Yuen Yuen Ang shows how states discover winners under uncertainty. Ken Opalo maps two structural barriers to industrial policy. Taken together, they point to another factor: professional formation.
GIFF is the most systematic version of the picking step — Lin and Monga provide governments with a methodology for identifying sectors where comparative advantage hasn't yet emerged. That part it handles well.
But the problem raised in the post sits downstream of that. Sector identification tells you where to look. The harder thing is what you do once you're looking: you have a promising sector, you've backed some firms, and now you have to figure out who's Shinjin and who's Hyundai before the market settles it for you. That means reading firm-level performance under political pressure, holding a view against those who want the subsidy to continue, and actually pulling support from those that don't perform. Sector-identification methodology is not designed to prepare officials to do that.
Japan illustrated this. The 1962 survey showed MITI working. However, the enforcement capacity behind it had been accumulating since before the war. The officials running postwar industrial strategy had two or more decades of consequential decisions behind them. The sector choices were the visible part. The specific enforcement underneath was what made them stick.
On electoral cycles: I think you're right that the incentive problem is structural, and it may be harder than it looks. Building a professional corps capable of analysing firm-level performance, not just sector potential, takes time. The government that makes that investment rarely gets to see the payoff. That's the thing none of the frameworks solves for.
Curious as to what you think about tools such as the Growth Identification and Facilitation Framework (GIFF) developed by Professors Justin Yifu Lin and Celestin Monga to proactively identify sectors that might have potential for growth and employment generation and then direct policy and incentives to them. The Japanese also seem to have used similar frameworks (well described in the Economist’s famous 1962 survey: “Consider Japan”) during their economic takeoff stage after WWII to identify which sectors to back and direct their companies towards.
The incentive problem is definitely a knotty one. Not sure how this can be resolved in democracies when electoral cycles seem to push political actors towards policies which they hope will pay off in time for the next election instead of playing the long game to build the capacity needed to properly administer incentive programs.
This is a very useful framework. One case I would suggest looking at is Hefei, China.
Hefei is interesting because it shows how industrial-policy judgment is formed not only inside bureaucratic institutions, but through repeated engagement with firms, technologies, capital markets, supply chains, and local fiscal risk. The city’s bets on display panels, semiconductors, new-energy vehicles, and especially NIO were not simply “picking winners” from above. They involved a local state learning how to read industrial signals, structure financing, absorb risk, coordinate land and infrastructure, and then scale around firms that generated real capability.
In that sense, Hefei may be a concrete example of the “reading problem” your essay identifies. The key issue is not just whether officials are professionally formed in the abstract, but how they become embedded in production systems deeply enough to distinguish real capability from rent-seeking projects. Hefei’s experience suggests that industrial-policy capacity can be formed through a long sequence of situated decisions: factory floors, investment negotiations, supply-chain mapping, engineering timelines, financing structures, and local government balance-sheet pressure.
It might be one of the most useful Chinese cases for studying how a local state learns to read winners before the outcome becomes obvious.
This emphasis on "professions" is super important. And your point they take time to build and yet can be destroyed quickly is a huge cautionary tale no one wants to hear.
Looking at this through a Complex Adaptive Systems lens, it makes sense that the fitness function can be reset, that feedback loops take a little while to embed, and that agent behaviour is the toughest to change (especially in the absence of external stimuli).
Very interesting piece! I am not sure about your argument that World Bank staff is not trained to promote IP. At the WB, and the IMF for that matter, many are now casting IP as a way to address certain market failures to stay close to neoclassical economics. For example, governments can promote industrial parks as a way to tackle coordination problems, a powerful type of market failure. All staff need to do is to update their priors: these coordination failures are more important in the context of economic development than previously believed. So, at least in some subareas of IP it’s about applying old thinking to a new domain.
Fair point - probably a more historic perspective. Although there may be a difference between such training and knowledge being accessible vs internalised. Although a second-order point to the argument about the length of time it takes to develop practical wisdom.
GIFF is the most systematic version of the picking step — Lin and Monga provide governments with a methodology for identifying sectors where comparative advantage hasn't yet emerged. That part it handles well.
But the problem raised in the post sits downstream of that. Sector identification tells you where to look. The harder thing is what you do once you're looking: you have a promising sector, you've backed some firms, and now you have to figure out who's Shinjin and who's Hyundai before the market settles it for you. That means reading firm-level performance under political pressure, holding a view against those who want the subsidy to continue, and actually pulling support from those that don't perform. Sector-identification methodology is not designed to prepare officials to do that.
Japan illustrated this. The 1962 survey showed MITI working. However, the enforcement capacity behind it had been accumulating since before the war. The officials running postwar industrial strategy had two or more decades of consequential decisions behind them. The sector choices were the visible part. The specific enforcement underneath was what made them stick.
On electoral cycles: I think you're right that the incentive problem is structural, and it may be harder than it looks. Building a professional corps capable of analysing firm-level performance, not just sector potential, takes time. The government that makes that investment rarely gets to see the payoff. That's the thing none of the frameworks solves for.
Curious as to what you think about tools such as the Growth Identification and Facilitation Framework (GIFF) developed by Professors Justin Yifu Lin and Celestin Monga to proactively identify sectors that might have potential for growth and employment generation and then direct policy and incentives to them. The Japanese also seem to have used similar frameworks (well described in the Economist’s famous 1962 survey: “Consider Japan”) during their economic takeoff stage after WWII to identify which sectors to back and direct their companies towards.
The incentive problem is definitely a knotty one. Not sure how this can be resolved in democracies when electoral cycles seem to push political actors towards policies which they hope will pay off in time for the next election instead of playing the long game to build the capacity needed to properly administer incentive programs.
This is a very useful framework. One case I would suggest looking at is Hefei, China.
Hefei is interesting because it shows how industrial-policy judgment is formed not only inside bureaucratic institutions, but through repeated engagement with firms, technologies, capital markets, supply chains, and local fiscal risk. The city’s bets on display panels, semiconductors, new-energy vehicles, and especially NIO were not simply “picking winners” from above. They involved a local state learning how to read industrial signals, structure financing, absorb risk, coordinate land and infrastructure, and then scale around firms that generated real capability.
In that sense, Hefei may be a concrete example of the “reading problem” your essay identifies. The key issue is not just whether officials are professionally formed in the abstract, but how they become embedded in production systems deeply enough to distinguish real capability from rent-seeking projects. Hefei’s experience suggests that industrial-policy capacity can be formed through a long sequence of situated decisions: factory floors, investment negotiations, supply-chain mapping, engineering timelines, financing structures, and local government balance-sheet pressure.
It might be one of the most useful Chinese cases for studying how a local state learns to read winners before the outcome becomes obvious.
Thanks for the suggestion - very helpful. I will certainly look it up!
This emphasis on "professions" is super important. And your point they take time to build and yet can be destroyed quickly is a huge cautionary tale no one wants to hear.
Looking at this through a Complex Adaptive Systems lens, it makes sense that the fitness function can be reset, that feedback loops take a little while to embed, and that agent behaviour is the toughest to change (especially in the absence of external stimuli).
Very interesting piece! I am not sure about your argument that World Bank staff is not trained to promote IP. At the WB, and the IMF for that matter, many are now casting IP as a way to address certain market failures to stay close to neoclassical economics. For example, governments can promote industrial parks as a way to tackle coordination problems, a powerful type of market failure. All staff need to do is to update their priors: these coordination failures are more important in the context of economic development than previously believed. So, at least in some subareas of IP it’s about applying old thinking to a new domain.
Fair point - probably a more historic perspective. Although there may be a difference between such training and knowledge being accessible vs internalised. Although a second-order point to the argument about the length of time it takes to develop practical wisdom.