I frequently read or hear a claim that the United States needs a more aggressive industrial policy to keep pace with China’s industrial policy. On its face, the claim is a curious one. After all, China’s “industrial policy” didn’t work very well from the 1950s up through the 1970s. It was only when China’s industrial policy started involving considerably less government control over the economy in the early 1980s that China’s impressive surge of economic growth began. There are a bunch of different ways to compare per capita GDP across countries (depending on the exchange rate between currencies that seems most relevant), but on World Bank estimates, China’s per capita GDP is still only about 20-25% of the US level.

Thus, from an overall perspective, China’s economic success looks a lot less like a carefully directed move forward and more like a bunch of political leaders trying to scramble up to the front of the economic parade so that they can claim to be leading it. And while I’m willing to take my economic lessons from wherever they emerge, the question of what lessons the US economy should be learning from an economy that has grown over four decades to reach maybe one-quarter of US per capita GDP are less than obvious.

But what if we dig down into the details of what China’s government has actually done to favor certain industries, and what lessons might be learned? Chad Bown interviews Lee Branstetter on this subject in “Is China’s industrial policy working?” (Trade Talks podcast, April 23, 2023, audio and transcript available).

Their discussion starts off with a quick overview of why even many market-oriented economists think it’s appropriate and productive for the government to play a role in supporting research and development (through some combination of intellectual property laws, subsidies and tax breaks), as well as education and infrastructure. Thus, the focus here is on “industrial policy” that focuses not on the overall economic climate, but on support for specific chosen industries. In addition, as Branstetter points out, China’s economic policy through the 1980s and 1990s was mostly about less government intervention, or on supportive conditions for extremely broad areas of the economy–like manufacturing. As Bown describes this time: “China reformed and became more market oriented. The government is still giving out a lot of subsidies, but in terms of industrial policy, those subsidies were not precisely targeted. And I suppose when you are subsidizing everything, you are preferencing and targeting nothing.”

But then China’s leadership becomes more involved in targeting specific industries. Branstetter describes the kind of research he has been doing in this way:

In 2007, China made it mandatory for companies listed on a Chinese stock exchange to disclose in their annual reports the subsidies that they received from the Chinese government. What that means is that starting in 2007, at least for China’s list listed companies we can get pretty rich firm-level data on the subsidies they received, and we can actually use that data along with all the other data disclosed by these firms to their investors to try and get a sense of how subsidies are correlated with firm characteristics, especially productivity. If the focus of Chinese industrial policy after the mid-2000s was really to strengthen the
innovative capacity of these national champions, then it should be making firms more productive.

It’s then possible to look at whether China is targeting firms that already have high productivity, and trying to push them ahead, or whether it is targeting firms with lower productivity to help them catch up. In either case, one can look at the connection from subsidies to later productivity gains. It turns out that subsidies were mainly going to lower-productivity firms, and didn’t seem to help their productivity. Instead, firms were being chosen for subsidies because they were large. Branstetter says:

We find that the Chinese government is not giving subsidies to initially more
productive firms. If anything, the statistical association is actually negative. The Chinese government is, on average, giving more subsidies to less productive firms. …

Chinese firm’s annual reports do often include language that describes what particular subsidies were for. But if we focus on that subset of subsidies that are meant to promote research and development, or the subset of subsidies that are meant to support upgrading of equipment, even for these specific subsidies, we find no relationship with productivity. It’s not the case that firms that are more productive are more likely to receive these subsidies in the first instance. And it’s not the case that firms that receive these subsidies become more productive later. …

Firms that are larger, as measured by total assets or employment, appear to be somewhat more likely to receive these subsidies. As we dug into this data, it became
increasingly clear to us that the subsidies provided to Chinese firms had lots of objectives, many of which were not connected to productivity. We see significant quantities of subsidies going into declining industries like mining. We see significant subsidies that appear to be designed to support employment in large firms. … It’s understandable why the government would pursue this objective, but it’s also crystal clear that the pursuit of this objective directly undermines the pursuit of turning the already more productive firms into super innovators. The money that’s given to prop up failing firms is money that cannot be given to support the technology leaders of the future, and the more you do the former, the less resources you have to do the latter.

In 2015, China announces a “Made in China 2025” policy, which is clearly seeking to have products produced with a wide range of technologies in other countries be produced in China instead. It turns out that this group of subsidy-receiving firms is more likely to have patents–which sounds as if industrial policy may be at work. But at least for the first few years of the program, the firms getting subsidies don’t seem to improve productivity and don’t seem to be taking out new patents at a faster rate. As Branstetter says:

And given how easy it is for firms to get patents in China, given the strong incentives they have with subsidies available at the local government level and elsewhere to take out these patent applications, it’s really surprising that we just don’t see any impact that these subsidies are making these farms more innovative or more productive. … When we try to evaluate the net benefits, if any, of Made in China 2025, it’s hard, at least on the basis of our analysis, to come to a very positive conclusion. Resources have been expended, but the desired innovative outcomes have not yet emerged.

These kinds of arguments may seem counterintuitive, because the conviction that that China’s industrial policy is a sweeping success has become so embedded. For those who would like to dig in more, the study on subsidies is available as NBER working paper #30699, the study on Made in China 2025 as NBER working paper #30676, and a draft of an overall essay on China’s industrial policy is also at the NBER website.

The fundamental point, of course, is that industrial policy can’t be evaluated by pointing to a few companies that are success stories and that also received subsidies. Conversely, industrial policy also can’t be evaluated by looking at a few spectacular failures of companies that got subsidies, either. You need to look at the full array of companies receiving subsidies, see what mixture of political and economic factors guided that choice, and then look at later outcomes for the group as a whole.