It’s a good thing that World Bank is finally OK with industrial policy, but policymakers shouldn’t let themselves be constrained by the coming prescriptive “best practices.”
Great post - agreed with all points made. This is a much needed line of argumentation. It remains a deep frustration to me that we have not as yet been able to develop demand-led and champion-responsive mechanisms to support governments in their learning-by-doing and crossing-the-river-while-feeling-the-stones approach to industrial policy. Such a support structure is key because it is terribly hard, and most low-income countries already suffer from rent capturers - such that the progressives trying industrial policy already start with a massive mountain to climb (not to mention challenges like climate change, unfavourable global financing, AI etc). We have frameworks and learnings from political economy, advisory lessons, delivery/management (inc PDIA etc), organisational capacity development, market and industrial economics, and political science (among other fields) to do this. And good partners that can be pulled in to support on key aspects, as needed by the gov't champions of industrial policy. The critical support needed is with the doing and implementing, not the upstream stuff on which sectors to prioritise, policy formulation etc (even though that element is valuable too).
I completely agree, and the "crossing the river while feeling the stones" framing is exactly right. The support gap you're pointing at is the real one.
One reason I have focused on a complex adaptive systems lens in this space is that it offers a way to match tools to governments. The same export discipline or credit allocation mechanism that built industrial capacity in Korea can actively entrench rent-seekers in a government that doesn't yet have the capacity to enforce the rules behind it. The tool isn't the issue; it's whether the government can actually make it work as intended.
That makes your rent-capture point more specific. Governments already surrounded by well-organised rent-seekers aren't just facing a harder climb. Some of the standard industrial policy instruments will give those rent-seekers more leverage, not less. Demand-led and champion-responsive support is exactly the right approach. But as a policy prescription, it needs a way to say: at this level of government capacity, these tools are viable, and these ones aren't yet. Without that, there's a real risk of handing a government tools designed for a different situation.
The iterative, learning-by-doing approach is the right process. The practical question is which stones a particular government can actually stand on, and which will give way. Getting that match between tools and capacity right is the hard part, but it's the question that needs to come before sector choice. Pick an instrument that the governing system can't back up, and it won't matter which industry is being targeted.
Perhaps also needed is some more capacity between the downstream and upstream of learning/doing scope and support plus structural/policy environment. Perhaps instead of a stolid “plan” that operates like an instruction manual or rubric, what might be more valuable or necessary are the set of just-in-time analytical tools that allow for widespread nimbleness/databases/measurable indicators or realistic/adjustable/just-in-time/performance-based work breakdown structures between these two structural conditions.
Typically, this nimbleness is found inside entrepreneurial and other private sector spaces — proprietary conditions that elicit tools that don’t necessary scale. The scalable, broadly useful post-planning version of these tools/resources are probably happening just at the edges of public-private spaces. Those spaces have to be protected to understand what comes after the “project” and industrial policy’s relegation to a series of project management tasks.
Yes indeed. There are some good books on this, for example Martin Williams' 'Reform as a Process' book which documents over 100 reforms by African governments. He shows that it is not top down plans or one-off projects that make the difference, but entrepreneurial leadership in the civil service and by cabinet members that catalysed, supported and added dynamism to ongoing effort. So external actors should help provide the tool box with multiple tools and options, and help enable them, under the leadership of the visionary progressives in governments. Private sector entrepreneurialism (pioneering magicians in the Deals and Development framework) are key, and so are entrepreneuring reformers in the government. When these two come together arounda focused agenda to develop a vertical or horizontal sector (or 2 or 3 or 4), change is possible. The evidence of industrial policy is clear on this.
The gap you name is telling: every field you list (PDIA, political economy, delivery management) is a framework, yet the thing missing is a mechanism that responds to demand rather than supplying its own agenda. "Feeling the stones" is inherently nonlinear, and 81 percent of the marine fish systems we manage turn out to be nonlinear too, but the models governing them still assume a straight line. The support structures fail the same way: they standardize the very groping that was supposed to stay adaptive.
Yes. We need philanthropic funding to provide base funding for such a mechanism based on context and complexity, because it is possible. There have been similar ones in the past (not perfect but good enough) albeit not for industrial policy.
Thanks for sharing this. "Building the requisite competencies as you go" is a really important point that you flag, and I think it could even be taken one level further.
The institutions behind the canonical industrial policy success stories (Korea's EPB, Taiwan's ITRI, Singapore's EDB) took two to three decades to form. The professional judgment capable of backing the right firms and disciplining the wrong ones accumulated over hundreds of real, consequential decisions, through a generation of bureaucratic practice where the feedback was genuine and the stakes were high.
You identify the political constraint: no durable developmentalist coalition, no serious industrial policy. But there is a formation constraint underneath the political one. The professional corps capable of carrying out serious industrial policy does not emerge from training programmes. These are officials who must hold a fifteen-year investment horizon and read sector dynamics well enough to tell a legitimate developmental bet from a rent play. They are also the ones who have to enforce discipline against connected firms when it matters — which is the hardest part. It forms through practice in environments where the feedback signal is real. A formation environment can be built over time; it cannot be purchased.
This creates a specific timing problem for any government that launches a serious industrial policy today. The political window to move is short. The professional formation timeline required for the next phase (enforcement, graduated exit, discipline at scale) typically outlasts that window by a generation. The risk is that the Bank's best-practice packages give governments the legitimacy to launch programmes they currently lack the professional depth to execute. That is not capacity building. Your reference to the Ajaokuta example is one version of where that leads.
Yuen Yuen Ang's recent piece on directed improvisation (https://polytunity.substack.com/p/industrial-policy-under-uncertainty) approaches the same problem from the discovery side: how states exercise influence rather than pick winners under genuine uncertainty. In the comments there (https://polytunity.substack.com/p/industrial-policy-under-uncertainty/comment/267004650), I raised the question of whether adaptive signal infrastructure is widely enough in place to run her grey-to-endorsement sequence. Yozma worked because foreign co-investors conducted the assessment; Nollywood's signals have no equivalent external validator. Reading emerging signals and acting on them credibly requires the same professional judgment that takes decades to build. That can't be imported either.
Where the evidence is genuinely thin relates to which countries are actually making the formation investment now, not in instruments but in the cadres who could carry serious industrial policy in a generation's time? That seems more consequential than the specifics of programme design.
Fantastic post. What would be your advice to a country (or for a development funder to a country) that clearly does not have “a durable developmentalist political coalition” and is stuck with two-bit rent seekers?
Great question because many countries in Africa are stuck with two bit rent seekers. But the Deals and Development framework, which built on Mushtaq Khan’s work, gave us insights on how to work in such a context. It requires a much longer time horizon than the usual 5 year planning and the identification of ways to move power broker & restores (rent capturers) business elite to ‘magician’ broker elite where their business model is less based on rents and more based on value addition, competing in a competitive or quasi competitive market and collaborating. Dangote is a famous example of such a shift - but you need a big enough cluster of shifters. The latter tends to mean business elite actually want state capacity to be strengthened (rather than making money from state extraction and state dysfunction). Similar work needs to be done on the political settlement side but that is harder. Taking a 20-30 year view is important. A certain form of industrial policy (let’s call it foundational, or early stage) is important to help with this. International development funders (esp philanthropy) need to support this.
How does this narrative still hold in a world that’s increasingly automating? Growth in developing countries through industrialization looks broken if (1) exports are the source of that growth like in the Asian Tiger cases; (2) developed countries can now produce commodities cheaply with autonomous labor.
the world bank report is vague, the bank is drifting in the wind, doesnt want to become seen as irrelevant but has no real guidance nor parameters for finance. there's the nuts and bolts of increasing each state's capacity that trends and agencies have to get in and do, ang yuen yuen's article cited in another comment is an example of an aspect in a context, and there is the structure of political coalition that 'keeps it sold' and helps adjust the policy. carlos oyas makes a constrast of ethiopia with purposive policy driving change vs angola drifting. a notable feature of ethiopia and indonesia is not only the policy planning and execution, but that it endures through successive administrations. every step of success brings different challenges, continuing the industrial development of the TPLF Abiy had to tackle the political structure, elaborating Medemer and suppressing separatist rebellions. the term 'industrial policy' is likely to always be insufficiently broad - it easily logically includes large harm and large help. the term state capacity is also too broad by itself though less tainted by bad history. the practice of states and critique of developmentalists is going to have to push further and draw lines of opposing paths with more specific titles.
This is quite timely and informative. It reinforces a long-standing view among many political economists and development experts: No poor country has ever "free-marketed" its way to economic prosperity, nor has any nation achieved rapid economic transformation solely by following World Bank, IMF, or WTO prescription. None.
History is clear. Whether we examine the 18th and 19th century examples like Britain, USA, France, Germany, Sweden, Belgium, the Netherlands, Russia, or the post-WWII success stories like Japan, Taiwan, and South-Korea, or the late 20th century “East Asian Miracle” economies - China, South-Korea, Indonesia - the pattern is consistent. Every country that achieved rapid development did so through strategic state intervention, industrial, trade, and technology policy (ITT), protection of infant industries, and deliberate technological upgrading, not laissez-faire markets.
Even the more recent members of BRICS economies - Brazil, Russia, Indian, China, South Africa – relied on state-guided development, not pure market fundamentalism.
Nevertheless, this does not mean the World Bank, IMF, WTO, and other multilateral institutions lack relevance in today's global order. They remain important actors. However, their advice must be interpreted differently by different countries, depending on where each nation stands in its journey of economic catch-up, structural transformation, or development maintenance.
In development, context is everything, and there is no one-size-fits-all. What accelerates growth in one country may constrain it in another. The challenge for policymakers is to understand where they are on the development ladder, recognized “ladder-kicking” developmental prescription, and apply global advice in ways that strengthen, rather than weaken, their long-term national capabilities and interests.
Great post - agreed with all points made. This is a much needed line of argumentation. It remains a deep frustration to me that we have not as yet been able to develop demand-led and champion-responsive mechanisms to support governments in their learning-by-doing and crossing-the-river-while-feeling-the-stones approach to industrial policy. Such a support structure is key because it is terribly hard, and most low-income countries already suffer from rent capturers - such that the progressives trying industrial policy already start with a massive mountain to climb (not to mention challenges like climate change, unfavourable global financing, AI etc). We have frameworks and learnings from political economy, advisory lessons, delivery/management (inc PDIA etc), organisational capacity development, market and industrial economics, and political science (among other fields) to do this. And good partners that can be pulled in to support on key aspects, as needed by the gov't champions of industrial policy. The critical support needed is with the doing and implementing, not the upstream stuff on which sectors to prioritise, policy formulation etc (even though that element is valuable too).
The trouble with assembling those frameworks (PDIA, political economy, delivery units) into a support structure is that the apparatus tends to standardize what is meant to stay adaptive. "The Straight Line" caught this in fisheries: 81% of fish populations behave nonlinearly, yet the management models drawing straight lines through bent data govern them anyway. Feeling-the-stones resists a template by design, which is exactly why the rent capturers find the templates easier to game.
Yes - the trouble isn’t that we can’t plan, it’s that our tools of “management” and “planning” — like traditional HR and learning practices (for anyone teaching in the university today)— are obsolete.
Great post - agreed with all points made. This is a much needed line of argumentation. It remains a deep frustration to me that we have not as yet been able to develop demand-led and champion-responsive mechanisms to support governments in their learning-by-doing and crossing-the-river-while-feeling-the-stones approach to industrial policy. Such a support structure is key because it is terribly hard, and most low-income countries already suffer from rent capturers - such that the progressives trying industrial policy already start with a massive mountain to climb (not to mention challenges like climate change, unfavourable global financing, AI etc). We have frameworks and learnings from political economy, advisory lessons, delivery/management (inc PDIA etc), organisational capacity development, market and industrial economics, and political science (among other fields) to do this. And good partners that can be pulled in to support on key aspects, as needed by the gov't champions of industrial policy. The critical support needed is with the doing and implementing, not the upstream stuff on which sectors to prioritise, policy formulation etc (even though that element is valuable too).
I completely agree, and the "crossing the river while feeling the stones" framing is exactly right. The support gap you're pointing at is the real one.
One reason I have focused on a complex adaptive systems lens in this space is that it offers a way to match tools to governments. The same export discipline or credit allocation mechanism that built industrial capacity in Korea can actively entrench rent-seekers in a government that doesn't yet have the capacity to enforce the rules behind it. The tool isn't the issue; it's whether the government can actually make it work as intended.
That makes your rent-capture point more specific. Governments already surrounded by well-organised rent-seekers aren't just facing a harder climb. Some of the standard industrial policy instruments will give those rent-seekers more leverage, not less. Demand-led and champion-responsive support is exactly the right approach. But as a policy prescription, it needs a way to say: at this level of government capacity, these tools are viable, and these ones aren't yet. Without that, there's a real risk of handing a government tools designed for a different situation.
The iterative, learning-by-doing approach is the right process. The practical question is which stones a particular government can actually stand on, and which will give way. Getting that match between tools and capacity right is the hard part, but it's the question that needs to come before sector choice. Pick an instrument that the governing system can't back up, and it won't matter which industry is being targeted.
Agreed, Stephen. We need to design and fund a mechanism to support governments in this way, demand led and context/ capability suitable.
Yes - need both.
Perhaps also needed is some more capacity between the downstream and upstream of learning/doing scope and support plus structural/policy environment. Perhaps instead of a stolid “plan” that operates like an instruction manual or rubric, what might be more valuable or necessary are the set of just-in-time analytical tools that allow for widespread nimbleness/databases/measurable indicators or realistic/adjustable/just-in-time/performance-based work breakdown structures between these two structural conditions.
Typically, this nimbleness is found inside entrepreneurial and other private sector spaces — proprietary conditions that elicit tools that don’t necessary scale. The scalable, broadly useful post-planning version of these tools/resources are probably happening just at the edges of public-private spaces. Those spaces have to be protected to understand what comes after the “project” and industrial policy’s relegation to a series of project management tasks.
Yes indeed. There are some good books on this, for example Martin Williams' 'Reform as a Process' book which documents over 100 reforms by African governments. He shows that it is not top down plans or one-off projects that make the difference, but entrepreneurial leadership in the civil service and by cabinet members that catalysed, supported and added dynamism to ongoing effort. So external actors should help provide the tool box with multiple tools and options, and help enable them, under the leadership of the visionary progressives in governments. Private sector entrepreneurialism (pioneering magicians in the Deals and Development framework) are key, and so are entrepreneuring reformers in the government. When these two come together arounda focused agenda to develop a vertical or horizontal sector (or 2 or 3 or 4), change is possible. The evidence of industrial policy is clear on this.
The gap you name is telling: every field you list (PDIA, political economy, delivery management) is a framework, yet the thing missing is a mechanism that responds to demand rather than supplying its own agenda. "Feeling the stones" is inherently nonlinear, and 81 percent of the marine fish systems we manage turn out to be nonlinear too, but the models governing them still assume a straight line. The support structures fail the same way: they standardize the very groping that was supposed to stay adaptive.
Yes. We need philanthropic funding to provide base funding for such a mechanism based on context and complexity, because it is possible. There have been similar ones in the past (not perfect but good enough) albeit not for industrial policy.
Thanks for sharing this. "Building the requisite competencies as you go" is a really important point that you flag, and I think it could even be taken one level further.
The institutions behind the canonical industrial policy success stories (Korea's EPB, Taiwan's ITRI, Singapore's EDB) took two to three decades to form. The professional judgment capable of backing the right firms and disciplining the wrong ones accumulated over hundreds of real, consequential decisions, through a generation of bureaucratic practice where the feedback was genuine and the stakes were high.
You identify the political constraint: no durable developmentalist coalition, no serious industrial policy. But there is a formation constraint underneath the political one. The professional corps capable of carrying out serious industrial policy does not emerge from training programmes. These are officials who must hold a fifteen-year investment horizon and read sector dynamics well enough to tell a legitimate developmental bet from a rent play. They are also the ones who have to enforce discipline against connected firms when it matters — which is the hardest part. It forms through practice in environments where the feedback signal is real. A formation environment can be built over time; it cannot be purchased.
This creates a specific timing problem for any government that launches a serious industrial policy today. The political window to move is short. The professional formation timeline required for the next phase (enforcement, graduated exit, discipline at scale) typically outlasts that window by a generation. The risk is that the Bank's best-practice packages give governments the legitimacy to launch programmes they currently lack the professional depth to execute. That is not capacity building. Your reference to the Ajaokuta example is one version of where that leads.
Yuen Yuen Ang's recent piece on directed improvisation (https://polytunity.substack.com/p/industrial-policy-under-uncertainty) approaches the same problem from the discovery side: how states exercise influence rather than pick winners under genuine uncertainty. In the comments there (https://polytunity.substack.com/p/industrial-policy-under-uncertainty/comment/267004650), I raised the question of whether adaptive signal infrastructure is widely enough in place to run her grey-to-endorsement sequence. Yozma worked because foreign co-investors conducted the assessment; Nollywood's signals have no equivalent external validator. Reading emerging signals and acting on them credibly requires the same professional judgment that takes decades to build. That can't be imported either.
Where the evidence is genuinely thin relates to which countries are actually making the formation investment now, not in instruments but in the cadres who could carry serious industrial policy in a generation's time? That seems more consequential than the specifics of programme design.
Fantastic post. What would be your advice to a country (or for a development funder to a country) that clearly does not have “a durable developmentalist political coalition” and is stuck with two-bit rent seekers?
Great question because many countries in Africa are stuck with two bit rent seekers. But the Deals and Development framework, which built on Mushtaq Khan’s work, gave us insights on how to work in such a context. It requires a much longer time horizon than the usual 5 year planning and the identification of ways to move power broker & restores (rent capturers) business elite to ‘magician’ broker elite where their business model is less based on rents and more based on value addition, competing in a competitive or quasi competitive market and collaborating. Dangote is a famous example of such a shift - but you need a big enough cluster of shifters. The latter tends to mean business elite actually want state capacity to be strengthened (rather than making money from state extraction and state dysfunction). Similar work needs to be done on the political settlement side but that is harder. Taking a 20-30 year view is important. A certain form of industrial policy (let’s call it foundational, or early stage) is important to help with this. International development funders (esp philanthropy) need to support this.
Welcome back! Hope all is okay.
How does this narrative still hold in a world that’s increasingly automating? Growth in developing countries through industrialization looks broken if (1) exports are the source of that growth like in the Asian Tiger cases; (2) developed countries can now produce commodities cheaply with autonomous labor.
the world bank report is vague, the bank is drifting in the wind, doesnt want to become seen as irrelevant but has no real guidance nor parameters for finance. there's the nuts and bolts of increasing each state's capacity that trends and agencies have to get in and do, ang yuen yuen's article cited in another comment is an example of an aspect in a context, and there is the structure of political coalition that 'keeps it sold' and helps adjust the policy. carlos oyas makes a constrast of ethiopia with purposive policy driving change vs angola drifting. a notable feature of ethiopia and indonesia is not only the policy planning and execution, but that it endures through successive administrations. every step of success brings different challenges, continuing the industrial development of the TPLF Abiy had to tackle the political structure, elaborating Medemer and suppressing separatist rebellions. the term 'industrial policy' is likely to always be insufficiently broad - it easily logically includes large harm and large help. the term state capacity is also too broad by itself though less tainted by bad history. the practice of states and critique of developmentalists is going to have to push further and draw lines of opposing paths with more specific titles.
Advise that you remain attentive to the thesis of Ha Joon Chang’s “Bad Samaritans”
Thanks, good point about learning by doing. As for the World Bank, they should move their HO to Beijing or Shanghai. Seriously.
This is quite timely and informative. It reinforces a long-standing view among many political economists and development experts: No poor country has ever "free-marketed" its way to economic prosperity, nor has any nation achieved rapid economic transformation solely by following World Bank, IMF, or WTO prescription. None.
History is clear. Whether we examine the 18th and 19th century examples like Britain, USA, France, Germany, Sweden, Belgium, the Netherlands, Russia, or the post-WWII success stories like Japan, Taiwan, and South-Korea, or the late 20th century “East Asian Miracle” economies - China, South-Korea, Indonesia - the pattern is consistent. Every country that achieved rapid development did so through strategic state intervention, industrial, trade, and technology policy (ITT), protection of infant industries, and deliberate technological upgrading, not laissez-faire markets.
Even the more recent members of BRICS economies - Brazil, Russia, Indian, China, South Africa – relied on state-guided development, not pure market fundamentalism.
Nevertheless, this does not mean the World Bank, IMF, WTO, and other multilateral institutions lack relevance in today's global order. They remain important actors. However, their advice must be interpreted differently by different countries, depending on where each nation stands in its journey of economic catch-up, structural transformation, or development maintenance.
In development, context is everything, and there is no one-size-fits-all. What accelerates growth in one country may constrain it in another. The challenge for policymakers is to understand where they are on the development ladder, recognized “ladder-kicking” developmental prescription, and apply global advice in ways that strengthen, rather than weaken, their long-term national capabilities and interests.
Great post - agreed with all points made. This is a much needed line of argumentation. It remains a deep frustration to me that we have not as yet been able to develop demand-led and champion-responsive mechanisms to support governments in their learning-by-doing and crossing-the-river-while-feeling-the-stones approach to industrial policy. Such a support structure is key because it is terribly hard, and most low-income countries already suffer from rent capturers - such that the progressives trying industrial policy already start with a massive mountain to climb (not to mention challenges like climate change, unfavourable global financing, AI etc). We have frameworks and learnings from political economy, advisory lessons, delivery/management (inc PDIA etc), organisational capacity development, market and industrial economics, and political science (among other fields) to do this. And good partners that can be pulled in to support on key aspects, as needed by the gov't champions of industrial policy. The critical support needed is with the doing and implementing, not the upstream stuff on which sectors to prioritise, policy formulation etc (even though that element is valuable too).
The trouble with assembling those frameworks (PDIA, political economy, delivery units) into a support structure is that the apparatus tends to standardize what is meant to stay adaptive. "The Straight Line" caught this in fisheries: 81% of fish populations behave nonlinearly, yet the management models drawing straight lines through bent data govern them anyway. Feeling-the-stones resists a template by design, which is exactly why the rent capturers find the templates easier to game.
Yes - the trouble isn’t that we can’t plan, it’s that our tools of “management” and “planning” — like traditional HR and learning practices (for anyone teaching in the university today)— are obsolete.