Mr Chips goes to Brussels: Semiconductors in Europe
Bob Hancké and Angela Garcia Calvo take a closer look at the initiatives to build semiconductors in Europe through the perspective of industrial policy successes and failures in Europe.
Is making semiconductors in Europe a good idea? With the introduction of the European Chips Act in early 2022 this is no longer an academic question. As with many of these questions, answering it depends in large part on what you think the problem is and where you would look for solutions. If you see the issue as how to secure the supply of chips to world-leading European industries in times of crisis, then (re)shoring some of that production seems to make a lot of sense. Because those arguments are almost intuitively self-evident, we won’t belabour them here. But if you think of the question as an industrial policy problem, the world looks quite different – not least because industrial policy, even in Europe, has a complicated and convoluted past.
In a recent article in Global Policy, we take an industrial policy perspective to evaluate current European plans for semiconductors production (or ‘fabs’ in the jargon) in light of a wider – but remarkably one-sided – debate on industrial policy in the advanced capitalist economies. Most of that literature has looked at policymakers, their targets, motivations, commitment, and resources. While governments obviously play a central role in policymaking, they do not – to paraphrase Marx – choose the conditions under which they play that role. Most industrial policy follows the optimistic motto that ‘with enough political will (and funding) we can find a way’. But a look back at industrial policy across post-war Europe suggests that this optimism is unwarranted. The otherwise quite formidable French state, for example, was unable to build a computer industry in the 1960s and 1970s, despite preferential procurement systems, vast resources spent on the industry, and world-class engineering schools. On the other side of the Rhine, Germany – no less competent a political economy – has few problems developing and sustaining world-class automotive and machine tool industries, but for the better part of three decades, German policymakers have lamented the absence of a German Apple, Microsoft or Google. And one of the most important industrial policy initiatives in 1970s Europe, targeting the survival and renaissance of the British car industry between the mid-1960s and the mid-1980s, ended in dramatic fashion, when the nationalised and heavily subsidised British Leyland disappeared.
Markets, firms, institutions and policies: the need for congruence
While much distinguishes these three economies, failure in crucial industrial policy areas is what they share. And a deeper look suggests that the lack of success has two basic reasons, none of which related to policymakers, policy targets, or resources. First, they tried to build an industrial policy in sectors where no or very few successful companies existed – the surprisingly ignored rationale for an industrial policy in the first place – and, second, the existing institutions that governed labour, capital markets, and innovation systems militated against the policy objectives. Developing and building computers required flatter hierarchies than the top-down French system typically produced, and more experimentation than the Cartesian ‘one-best-way’ doctrine allowed. German labour and capital market institutions help local firms excel in mature, medium-tech, slow-moving sectors, but have difficulties handling rapid shifts that (may) destroy deep, technology-specific skills and long-term investments. And both British managers and unions were locked in adversarialism when success in the automotive industry required cooperation and negotiation.
To be sure, industrial policy has worked in Europe: think of the success of Airbus after 1970, the French nuclear industry in the 1970s and 80s, or the Swedish and Finish mobile telephony sector in the 1990s. In each of those cases, the policies worked well because of their articulation with the existing institutions and industries. Where potentially competitive firms existed, the governments ‘produced’ the broader institutional context to support them; conversely, where institutions existed but firms were reluctant to enter, industrial policy encouraged entrepreneurship.
The institutional conditions for industrial policy
Instead of focusing on policymakers, as most students of industrial policy have done, we therefore draw attention to two essential ‘first-order’ conditions that determine the chances of industrial policy success: the existence of potentially competitive incumbent firms, and the alignment of the policy objectives with the prevailing institutional framework. Where the policies are aligned with both, industrial policy is, at most, a matter of setting broad frameworks and let firms get on with their business. When one is missing, industrial policy first has to think about developing the weak or missing condition before considering targets and resources. Where both are absent – i.e., where existing firms are very weakly positioned in markets and institutions push in a very different direction than the policy – industrial policy initiatives are almost certain to fail.
So, what about chips?
What, then, does this tell us about semiconductors and fabs in Europe? In a nutshell, this framework suggests that an expansive strategy for mature semiconductor production makes little economic sense in Europe. Set-up costs of chips plants are massive, yet returns are tiny: fabs, with expensive clean rooms, cost a small fortune to build and then need to run at 95% of capacity or more to be profitable. But the market is extremely volatile – the main reason why we are talking about building chips in Europe in the first place – a characteristic of this market that dramatically increases investment risks. Mature chips are also almost commodities and require little skilled labour – a configuration that is at odds with the prevailing European industrial strategies of a high-skill, high-wage economy. And pure economics matters too: several decades of increasingly open global markets have produced complex systems of trade interdependence, by and large built on specialisation and comparative advantages, which have kept prices low and supplies stable.
But most importantly, perhaps, there are few existing European chips producers, and after several decades of industrial policy in the area (the first notions of a European Chips strategy date back to 1982), Europe’s global share of chip production has actually fallen. It is hard to imagine that another go in 2022 will lead to a fundamentally different outcome, not least because Europe’s goals imply a quadrupling of local capacity to reach a 20% global share. In sum, producing mature semis, the low-end segment of a high value-added market, in Europe is not a good idea.
Semiconductor fabs for European skills and vice versa
Instead, we think that the EU should concentrate on market segments that leverage existing high skills in the workforce, its world-class research system, and in which Europe has either already developed a comparative advantage or will not find itself competing with well-established rivals. Europe’s future in the industry lies in developing and manufacturing things such as advanced AI chips, lithographic equipment, and similar products, where the markets are only just emerging, and very few (non-European) firms have a strong position.
These types of products are also perfectly matched to Europe’s existing comparative institutional advantages and competitive strengths. You could think of them as ‘co-specific’ assets, since other countries, which produce chips and computers, need them. Europe could therefore leverage its own version of a bottleneck in these markets to secure more than just best efforts from others. Its quasi-monopoly in these product lines more than offsets its reliance on others.
Since Adam Smith and Ricardo, we have always seen gains as following organically from trade: the butcher and the baker need each other and benefit from their mutual presence, in the same way that Britain and Portugal benefited from trade in wine and textiles. The targeted policy in the semiconductor sector that we suggest here adds a more calculated, strategic dimension to that old story of the wealth of nations.
Bob Hancké, Associate Professor of Political Economy, LSE. Angela Garcia Calvo, Assistant Professor in International Business and Strategy, University of Reading.
Photo by Sergei Starostin