What is industrial policy for?

2

Industrial policy is back in vogue in rich countries. For a good example of its revival in US politics, watch the discussion my colleague Rana Foroohar had with Brian Deese, director of President Joe Biden’s National Economic Council, a few weeks ago.

The motivations tend to be compound, sometimes conflicting, and potentially confused. Industrial policy, at least when taken to mean boosting domestic manufacturing capacity, could aim at any of the following: to maintain a technological leading edge; to increase the number of factory jobs in the hope of helping left-behind people and places; to overcome productivity stagnation; to safeguard sufficient medical and other critical supplies in emergencies; or to concentrate supply chains within one’s jurisdiction for geopolitical reasons.

Not all these goals sit easily together. Technological edge and productivity are not helped by factory fetishism, and international supply chains may be more resilient than purely domestic ones.

So here are three corrections of sorts to any too-facile fashion for industrial policy, focusing on the much-discussed case of semiconductors. The sector is badly hit by shortages, with big repercussions for industrial goods at large, from cars to TV displays.

How bad is the situation really? Here are some numbers on semiconductors from a recent White House report on US supply chain resilience.

First, do not take a bygone era as the norm. Deese highlights the fall in the US share of global semiconductor production. According to the White House report, in 1990 US companies produced 37 per cent of the global supply; today it is 12 per cent. But the report also shows they dominate the world market for chip design — almost totally so in the case of several important chip types such as computer central processing units. The US also has a higher share of the global production of integrated manufacturers that design their own chips, as opposed to contract manufacturers.

But how much is “enough”? The US accounts for 24 per cent of the global economy, according to IMF numbers, whose trend suggest the share falling to 22 per cent in the next five years. Industrial policy advocates should ask why the US share of global chip production should be more than this (and similarly for other countries). If the point of market dominance is productivity and technological edge, then it may make sense to seek a more-than-proportionate share of design of the most advanced chip fabrication, or of chipmaking equipment, and reap economies of scale from large export sales. But in some, if not all, cases this is more or less the situation for the US.

Second, give a thought to the composition of the trade balance. If you aim to produce more semiconductors (or anything else) at home, whether for domestic supply or for exports, what else are you then willing to import more of or export less of? Industrial policy is not going to change the overall trade balance, which is determined by macroeconomic factors. It has proved very hard for the US to reduce its trade deficit. As for the EU and especially Germany, they can hardly increase their already big surpluses further.

So greater self-sufficiency — reduced imports — will be offset by either smaller exports or greater imports of something else. There is no automatic reason why the new combination should in total be better for productivity, security or the left-behind than the old. You have to make policy in the round.

Third, self-sufficiency and resilience are not the same thing. The current chip shortages have very little to do with the fact that the supply chain crosses borders. As Chad Bown explains:

“Carmakers overreacted to the initial shock of Covid-19 and, in early 2020, slashed orders for chips. By the time the auto companies realized their mistake, chipmakers were already at capacity supplying the suddenly booming market for work-from-home goods. The perfect storm only got worse: Arctic weather in Texas, a drought in Taiwan, and an earthquake and fire in Japan all worked to slow production.”

Add to this the direct disruption from coronavirus itself: a Covid outbreak in Taiwan is threatening to delay chip shipments further.

None of these problems is inherent to the cross-border supply chain, which may even have prevented worse problems (imagine if all the world’s chips were being made in Texas during the freeze).

The one way in which the cross-border nature of the supply chain does matter has to do with Donald Trump’s attempt to hamper Huawei’s 5G dominance through export controls on US-made semiconductors and chipmaking equipment, and his imposition of tariffs on chip imports as part of a broader trade war. As Bown explains, these led to a hoarding of chips and forced global customers to turn to non-US suppliers of chips and chipmaking tools, adding to the shortage for consumers and to losses of the sales that funded US producers’ technological development.

In sum, trying to repatriate supply chains may threaten domestic resilience and technological leadership more than it contributes to it.

Many policymakers know this, at least in theory. As Deese says:

“Resilience does not mean closing ourselves to the rest of the world. Partnerships with our allies that promote more stable access to key inputs while improving environmental sustainability and workers’ rights is essential.”

In practice, the jury is out. Bown recommends shifting the focus from export control to common regulation among geopolitical partners, so that markets remain open, competitive and plural, working on standards that help secure cyber security. Here, too, there is an echo in Deese’s remarks, which mention “shaping markets”.

The good news is that the recent EU-US summit strengthened the resolve to do precisely this. The bad news is that in much political debate, the task still seems to be seen as ensuring that one’s own country or region produces as much as possible on its own.

Other readables

  • The good, the bad and the ugly of international corporate tax reform.

  • To prepare for the big economic changes ahead of us, Gavin Kelly says we should pay more attention to the 1960s in the UK, a period of great labour market churn combined with productivity increases. “Contrary to the received wisdom that would emerge fifteen years later, the belief was that greater labour market dynamism required more, not less, collective insurance and social support,” Kelly writes.

  • The pandemic has worked wonders for Italy’s lagging digitisation.

  • Trials of a four-day working week in Iceland seems to show that working time can be cut while maintaining output and pay.

Numbers news

View original post