A recent working paper by Jaedo Choi and Andrei A. Levchenko claims that industrial policy in South Korea has had sizeable long-term benefits:

This paper provides causal evidence of the impact of industrial policy on firms’ long-term performance and quantifies industrial policy’s long-term welfare effects. Using a natural experiment and unique historical data during the Heavy and Chemical Industry (HCI) Drive in South Korea, we find large and persistent effects of firm-level subsidies on firm size. Subsidized firms are larger than those never subsidized even 30 years after subsidies ended. … Counterfactual analysis implies that the industrial policy generated larger benefits than costs. If the industrial policy had not been implemented, South Korea’s welfare would have been 22–31% lower, depending on how long-lived are the productivity benefits of LBD.

This seems like a win for industrial policy proponents, but the paper is far from a slam dunk.

The authors note that the HCI Drive happened against the backdrop of the Foreign Capital Inducement Act (FCIA) of 1966. Under the FCIA, the Korean government rationed firms’ access to cheaper foreign credit, only selectively allowing some companies to borrow abroad. Once firms got government authorization to access foreign loans, the Korean government would guarantee them, eliminating risks of firm default and exchange rate depreciation. This meant these firms could borrow at favorable – even negative – real interest rates. The HCI Drive, therefore, was a policy that used the government’s discretion to allocate credit (and implicit subsidies) under the FCIA to heavy and chemical industries in specific regions.

This means at least two simultaneous factors were at play: the access to foreign credit markets, previously restricted under the FCIA; and the government subsidies in the form of guarantees. The authors do show that the firms granted access to foreign credit witness a higher increase in sales. This could, however, result from the easing of previous credit restrictions, in which case government policy was likely hampering development. So attributing the improved performance to subsidies alone is unpersuasive.

Additional puzzling evidence comes from data on productivity (how efficiently firms use inputs). Choi and Levchenko show that HCI Drive-targeted firms experienced a large decrease in productivity, something other research, using the same data, has highlighted. The two pieces of evidence – the apparent HCI policy success in growing sales, and the simultaneous decrease in productivity –could indicate that South Korea’s HCI Drive was a success to the extent it circumvented previous foreign credit restrictions, leading to an increase in sales. But subsidies also led to excessive borrowing, bloated firms, and lower productivity. In this scenario, simply repealing the FCIA would have produced better results.

Hardly an industrial policy success story.