Countries have been experimenting with industrial policy for centuries to promote their economic development. Recent initiatives such as the U.S.’s “CHIPS for America Act” or China’s “Made in China 2025” signal a re-emergence of industrial policy. In light of the current spate of new large-scale initiatives, gaining a better understanding of global industrial policies is crucial. Yet the return of industrial policy onto the global agenda has exposed the lack of comparable data on international policy practice.
We develop a novel approach to address this shortcoming and use machine learning techniques that automatically classify industrial policies based on policy descriptions.1 Instead of using policy measures themselves, our methodology focuses on the language of policy descriptions to identify those that include goal-oriented actions aimed at changing the composition of economic activity ‒ our definition of industrial policy. We apply our classification algorithm to a large international dataset of commercial policies, the Global Trade Alert (GTA)2, and obtain counts of new industrial policy use at the country-sector-year level. In this article, we highlight four key findings from the analysis.
Industrial policy is on the rise
Industrial policy has increased considerably over the past decade, growing more than twofold. The figure below presents the share of trade policies our model has classified as industrial policy throughout the 2010s. Around 20 per cent of all policies in the GTA in 2009 were labelled ‘industrial policy’, while in 2019, that same share was at nearly 50 per cent. This finding provides empirical support to the widely held view among experts that industrial policy is on the rise and may actually be accelerating.
Industrial policy is unevenly used across countries, with high-income industrialized countries taking the lead
Industrial policy skews heavily towards high-income industrialized countries (see next figure). High-income countries implement about five times as many industrial policies, on average, as low- to middle-income economies. Among the latter, it is primarily middle-income industrial economies (such as BRICS) that use industrial policies. We found virtually no industrial policy in low-income countries.
Industrial policy targets similar industries across the income distribution
Although the extent of industrial policy varies across poor and rich countries, we find that countries generally tend to focus on similar types of industries. When comparing the top 10 industries targeted by industrial policy in high-income economies (figure below, left side) and in low- to middle-income economies (figure below, right side), the overlap is striking. Countries across the income distribution target machinery and transport equipment as well as some heavy industry.
Interestingly, one exception to the above are policies directed at green energy and in particular at clean electricity generation, which are far more prevalent in high-income economies. This activity is assigned to “Mineral oils and products” in the figure. The majority of industrial policies in rich countries targets this specific sector, but it is further down the list in low- to middle-income economies (ranking 8th). As this sector includes both hydrocarbons and clean sources of electricity generation, it is important to distinguish between the two types of activities. In high-income economies, most industrial policies in this sector (68 per cent) are directed at electricity generation from renewable sources. In low- to middle-income economies, we observe the reverse: only 28 per cent of industrial policies in the sector target clean energy while the remainder focus on hydrocarbons.
Contemporary industrial policy is converging towards a common set of policy measures – with one exception
As regards specific instruments used for industrial policy, we again find a substantial overlap between rich and poor countries (see figure below). Countries across the income distribution heavily rely on trade finance, state loans, financial grants and local sourcing requirements. There is one striking difference, however, namely the use of import tariffs in countries in the lower parts of the income distribution (where it is the third most frequently used tool). Import tariffs are mostly absent in high-income countries’ industrial policy. Instead, financial assistance in foreign markets, public procurement and capital injection and equity stakesare more widely used. Unlike non-tariff measures, tariffs increase fiscal revenue and require lower administrative capacity. This is likely one of the reasons why less developed countries use them more intensively.3
One last aspect worth noting is that 60 per cent of industrial policies are targeted at specific firms4. This is consistent with the type of policy measures we find to be most commonly applied to implement industrial policy, trade financing, state loans and financial grants. Taken together, the extensive use of intricate, non-tariff and firm-specific measures characterizes contemporary industrial policy, which is likely to require a substantial level of state capacity.
Government capabilities in developing countries urgently need to be strengthened
As the world moves towards more overt industrial policy, lower income countries may find it increasingly difficult to compete in international markets. Simply put, high-income countries have the fiscal and administrative capacity to more readily deploy modern industrial policies in ways that may be challenging for lower income countries to emulate. This underscores the urgent need to strengthen government capabilities in developing countries.
Disclaimer: The views expressed in this article are those of the authors based on their experience and on prior research and do not necessarily reflect the views of UNIDO (read more).
- See Juhász, R. & Lane, N. & Oehlsen, E. & Pérez, V. C., (2022). "The Who, What, When, and How of Industrial Policy: A Text-Based Approach".
- This dataset spans from 2009 until today and contains English language descriptions of each policy, thus allowing us to use standardized brief policy descriptions as inputs in our machine learning model. See https://www.globaltradealert.org/
- The GTA only focuses on the flow of policies, not on the stock. For that reason, our measures may underestimate the number of policies implemented in lower income countries if they fall back on their “stock” of tariffs to meet industrial policy ends.
- For further details please consult: Juhász, R. & Lane, N. & Oehlsen, E. & Pérez, V. C., (2022). "The Who, What, When, and How of Industrial Policy: A Text-Based Approach".