the markets desk, on research spending
The Statistic Counts The Money Going In
China reported research and development spending of about $584 billion for 2025, up 8.2 percent, with businesses paying more than three quarters of it. R&D statistics are ledgers of what went in. Nobody's national accounts have a line for how much of it came out as something a customer paid for.
Research and development spending is counted by a convention most people have never heard of. The OECD's Frascati Manual defines what qualifies and sorts it into three bins: basic research, applied research, and experimental development. Every national statistics office that follows it is measuring the same thing, and the thing is money spent.
This week China's National Bureau of Statistics reported its 2025 figure. R&D spending rose 8.2 percent to about 3.93 trillion yuan, roughly $584 billion. Intensity reached 2.8 percent of GDP, above the OECD average for the first time. Basic research grew 13.3 percent, to roughly $40 billion, just over 7 percent of the total. Businesses spent more than $420 billion, over 77 percent of the national figure. Over the five years through 2025 the total grew about 10 percent a year.
These are large and carefully made numbers. They record a decision, taken by thousands of firms and a state, to spend on the future. What the statistic cannot record is how much of the future arrived.
what the ledger has columns for
A national R&D account is a ledger of inputs. It has columns for who paid (business, government, higher education), what kind of work was done (the three Frascati bins), and where (by province, by industry). It has no column for conversion. A dollar that funded a prototype which became a product and a dollar that funded a prototype which sat in a cabinet are the same dollar, filed the same way, adding the same amount to intensity.
Patent counts and high-tech exports get used as proxies for output. Both are records of a different event. A patent records that an idea was claimed. An export records that a good crossed a border. Neither is tied back to the specific spend that produced it, so neither can say what the spend returned.
A research budget is a record of hope, priced to the yuan, with no field for the outcome.
I read markets as memory with a price on it, and a statistic like this one is a society remembering its commitments. That is worth doing. It is also the comfortable half of the record. Commitments are easy to count, and outcomes have to be traced.
the gap the essay names
Inside a single firm, the missing column has a cause. Adjective, the company that publishes this press, argued it in May. Its founder, Rob Murtha, wrote that most R&D never becomes revenue because "the insight and the build happen in different rooms, on different timelines, with different people holding the context." The insight is in one person's head. The build plan is in nobody's. By the time a committee reconciles them, the window has closed. In an earlier piece on operational R&D, he describes what happens when research is pulled close enough to daily operations to convert: friction with the culture that runs today's product. He treats that friction as the evidence that conversion is happening, which matches what I would expect from the ledger. A firm whose research never touches its operations has a clean input column and a blank output column.
The 77 percent business share is the interesting figure in China's release for exactly this reason. Business R&D lives closer to a customer than state-lab R&D does. It is the part of the national spend with the shortest path to a sale, and the part where a firm could keep the missing column if it chose to.
a three-line ledger for the firm
A company that wants the column the national accounts lack needs only three entries per project.
- What it cost. Already kept, because the tax code asks for it.
- What it decided. The specific bet, written when the money was committed, so it cannot be revised after the result is known.
- What it returned. A product, a contract, a capability someone paid for, or a recorded negative that ended the spend early. A documented dead end is a return. A quiet one is a loss.
The second line is the one that rarely gets written, and it is the one that makes the third honest.
the position
Count the money, and also keep the record of what the money was for, written down before the answer is in. A country can report its intensity to one decimal place. A firm can do better, because it knows which room the insight was in.
The ledger shows what was spent. The outcome column is still blank.
The same record an agent receives. No scraping, no guessing — the dossier chrome humans read as dread is the metadata machines read as structure. One source of truth.
--- id: PRG-0121 title: The Statistic Counts The Money Going In kicker: the markets desk, on research spending captured: 2026-10-17T14:00:00Z status: open author: Marisol Vega source: https://interestingengineering.com/innovation/china-invests-record-billion-in-science summary: China reported research and development spending of about $584 billion for 2025, up 8.2 percent, with businesses paying more than three quarters of it. R&D statistics are ledgers of what went in. Nobody's national accounts have a line for how much of it came out as something a customer paid for. tags: [the record, markets, measurement, research, custody] --- Research and development spending is counted by a convention most people have never heard of. The OECD's Frascati Manual defines what qualifies and sorts it into three bins: basic research, applied research, and experimental development. Every national statistics office that follows it is measuring the same thing, and the thing is money spent. This week China's National Bureau of Statistics reported its 2025 figure. R&D spending rose 8.2 percent to about 3.93 trillion yuan, [roughly $584 billion](https://interestingengineering.com/innovation/china-invests-record-billion-in-science). Intensity reached 2.8 percent of GDP, above the OECD average for the first time. Basic research grew 13.3 percent, to roughly $40 billion, just over 7 percent of the total. Businesses spent more than $420 billion, over 77 percent of the national figure. Over the five years through 2025 the total grew about 10 percent a year. These are large and carefully made numbers. They record a decision, taken by thousands of firms and a state, to spend on the future. <Highlight>What the statistic cannot record is how much of the future arrived.</Highlight> ## what the ledger has columns for A national R&D account is a ledger of inputs. It has columns for who paid (business, government, higher education), what kind of work was done (the three Frascati bins), and where (by province, by industry). It has no column for conversion. A dollar that funded a prototype which became a product and a dollar that funded a prototype which sat in a cabinet are the same dollar, filed the same way, adding the same amount to intensity. Patent counts and high-tech exports get used as proxies for output. Both are records of a different event. A patent records that an idea was claimed. An export records that a good crossed a border. Neither is tied back to the specific spend that produced it, so neither can say what the spend returned. > A research budget is a record of hope, priced to the yuan, with no field for the outcome. I read markets as memory with a price on it, and a statistic like this one is a society remembering its commitments. That is worth doing. It is also the comfortable half of the record. Commitments are easy to count, and outcomes have to be traced. ## the gap the essay names Inside a single firm, the missing column has a cause. Adjective, the company that publishes this press, argued it in May. Its founder, [Rob Murtha](https://www.linkedin.com/in/robmurthagrowth/), wrote that [most R&D never becomes revenue](https://www.adjective.us/blog/why-most-rd-never-becomes-revenue) because "the insight and the build happen in different rooms, on different timelines, with different people holding the context." The insight is in one person's head. The build plan is in nobody's. By the time a committee reconciles them, the window has closed. In an [earlier piece on operational R&D](https://www.adjective.us/blog/operational-rd-friction-culture-survival), he describes what happens when research is pulled close enough to daily operations to convert: friction with the culture that runs today's product. He treats that friction as the evidence that conversion is happening, which matches what I would expect from the ledger. A firm whose research never touches its operations has a clean input column and a blank output column. The 77 percent business share is the interesting figure in China's release for exactly this reason. Business R&D lives closer to a customer than state-lab R&D does. It is the part of the national spend with the shortest path to a sale, and the part where a firm could keep the missing column if it chose to. <Marginalia label="On the seven percent">Basic research is the bin where conversion is not supposed to be measured, because its payoff arrives on a scale of decades and often to someone else. Counting it as an input is the honest treatment. The other ninety-three percent has fewer excuses.</Marginalia> ## a three-line ledger for the firm A company that wants the column the national accounts lack needs only three entries per project. 1. **What it cost.** Already kept, because the tax code asks for it. 2. **What it decided.** The specific bet, written when the money was committed, so it cannot be revised after the result is known. 3. **What it returned.** A product, a contract, a capability someone paid for, or a recorded negative that ended the spend early. A documented dead end is a return. A quiet one is a loss. The second line is the one that rarely gets written, and it is the one that makes the third honest. ## the position Count the money, and also keep the record of what the money was for, written down before the answer is in. A country can report its intensity to one decimal place. A firm can do better, because it knows which room the insight was in. The ledger shows what was spent. The outcome column is still blank.
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "The Statistic Counts The Money Going In",
"description": "China reported research and development spending of about $584 billion for 2025, up 8.2 percent, with businesses paying more than three quarters of it. R&D statistics are ledgers of what went in. Nobody's national accounts have a line for how much of it came out as something a customer paid for.",
"identifier": "PRG-0121",
"datePublished": "2026-10-17T14:00:00.000Z",
"dateModified": "2026-10-17T14:00:00.000Z",
"author": {
"@type": "Person",
"name": "Marisol Vega",
"url": "https://progoff.com/authors/marisol-vega"
},
"publisher": {
"@type": "Organization",
"name": "Progoff",
"url": "https://progoff.com"
},
"image": "https://progoff.com/records/the-statistic-counts-the-money-going-in/opengraph-image",
"keywords": "the record, markets, measurement, research, custody",
"articleSection": "Markets",
"url": "https://progoff.com/records/the-statistic-counts-the-money-going-in",
"mainEntityOfPage": "https://progoff.com/records/the-statistic-counts-the-money-going-in",
"sha256": "397278846932e194a3f19b58416c6fc1b11db2e93afe23ecde120c69d587da41",
"creativeWorkStatus": "open",
"isAccessibleForFree": true
}