Daniel sent us this one — he's pointing out that in the AI-saturated moment we live in, it's easy to forget an entire economy exists beyond knowledge work. Countries still classify their economic activity into these big buckets: agriculture, industry, services. But that three-part model is incredibly crude. He's asking how different countries actually break down their economic activity by sector, and then specifically — where does Israel fit? Because Israel has this massive knowledge sector now, this tech brand that everyone knows, but how much of the country's economy is still in everything else? It's a good question. The tech narrative swallows the whole story.
It absolutely does, and the numbers are genuinely surprising if all you know about Israel is the start-up nation branding. But before we get to Israel specifically, you're right that the three-sector model itself needs some poking at. I was looking at the World Bank's classification framework, and they still use agriculture, industry, and services as their primary buckets — but they've been subdividing services for decades now because the category became this giant grab bag. Everything from a guy cutting hair to a team building a large language model both land in services. That's the problem.
The haircut and the LLM, same bucket. It's like calling both a tricycle and an F-35 "vehicles.
And the distortion has gotten worse over time. In eighteen seventy, most economies were agricultural — the industrial revolution was still spreading. By nineteen fifty, industry dominated in developed nations. Now services account for something like sixty-five percent of global GDP, but that number is almost meaningless because it's so heterogeneous. The OECD started pushing member countries to break services into subcategories — market services versus non-market services, knowledge-intensive versus labor-intensive — specifically because the aggregate number was hiding more than it revealed.
When someone says "the US is eighty percent services," what are they actually saying?
They're saying that four out of every five dollars of economic output comes from something that isn't growing a crop or manufacturing a physical object. But that includes real estate, which is about thirteen percent of US GDP on its own. It includes healthcare — another eighteen percent. It includes finance, insurance, retail, transportation, information technology, professional services, hospitality, education. The composition matters enormously. A country that's eighty percent services because it has a massive tourism industry looks nothing like a country that's eighty percent services because it's a global financial hub, which looks nothing like a country that's eighty percent services because it's a tech export powerhouse.
The three-sector model doesn't catch any of that. It's the economic equivalent of describing a meal as "contains solids and liquids.
So let's actually put some numbers on this. The CIA World Factbook maintains sector composition data for every country, and the World Bank does its own estimates. If you look at the most recent figures — these are twenty twenty-three and twenty twenty-four estimates — the global average is roughly four percent agriculture, twenty-eight percent industry, and sixty-eight percent services. But those averages mask staggering variation.
Give me the extremes.
Sierra Leone — agriculture is over sixty percent of GDP. Somalia, about the same. These are economies where most people are subsistence farmers. On the other end, Singapore — agriculture is zero percent. They import virtually all their food. Hong Kong, zero point one percent. The UK, about zero point seven percent. So you have this spectrum from economies that are still fundamentally agrarian to economies where agriculture has essentially vanished as a share of output.
Which countries are still heavy on manufacturing and construction?
China is about thirty-nine percent industry, which is enormous for an economy of its size. But the real outliers are the petrostates and mineral exporters. Libya — industry is over fifty-eight percent of GDP, almost all of it oil and gas. Kuwait, fifty-one percent. Qatar, about forty-nine percent. Saudi Arabia has been trying to diversify, but industry is still around forty-four percent. And then you have countries like South Korea at thirty-two percent industry, which is significant because it reflects genuine manufacturing — ships, semiconductors, automobiles — not just resource extraction.
The industrial category has its own internal problem. An oil well and a semiconductor fab are both "industry.
Exactly the same issue as services. And the policy implications are completely different. An economy built on resource extraction faces Dutch disease — your currency appreciates, your other sectors become uncompetitive, and when the resource runs out or the price collapses, you're in trouble. An economy built on advanced manufacturing faces automation risk, supply chain concentration risk, but it's building transferable engineering capabilities. Same bucket, totally different futures.
Named after the Netherlands discovering natural gas in the nineteen sixties and watching their manufacturing sector wither.
That's the one. The canonical example. And it's still happening — look at what oil did to Venezuela, what it's doing to Nigeria. The three-sector model can't distinguish between an industrial base that's a strategic asset and one that's a resource curse waiting to happen.
Let's bring this to Israel. The prompt asks specifically where Israel fits. What's the official breakdown?
According to Israel's Central Bureau of Statistics — the most recent data from late twenty twenty-four and early twenty twenty-five — agriculture is about one point two percent of GDP. Industry, including construction, is about seventeen percent. And services are about eighty-two percent.
Eighty-two percent services. That sounds like a post-industrial knowledge economy, case closed.
That's what the headline number says. But here's where it gets interesting. Within that seventeen percent industry, Israel has a significant manufacturing sector that most people don't associate with the country at all. The big one is chemicals and pharmaceuticals. Teva, the generic drug manufacturer, is one of the largest companies in Israel by revenue. Israel Chemicals Limited — ICL — is a major producer of potash, bromine, and phosphate-based fertilizers. They're extracting minerals from the Dead Sea. That's heavy industry. That's not start-ups in a WeWork.
Dead Sea minerals. So part of Israel's industrial sector is basically mining.
Mining and quarrying is about one and a half percent of GDP on its own. Then there's the defense manufacturing sector — Rafael, Israel Aerospace Industries, Elbit Systems. These are large industrial employers producing physical hardware. Drones, missile defense systems, armored vehicles, avionics. That's manufacturing. And then there's the food processing industry — Strauss, Tnuva, Osem. These are major domestic manufacturers.
The seventeen percent industry figure is doing real work. It's not a rounding error.
It's not a rounding error, and it's strategically significant in ways that the percentage alone doesn't capture. Defense manufacturing in particular — that's not just an economic activity, it's a national security capability. You can't outsource your missile defense production to a country that might cut you off during a war. So the composition of that seventeen percent matters enormously.
The services side? Eighty-two percent is a big number, but you said it's heterogeneous.
Let me break it down. The largest subcategory within Israeli services is what the CBS classifies as "public and community services" — that's government, education, healthcare, social services. That's about twenty-two percent of GDP. Then you have what they call "business and financial services," which includes the tech sector, banking, insurance, legal services, consulting — that's roughly twenty-eight percent. Wholesale and retail trade, accommodation, and food services — that's about twelve percent. Transportation, storage, and communications — about seven percent. Real estate — about ten percent.
The tech sector — which is what everyone thinks of when they think of Israel's economy — is a subset of that twenty-eight percent business and financial services bucket.
And within that, the actual high-tech sector — software, R and D, semiconductors, cybersecurity, AI — is estimated at roughly fifteen to eighteen percent of GDP depending on how you define it. It's the single largest component of business services, but it's not the majority of the economy. The narrative that Israel is "the start-up nation" and therefore the entire economy is tech is just wrong. It's a major sector, it's the growth engine, it generates a disproportionate share of exports — but in terms of domestic economic activity, most Israelis do not work in tech.
What's the employment breakdown?
This is where the sector composition gets really revealing. According to the CBS labor force survey from early twenty twenty-five, total employment in Israel is about four point three million people. The high-tech sector employs roughly three hundred and seventy thousand people. That's about eight point six percent of the workforce.
Less than nine percent of workers.
Less than nine percent. And that figure has been relatively stable for the past few years. The sector is incredibly productive — those three hundred seventy thousand workers generate a wildly disproportionate share of exports and GDP — but it's not the mass employer that the branding suggests. The largest employers in Israel are the public sector, the healthcare system, the education system, retail, construction, and manufacturing.
The average Israeli worker is more likely to be a teacher or a nurse or a supermarket cashier than a software engineer.
By a large margin. And this creates a structural tension in the economy that doesn't show up in the sector breakdown. You have a high-productivity, high-wage export sector that employs less than one in ten workers, and then you have a much larger domestic services sector where productivity growth has been sluggish and wages have been relatively flat. The OECD has been flagging this for years — Israel has what they call a "dual economy" problem. The tech sector is world-class, competing with Silicon Valley on productivity and wages. The rest of the economy looks more like southern Europe.
That split maps onto some of Israel's social divisions, I imagine.
It maps very directly. The tech workforce is disproportionately secular, disproportionately Jewish, disproportionately concentrated in the Tel Aviv metropolitan area and a few other coastal cities. Arab Israelis, Haredi Jews, and residents of the periphery — the Galilee, the Negev — are significantly underrepresented in the tech sector. Their employment is concentrated in manufacturing, construction, agriculture, retail, and public services.
The sector composition isn't just an accounting exercise. It's tracking social and political fault lines.
It's the economic geography of some of the country's deepest tensions. And this is where the three-sector model is particularly useless. It tells you Israel is eighty-two percent services. It doesn't tell you that within services, you have a bifurcation between a globally competitive knowledge sector and a domestically oriented, lower-productivity services sector. It doesn't tell you that this bifurcation has ethnic, religious, and geographic dimensions. It doesn't tell you that the growth engine of the economy is disconnected from the lived experience of most of the population.
That's the thing that always strikes me about these aggregate numbers. They can be technically accurate and substantively misleading.
The technical term for what we're describing is "economic dualism," and it's not unique to Israel. You see versions of it in many countries. India has a world-class IT services sector and a vast agricultural and informal economy. Italy has a highly productive industrial north and a less developed south. But Israel's version is particularly stark because the gap between the tech sector and the rest is so wide, and because the country is small enough that the tension is inescapable.
What about agriculture? One point two percent of GDP — that's tiny. But I know Israel has this outsized reputation in agricultural technology.
This is one of the ironies. Israel's agricultural sector is tiny in terms of GDP share and employment — about one percent of the workforce — but it's globally influential in agricultural technology. Drip irrigation was invented in Israel. Desalination technology, water recycling, precision agriculture, crop genetics — Israel is a major exporter of agritech. So the sector's economic footprint is small, but its technological footprint is large. The knowledge generated in agriculture gets exported, even though the domestic production is modest.
The agriculture sector produces technology that's more economically significant than the crops themselves.
And that's another limitation of the three-sector model — it classifies economic activity by the output, not by the knowledge embedded in the output. A company that develops irrigation sensors is classified under manufacturing or tech services, not agriculture, even though its entire business is agriculture-adjacent. The sector boundaries blur in ways the model can't capture.
Let's zoom out to the international comparison for a moment. How does Israel's sector composition compare to other developed economies?
Israel's breakdown — roughly two percent agriculture, seventeen percent industry, eighty-two percent services — is actually quite similar to other developed economies. The US is about one percent agriculture, eighteen percent industry, eighty-one percent services. The UK is zero point seven percent agriculture, seventeen percent industry, eighty-two percent services. France is about one point six percent agriculture, seventeen percent industry, eighty-one percent services. So at the aggregate three-sector level, Israel looks like a typical advanced economy.
Which is exactly the problem with the model. It makes every developed economy look the same.
The three-sector model was developed in the nineteen thirties and forties — Allan Fisher, Colin Clark, Jean Fourastié. It was designed to track the transition from agrarian to industrial to service-based economies. It was useful for understanding the broad arc of development. But once you reach the service-dominated stage, which most developed economies did by the nineteen eighties or nineties, the model stops discriminating. It tells you you've arrived, but it doesn't tell you what kind of economy you've arrived at.
The model's job is done once you hit the services plateau. After that, you need different tools.
Those tools exist. The OECD's STAN database breaks down economic activity into dozens of sub-sectors. National statistical agencies produce input-output tables that show how sectors interconnect. There are classifications like NACE in Europe and NAICS in North America that go into granular detail. The problem isn't that we can't measure economic composition more precisely — it's that the three-sector summary is so convenient for headlines and political rhetoric that it persists long after its useful life.
The "we're a services economy" line is the economic equivalent of saying "I'm a person who eats food." It's true but it tells you nothing.
It can actually obscure important structural shifts. For example, there's been a lot of discussion in recent years about "reindustrialization" — countries trying to rebuild domestic manufacturing capacity after decades of offshoring. The US CHIPS Act, the European Chips Act, the various supply chain reshoring initiatives. If you're just looking at the three-sector breakdown, you might not see these shifts for years because they're happening within the industrial category — the composition of industry is changing from low-value assembly to high-value semiconductor fabrication, but the aggregate share might not move much.
Even the industrial category is experiencing internal transformation that the model masks.
The services category is experiencing even more dramatic internal transformation. The rise of AI is going to reclassify a lot of economic activity. If a law firm replaces half its associates with AI tools that do document review, the output might still be classified as legal services, but the nature of the work, the employment structure, the productivity dynamics — all completely different. The sector label stays the same while the sector transforms.
Which brings us back to Daniel's point about the AI-centric era. We're in a moment where the most economically significant transformation is happening within a category that the standard model doesn't subdivide.
That transformation is going to hit different economies differently. Countries where a large share of services employment is in routinized cognitive work — data entry, basic accounting, call centers, paralegal work — those jobs are the most exposed to AI automation. Countries where services are dominated by in-person, non-routinized work — elder care, hospitality, construction — are less exposed in the near term. Same "services" label, completely different vulnerability profiles.
Where does Israel sit on that spectrum?
It's a mixed picture. The tech sector is obviously going to benefit from AI — Israel is a major AI research and development hub. But a significant portion of Israel's services employment is in sectors that are relatively shielded from near-term AI disruption. Healthcare, education, and social services are large employers where the work is fundamentally interpersonal. Tourism and hospitality — also significant in Israel — requires physical presence. The sectors most exposed to AI disruption in Israel are probably finance and insurance, some business services, and certain government administrative functions.
Israel's services composition might actually be somewhat resilient to the first wave of AI-driven labor market disruption, relative to economies that are heavier on back-office processing and call centers.
That's a reasonable reading. India and the Philippines, for example, have built large business process outsourcing sectors that are directly in the line of fire. Israel doesn't have that same exposure. The tech sector is likely to grow, and the domestic services sector is weighted toward activities that are harder to automate.
Let's go back to something you mentioned earlier — the dual economy problem. The OECD has been warning about this. What do they actually recommend?
The OECD's economic surveys of Israel — the most recent ones from twenty twenty-three and twenty twenty-four — consistently recommend a few things. First, improve education and training for populations that are underrepresented in the tech sector. That means investing in STEM education in Arab schools, in Haredi communities, in the periphery. Second, reduce regulatory barriers that protect low-productivity domestic sectors from competition — Israel has some of the highest non-tariff barriers to imports in the OECD, which keeps prices high and reduces pressure on domestic firms to improve. Third, invest in infrastructure — public transportation, digital connectivity — to better connect peripheral regions to the economic center.
How much of that is actually happening?
Some of it. The government has expanded tech training programs for underrepresented groups — there are coding bootcamps, there are initiatives to integrate Haredi men and women into the tech workforce. The results have been mixed. The Haredi employment rate has increased significantly over the past decade, from around fifty percent to over fifty-five percent for men and substantially higher for women, but most of that employment is not in high-tech — it's in education, administration, and services. The Arab tech employment rate has grown but remains very low, around three to four percent of the tech workforce despite Arabs being about twenty-one percent of the population.
The dual economy is proving stubborn.
And part of the reason is that the forces driving the dual economy aren't just about skills and training. There are network effects in tech employment — people hire people they know, people they served with in the military's technology units, people from their social circles. There are cultural barriers. There are geographic barriers — most tech companies are in the Tel Aviv area, and commuting from the periphery is difficult given Israel's congested transportation infrastructure. The sector composition numbers are describing a set of structural realities that can't be changed with a few training programs.
It's interesting that the three-sector model was originally developed partly as a tool for understanding development — the idea that economies progress from agriculture to industry to services as they develop. But Israel's case shows that reaching the services stage doesn't resolve the development challenge. You can have a highly developed tech sector and still have large populations that are effectively in a different economic era.
That's why Amartya Sen and others pushed back against GDP-centric development metrics and pushed for the Human Development Index and the capabilities approach. The sector composition tells you what the economy produces, but it doesn't tell you who participates in that production or who benefits from it. Israel ranks very high on the HDI — twenty-second globally in the most recent rankings — but that aggregate masks significant internal variation. The HDI for the Arab population is lower than the national average. The HDI for Haredi communities is lower. The aggregate is a weighted average that hides the distribution.
The distribution question is really what the prompt is getting at, I think. It's not just "what are the sectors," it's "how much of the economy is still in other forms of activity" — meaning, beyond the knowledge sector that dominates the international perception. The answer turns out to be: most of it.
Most of it in employment terms, and a significant share in output terms. The tech sector is about fifteen to eighteen percent of GDP and less than nine percent of employment. The rest of the economy — manufacturing, construction, agriculture, retail, healthcare, education, government, tourism, transportation, real estate — that's over eighty percent of GDP and over ninety percent of employment. The start-up nation is a real thing, but it's not the whole thing. Not even close.
I'm curious about the export picture, because that's where the tech dominance really shows up. What's the export composition?
This is where the contrast is sharpest. According to the Israel Export Institute and the Central Bureau of Statistics, high-tech exports — that's software, R and D services, semiconductors, electronics, pharmaceuticals, and advanced manufacturing — account for about fifty to fifty-five percent of Israel's total exports. In some years it's been as high as sixty percent. The rest is split among polished diamonds, which is a separate category and has been declining, chemicals and minerals, agricultural products, and traditional manufacturing.
The tech sector is less than a fifth of GDP but more than half of exports.
And that's the duality in a nutshell. The tech sector is globally integrated and export-oriented. The rest of the economy is largely domestically oriented. When people abroad think of the Israeli economy, they see the export profile — cybersecurity firms, AI start-ups, semiconductor design, pharmaceutical patents. They don't see the domestic economy — the supermarkets, the hospitals, the construction sites, the government offices. But that domestic economy is where most Israelis work and where most of the consumption happens.
It's like judging the US economy entirely by Silicon Valley and Wall Street.
It's exactly like that. And people do that too. The perception gap between a country's export brand and its domestic economic reality is a general phenomenon, but it's particularly pronounced for Israel because the export brand is so strong and so concentrated in a few highly visible sectors.
What about the construction sector? You mentioned it's part of industry. I know Israel has been building at a rapid pace.
Construction is a significant part of the Israeli economy — about five to six percent of GDP, which is higher than in most OECD countries. Israel has one of the highest population growth rates in the developed world, driven by relatively high fertility rates. The population is growing at about two percent per year, which means you need a lot of housing, a lot of infrastructure, a lot of schools. Construction employment is around two hundred thousand workers, and a large share of those workers are Arab Israelis and foreign workers, particularly in the manual trades.
Construction is a major employment sector for populations that are underrepresented in tech.
It's one of the primary economic bridges for Arab Israeli communities. And it's a sector that's very sensitive to the economic cycle and to security situations. When there's a conflict, construction often slows down because of restrictions on Palestinian workers entering Israel, who have historically been a major part of the construction workforce. The sector composition data captures the output, but it doesn't capture the complex labor dynamics and the security-economy linkages.
Let's talk about one more thing that the three-sector model misses entirely — the informal economy. How big is it in Israel?
The informal economy in Israel is estimated at about fifteen to twenty percent of GDP, which is actually lower than in many other countries — in some Mediterranean economies it's over twenty-five percent — but it's not trivial. It's concentrated in certain sectors: domestic services, small construction and renovation, some agricultural work, tutoring, cash-based businesses. The CBS does attempt to estimate it and include it in the national accounts, but by its nature it's hard to measure precisely.
The informal economy doesn't show up in the sector breakdown at all, or it's imputed based on assumptions.
It's imputed. The official sector composition data is based on a combination of tax records, business surveys, and statistical estimation. The informal economy is by definition off the books, so it has to be estimated indirectly — through currency demand models, through discrepancies between income and expenditure in household surveys, through labor force participation data. It's a best guess, and it varies in quality across countries.
When we say Israel is eighty-two percent services, that number is constructed from a mix of hard data and statistical inference, with a margin of error that's probably a few percentage points.
And that's true for every country. The sector composition data is useful for broad comparisons and trend analysis, but it shouldn't be treated as precise accounting. It's an estimate built on methodology that varies across national statistical agencies. Comparing Nigeria's sector breakdown to Norway's requires understanding that the underlying data quality and estimation methods are completely different.
Which is another reason the three-sector model is crude. It implies a comparability that may not exist.
The World Bank and the IMF do attempt to harmonize the data, but there are limits. Some countries haven't updated their sector composition estimates in years. Some use different classification systems. Some have large informal sectors that are estimated differently. The numbers look precise — seventy-two point three percent services — but they're approximate.
Let's pull this together. Daniel's prompt asked two things: how countries classify economic activity by type, and where Israel fits. On the classification side, the answer is that the dominant framework is still the three-sector model, but it's increasingly inadequate for understanding modern economies. It lumps together vastly different activities, it masks internal dualism, and it doesn't capture the knowledge-intensity or the distribution of economic participation.
On the Israel side: the official breakdown is about one percent agriculture, seventeen percent industry, and eighty-two percent services. That looks like any other advanced economy at the aggregate level. But the composition within those categories reveals a dual economy — a globally competitive, export-oriented tech sector that employs less than nine percent of the workforce and generates over half of exports, and a much larger domestic economy that looks more like a typical middle-income country in terms of productivity and wages. The tech brand is real, but it's not the lived economic experience of most Israelis.
The dualism maps onto Israel's social divisions — the tech sector is concentrated geographically in the Tel Aviv area and demographically among secular Jews, while Arab Israelis, Haredi Jews, and residents of the periphery are concentrated in the domestic services, manufacturing, and construction sectors.
The sector composition data, properly disaggregated, is basically a map of Israel's economic and social structure. The headline numbers hide more than they reveal, but the detailed sub-sector data tells a rich story about who does what, where, and for what reward.
One last thing I want to touch on — you mentioned the OECD's reindustrialization push, the semiconductor subsidies, the reshoring. Is Israel participating in that trend?
Israel is in an interesting position. It doesn't have the kind of large-scale consumer electronics manufacturing that China and Southeast Asia dominate. But it does have Intel's largest fabrication plant outside the US — Intel has been in Israel for decades and has invested over fifty billion dollars there cumulatively. The Kiryat Gat fab is one of the most advanced semiconductor manufacturing facilities in the world. And Israel has a growing cluster of chip design firms — not just Intel, but also NVIDIA, Apple, Amazon, and many Israeli start-ups doing chip design for AI and other applications.
Israel is part of the semiconductor supply chain, but more on the design and specialized fabrication side than on the mass-production side.
And that's consistent with the broader pattern — Israel's industrial sector is concentrated in high-value, knowledge-intensive manufacturing rather than in labor-intensive assembly. The sector composition says "seventeen percent industry," but it's a very specific kind of industry. It's not the industry of the nineteenth-century textile mill or the twentieth-century automobile assembly line. It's the industry of the twenty-first-century clean room and the missile guidance system.
The three-sector model was born in an era when industry meant smokestacks. It still carries those connotations even though the actual industrial activity has transformed.
That's why I think the model persists despite its limitations. It provides a familiar vocabulary. People know what you mean when you say "agriculture," "industry," and "services." The problem is that they think they know more than they actually do. The words are familiar, so the concepts feel clear, even when they're not.
The illusion of understanding through familiar labels.
That's the human condition in a lot of domains.
I think we've given this a proper going-over. The three-sector model — useful as a first approximation, misleading as a final answer. Israel — a services economy on paper, a dual economy in reality, with a tech sector that punches far above its weight in exports and global visibility but doesn't employ most of the country. And the real story is in the sub-sector detail, which is where the interesting questions about productivity, inclusion, and economic resilience actually live.
I'd add — this is a live issue. The government's economic planning, the Bank of Israel's monetary policy, the allocation of investment — all of these depend on understanding the structure of the economy. If the model you're using groups together things that shouldn't be grouped together, you're going to make policy mistakes. The dual economy isn't just an academic observation — it has real consequences for interest rates, for exchange rate management, for education policy, for social cohesion.
The classification isn't just accounting. It's the lens through which decisions get made.
If the lens is blurry, the decisions will be too.
Now: Hilbert's daily fun fact.
Take it away, Hilbert.
Hilbert: In the eighteen forties, the French mathematician Joseph-Alphonse Adhémar published a theorem about the precession of the equinoxes that was almost entirely ignored for a hundred and fifty years. When it was finally rediscovered in the nineteen nineties, researchers estimated that at any given time, there are approximately two thousand seven hundred proven but completely forgotten mathematical theorems sitting in obscure nineteenth-century journals, most of them in European library basements. The lost-to-rediscovered ratio for the eighteen forties alone is about nine to one.
Nine to one. So for every theorem we remember, nine are just...
In a basement. In Chad, apparently, but mostly Europe.
The global supply chain of forgotten math.
We should check the basement.
We absolutely should not. This has been My Weird Prompts, with thanks to our producer Hilbert Flumingtop. If you enjoyed this, leave us a review wherever you get your podcasts — it helps. And you can find every episode at myweirdprompts dot com.
See you next time.
