Costa Rica in twenty twenty-four, mostly coffee and medical instruments. That's a real number you can look up. Denmark to Costa Rica, Costa Rica to Denmark — there's no publicly available breakdown at that level of granularity.
Which means the honest answer to "does Denmark trade services with Costa Rica" is "probably, but we can't prove it.
That's the thing. If you think about it for five seconds, of course there's services trade between Denmark and Costa Rica. A Danish tourist goes to Costa Rica and stays in a hotel — that's Costa Rica exporting accommodation services to a Danish consumer. A Costa Rican freelancer does a fifty dollar design job for a Danish company on Upwork — that's a services export. A Danish shipping company insures a vessel through a Costa Rican intermediary. None of this shows up in bilateral services trade statistics.
We've got two problems stacked on top of each other. Problem one: goods trade data misses anything below reporting thresholds. Problem two: services trade data misses almost everything at the bilateral level. Put them together and "zero trade" between two countries is basically a statement about data collection, not about economic reality.
That brings us to the de minimis concept. What counts as negligible? For Ireland, the trade volume with Israeli settlements is estimated at under ten million euros annually. Compare that to Ireland's total goods trade with Israel proper — about two point five billion euros a year. The settlement trade is less than half a percent of that. It's a rounding error in Ireland's two hundred billion plus total goods trade.
Is it zero? It's small, but it's not zero. And the Irish bill proves it's not zero, because they went to the trouble of banning it. You don't ban something that doesn't exist.
And this is where the political dimension crashes into the statistical one. The Irish bill is unique precisely because the trade volume is so small. Ireland can afford to ban settlement goods because it costs them almost nothing in economic terms. The political symbolism is enormous, but the economic impact is negligible.
Which is the opposite of how trade sanctions usually work. Normally you sanction a country because the trade volume is large enough to hurt. Here, the trade volume is so small it's almost cost-free to ban it. It's virtue signaling with customs paperwork.
The bureaucratic mechanism makes it possible. The bill works because Israel has a system of origin labeling that distinguishes products from settlements. The Irish customs authority can look at a product, see the origin code, and say "this is from a settlement, banned." Without that labeling infrastructure, the bill would be unenforceable.
Which is why services got left out. You can't label a service. You can't look at a consulting invoice and determine the geographic origin within a disputed territory. The data infrastructure doesn't exist.
That's the knock-on effect Daniel's question points toward. The Irish bill reveals a fundamental asymmetry in trade law. Goods are boycottable because they're traceable. Services are harder to boycott because they're harder to trace. But that asymmetry isn't about economics — it's about data collection. The goods that get boycotted are the goods we can see.
Let's step back to the big question. Does every country trade with every other country? The short answer is no, not in any measurable sense.
Let me give you a concrete example. Total goods exports: about two hundred thousand dollars a year, mostly to Fiji and Australia. Trade with Denmark? Effectively zero in the official data. There might be the occasional handicraft sold to a Danish tourist passing through, but it's not showing up in COMTRADE. For all practical purposes, the Tuvalu-Denmark trade relationship does not exist in goods.
If a Tuvaluan sells a digital artwork as an NFT to someone in Denmark, that's trade. It won't appear in any official statistic, but economically, it happened.
And that's the power-law distribution of global trade. A handful of country pairs account for the vast majority of trade. The US and Canada, China and the US, Germany and France — these are enormous flows measured in hundreds of billions. Then you've got a long tail of medium-sized relationships. Then you've got a very long tail of near-zero pairs where trade is either tiny or completely unmeasured.
The yardstick for "negligible" is not global — it's country-specific. For Tuvalu, a ten thousand dollar export is significant. For China, a ten million dollar trade flow might not even be worth noting. The de minimis threshold scales with the size of the economy.
China's an interesting case here. Even with the smallest partners — places like San Marino or the Cook Islands — there's usually some recorded flow. Chinese customs is thorough, and Chinese firms export everywhere. But for a small island nation in the Pacific, trade with anyone outside their immediate region might be de minimis by their own standards.
The "complete network" assumption — the idea that every country trades with every other — is false for goods if you're looking at official data. But it might be true for services if you include digital services, because anyone with an internet connection can trade services with anyone else.
That's the wild card. A freelancer in Costa Rica doing a fifty dollar design job for a Danish company — that's services trade. A Kenyan developer contributing to an open source project maintained by a German nonprofit — is that trade? What about a Brazilian musician whose song gets streamed by someone in Lithuania and generates a fraction of a cent in royalties?
These are real economic flows. Money moves across borders. But they're invisible to trade statistics because they don't pass through customs. They pass through payment processors and streaming platforms and freelance marketplaces, none of which report bilateral trade data in a standardized way.
This is where the Irish bill's goods versus services debate becomes a preview of future trade conflicts. If you want to boycott a country or a territory, goods are easy — you block the containers. Services are a nightmare. How do you stop your citizens from hiring freelancers in a boycotted territory? How do you block streaming royalties? How do you prevent consulting contracts?
You'd need an entirely different enforcement mechanism. The Irish bill works because customs officials can inspect physical products. To extend it to services, you'd need to monitor financial transactions, internet traffic, contract law — it would be a completely different kind of state apparatus.
That's the escalation Daniel's question implies. If the Irish approach were generalized — if every country passed similar laws for every conflict zone they cared about — the trade data problem would shift from academic to operational overnight. Customs agencies would need to answer questions about product origins that they've never had to answer before. They'd need to distinguish not just between countries but between communities within disputed territories.
Which is basically impossible at scale. The Irish bill works because it targets one specific, well-labeled set of products from one specific territory. It's a bespoke solution. Generalize it and the system collapses under its own complexity.
That's before you even get to the question of whether it's legal under WTO rules. But that's a different episode.
Where does this leave us? Let's pull out some practical takeaways for how to think about trade statistics — and what the Irish bill teaches us about the limits of trade data.
First takeaway: when you see a claim that country X doesn't trade with country Y, be skeptical. It almost always means "no trade above reporting thresholds" or "services trade is unmeasured." The absence of evidence is not evidence of absence. Two countries can have a real economic relationship that's completely invisible to official statistics.
Second: the de minimis threshold is not universal. It depends on the size of the economy and the reporting capacity of the customs agency. For small economies, "zero trade" with distant partners is common in the data. For large economies, it's rare. You have to know who's doing the measuring.
Third: the goods versus services distinction isn't just an accounting technicality. It's becoming a frontline in trade law and trade politics. Goods are boycottable because they're visible. Services are harder to target because they're invisible. As the global economy shifts toward services and digital products, our ability to enforce trade restrictions is eroding.
Fourth, the Irish bill is a perfect case study in how trade data becomes political. The trade volume is negligible — under ten million euros in a multi-billion-euro trade relationship. But the data infrastructure that makes the boycott possible — the origin labeling system — is what makes it legally and bureaucratically feasible. The bill isn't about economics. It's about using customs mechanisms to make a political statement.
That's the thing to watch going forward. If this bill survives legal challenges — and there will be challenges, both in Irish courts and potentially at the European level — expect more countries to attempt similar targeted trade legislation. Not because the trade volumes matter economically, but because the customs mechanism provides a way to signal political alignment without bearing real costs.
You're using trade data infrastructure to distinguish between communities based on ethnicity or religion, in the same geographic territory. That's not how customs systems are supposed to work.
Customs systems are designed to distinguish between countries, not between neighborhoods. The Irish bill is essentially asking customs officials to do something the system was never built for — to look at a product from the West Bank and determine not just which territory it came from, but which community within that territory produced it.
Which brings us back to Daniel's question about the yardstick for negligible trade. The Irish bill proves that "negligible" is a political judgment, not an economic one. Ten million euros is negligible by any normal trade standard. But it's not negligible if you're trying to make a point.
That's the open question going forward. As digital services grow, will the concept of "zero trade" become obsolete? Every country with internet access trades services, even if invisibly. The Costa Rican freelancer, the Kenyan developer, the Brazilian musician — they're all part of global trade flows that our statistical systems can't see.
The data infrastructure is lagging behind the economy it's supposed to measure. We're still counting containers while the real growth is in bits and bytes.
That's not going to fix itself. The WTO knows this — they've been talking about the services data gap for years. But fixing it requires international coordination on a massive scale, and the political will just isn't there. Countries don't want to invest in better trade data collection because better data might reveal things they'd rather not measure.
Like services trade with settlements they've just banned.
The Irish bill is silent on services not because services trade with settlements doesn't exist, but because measuring it would be politically inconvenient and bureaucratically impossible. It's easier to pretend it's zero.
The answer to Daniel's question — does every country trade with every other country — is no, if you mean measurable goods trade above reporting thresholds. But yes, probably, if you include services and digital flows and tiny transactions that fall below the statistical radar. The "zero" in trade data is almost always an artifact of measurement, not a fact about the world.
The yardstick for negligible trade is specific to each country, scaled to the size of its economy and the capacity of its customs agency. There's no global threshold. What's negligible for China is significant for Tuvalu. What's invisible to COMTRADE might be perfectly real to the people doing the trading.
One last thing worth noting. The Irish bill passed on July seventh. It's now July thirteenth. In less than a week, we've already seen diplomatic fallout — and the trade data question has gone from academic hypothetical to operational reality. Irish customs officials are now legally required to distinguish settlement goods from non-settlement goods. That's a real bureaucratic process that didn't exist a week ago.
It's going to generate data. Ironically, the bill might produce the most detailed dataset on settlement goods trade that's ever existed — because now Irish customs has to track it. The boycott creates the very data that makes the trade visible.
Which is a neat little paradox. The thing you ban becomes the thing you measure most carefully.
And now: Hilbert's daily fun fact.
Hilbert: In seventeen twenty-three, the Mongolian navy — consisting of a single captured Chinese junk — nearly defeated the entire Russian Baltic fleet when a sudden storm in Lake Baikal drove the Russian ships onto rocks while the Mongolian vessel, crewed by horsemen who had never sailed before, simply rode out the weather by tying themselves to their horses on deck. The Russians lost three ships. The Mongolians lost zero. It remains the only naval engagement in history where the winning side had more horses than sailors.
I have so many questions, and I'm choosing to ask none of them.
The Mongolian navy. Thank you, Hilbert.
Here's what I'm left thinking about. The Irish bill might be a test case for a new kind of trade restriction — one that's politically symbolic, economically negligible, and bureaucratically innovative. If it survives legal challenges, expect more countries to try similar things. And each one will force customs systems to answer questions they weren't designed for — about product origins, about community distinctions, about the granularity of trade data. The de minimis question stops being academic and starts being operational.
The services gap is only going to get bigger. Every year, more trade shifts from physical goods to digital services. Our statistical systems are still designed for the twentieth century economy. By the time we fix the data, the economy will have moved on again.
Thanks to our producer Hilbert Flumingtop. This has been My Weird Prompts. If you enjoyed this episode, tell someone who's ever looked at a trade statistic and wondered what it actually means. We're at my weird prompts dot com.
See you next time.
