- Capital Gains
- Posts
- When Does Industrial Policy Work?
When Does Industrial Policy Work?
Sometimes you want a plan, sometimes a plan is a tax on responding to new information
Know someone who might like Capital Gains? Use the referral program to gain access to my database of book reviews (1), an invite to the Capital Gains Discord (2), stickers (10), and a mug (25). Scroll to the bottom of the email version of this edition or subscribe to get your referral link!
The US is going through one of those fun political inversions where a President elected on a right-of-center platform has, as one of his signature policy moves, started having The People acquire ownership of the Means of Production (of chips and critical minerals). There are older precedents; under the deregulatory Carter presidency, the US bailed out Chrysler and took warrants; Reagan's Treasury Department ended up owning most of Continental Illinois; and in the last few months of his presidency, Bush managed to take control of most of AIG.1 Governments always have some kind of view on what the economy should look like, which they express with varying levels of finesse. Tariffs, depreciation rules, permitting processes, education funding, zoning, spectrum auctions, not to mention interest rates and deficits—all of these are, implicitly, statements about what direction the government thinks the economy should go.
There are definitely disaster stories to worry about—and some of the successes are qualified ones. China, for example, shows that some level of industrial policy can lead to massive GDP growth, and lift hundreds of millions of people out of poverty. On the other hand, one of the reasons those people were so impoverished in the first place was China's previous round of industrial policy—backyard steel production, mass relocation from rural areas to cities, and an attendant famine. China had also adopted a policy of locating some manufacturing in hard-to-reach places so it would be difficult for enemies to seize during a war. But this also meant that it was difficult to reach, for workers and materials, during peacetime, and China had more of that than they'd expected.
One old model of US industrial policy echoes some modern vendor financing: the government granted land to railroads in return for their construction. The land in question was worth a lot more conditional on being next to a railroad, but the government could hold back some land that would also rise in value thanks to being adjacent to railroad-adjacent land—or because the railroads, and earlier canals, got every other square mile. There was actually a second layer of vendor financing to this, where railroads encouraged experienced farmers to buy the land, and sold it on generous terms, so they could convert the land into a stream of freight revenue.
But the story of railroads also illustrates how hard it is to calibrate a subsidy like this. A transcontinental railroad is a lot more valuable than two nearly-transcontinental railroads that can't send goods or passengers on an uninterrupted trip, so it makes some sense to overshoot. But a bigger-than-necessary subsidy is only going to get returned if both the railroad operators and politicians are pretty upstanding, and neither of these conditions really obtained at the time. So the Union Pacific's construction cost was nearly twice the actual cost, with much of the difference going to its backers and some of it going to politicians. (Any resemblance to the suspiciously high price of constructing new subways in the US today, and the opaque set of interest groups involved, is, presumably, coincidental.)
Japan's approach early on was very interesting: they knew that steel was a complement to many other value-added industries, and realized that declining shipping costs would make it a global market and that they had plenty of cheap labor. But steel is also a scale business, and it's a bad idea for a country's first new steel mill to be at a significantly higher scale than what they're used to. So they structured their subsidy program with export credits that roughly matched the cost gap from running a bigger steel mill. That meant that Japanese producers still had to compete head-to-head with the US-based ones—but they had to compete on quality. Tying things up, Japanese industrial planners put pressure on banks to lend to the steel industry so it could expand capacity. As it expanded, those subsidies declined. So the implicit message the government was sending these companies was: we'll give you the economics you'll eventually get, but you have to retroactively earn them.
This approach is expensive, but from the government's perspective it's basically making an investment. The government doesn't get literal equity in this case, but the money they funnel into a steel industry early on starts to trickle back in the form of the taxes paid by companies and workers—and not just the ones at the steel company, but its suppliers and especially its customers.
That's the thing industrial policy can do very well, albeit in competition with financial markets: figure out a potential equilibrium in which there's more economic activity and it's higher value-added. Estimate the benefit of that, and the cost of getting there, and then, if the numbers work, spend enough to make it happen. This works very well for companies that are in the middle of the supply chain, where they're processing goods that then get sold to other companies before eventually being turned into final goods for consumers. It's harder for a consumer-facing product, in part because those products can be so attractive to elites that there's an incentive to overfund them—many countries have a better airline than they really need, and a few unfortunate ones have a bigger fleet than they need but no ability to keep the planes in the air.
So the heart of industrial policy is a kind of ruthless cynicism. A big, rich country can subsidize its way to having almost any industry—if the US wants to lead the world in sock production, we can offer wage subsidies that more than offset labor cost advantages in poor countries, freely lend to companies that want equipment, set up a 400% tariff on imported socks, and soon enough most of the world's socks would be made in America. It might help the domestic textile equipment business a bit, and perhaps American clothing retailers would also see some upside from cheaper shipping and faster turnaround for whatever new fashions hit the sock world. But these socks would be made by Americans who could produce more value in other jobs, and made for Americans whose after-tax income was now lower because of the cost of this subsidy. The categories that work well are the ones that create upstream and downstream jobs, and mix physical capital with tacit knowledge. The other option, currently dominated by the private sector but historically something the US government actually did quite well, is to identify an industry that doesn't exist yet, could probably only exist in one country, and, having solved for "where," focus funds on "whether" and "when."
We’ve looked at industrial policy in a variety of contexts at The Diff.
One case study in successful industrial policy is TSMC ($). (Disclosure: Long.)
Industrial policy tends to create national champions ($), some of whom have an advantage mostly in getting government protection.
And they also create higher returns from lobbying ($).
One reason US industrial policy is hard is that we're swimming upstream against the strong dollar, but other countries face different constraints in trying to supplant the dollar ($).
Share Capital Gains
Subscribed readers can participate in our referral program! If you're not already subscribed, click the button below and we'll email you your link; if you are already subscribed, you can find your referral link in the email version of this edition.

Join the discussion!
1 On the opposite end, but still a Nixon-in-China moment, it was LBJ, architect of the Great Society, who firmly set Fannie Mae on the path from being a government agency that subsidized home ownership to being a big rates-trading house that could have spent more time on risk management.
Reply