How has China’s rise as a science, research and technology powerhouse reshaped how research gets done across the world? One admittedly simplistic way to track this is to look at a single widely available statistic: R&D spending. (It’s worth keeping in mind that R&D spending is not precisely “science”–it does include spending on research projects by academic institutions, but most of it is actually corporate expenditures.) The story those numbers tell is less alarming for the US than you might assume from a lot of reporting, but the shift in the global distribution of research does create new issues.

While the OECD is the usual go-to source for cross-country data on R&D spending, I am not sympathetic with all of the technical choices they make. In particular, they present all their figures in terms of purchasing power parity, which in China’s case means R&D spending is converted to US dollars at an exchange rate of around 4, instead of 6-7. While the OECD charts show China’s total R&D spending overtaking that of the EU and closing in on the US, that’s mostly an artifact of this particular methodological choice.

Using purchasing power parity certainly sounds all proper and economist-y but it’s not automatically appropriate for every purpose. The PPP exchange rates were developed to compare the living standards of ordinary people across countries by laboriously comparing prices for a basket of the goods and services they consume; since a lot of these are not traded across borders their prices can vary widely. It’s not clear this is the right choice for comparing R&D expenditures, which are going to be mainly on salaries of highly skilled staff and specialized equipment. To me it seems more likely that there is in fact a global market for top researchers and their gear, and that therefore market exchange rates are appropriate.

To come up with more accurate charts, I made my own cross-country comparison of R&D spending. The procedure is simple: take the R&D share of GDP reported for a country in the World Bank’s World Development Indicators database, multiply that by its annual GDP, and convert to US dollars at market exchange rates. Since the R&D share of GDP doesn’t change a lot from year to year, in cases where it hasn’t been updated yet I use the previous year’s share times the current year’s actual GDP.

I ended up with 56 major economies where the World Bank has more than scattershot data on R&D spending, with data up to 2021. That’s not a complete sample of the world of course, but if a country can’t report its R&D spending to international organizations consistently then it probably isn’t capable of doing a lot of R&D spending anyway. Aggregating the countries into large regions generates the following result in terms of absolute values:

The rise of China is indeed pretty dramatic. In particular it has overtaken the combined R&D spending of developed countries in its own region of Asia (Australia, New Zealand, Japan, South Korea, Taiwan, Singapore), which has been somewhat stagnant for the past decade. R&D spending elsewhere in the world has continued to grow at a decent clip though. So it’s also useful to look at the relative shares:

To me this chart is even more interesting. Back in 1995, when data for most countries becomes available, global R&D spending was roughly equally distributed across three groups of developed economies, those in North America, Europe and Asia-Pacific. China’s rise has come mainly at the expense of Europe and developed Asia, whose R&D spending has not grown as rapidly in dollar terms. The US has done better and has actually kept a high share of global R&D spending, with its share rising not falling in the last several years. Other developing countries (admittedly not as well represented in this sample) have attained a marginally higher share of global R&D over the past decade, but nothing like China.

Based on recent trends, it seems that global R&D spending is becoming less evenly distributed across the world, and is increasingly concentrated in the two hubs of the US and China. That does line up with anecdotal impressions: the US and China are home to the two main clusters of large internet companies, and are also the two leading locations for artificial-intelligence research.

These two hubs are, obviously, not talking to each as much as they used to. There has been only minimal travel in and out of China for the past three years, and the recent political climate in the US has made collaboration with Chinese researchers much more fraught. If ties between the US and China stay troubled, then this more polarized distribution of global R&D spending might turn out to be a less efficient allocation of resources.

Those research dollars will do the most good for technological progress in the world as a whole if they are spent in complementary ways. If instead the two hubs are pursuing conflicting or duplicative agendas, then the same global sum of R&D spending could produce fewer results. This is speculative, of course, as in general it’s hard to know how inputs of R&D spending translate into the output of actual productivity gains. But it is clearly the case that the main global locations for R&D spending are different than they were two or three decades ago, and that the relationships among those locations are more complex.