A version of this post was published in Business Day on March 1, 2023.

“Industrial policy” usually means some government intervention that aims to permanently alter the patterns of production in an economy. This definition is vague, because many types of actions attract this label. For example, even subsidies to university researchers are sometimes considered “industrial policy” (the hope being that they stimulate technological innovations). But for developing countries, it most often means the protection of local manufacturing firms from import competition, subsidies for such firms, or both.

The best case one can make for why these types of policies might work is an “infant industry” argument. (By “best case”, I mean one that holds out at least some promise of delivering gains to the economy as a whole. Another justification might be that industrial policies allow politicians to redistribute wealth to certain favoured groups without attracting the scrutiny that explicit cash handouts would - but that would be cynical.)

So how might this work? The argument goes that a given industry may not be efficient enough to compete yet, but it might be in the future. If the obstacles it faces can be overcome by expanding in scale - perhaps there are some large fixed costs, or acquiring more efficient technology and processes requires some “learning by doing” - then temporarily protecting it against foreign competition might allow this “infant” to grow up, and stand on its own.

It’s not hard to see potential holes in this argument. Why is the local market important? Most countries, even rich ones, are a tiny fraction of the global economy. (South Africa accounts for about 0.5% of world GDP; even a country as rich as Canada is about 2%.) So if achieving scale is important for future productivity, why not just seek export markets? And if a firm has the potential to be profitable, why is the private sector unwilling or unable to finance that initial growth? After all, investors routinely take losses in pursuit of future profits. So what does the government know that the private sector doesn’t that allows it to “pick winners”, as the argument suggests?¹

Despite these reasons for scepticism, two high-quality papers have recently documented successful examples of infant industry protection. Reka Juhasz (Juhász (2018)) studies the consequences of a natural experiment - the French embargo against British imports during the Napoleonic Wars. From 1803 to 1815, the French navy refused entry of British-made goods to all European ports, not just French ones. However, the strength of this shock to trade costs varied across France - because enforcement wasn’t perfect at all ports (Lisbon vs Rotterdam, say), and because different regions of France are differentially close to those ports (eg Paris vs Lyon). She shows that regions within France which experienced larger effective tariffs tended to develop more productive cotton-spinning factories than those which were able to import British cotton more readily. Even more impressive is the fact that that these productivity differences persisted seemingly long after the war, and the trade embargo, ended.

Closer to present-day debates, Nathan Lane (Lane (2023)) looks at the effects of South Korea’s “Heavy and Chemical Industry Drive”, a period of intensified support to certain industries - among them shipbuilding, petrochemicals, steel and electronics manufacturing. This support mostly took the form of subsidised credit, although interestingly the targeted industries were exempt from tariffs on imported inputs. He argues that this policy was pursued mainly for military, not economic reasons - the withdrawal of American troops in 1969 created a a lot of fears of an imminent North Korean invasion. Nevertheless, Lane is able to show that output, productivity, and exports increased in the targeted industries relative to untargeted ones; and further, the policy seemed to “spill over” by creating faster productivity growth in sectors closely related to the targeted ones.

These are excellent papers, and one can only be in awe the efforts of their authors (especially since the data they use had to be newly digitised, often from obscure hand-written records). But it is important to understand why these studies provide good evidence beyond the anectdotal. These authors both engage directly with the major concerns about infant industries - that the benefits of protection would be temporary; that the costs to other sectors could outweigh the benefits; and that the apparent successes are due to selection bias (e.g. if an industry that is growing anyway gets support, can we really credit the policy?). Worryingly, in a report by the CDE, Altbeker (2021) points out that much of the data that would be required to evaluate South Africa’s various “localisation” policies are not even collected.

We should remember that there have been many failures of industrial policy, too. Far from creating the next Samsung, the experience of many Latin American countries is that “import substitution” has led to higher prices for consumers and downstream industries, and an industrial landscape populated by inefficient “infant” firms that somehow never grow up. (Slaughter (2004) reviews some of the evidence on the failures of “import substitution” as of the mid-1990s.)

But the downsides to protectionism go beyond just their immediate impact on raising prices, in a subtler and ultimately more costly way. By protecting local firms from foreign competition, one necessarily creates market power. And rather than serving their customers, these firms will have an incentive to serve their political masters and protectors. In case this sounds too far-fetched, consider the evidence of Akcigit, Baslandze, and Lotti (2023). They study the political connections of Italian firms from 1993 - 2014, and show that although politically connected firms tend to enjoy more revenue and employment growth, but have lower productivity and innovate less. Similarly, Bertrand et al. (2018) show how French firms whose CEOs are closely connected to regional politicians employ more people (and they have fewer layoffs during election years); yet, they are less profitable than otherwise comparable firms.

The starkest way to see this mechanism - “free trade as competition policy” - at work is to read Schmitz (2005). There, he documents how, in the early 1980s, the sudden entry of cheaper Brazilian iron ore into the US and Canadian markets forced local firms to adjust. While productivity had stayed constant in the 15 years before, by 1990 it had doubled for US and Canadian iron ore producers - primarily, it seems, as a result of making work practices more flexible.

Industrial policy can indeed sometimes work. But it is a delicate case to make, and the arguments in favour of free trade remain underrated.

References