# Nathan Lane, PhD > Nathan Lane is an economist and data scientist at the London School of Economics studying industrial policy, economic development, and economic history. ## Defense Promotion and Industrial Development https://nathanlane.info/research/defense-promotion-industrial-development/ ### Abstract Major economies are rearming, and rearmament is itself industrial policy. I ask whether defense promotion—especially innovation policy—can drive industrial development beyond the defense sector. I synthesize evidence on supply-side instruments (including R&D tax credits and grants) and demand-side procurement, tracing each from its effects on firms’ own R&D investment through innovation and productivity outcomes to spillovers beyond the recipient firm. The evidence is conditionally positive: public support more often crowds private effort in than out, but returns concentrate among particular firms, technologies, and designs, and weaken as one moves from investment to outputs and from civilian to defense settings. Defense spillovers, in particular, cannot be assumed. I draw five lessons for designing defense industrial policy that delivers broader economic returns, with the greatest weight on Europe’s rearmament. The developmental promise of defense promotion is real but conditional: whether rearmament also strengthens the wider industrial economy turns on how it is designed. ### Resources - [Kiel Working Paper](https://www.kielinstitut.de/fileadmin/Dateiverwaltung/IfW-Publications/fis-import/502d503a-c601-48b1-82ef-4440655253ed-Kiel_Report_Nr._11.pdf) - [Working Paper](https://nathanlane.info/research/defense-promotion-industrial-development/) - [Full text (PDF)](https://nathanlane.info/research/defense-promotion-industrial-development/defense-promotion-industrial-development.pdf) ## Geopolitics and Export Miracles: Firm-Level Evidence from U.S. War Procurement in Korea https://nathanlane.info/research/geopolitics-export-miracles/ ### Abstract How did geopolitics shape East Asia’s economic development? We show that U.S. war procurement during the Vietnam War — a fiscal shock which peaked at 2.9 percent of South Korea’s GDP, rivaling the Marshall Plan — catalyzed Korea’s export-led industrialization. We construct a new firm-level dataset linking Korean export records with U.S. procurement contracts (1966–1974) to estimate the causal impact of winning a contract on export performance. Winning an initial contract increases a firm’s likelihood of exporting by 46 percentage points and triples its export value. These effects extend beyond sales to the United States: treated firms also expanded into third-country markets. We validate our research design using unique, contemporaneous firm-level export targets, showing that contracts were not anticipated and unrelated to export shocks. The policy had lasting effects. We find that firms treated in the 1960s responded more strongly to South Korea’s heavy and chemical industry drive of the 1970s, indicating that U.S. procurement and domestic industrial policy were complementary. Our findings reveal a neglected channel through which Cold War geopolitics helped shape the East Asian economic miracle. ### Resources - [Full text (PDF)](https://nathanlane.info/research/geopolitics-export-miracles/geopolitics-export-miracles.pdf) ## Industrial Policy in the Global Semiconductor Sector https://nathanlane.info/research/semiconductor-industrial-policy/ ### Abstract This paper documents and quantifies the magnitude of industrial policies in the global semiconductor sector since the passage of the CHIPS Act in the United States. We catalogue the universe of semiconductor policies announced since 2020, develop a methodology to quantify their financial magnitude, and examine their stated objectives. We find that governments have committed over $400 billion in semiconductor subsidies globally, with the largest programs in China, the United States, and the European Union. We analyze the economic rationale for these policies, including national security concerns, supply chain resilience, and technological leadership goals. Using a quantitative trade model, we assess the potential impacts of these policies on global semiconductor production, prices, and welfare. ### Resources - [NBER Working Paper](https://www.nber.org/papers/w32651) - [SSRN Working Paper](https://papers.ssrn.com/abstract=4878912) - [OSF Preprint](https://osf.io/preprints/osf/ny3eq) - [Full text (PDF)](https://nathanlane.info/research/semiconductor-industrial-policy/semiconductor-industrial-policy.pdf) - [NBER Digest](https://www.nber.org/digest/202410/global-perspective-industrial-policy-and-semiconductor-industry) - [NBER Digest (PDF)](https://github.com/nathanlane/nathanlanespdfrepo/raw/main/writing/GoldbergEtAl_NBERSemiConductorDigest_2024.pdf) ## The Value of Names: Civil Society, Information, and Governing Multinationals https://nathanlane.info/research/value-of-names/ ### Abstract Does the human rights spotlight impact multinationals? We evaluate the effect of publicizing human rights violations on firm value, focusing on salient events at the center of international campaigns: the assassination of environmental activists. Collecting 20 years of data on activist deaths, we use financial event study methodology to estimate the impact of the human rights spotlight on the stock price of firms associated with violence. We find that the effect of the human rights spotlight is substantial. Firms named in assassination coverage have large, negative abnormal returns after events, and imply a median loss in market capitalization of 100 million USD. We show that the media plays a key role in these effects; the negative impact of assassinations is strongest when they coincide with calm news cycles versus saturated news cycles, where news is less likely to reach investors. Our study highlights economic over non-pecuniary mechanisms. Association with violence negatively impacts supply chain contracts and also inspires negative reactions by institutional investors. Lastly, we show that assassinations are positively related to the royalties paid by mining projects to domestic governments. ### Resources - [Full text (PDF)](https://nathanlane.info/research/value-of-names/value-of-names.pdf) - [SocArXiv Preprint](https://osf.io/preprints/socarxiv/aw7sq) - [SSRN Working Paper](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3751162) - [synthReturn R package](https://github.com/davidkreitmeir/synthReturn) ## Manufacturing Revolutions: Industrial Policy and Industrialization in South Korea https://nathanlane.info/research/manufacturing-revolutions/ ### Abstract I study the impact of industrial policy on industrial development by considering an important episode during the East Asian miracle: South Korea’s heavy and chemical industry (HCI) drive, 1973—1979. Based on newly assembled data, I use the introduction and termination of industrial policies to study their impacts during and after the intervention period. (1) I reveal that the heavy-chemical industrial policies promoted the expansion and dynamic comparative advantage of directly targeted industries. (2) Using variation in exposure to policies through the input-output network, I demonstrate that policy indirectly benefited downstream users of targeted intermediates. (3) The benefits of HCI persisted even after it ended, some of which took time to manifest. These findings suggest that the temporary drive shifted Korean manufacturing into more advanced markets and supported durable change. This study helps clarify the lessons drawn from the East Asian growth miracle. ### Resources - [Quarterly Journal of Economics Article (Open Access)](https://academic.oup.com/qje/article/140/1/403/7926978) - [Full text (PDF)](https://nathanlane.info/research/manufacturing-revolutions/manufacturing-revolutions.pdf) - [arXiv Preprint](https://arxiv.org/abs/2505.20566) - [SocArXiv Preprint](https://osf.io/preprints/socarxiv/6tqax) - [SSRN Working Paper](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3890311) ## Measuring Industrial Policy: A Text-Based Approach https://nathanlane.info/research/measuring-industrial-policy/ ### Abstract Since the 18th century, policymakers have debated the merits of industrial policy (IP). Yet, economists lack basic facts about its use due to measurement challenges. We propose a new approach to IP measurement based on information contained in policy text. We show how off-the-shelf supervised machine learning tools can be used to categorize industrial policies at scale. Using this approach, we validate long-standing concerns with earlier measurement approaches that conflate IP with other types of policy. We apply our methodology to a global database of commercial policy descriptions, and provide a first look at IP use at the country, industry, and year levels (2010-2022). The new data on IP suggest that i) IP is on the rise; ii) modern IP tends to use subsidies and export promotion measures as opposed to tariffs; iii) rich countries heavily dominate IP use; iv) IP tends to target sectors with an established comparative advantage, particularly in high-income countries. ### Resources - [NBER Working Paper](https://www.nber.org/papers/w33895) - [Full text (PDF)](https://nathanlane.info/research/measuring-industrial-policy/measuring-industrial-policy.pdf) - [SocArXiv Preprint](https://osf.io/preprints/socarxiv/uyxh9) - [Industrial Policy Data](https://www.industrialpolicydata.com) - [Teaching materials](https://www.industrialpolicydata.com) - [Industrial policy classifier (model weights)](https://huggingface.co/industrialpolicygroup/industrialpolicy-classifier) ### Related Work This paper is part of the Industrial Policy Group's broader research agenda on understanding and measuring industrial policies globally. See our [main project page](https://www.industrialpolicygroup.com) for more information. ## The New Economics of Industrial Policy https://nathanlane.info/research/new-economics-industrial-policy/ ### Abstract We discuss the considerable literature that has developed in recent years providing rigorous evidence on how industrial policies work. This literature is a significant improvement over the earlier generation of empirical work, which was largely correlational and marred by interpretational problems. On the whole, the recent crop of papers offers a more positive take on industrial policy. We review the standard rationales and critiques of industrial policy and provide a broad overview of new empirical approaches to measurement. We discuss how the recent literature, paying close attention to measurement, causal inference, and economic structure, is offering a nuanced and contextual understanding of the effects of industrial policy. We re-evaluate the East Asian experience with industrial policy in light of recent results. Finally, we conclude by reviewing how industrial policy is being reshaped by a new understanding of governance, a richer set of policy instruments beyond subsidies, and the reality of deindustrialization. ### Resources - [Annual Review of Economics Article (Open Access)](https://www.annualreviews.org/content/journals/10.1146/annurev-economics-081023-024638) - [NBER Working Paper](https://www.nber.org/papers/w31538) - [SocArXiv Preprint](https://osf.io/preprints/socarxiv/gsyq4) - [Full text (PDF)](https://nathanlane.info/research/new-economics-industrial-policy/new-economics-industrial-policy.pdf) - [Teaching materials](https://www.annualreviews.org/content/journals/10.1146/annurev-economics-081023-024638) ## The Political Economy of Industrial Policy https://nathanlane.info/research/political-economy-industrial-policy/ ### Abstract We examine the ways in which political realities shape industrial policy through the lens of modern political economy. We consider two broad “governance constraints”: i) the political forces that shape how industrial policy is chosen and ii) the ways in which state capacity affects implementation. The framework of modern political economy suggests that government failure is not a necessary feature of industrial policy; rather, it is more likely to emerge when countries pursue industrial policies beyond their governance capacity constraints. As such, our political economy of industrial policy is not fatalist. Instead, it enables policymakers to constructively confront the challenges of policy design. ### Resources - [Journal of Economic Perspectives Article (Open Access)](https://www.aeaweb.org/articles?id=10.1257/jep.38.4.27) - [Full text (PDF)](https://nathanlane.info/research/political-economy-industrial-policy/political-economy-industrial-policy.pdf) - [SSRN Working Paper](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4823810) - [OSF Preprint](https://osf.io/preprints/osf/y74uh) ## The New Empirics of Industrial Policy https://nathanlane.info/research/new-empirics-industrial-policy/ ### Abstract Nations have and will continue to shape their economies through industrial policy. Nevertheless, the empirical literature on these interventions is thin, dwarfed by the attention industrial policies receive from policymakers across the world. In this paper, I discuss the difficulties of empirically studying industrial policy and review how new econometric work is confronting these issues. Through careful research design and attention to institutional detail, I argue that emergent studies are rapidly expanding what we know—and updating what we thought we knew—about these policies. As well, I argue tools from policy evaluation allow us to study the impact of endogenous industrial interventions. This review is a proposal to take industrial policy, along with their complexities, more seriously as objects of inquiry. Doing so requires not only more serious evaluations of past policy but also a reevaluation of past empirical work and consensus. ### Resources - [Journal of Industry, Competition and Trade Article](https://link.springer.com/article/10.1007/s10842-019-00323-2) - [Full text (PDF)](https://nathanlane.info/research/new-empirics-industrial-policy/new-empirics-industrial-policy.pdf) - [SocArXiv Preprint](https://osf.io/preprints/socarxiv/tnxg6) ## The Historical State, Local Collective Action, and Economic Development in Vietnam https://nathanlane.info/research/historical-state-development-vietnam/ ### Abstract This study examines how the historical state conditions long-run development, using Vietnam as a laboratory. Northern Vietnam (Dai Viet) was ruled by a strong, centralized state in which the village was the fundamental administrative unit. Southern Vietnam was a peripheral tributary of the Khmer (Cambodian) Empire, which followed a patron-client model with more informal, personalized power relations and no village intermediation. Using a regression discontinuity design, the study shows that areas exposed to Dai Viet administrative institutions for a longer period prior to French colonization have experienced better economic outcomes over the past 150 years. Rich historical data document that in Dai Viet villages, citizens have been better able to organize for public goods and redistribution through civil society and local government. We argue that institutionalized village governance crowded in local cooperation and that these norms persisted long after the original institutions disappeared. ### Resources - [Econometrica](https://onlinelibrary.wiley.com/doi/abs/10.3982/ECTA15122) - [Full text (PDF)](https://nathanlane.info/research/historical-state-development-vietnam/historical-state-development-vietnam.pdf) ## New measures reveal a growing industrial policy divide https://nathanlane.info/writing/new-measures-reveal-a-growing-industrial-policy-divide/ ### New measures reveal a growing industrial policy divide Long the preserve of developing economies, industrial policy is staging a comeback in the rich world. From Washington to Brussels, governments are increasingly intervening in markets with the goal of making certain domestic activities more attractive. As economists reconsider industrial policy (Juhász et al. 2023), one thing is missing: data. For all the attention industrial policy now receives, we know surprisingly little about what governments are actually doing. Is it true advanced economies dominate industrial policy today—or are we simply paying more attention when they do it? In recent work (Juhász, Lane, Oehlsen and Pérez 2025), we aim to measure industrial policy more systematically to identify patterns in its implementation and support evidence-based evaluation. Our approach analyses the text of policy announcements using a large language model (LLM) to determine whether a policy qualifies as industrial policy. This differs from traditional approaches that classify policies based on the tools they use. This allows us to include, for example, a Brazilian import tariff on IT and telecommunications equipment aimed at supporting the domestic industry, while excluding a similar tariff in Pakistan that is primarily intended to raise government revenue (Table 1 contains further examples). [Read full article here on VoxDev](https://voxdev.org/topic/methods-measurement/new-measures-reveal-growing-industrial-policy-divide). ## A New Economics of Industrial Policy https://nathanlane.info/writing/a-new-economics-of-industrial-policy-imf-finance-and-development/ [Read on IMF Finance & Development. Full link here.](https://www.imf.org/en/Publications/fandd/issues/2024/06/A-New-Economics-of-Industrial-Policy-Reka-Juhasz-and-Nathan-Lane) ## Not a ‘side dish’: New industrial policy and competition https://nathanlane.info/writing/not-a-side-dish-new-industrial-policy-and-competition-cepr/ Not a ‘side dish’: New industrial policy and competition [from VoxEU/CEPR](https://cepr.org/voxeu/columns/not-side-dish-new-industrial-policy-and-competition) *By Cristina Caffarra and Nathaniel Lane* ### Competition Policy is Not a Side Dish Speaking recently to an audience in Brussels (at [an event](https://www.bruxconference2024.com/event/f79d717d-15c4-485a-a60b-4c718e9490b0/summary) that has since become known as the 'Anti-Davos')¹ the European Commission's top competition enforcer provocatively remarked that, "[w]hen it comes to the big issues of our times, I am afraid competition policy is neither the problem, nor the solution – it's a side dish". The statement triggered a cascade of reactions and responses. Weeks away from a European election that may change Europe's political landscape, in this column we discuss why competition policy has a broader role than the comment credits it for. Far from being a 'side dish', antitrust will be important to the transformative role that resurgent industrial policy will need to play, particularly in Europe. #### Beyond digital regulation Digital enforcement ('taming Big Tech') has assumed totemic value in Europe: it is the test of our credibility, resolve, and effectiveness in confronting surveillance business models and the entrenched market power of Big Tech. As antitrust enforcement in digital has failed, regulation is now the beach on which we fight. If Europe can achieve results here, it shows the world what can be done. But digital enforcement is also getting too much attention in relation to the scale of Europe's real structural problems. While we pride ourselves of our efforts to 'tame Big Tech', policymakers must urgently confront Europe's underwhelming economic performance across sectors. Europe is falling behind on multiple fronts: productivity, competitiveness, R&D, IT investments, and more. This view is not controversial: many have been sounding the alarm, from the ECB (Schnabel 2024), to the Bruegel think tank with their recent report for the Commission's EGov Directorate (the 'Bruegel EGOV report'; Pinkus et al. 2024).² Digital enforcement will do its thing, and it is underway. Yet on the eve of a European election, deep structural problems are urgently on the table for the incoming 2024-29 Commission. The European Commission has tasked two former Italian prime ministers, Mario Draghi and Enrico Letta, to report on competitiveness and progress towards the Single Market, respectively. Their reports will inevitably discuss the causes of Europe's fragmentation – its multitude of languages, cultures, rules, financial markets, capital markets, and economic trajectories – and what can be done to reduce those barriers and address our 'competitiveness crisis' (Schnabel 2024). The reports are also expected to highlight the need for massive ramp-up of investment in multiple strategic sectors, to promote green transition and digitalisation, to increase Europe's resilience in the global economy, and curb de-industrialisation. The future demands and scope of these investments will be unprecedented. While Europe has experience with large-scale government spending, both historically and with the pandemic and its aftermath (e.g. the EU Recovery and Resilience Facility), calls for state-led investment are now much more expansive. The confluence of post-pandemic emergencies and the waning of the Washington Consensus have created a further sense of urgency in two dimensions. First, a global resurgence of economic interest in the design of appropriate industrial policies (e.g. Juhász et al. 2022); and second, a significant pivot in the US away from the traditional aversion to the state playing a role in markets (Armstrong and Wu 2024). What does this all mean for Europe, and the role of competition policy? #### The 'New Industrial Policy' Industrial policy has returned as a major object of interest, with a proliferation of new thinking over the last five years by academics and practitioners (Rodrik et al. 2023). Questions around industrial policy have turned from 'whether' (i.e. 'should governments carry out industrial policy?') to 'how' ('how should industrial policy be carried out?'). A recent wave of research, the 'New Economics of Industrial Policy', has generated more nuanced views, and nascent work is tempering historical concerns that industrial policies are necessarily harmful: because 'losers pick governments', and state support necessarily produces 'zombie companies' – inefficient national champions. While these risks are real, recent empirical work has also established key episodes where industrial strategies likely "shifted resources in the desired direction, often producing large long-term effects in the structure of economic activity" (Juhász et al. 2023). A diverse community of industrial policy thinkers is coalescing around critical themes. There are undoubtedly differences in the 'how' (from Mazzucato 2020, urgent advocacy for mission-oriented 'moonshots', to more mainstream economic theories of intervention in Juhász et al. 2023), but there are also key commonalities: the importance of focusing on strategic sectors, the need to go beyond blunt instruments of post-war policy, and a focus on collaborative and deliberative policymaking, with input from the private sector. Scholars are also emphasising the importance of averting government failures through the design of guard rails and conditionalities (Lane 2021, Mazzucato and Rodrik 2023). What makes the rethink all the more salient is the big 'pivot' of the current US administration towards 'industrial strategy', with large public funds being allocated and disbursed to support a variety of goals: green transition, rebuilding domestic capacity offshored in the neoliberal era, supporting deindustrialised areas, reducing dependency on concentrated and brittle supply chains, and 'crowding in' complementary private investments. Foroohar (2024) argues this does not yet amount to a fully coherent industrial policy, but we would be inclined to be indulgent given the magnitude of the pivot. #### 'Antimonopoly' thinking as foundational The focus on averting the misadventures of past industrial policy (in particular, support for 'national champions') is an important reason why antitrust thinking has a major role to play in the new landscape. In the US, the worlds of industrial policy and antitrust have recently been colliding. With the major shift in antitrust thinking in Washington over the past few years has come recognition that antimonopoly values (fairness, equality, citizenry) must pervade and motivate other economic policy tools – including trade an industrial policy. As the Chair of the Federal Trade Commission, Lina Khan, recently articulated, "we are hearing arguments that America must protect its domestic monopolies to ensure that we stay ahead on the global stage. (…) we should be extraordinarily sceptical of these arguments, and instead recognize that monopoly power is a major threat to America's national interest." Further, "the choice to bring antitrust lawsuits against AT&T and IBM ended up fostering waves of innovation" (Khan 2024). And yet further: "competition policy will be a key complement to achieve industrial policy goals. As we're handing out subsidies, are there going to be strings attached, that create trajectories on an open and competitive path, rather than a closed and monopolistic path? If the industrial policy vision is one of government as a more active participant in "market making" and "market shaping", we need to make sure that our values and our vision around competition policy are wholly a part of that decision making."³ Tim Wu, a key architect of Biden-era thinking on antitrust, also describes antitrust (and, in particular, past lawsuits breaking up monopolies) as 'industrial policy'. Indeed, where antimonopoly promotes intentional economic change, it is, by definition, "an industrial policy" (Juhász et al. 2022). Making antimonopoly thinking part of the industrial policy toolbox can help break with the past: there is clear recognition among industrial policy scholars that where strategic investments are made, markets must remain 'oxygenated' – not favour dominant players; and that the more successful industrial policies are those which have supported competition (Aghion et al. 2015, Nahm 2021). Just like the 'efficiency paradigm' of the neoliberal era has been superseded in antitrust, efficiency goals may not sufficiently capture the broader aims of an industrial policy – for suppliers, regional economies, communities, citizens, and more. 'Antimonopoly', the fight against market power and its pathologies, is a fundamental value that must underpin also the direction of investments to lift entire sectors and communities. What may not be 'efficient' may have other social benefits. #### The European predicament Europe has responded to the pandemic and the 'polycrisis' also with a large increase in public spending initiatives, but we continue to face a large and widening gap in economic performance with the US and other blocks. This reflects in part a deep structural problem of persistent fragmentation along national borders, which has been Europe's 'Sisyphean rock' for decades notwithstanding major past efforts (Pinkus et al. 2024). Confronting our declining economic performance will require a major increase in public spending in selected strategic sectors, which is hard for a collection of sovereign countries, with limited federal-level resources and persistent fears that common public spending could benefit some countries more than others. The Bruegel EGOV Study suggests as a possible way forward what they describe as "coordination for competitiveness" – the European Commission performing a central coordination role to "cooperate in areas that offer the greatest gains on a sector-by-sector basis, supported by some EU-level funding. Energy policy coordination and an EU Advanced Research Projects Agency (ARPA) are two examples."⁴ #### What should competition policy do? What should be the role of competition policy in designing these policies? Pushing back against 'national champions' is certainly not new in Europe, where European Commission state aid policy has been traditionally tasked to control excess spending by national governments, and their incentives to prop up their own 'zombie firms' with state funds. State aid is a large part of Directorate-General (DG) for Competition, systematically vetting national schemes dreamt up by member states to support local interests, with the objective of avoiding distortions to "competition in the Internal Market". The traditional requirement for state support not to fall foul of state aid rules is that it "addresses a market failure" in the "most efficient way possible". It is thus unsurprising that in a recent contribution to the debate on the need for more industrial policy, senior DG Competition officials drawing from their state aids experience recommended that each industrial policy intervention be justified by a specific market failure, and adopted measures be "efficiency-enhancing" (Piechucka et al. 2023). The paper mentions "efficiency" over 100 times, "efficiency-enhancing" over 30 times, and "market failure" 80 times. While of course we want to avoid wasteful effort, this focus seems out of line with the evolution of current thinking both in antitrust and industrial policy. For instance, efficiency criteria are at odds with the rationales driving policy discussions on place-based policies and 'good jobs', which are aimed to produce larger social benefits. Efficiency criteria are also empirically incomplete: how do we prove something is 'efficiency enhancing', particularly in the case of policies with long gestation periods and whose benefits are borne in the future (Lane 2020). But more fundamentally, the usefulness of 'efficiency' as a principled goal has come deeply into question. As put by Deaton (2024), "we valorize (efficiency) over other ends. Many subscribe to (the vision) that says economists should focus on efficiency and leave equity to others, politicians or administrators. But the others regularly fail to materialise, so when efficiency comes with upward redistribution – frequently though not inevitably – our recommendations become little more than a license for plunder." Simply put, extending traditional 'state aid' thinking to industrial policy is undesirable at a time when thinking and practice around interventions are evolving. We don't need to reassure ourselves we are being 'orthodox' by casting everything as a market failure (which is easy to do, in any event, if one tries – but so what?). Nor is efficiency the 'north star' we need to be pursuing. We will need major increases in spending from the centre, and coordination of spending at the national level to ensure that collective objectives are not undermined. But if the aim is to build capacity and improve performance in Europe, industrial policy intervention that benefits citizens (not merely as consumers) cannot be held to a 'market failure/efficiency-enhancing' paradigm. Competition insights and capabilities can and should be involved in industrial policy design to provide not just an assessment of state plans along traditional state aids lines, but also affirmative values of antimonopoly, de-concentration, fairness, and distribution. Traditionalists will say 'but what is the limiting principle?' – this the generic objection to everything by those who want no change. We cannot be stuck with 'efficiency' when we need to accomplish so much more, and neoliberal 'trickle down' has been shown to be a chimera. #### Urgent proof of concept: A digital industrial policy Major focus needs to be placed on powering Europe's digital infrastructure. Europe has set huge store by its ability to 'tame Big Tech' via multiple laws: the Digital Markets Act (DMA), Digital Services Act (DSA), Data Act, and AI Act. Whether this effort will truly enable European challengers to acquire more than a marginal role remains to be seen. But we need more than extracting from Big Tech concessions to provide better deals to app developers, search rivals, ad-tech companies and e-commerce sellers, and create access regimes to platforms that are now critical infrastructure. We also need to invest locally in an independent, federated, decentralised digital infrastructure on which Europeans can build. Europe has lower advanced technology adoption than the US, and the productivity divergence between high- and low-productivity firms has widened more in digital-intensive sectors (Criuscolo 2021). On the positive side, the number of EU-based start-ups is high, and there is vitality in terms of researchers, digital skills, and emerging technologies (Meyer 2024). Yet Europe's fragmentation and its dependencies on US giants make it challenging to implement, commercialise, and scale hi-tech activity. We thus need a robust digital industrial policy alongside the existing diet of EU regulation and innovation policies. This means coordinating national and EU-level efforts to create autonomous infrastructures and reduce the dependency on Big Tech. These goals also align with narratives on 'sovereignty'. As summarised by Bria (2023), "to strengthen our economic and political sovereignty in a complex geopolitical environment, Europe needs a combination of regulatory frameworks and active digital industrial policies. This objective goes beyond merely crafting regulations. It's about building new markets and industries, creating innovative institutions, and fostering ecosystems that truly serve the public interest." Investing in a 'Europe stack' tech ecosystem should be an attractive candidate for EU-level funding because the cross-border externalities are high. Bria (2023) suggests a €10 billion EU Digital Sovereignty Fund, which would "blend grants and equity investments, fostering pan-European collaboration among our national innovation agencies (…) to establish robust digital public infrastructures and digital commons, offering viable alternatives to current monopolistic digital platform models, supporting the deployment of open AI models and decentralized applications, sovereign data spaces, open knowledge tools and content, privacy-preserving digital IDs, and digital payment systems." The Bruegel EGOV Study (Pinkus et al. 2024) suggests an 'EU ARPA' involving the creation of an independent agency with a €5 billion budget to pool investment projects and coordinate spending at national level, to be topped up with additional funds from EU programmes. Objectives would be set by the EU Council and the European Parliament, but the agency would be autonomous in policy implementation. While prior initiatives have proven insufficient for multiple reasons (for example, the Juncker Plan of 2015 and the European Fund for Strategic Investments), and significant obstacles remain – not least risk-taking appetite and competencies – we need to double down now that we have more scholarship, experience, and expertise. Critically, the experience and expertise of DG Competition in digital markets will be critical here: successful investment in federated decentralised infrastructures requires understanding of the regulatory environment, and of competitive dynamics which can facilitate private complementary investment and innovation on these infrastructures. Overall, getting Europe to improve performance will require a massive, concerted effort at national and EU levels to identify strategic sectors and disburse funds in a targeted way that will crowd in private investment. Competition thinking has a key role to play, not to enforce narrow and nebulous efficiency goals, but to ensure initiatives are consistent with antimonopoly values, fairness, and opportunities. Not a 'side dish'. #### References Aghion, P, J Cai, M Dewatripont, L Du, A Harrison and P Legros (2015), "Industrial Policy and Competition", *American Economic Journal: Macroeconomics* 7(4): 1-32. Armstrong, R and E Wu (2024), "[Dani Rodrik: doing industrial policy right](https://www.ft.com/content/34872d9a-3587-4b27-a01d-2905f8e23408)", *Financial Times*, 9 February. Bria, F (2023), "[Open, Sovereign, Independent AI: Europe's greatest challenge?](https://medium.com/@francescabria/open-sovereign-independent-ai-europes-greatest-challenge-6c8a899041ec)", Speech at the European Parliament, 10 December. Caffarra, C (2021), "[What are we regulating for?](https://cepr.org/voxeu/blogs-and-reviews/what-are-we-regulating)", VoxEU.org, 3 September. Caffarra, C (2023), "[Europe's Tech Regulation Is Not an Economic Policy](https://www.project-syndicate.org/commentary/european-union-digital-markets-act-will-not-tame-big-tech-by-cristina-caffarra-2023-10)", Project Syndicate, 11 October. Caffarra, C (2024a), "[Antitrust and the political economy: Part 1](https://cepr.org/voxeu/columns/antitrust-and-political-economy-part-1)", VoxEU.org, 5 January. Caffarra, C (2024b), "[Antitrust and the political economy: Part 2](https://cepr.org/voxeu/columns/antitrust-and-political-economy-part-2)", VoxEU.org, 6 January. Caffarra, C and B Kilic (2024), "[Re-joining trade with antitrust](https://cepr.org/voxeu/columns/re-joining-trade-antitrust)", VoxEU.org, 7 March. Criscuolo, C (2021), "[Productivity and Business Dynamics through the lens of COVID-19: the shock, risks and opportunities](https://www.ecb.europa.eu/press/conferences/ecbforum/shared/pdf/2021/Criscuolo_paper.en.pdf)", European Central Bank Forum on Central Banking, September. Deaton, A (2024), "[Rethinking My Economics](https://www.imf.org/en/Publications/fandd/issues/2024/03/Symposium-Rethinking-Economics-Angus-Deaton)", IMF Finance & Development, March. Foroohar, R (2024), "[America (still) has no industrial policy](https://www.ft.com/content/6c72cfe0-445a-4d88-b674-bb7fe41b9826?accessToken=zwAAAY5uQPwNkc9scs_gRFpNiNO2dLt_5BuYJg.MEUCIQDEG5mDm4w5xi0UL9ylZIllRvjo0bPTgUpKD-hz2q3GOQIgNx88uYydbNlIMLjx01K0WLq4xB-23fTXRiZTx_lRm3c&segmentId=e95a9ae7-622c-6235-5f87-51e412b47e97&shareType=enterprise)", *Financial Times*, 18 March. Juhász, R, N Lane, E Oehlsen and V C Pérez (2022), "[The Who, What, When, and How of Industrial Policy: A Text-based Approach](https://osf.io/preprints/socarxiv/uyxh9)", SocArXiv, 25 August. Juhász, R, N J Lane and D Rodrik (2023), "[The New Economics of Industrial Policy](https://www.nber.org/papers/w31538)", NBER Working Paper 31538. Khan, L (2024), "[America Has a Resilience Problem](https://foreignpolicy.com/2024/03/20/lina-khan-ftc-trade-united-states-economy-tech-monopoly-national-security-boeing/?tpcc=recirc_latest062921)", *Foreign Policy*, 20 March. Lane, N (2020), "[The New Empirics of Industrial Policy](https://doi.org/10.1007/s10842-019-00323-2)", *Journal of Industry, Competition and Trade* 1(2): 1–26. Lane, N (2021), "[A Flight Plan That Fails](https://www.bostonreview.net/forum_response/a-flight-plan-that-fails/)", *Boston Review*, 15 September. Mazzucato, M (2020), *Mission Economy: A Moonshot Guide to Changing Capitalism*, Allen Lane. Mazzucato, M and D Rodrik (2023), "[Industrial Policy with Conditionalities: A Taxonomy and Sample Cases](https://www.ucl.ac.uk/bartlett/public-purpose/sites/bartlett_public_purpose/files/mazzucato_m._rodrik_d._2023_industrial_policy_with_conditionalities_a_taxonomy_and_sample_cases.pdf)", UCL Institute for Innovation and Public Purpose Working Paper 2023/07. Meyer, Z (2024), "[Helping Europe's Digital Economy Take Off: An Agenda for the Next Commission](https://www.cer.eu/publications/archive/policy-brief/2024/helping-europes-digital-economy-take-off)", Centre for European Reform, 20 February. Nahm, J (2021), *Collaborative Advantage: Forging Green Industries in the New Global Economy,* Oxford University Press. Piechucka, J, L Saurí-Romero and B Smulders (2023), ["Industrial Policies, Competition and Efficiency: The Need for State Aid Control"](https://academic.oup.com/jcle/article-abstract/19/4/503/7334444), *Journal of Competition Law & Economics* 19(4): 503-526. Pinkus, D, J Pisani-Ferry, S Tagliapietra, R Veugelers, G Zachmann and J Zettelmeyer (2024), "[Coordination for EU competitiveness](https://www.europarl.europa.eu/RegData/etudes/STUD/2024/747838/IPOL_STU(2024)747838_EN.pdf)", Bruegel EGOV Report, March. Rodrik, D (2014), "[Green Industrial Policy](https://drodrik.scholar.harvard.edu/files/dani-rodrik/files/green_industrial_policy.pdf)", *Oxford Review of Economic Policy* 30(3): 469-491. Rodrik, D, R Juhasz and N Lane (2023), "[Economists Reconsider Industrial Policy](https://www.project-syndicate.org/commentary/new-economic-research-more-favorable-to-industrial-policy-by-dani-rodrik-et-al-2023-08)", Project Syndicate, 4 August. Schnabel, I (2024), "[From laggard to leader? Closing the euro area's technology gap](https://www.ecb.europa.eu/press/key/date/2024/html/ecb.sp240216~df6f8d9c31.en.html)", Inaugural Speech: European and Monetary Union Lab, 16 February. --- #### Footnotes 1. https://www.politico.com/news/2024/02/16/how-some-of-the-worlds-most-powerful-regulators-are-trying-to-upend-the-economic-system-00141802 2. The analysis of the causes of the gap in the Bruegel EGOV report includes: low labour and total factor productivity growth relative to the US especially since 2020, with large intra-EU differences; especially dramatic difference in productivity with the US in information and communication technology (ICT); slower accumulation of IT capital and better technology adoption in the US; much lower intensity of R&D spending especially in three key sectors – pharma/biotech, software and IT; significantly higher industrial electricity retail prices than the US; higher hourly labour costs; much higher cost of equity finance and lower volume of venture capital funding and therefore much greater restrictions in accessing risk capital; and finally, much lower trade across national borders than one would expect to see given past effort at market integration. 3. https://cepr.org/events/competition-policy-rpn-reinvigorating-antitrust-citizens-not-just-consumers 4. The reference is to the US ARPA, which has been instrumental in mobilising resources and investing them in high-risk, high-reward projects. ## Economists Reconsider Industrial Policy https://nathanlane.info/writing/economists-reconsider-industrial-policy-project-syndicate/ [Read on Project Syndicate. Full link here.](https://www.project-syndicate.org/commentary/new-economic-research-more-favorable-to-industrial-policy-by-dani-rodrik-et-al-2023-08) ## Economics Must Catch Up On Industrial Policy https://nathanlane.info/writing/economics-must-catch-up-on-industrial-policy-promarket-series/ ### Economics Must Catch Up On Industrial Policy (ProMarket Series) *[This article [original here]](https://www.promarket.org/2023/03/14/economics-must-catch-up-on-industrial-policy/) is part of [ProMarket's series on industrial policy](https://www.promarket.org/category/news/series/industrial-policy/).* As countries openly embrace industrial policies, social science—and economics in particular—is faced with a knowledge deficit. Put simply, scholarship around industrial policy has left us bereft of a systematic understanding of these policies. We are missing an empirical roadmap for how to perform industrial policy. For starters, systematic evidence on best practices is incomplete. Where industrial policies are multidimensional and complex, our understanding is inadequate across nearly every dimension. The abandonment of industrial policy as a serious object of study has left us with an oddly piecemeal understanding of practice. Given the renewed international interest and national experimentation with industrial policy, we need an empirically grounded conversation now more than ever. We need evidence, data, and concreteness, and we needed it yesterday. After a decades-long intellectual hiatus, industrial policy must be taken seriously as an object of study by the economics discipline. In doing so, scholars need to interrogate how industrial policy is practiced *in* practice. Our recent work is a small step in this direction, and an attempt to measure industrial policy in the real world. By building descriptive facts surrounding contemporary practice, we join emerging work on industrial policy and contribute to our basic understanding of these levers. #### Going Underground, Impossibility Theorems, and the Rise of Empirics For a subject wrought with controversy, one fact is not controversial: since the 1980s, industrial policy has receded as a serious object of inquiry in economics. The height of globalization and, with it, the triumphalism of the Washington Consensus, signaled the nadir of industrial policy as a serious subject of study within mainstream economics. Academic economics saw industrial policy as *old hat*, part of the structural developmentalism of the previous generation of economics. Yet while industrial policy was declared dead, practice hummed along anyways. Academic economics may have turned away from policymaking, but the exigencies of the real world meant that industrial policy never truly went away; it merely faded from the view of mainstream academia, and public oversight. In the early 1980s, Robert Reich [observed](https://hbr.org/1982/01/why-the-us-needs-an-industrial-policy) that governments always end up doing industrial strategy, only without a coherent strategy, pursuing promotion with "no rhyme or reason" the policies most vulnerable to special interest pressure. The rise of international agreements constrained industrial policy in its most conspicuous forms, yet these interventions never went away. Between the 1980s and 1990s, a steady march of multilateral agreements reduced old tools of protectionism, such as tariffs, but industrial policy merely mutated and [took new forms](https://jasongarred.org/Garred2018TradePolicyChina.pdf). As if summoned by the economist's scepter of unintended consequences, industrial policy merely went underground. I call these the 'wilderness years' for industrial policy research. In these wilderness years, where academic economics *did* engage with industrial policy, it was relegated to whimsical thought experiments. Industrial policy was simply a curiosity. Gary Becker perhaps best exemplified the spirit of the debate in 1985: "The best industrial policy is none at all." Here, much intellectual engaging in industrial policy largely focused on high-level debates surrounding the impossibility of industrial policy. As if the question of industrial policy was a hypothetical binary variable, untangled from the complexities of practice on the ground. To this day, much work in economics has been dedicated to demonstrating that industrial policy does not work, even empirically. This "impossibility theorem" view of industrial policy permeates the conversation surrounding industrial policy to this day. Inspired by debates about central planning, some of the most vocal critics have followed a Hayekian and Austrian route, [claiming that](https://www.promarket.org/2023/01/24/industrial-policy-is-a-seductive-mirage/) "industrial policy is literally impossible." Confronting the details of actual practice is largely second order. Nevertheless, it is a tall order to claim that industrial policy is, literally, impossible. Whilst important for intellectual debates, it is often lateral to the realities of practice. These claims are also open to being falsified by empirical work. Historically, for example, they require social scientists to confront episodes where industrial policy featured conspicuously in post-war development, including the experiences of Taiwan, Singapore, Japan, South Korea, and, more recently, China. Today, these claims implore us to erroneously reject emerging work on the efficacy of [place-based policies](https://www.aeaweb.org/articles?id=10.1257/jep.34.3.99), [or innovation policy and R&D](https://www.nber.org/papers/w26483), two literatures where empirical evidence has rapidly expanded. In the wilderness years of industrial policy, the scattered empirical work that *did* exist, thus, centered on [rejecting the potential](https://www.jstor.org/stable/3986500) that industrial policy played a role in the postwar miracle episodes. To the extent that the mainstream economics permitted the role of industrial policy, such as the [World Bank's post-mortem of the East Asian growth miracle](https://www.google.co.uk/books/edition/The_East_Asian_Miracle/k3HuPwOlPXoC?hl=en), it nevertheless located success with largely getting market fundamentals right. It was largely left to political scientists, comparative sociologists, and heterodox economists in seminal qualitative work to unpack the practice and role of industrial policy in the [East Asian growth miracle](https://press.princeton.edu/books/paperback/9780691117294/governing-the-market). However, the empirical turn in development economics has slowly challenged the impossibility theorem-style arguments against industrial policy. As industrial policy re-emerged over the past decade, empirical work has begun evaluating the tenants of these policies. A trickle of well-identified, econometric [studies have started to pick apart](https://osf.io/preprints/socarxiv/tnxg6/) and unpack the experience of specific case studies. For example, Juhász's (2018) paper on the Napoleonic blockade applied the tools of causal inference and natural experiments—in this case, a blockade that mimicked protection of French textile manufacturers—to examine the mechanisms of the infant industry hypothesis. When subject to serious empirical study, it turns out that the theory of industrial policy is very much relevant. For the East Asian miracle, [the work of Lane (2022)](https://osf.io/preprints/socarxiv/6tqax/) targeted the Heavy Chemical and Industry Drive of South Korea. Largely, empirical work on the South Korean episode has made room for a formative role of industrial policy, as least in one canonical episode. #### Back To Basics: Measurement in an Imperfect World Yet, although the empirical turn in economics has finally reached the industrial policy debate, practitioners of industrial policy still lack fundamentals. Not only is the econometric work on industrial policy still in its infancy, the policy world lacks something more fundamental: empirical basics. Specifically, we do not yet understand the descriptive facts, nor do we have measures of contemporary policy practice. Our recent work, joint with Emily Oehlsen and Veronica Perez, is a [new attempt to establish](https://steg.cepr.org/publications/who-what-when-and-how-industrial-policy-text-based-approach) empirical fundamentals of industrial policy. In doing so, we join researchers who have started doing the hard work of providing basic insights surrounding industrial policy, from considering the [conceptual foundations](https://www.promarket.org/2023/01/17/a-new-framework-for-better-industrial-policies/) of industrial policy, to accounting for industrial policy practice in [primary source documents](https://www.csis.org/programs/chinese-business-and-economics/projects/chinese-industrial-policy). Our proof of concept uses machine learning techniques that automatically classify industrial policies using textual policy descriptions. Instead of using quantitative policy measures, which may be used for many purposes, 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 ‒ the definition of industrial policy. We apply our classification algorithm to a large international database of commercial policy text, the Global Trade Alert (GTA) project, and obtain counts of new flows of industrial policy use at the country-sector-year level. Creating measures of industrial policy allows us, for the first time, to describe the rough contours of industrial policy practice, globally. Even this prototype dataset yielded results that surprised us. We see that the industrial policies used today are a far cry from caricatures of overt post-war protectionism. Policies tend to take the form of financial lending and export support, and much less so tariffs. These industrial policy levers also tend to be granular, often targeted toward specific clusters of industries. Although, like the past, manufacturing is an omnipresent feature of targeting: economies across the income distribution target capital goods, as well as heavy industry. Given the seemingly technocratic nature of industrial policy that emerges, we find a strong relationship between the flow of new industrial policy and a country's income. According to our data, high-income countries implement about five times as many industrial policies, on average, as low- to middle-income economies. Among these latter countries, it is mostly industrialized, middle-income economies using industrial policies ([UNIDO](https://iap.unido.org/articles/trends-global-industrial-policy)). Furthermore, wealthy nations tend to target commodities for which they have high comparative advantage in world export markets. Whatever the source of these patterns, they beg the question: developing countries will face hurdles if they leverage industrial policies to move into competitive manufacturing markets. Simply put, advanced, high-income countries have the state capacity to deploy technocratic industrial policies, and doing so may be challenging for lower-income countries. Thus, the patterns in our data suggest the need to invest in state capacity, as well as the need to understand the institutional demands of advanced industrial policies. Currently, we lack systematic quantitative evidence on the political economy of successful industrial policies. Our research is a first stab into establishing basic cross-country facts about industrial policy. For practitioners and researchers, establishing an empirical picture of industrial policy is necessary to have constructive conversations about how to perform industrial policy. After all, industrial policy never went away—it's been alive and well in the developed world, even while academic economists read its eulogy. As the world starts to take industrial policy seriously, economics should take these policies seriously, too, as an object of study. Economists must perform the hard work of establishing the basic facts surrounding these controversial policies. We're already late. ## A Flight Plan That Fails https://nathanlane.info/writing/a-flight-plan-that-fails-boston-review/ ### A Flight Plan That Fails By Nathan Lane *September 15, 2021* *From the Boston Review's Forum on industrial policy [[Original post here](https://www.bostonreview.net/forum/industrial-policys-comeback/a-flight-plan-that-fails/)]. [A debate](https://www.bostonreview.net/forum/industrial-policys-comeback/a-flight-plan-that-fails/) with Mariana Mazzucato, Rainer Kattel, Josh Ryan-Collins.* It's a weird time to be an industrial policy researcher. "[The Return of the Policy That Shall Not Be Named](https://www.imf.org/en/Publications/WP/Issues/2019/03/26/The-Return-of-the-Policy-That-Shall-Not-Be-Named-Principles-of-Industrial-Policy-46710)" caught us off guard, and the demand for prescriptions looms over an embarrassingly scant body of knowledge—especially within the field of economics, which for the past few decades has offered little more than Gary Becker's 1985 quip that "the best industrial policy is none at all." At a time when governments are returning to industrial policy, we are largely clueless about how to make it work. > At a time when governments are returning to industrial policy, we are largely clueless about how to make it work. In this climate of ignorance, there is something a little jarring about Mazzucato, Kattel, and Ryan-Collins's wish to lead us from today's tepid practice to a bolder paradigm. In order to meet the vast challenges we face today, we are implored to reject our current "market failure approach" and think grander. Instead of surgical policies, we must marshal state, civil society, and markets all in service of a mission-driven cause. While this call for moonshots is stirring, it ultimately says too little about how to turn this vision into reality. When it comes to the conceptual issues facing industrial policymakers, Mazzucato and colleagues are mostly right, albeit at a high level. Industrial policy certainly requires more institutional capacity (a vital point that is too easily lost in technocratic debates). We do need new tools for policy evaluation—since the criteria we wield today are often meant to declaw industrial policy rather than assess it—as well as new strategies for sharing risk and reward. But this abstract framework leads us astray in three ways, particularly for the United States. First, this paradigm largely avoids any specificity about the hard decisions to be made—a blind spot created in part by their caricature of industrial policy's past. Second, they say almost nothing about politics, which has prevented even the "market failure approach" they reject from being implemented robustly. And third, they dismiss some of the powerful tools progressives already have at this critical juncture. Without grappling with the details, vaulting ambition—however inspiring—risks leading us only further into the wilderness. Start with Mazzucato and colleagues' portrait of industrial policy's past. They are right that the Washington Consensus viewed industrial policy as a failed vestige of postwar developmentalism. In the last two decades of the twentieth century, most academic economists who had anything to say about industrial policy quibbled over its theoretical flaws. But even as the discipline deemed industrial policy implausible, the world has kept using it—and we are finally learning more about the complexity of these real world efforts in the wake of our field's "empirical revolution." Budding work over the last few years [has painted a rich, varied picture](https://nathanlane.info/research/new-empirics-industrial-policy/) of industrial policy in practice, not just in theory. Nearly every paper complicates received wisdom about how industrial policy worked, or didn't. As an empiricist and economic historian who has learned a great deal from this work, I don't recognize the tepidness that Mazzucato and colleagues think characterize postwar policies across the board: those Cold War interventions fixated on "narrowly defined technological goals and specific sectors" that they take as a foil. > Industrial history tells us that grand, mission-oriented policies do not absolve us from having to think about details. The truth is that postwar industrial policy was hardly devoid of grand visions. Wading through forgotten Five-Year Plans across the (non-communist) developed and developing world, you will not see a timid focus on particular industries or lack of multi-sectoral missions. Ideas like the "Big Push," the postwar concept that investment may need to be coordinated, were multi-front investment drives and pursued with a myriad of policy levers. And in other dimensions, Lyndon B. Johnson's Great Society initiative embodied an ambitious social mission in the United States. Industrial history also tells us that grand, mission-oriented policies do not absolve us from having to think about details. Postwar East Asia, the paragon of postwar policymaking, has hardly been characterized by timid policies or narrow focus; their successes don't seem distant from the sweeping missions Mazzucato and colleagues champion. At the same time, their grandness required specific interventions, and they were crucial to their working. Moreover, past efforts didn't fail for lack of ambition to bring together multiple sectors of the economy, much less society. They typically failed due to the [realities of their state capacity and political context.](https://theconversation.com/ghana-lessons-from-nkrumahs-fallout-with-his-economic-adviser-53233) That brings us to the second weakness of this framework: its failure to grapple with the realities of political economy. We might read Mazzucato, Kattel, and Ryan-Collins as urging a profound shift in our preferences, yet they fail to elucidate the constraints, much less the political game they are bound by. Policy and institution-building must contend with discordant interest groups and the realities of the political business cycle. In the United States, in particular, industrial policy must contend with staunch Republican opposition, whose whole raison d'être is to shrink the state to such an extent that it can be drowned in a bathtub, as Grover Norquist put it. Any moonshot will have to survive the punishing magnitude of U.S. political gravity. Perhaps it is precisely because of this political climate that ambition is necessary: as a tactical matter, policies must be boldly formulated in order to survive the political bargain that will inevitably gut them. (Congressional winnowing of Biden's grand visions for infrastructure may be a case in point.) But this kind of tactical ambition should not come at the cost of misdiagnosing the problem. Mazzucato and colleagues wish to save us from the tepidness of a "market failure approach"—"find the market failure; fix it with a support instrument"—as if it has actually been taken seriously in U.S. economic policy over the last forty years. If only. We should not conflate these political failures with the concept as such. In reality, market failure remains an indispensable tool. Quantitative work in economics has shown that market imperfections can be a powerful guide for the allocation of policy. Ernest Liu, for example, [has recently demonstrated](https://www.youtube.com/watch?v=H2xwdTqqQRU) how analysis of distortions that ripple through the industrial network can help pinpoint sectors most ripe for policy. Closely related to network economies, concepts such as "granularity" reveal another interaction between market imperfection and industrial strategy. In a world rife with imperfections, a small number of firms can constitute the lion's share of economic activity in a given market—domestically and internationally. As a result, a handful of firms [may shape](https://voxeu.org/article/superstar-firms-and-comparative-advantage-countries) the comparative advantage of a national export industry. The implications of this form of market failure are important when it comes to industrial policy and regulation, as [Cecile Gaubert, Oleg Itskhoki, and Maximilian Vogler](https://www.sciencedirect.com/science/article/pii/S0304393221000416) have recently shown. > Beneath any moonshot must lie launchpad scaffolding, but we lack the messy, hard, complicated details of practical deployment. Or consider industrial policies in oligopolistic markets (another form of market imperfection). In a study of China's shipbuilding push, [Myrto Kalouptsidi, Panle Jia-Barwick, and Nahim Bin Zahur](https://voxeu.org/article/industrial-policy-lessons-china) have shown how industrial policies interact with the complexities of market structure. Among other things, this work can speak to the multitude of policy levers mobilized by the Chinese Communist Party. These are undoubtedly useful tools for understanding the multitude of incentives that industrial policy can wield. If Mazzucato, Kattel, and Ryan-Colins are wrong to dismiss the value of attending to market failures, they are correct about our need for new tools. The tools they promote, however—ones focused on dynamic efficiency—aren't mutually exclusive of market failures. Notions of dynamic efficiency are indeed important for evaluating industrial policy: as the authors note, when long-haul missions are meant to pay future dividends, short run attention to static efficiency can be misleading. But dynamic efficiency alone can't save us. In fact, the move the authors make—rejecting the importance of market failure in favor of dynamic efficiency—has often been deployed in arguments *against* government intervention, from Harold Demsetz's famous [critique](https://www.jstor.org/stable/724977?seq=1#metadata_info_tab_contents) of Kenneth Arrow's [argument](https://www.nber.org/system/files/chapters/c2144/c2144.pdf) for government investment to Robert Bork's limited conception of the scope of antitrust law and the [intellectual toolkit of Chicago School deregulation](https://www.springer.com/gp/book/9789024737925). Dynamic efficiency has even been deployed in the defense of [Jeff Bezos and the necessity of monopoly power](https://www.econlib.org/profit-for-progress/). It's not that dynamic efficiency is useless. But, unmoored from attention to market failures, it will not get us to where we need to be. *** For all these reasons, the picture offered by Mazzucato and colleagues fails as a guide to the future of industrial policy. Beneath any moonshot must lie launchpad scaffolding, but we lack the messy, hard, complicated details of practical deployment. Grand ambitions should not blind us to the granular, technical decisions they will entail. Industrial policy does need to be ambitious, but even more urgently, it needs to be detailed, practical, and precise. ## Industrial Policy - A Round Up of Historical Case... https://nathanlane.info/posts/industrial-policy-a-round-up-of-historical-case-studies-and-beyond/ This is a short literature review of recent papers on industrial policy (and related topics), focusing on work coming from economic historians. It is taken somewhat--though not entirely--from an upcoming literature review I have written on the empirics of industrial policy. (Stay tuned for more contemporary policy discussions.) **I. Output Protection: Temporary and Otherwise.** ![Industrial Policy](https://nathanlane.info/images/blog/uploads/2019/01/17/e010782410-v8.jpg) _A. France._ First things first. Reka Juhasz (2018) study has been widely discussed--for good reason. It is one of the first attempts to unpack the impact of infant industry policy using contemporary causal inference. In this case using the Napoleonic blockade of England as a short-term shock that a) mimicked a temporary protection policy of French textile producers, the spatial scope of which meant b) different regions of French producers were subjected to different amounts of protect. See the [AEA's writeup here](https://www.aeaweb.org/research/infant-industry-napoleonic-blockade). Following up on this, a new paper Bécuwe and Blancheton (2018) considers the role of protection and the long-run impact on British textiles, albeit from a more case-study business history approach. > Bécuwe, S., and Blancheton, B. (2018). French textile specialisation in long run perspective (1836–1938): trade policy as industrial policy. Business History, 1–21. [https://doi.org/10.1080/00076791.2018.1494732](https://doi.org/10.1080/00076791.2018.1494732) \[Ungated: \] > > Juhasz, R. (2018). Temporary Protection and Technology Adoption: Evidence from the Napoleonic Blockade. _American Economic Review_, _108_(11), 3339–3376. [https://doi.org/10.1257/aer.20151730](https://doi.org/10.1257/aer.20151730 "https://doi.org/10.1257/aer.20151730") \[Ungated: [http://www.rjuhasz.com/research/juhasz_blockade.pdf](http://www.rjuhasz.com/research/juhasz_blockade.pdf)\] _B. Canada._ Industrial policy is often associated with postwar East Asian economies and the input substitution industrialization (ISI) of Latin American. Of course, most Western nations pursued significant industrial policies throughout their development path. One canonical example is Canada's National Policy in the late 19th century. An interesting wave of papers have explored the development impact of the National Policy on Canadian industrialization: Alexander and Keay (2018), Harris, Keay, and Lewis (2015), and Inwood, and Keay, (2013). > Alexander, P. D., and Keay, I. (2018). A general equilibrium analysis of Canada's national policy. Explorations in Economic History, 68, 1–15. [https://doi.org/10.1016/j.eeh.2017.11.002](https://doi.org/10.1016/j.eeh.2017.11.002 "https://doi.org/10.1016/j.eeh.2017.11.002") \[Ungated: [https://www.bankofcanada.ca/wp-content/uploads/2017/05/swp2017-18.pdf](https://www.bankofcanada.ca/wp-content/uploads/2017/05/swp2017-18.pdf "https://www.bankofcanada.ca/wp-content/uploads/2017/05/swp2017-18.pdf")\] > > Harris, R., Keay, I., and Lewis, F. (2015). Protecting infant industries: Canadian manufacturing and the national policy, 1870–1913. Explorations in Economic History, 56, 15–31. [https://doi.org/10.1016/j.eeh.2015.01.001](https://doi.org/10.1016/j.eeh.2015.01.001 "https://doi.org/10.1016/j.eeh.2015.01.001") \[Ungated: [https://pdfs.semanticscholar.org/d357/cb822e40abea4b4d56d4383824738907fbc9.pdf](https://pdfs.semanticscholar.org/d357/cb822e40abea4b4d56d4383824738907fbc9.pdf "https://pdfs.semanticscholar.org/d357/cb822e40abea4b4d56d4383824738907fbc9.pdf")\] > > Inwood, K., and Keay, I. (2013). Trade policy and industrial development: iron and steel in a small open economy, 1870−1913. Canadian Journal of Economics/Revue Canadienne d'économique, 46(4), 1265–1294. [https://doi.org/10.1111/caje.12048](https://doi.org/10.1111/caje.12048 "https://doi.org/10.1111/caje.12048") \[Ungated: [https://web.archive.org/web/20180529214458/http://www.econ.queensu.ca/files/other/inwoodkeay13.pdf](https://web.archive.org/web/20180529214458/http://www.econ.queensu.ca/files/other/inwoodkeay13.pdf "https://web.archive.org/web/20180529214458/http://www.econ.queensu.ca/files/other/inwoodkeay13.pdf")\] **II. Temporary Subsidies and Spillovers** Besides temporary protection of output _per se_, a number of studies have considered the effect of other temporary interventions. For example, Hanlon (2018), considers how the temporary cost advantages related to steel jump-started the U.K.'s lead in metal ship building, and the role labor market spillovers played in maintaining their leadership. On the other side of the Atlantic, Hanlon also focuses on how interventions allowed U.S. producers to transition from wood to steel. Mitrunen (2019) also looks at the labor market spillovers of a particularly interesting intervention, focusing Finland's postwar reparations to the Soviet Union. The reparations policy mandated that Finland pay Stalin in the form of heavy industrial commodities. Accordingly, Mitrunen studies reparations as a type of forced export industrial policy, focusing on the effects on skilled labor--contemporaneous and intergenerational effects. Jaworski and Smyth (2018) considers how U.S. government purchases during wartime helped the "big names" of the U.S. aeronautical industry. In particular, these purchases allowed U.S. manufacturers survive later turbulent periods of shakeout. In this bucket, is also my own study on South Korea's heavy chemical and industry big push. Here (Lane 2017), I focus on the impact of a temporary big push policy, emphasizing linkage effects of industrial targeting. \[[Read my piece on this here.](https://voxdev.org/topic/firms-trade/manufacturing-revolutions-role-industrial-policy-south-korea-s-industrialisation)\] Similarly, Pons-Benaiges (2017) job market paper explores the composition and scope of Japan's postwar industrial policy. His paper is, in part, [a reply to a classic piece by Beason and Weinstien (1996)](https://econpapers.repec.org/article/tprrestat/v_3a78_3ay_3a1996_3ai_3a2_3ap_3a286-95.htm), which asked whether Japanese interventions targeted sectors with increasing returns to scale. > Hanlon, W. (2018). The Persistent Effect of Temporary Input Cost Advantages in Shipbuilding, 1850-1911. Working Paper. \[Ungated: [walkerhanlon.com/papers/hanlon_shipbuilding.pdf](https://walkerhanlon.com/papers/hanlon_shipbuilding.pdf)\] > > Jaworski, T., and Smyth, A. (2018). Shakeout in the early commercial airframe industry. The Economic History Review, 71(2), 617–638. [https://doi.org/10.1111/ehr.12430](https://doi.org/10.1111/ehr.12430) > > Lane, N. (2017). Manufacturing Revolutions - Industrial Policy and Networks in South Korea. Working Paper. \[Ungated: [Manufacturing Revolutions](https://nathanlane.info/research/manufacturing-revolutions/)\] > > Mitrunen, M. (2019). War Reparations, Structural Change, and Intergenerational Mobility. Working Paper. \[Ungated: [https://www.dropbox.com/s/ohl6zqyi4ctkclw/Mitrunen_JMP.pdf?dl=0](https://www.dropbox.com/s/ohl6zqyi4ctkclw/Mitrunen_JMP.pdf?dl=0 "https://www.dropbox.com/s/ohl6zqyi4ctkclw/Mitrunen_JMP.pdf?dl=0")\] > > Pons-Benaiges, O. (2017). Did Government Intervention Target Technological Externalities? Industrial Policy and Economic Growth in Postwar Japan, 1964-1983. Working Paper. [https://web.archive.org/web/20200309004942/https://economics.stanford.edu/sites/g/files/sbiybj9386/f/oriol_jmp.pdf](https://web.archive.org/web/20200309004942/https://economics.stanford.edu/sites/g/files/sbiybj9386/f/oriol_jmp.pdf) **III. Managerial and technological interventions** Giorcelli (2019) is a fantastic take on a large industrial intervention. The factor accumulation promoted by the Marshall Plan across Western Europe looms large as an explanation behind "Les Trente Glorieuses"—the decades of sustained economic growth following World War II. Giorcelli focuses on the long-run implications of a particularly successful Marshall Plan project in Italy, the United States Technical Assistance and Productivity Program (1952–58). Specifically, the policy focused on both transferring technology--and more importantly--modern managment techniques to Italian firms: > Giorcelli, M. (2019). The Long-Term Effects of Management and Technology Transfers. American Economic Review, 109(1), 1–33. [https://www.aeaweb.org/articles?id=10.1257/aer.20170619](https://www.aeaweb.org/articles?id=10.1257/aer.20170619 "https://www.aeaweb.org/articles?id=10.1257/aer.20170619") \[Ungated: [www.giorcellimichela.com/papers/giorcelli_productivity_program.pdf](https://www.giorcellimichela.com/papers/giorcelli_productivity_program.pdf)\] ## A Delightful (Informal) Romp Through Implementation Theory https://nathanlane.info/posts/a-delightful-informal-romp-through-implementation-theory/ As an undergraduate suffering through a graduate PhD micro course, I recall my professor bringing up the "hot new thing" that was implementation theory. I tried hard at the time to ascertain what this literature was. The other day a colleague of mine mentioned that he was relieved by the direction network economics as going, with more applied studies and an increasing number of empirically-grounded studies. "If that hadn't happened," he said, "it may have gone the way of implementation theory." With that, it hit me: _what the hell happened to implementation theory?_ So I tweeted into the Econ Twitter black hole. ![]()And received this patient reply from good ol' Ben Golub: ![]() Thus starts his thread with a fantastic piece by Matt Jackson (of current network theory fame): A Crash Course in Implementation Theory (2001). Which is a glimpse into the (then) state of the art. > [Jackson, M. O. (2001). A crash course in implementation theory. Social Choice and Welfare, 18(4), 655–708.](https://link.springer.com/content/pdf/10.1007/s003550100152.pdf) \[Ungated: [link](https://authors.library.caltech.edu/79990/1/sswp1076.pdf)\] With a also a little lit review. ![]() The two articles are from Matt Jackson (1992), and Maskin and Jean Tirole (1999): > [Jackson, M. O. (1992). Implementation in Undominated Strategies: A Look at Bounded Mechanisms. The Review of Economic Studies, 59(4), 757–775](http://dx.doi.org/10.2307/2297996). \[Ungated: [Link](https://www.kellogg.northwestern.edu/research/math/papers/833.pdf)\] > [Maskin, E., & Tirole, J. (1999). Unforeseen Contingencies and Incomplete Contracts. The Review of Economic Studies, 66(1), 83–114.](http://dx.doi.org/10.1111/1467-937X.00079) \[Ungated: [Link](https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&cad=rja&uact=8&ved=2ahUKEwigj5SWpdbfAhXPdN4KHVzACaAQFjAEegQIBhAC&url=https%3A%2F%2Fwww.sss.ias.edu%2Ffiles%2Fpapers%2Feconpapereight.pdf&usg=AOvVaw2XzSxGBWsgdl7DABnMe9s2)\] But Ben then gives a nice synopsis of what happened. ![]() ## Saying the Quiet Part https://nathanlane.info/posts/saying-the-quiet-part-1/ ![](https://nathanlane.info/images/blog/uploads/2018/12/27/c42145-20.jpg) It's often revealing when someone when someone "says the quiet part loud." I thought this was particularly resonant: > We are fortunate that, thanks to globalisation, policy decisions in the US have been largely replaced by global market forces. National security aside, it hardly makes any difference who will be the next president. The world is governed by market forces. > > \-Alan Greenspan, 2007, interview with a Swiss newspaper. > > op cit. Tooze (2018) "Crashed: How a Decade of Financial Crises Changed the World" I suspect this may be more true for the developing world than OECD countries. Regardless, though, I think do not think it is a radical statement there is a conflict between domestic political goals (globally) and the rules governing the international economic system. Just ask the architects. The is a point captured in the recent book, "Globalists: The End of Empire and the Birth of Neoliberalism" by Quinn Slobodian. (Coincidentally, quote is also in the introduction.) ## Building a Dataset from UN Comtrade's API https://nathanlane.info/posts/uncomtrade/ The following is a simple program for building data sets using the United Nation's COMTRADE API. Note: this code does not apply to the "Bulk download" API, which allows users to download entire chunks of annual or monthly data. Instead, this program is for making specific, repeated queries for import (export) data between certain countries. ### 1. First Things First: Some Prepation. My script loops over lists of reporting countries and their counterparts. I have these lists saved as CSV files. The UN Comtrade system uses its own coding scheme. Thus, we do not make queries based on country names, but using their identifiers found here coding scheme. I have list of countries and their corresponding codes saved as.csv files. I have a trade partner list (uncomtrade_partners.csv) and list of reporting countries (uncomtrade_reporting.csv), both of which have the following layout:
![Image](https://nathanlane.info/images/blog/assets/comtradelist.png)
The countries I wish to query have a column (data.frame$V3) with a 1 indicating the country I wish to include in my query. The following code sets takes these.csv files, and creates two lists of UN COMTRADE codes that correspond to the reporting countries I want and the partner countries I want. ```R ## TOP MATTER # Libraries. library(rjson) library(magrittr) library(httr) library(RCurl) library(rvest) library(data.table) # A directory where we save things. /your/file/path -> project_directory ## PREPARE COUNTRY LISTS # Deal with partner country list. "uncomtrade_partners.csv" %>% file.path( codelist_path,. ) %>% read.csv(. ) -> partners_dummies # Get partner list. partners_dummies[ which( partners_dummies$V3==1 ),] %>% .$V1 %>% as.vector(. ) -> partners_slist # Deal with reporting country list. "uncomtrade_reporting.csv" %>% file.path( codelist_path,. ) %>% read.csv(. ) -> reporting_dummies # Get reporting country list reporting_dummies[ which( reporting_dummies$V3==1 ),] %>% .$V1 %>% as.vector(. ) -> reporting_list ``` ### 2. The Meat: Querying and Saving Data Chunks Next I define two programs. The first program is a file for retreiving the UN COMTRADE data in (.csv) form from the UN COMTRADE's bulk download API. The following code is based loosely on COMTRADE's own R code One important distiction between this code and mine is that I found that on non-Windows computers, the UN's code didn't handle HTTPS very well. In fact, using their code I tended to get connection errors, in particular, ```R Error in curl::curl_fetch_memory(url, handle = handle): Peer certificate cannot be authenticated with given CA certificates ``` Yup, an HTTPS related error. Thus, the line set_config( config( ssl_verifypeer = 0L ) ) deals with R's HTTPS issues (As reported here). Finally, read.csv( text = content( GET(string), as="text", type = "text/csv", header = TRUE )) _properly_ retrieves a concatenated HTTPS URL stored in the string, and parsing the retrieved text content into a coherent.CSV table. (The function deals with.JSON-based queries in similar ways, but I just prefer.CSV data.) ```R ## PROGRAM 1. GET DATA FROM THE UN COMTRADE API. # First deal with HTTPS. httr::set_config( config( ssl_verifypeer = 0L ) ) # Define UN COMTRADE RETRIEVAL FUNCTION. get.Comtrade <- function( url="https://comtrade.un.org/api/get?" , maxrec=50000 , type="C" , freq="A" , px="HS" , ps="now" , r , p , rg="all" , cc="TOTAL" , fmt="json" ) { # Tell us where we are. paste( "Querying reporting country", r, "trade with", p, sep = " " ) %>% print(. ) # The URL string constructed from function arguments. string <- paste(url ,"max=",maxrec,"&" #maximum no. of records returned ,"type=",type,"&" #type of trade (c=commodities) ,"freq=",freq,"&" #frequency ,"px=",px,"&" #classification ,"ps=",ps,"&" #time period ,"r=",r,"&" #reporting area ,"p=",p,"&" #partner country ,"rg=",rg,"&" #trade flow ,"cc=",cc,"&" #classification code ,"fmt=",fmt #Format ,sep = "" ) # How to process.csv OR the.json data. if( fmt == "csv" ) { # Read.csv. data <- read.csv( text = content( GET(string), as="text", type = "text/csv", header = TRUE )) # Make dataset into datatable. save.Comtrade( data, px, ps, r, p ) } else { if(fmt == "json" ) { # Access URL properly and retrieve JSON. retrieved_url <- content( GET( string ), as ="text", type = "text/html" ) raw.data <- jsonlite::fromJSON( retrieved_url ) # Turn into dataframe data <- as.data.frame( raw.data ) colnames( data ) <- names( data ) # Make dataset into datatable. save.Comtrade( data, px, ps, r, p ) } } } ``` The second program simply saves each date query. Specifically, within the get.Comtrade() function above, we dumped our queried.CSV tables into a local directory. (As opposed to holding them in memory, given R's funky memory issues). This was done using our function, save.Comtrade(), which takes an individual data.frame and saves it using write.csv(). ```R ## PROGRAM 2. SAVE PREPARED DATA.FRAME AS CSV. # Path for raw UN comtrade queries. file.path( working_path, "data", "uncomtradequeries" ) -> rawcomtrade_path ##... Takes argumenents from main function get.Comtrade save.Comtrade <- function( dataset_argument, px, ps, r, p ){ # Prepare file name, pass to fast writing CSV method. paste( "trade", px, ps, r, p, "raw.csv", sep = "_" ) -> filename # Tell us where we are. paste( "Saving dataset for reporting country", r, "trade with", p, sep = " " ) %>% print(. ) # Concatenate the project directory and filename. filename %>% file.path( project_directory,. ) %>% # Save the dataset as a CSV. write.csv( dataset_argument, file =. ) } ``` ### 3. Making Multiple Queries Using mapply() The mapply() function runs the get.Comtrade() UN COMTRADE retrieval program, doing so for each combination of countries from the __reporting_list__ and the __partners_list__ generated in Part 1--holding a list of parameters constant with MoreArgs. ```R ## EXECUTE QUERY FOR ALL COUNTRY PAIRS. # Build array (actually, data.frame) of arguments for function: argument_array <- expand.grid( partner_arguments = partners_list, reporting_arguments = reporting_list ) # Show arguments array: argument_array # Take multiple arguments and apply the function. mapply( get.Comtrade, r = argument_array$reporting_arguments, p = argument_array$partner_arguments, MoreArgs = list( ps = "all", fmt = "csv", rg = "1", px = "h0", cc = "all") ) ``` Again, mapply() is like a loop, applying the API querying function to each reporting+partner country combination. However, we cannot directly use the _reporting_list_ and the _partners_list_ as the two argument lists in mapply() (doing so won't generate a complete list of combinations). Instead, we use the expand.grid() function to generate all reporting+partner country combinations; and this table gives us two columns, argument_array$reporting_arguments and argument_array$partner_arguments, which are the actual country argument lists used by mapply(). While mapply()'s main value added is cycling over many argument lists, we want to hold some get.Comtrade() arguments constant. To do so, we supply a list of arguments to MoreArgs. The parameters we hold costant are in the list, list( ps = "all", fmt = "csv", rg = "1", px = "h0", cc = "all"). ### 4. Stacking Saved Data Chunks into a Dataset Together, mapply() and get.Comtrade() repeatedly queried and saved each chunk of queried data to a directory. However, 100s of data chunks are hardly useful as a dataset. We was to assemble the pieces into a single coherent dataset. Thus, the following program assembles our saved queries into a coherent dataset. ```R ## MAKE DATASET OUT OF ALL SAVED FILES ## Make list of files we saved locally to the project path. csvfile_list <- list.files( path = project_directory, pattern = "trade_h0.*.csv", all.files = FALSE, full.names = TRUE, recursive = FALSE ) # Lapply the rest of the files to the main.dt file 1. lapply( csvfile_list, fread, sep = "," ) %>% rbindlist(., fill = TRUE ) -> main.dt # Save the stacked CSV. write.csv( main.dt, file = file.path( project_directory, "thebigdataset.csv" ) ) ``` The above code makes a list of all the files in _/your/file/path_ that match our naming scheme, _trade_h0_all_[REPORTING COUNTRY]_[PARTNER COUNTRY]_raw.csv_. Next, we efficiently open all these files and stack them into a coherent dataset. We use lapply() to loop over the list of.csv files and open each using nimble fread() function (from the wonderful __data.table package). The list of opened datasets are then "stacked" into a single dataset using rbindlist(). Finally, we we save the assembled dataset to our local direction, _/your/file/path_. You can use your preferred method of course, especially if you run into memory problems writing large.CSV files with R (I prefer using the experimentalfwrite() function in the data.table package, __which I discuss here.__). ## Quick Note: Writing large .csv files in R with fwrite()!... https://nathanlane.info/posts/writingwithfwrite/ ### The problem. R can be nasty when it comes to reading and writing "large" datasets. As practitioners, we often appeal to hacky practices, emergent libraries, alternatives methods to avoid crashing our systems during heavy jobs. One alternative [I had the appealed to]({{ base }}/tutorial/2016/01/27/ffasthackforbigdata.html) was the __[ff library](https://cran.r-project.org/web/packages/ff/index.html)__, a great library for processing large data. Importantly, ff had an awesome function for quickly saving multi-gig.csv files (write.table.ffdf). Recently, however, write.table.ffdf (and other go-to methods) seemed to constantly crash large jobs on my Linux machine. I'd come back to my computer to only to find, ![Image](https://nathanlane.info/images/blog/assets/fatality.png) ### fwrite() to the rescure. (Re-)enter the data.table package. Like ff, data.table is useful in its own right for processing big data. Though the library __had__ great methods for quickly opening large files with its fread() command, it lacked a comparable fwrite() command. Until now... data.table author, __[Matt Dowle](https://github.com/mattdowle)__, is experimenting with adding (See his extensive blog post [here](http://blog.h2o.ai/2016/04/fast-csv-writing-for-r/)) an fwrite() functon (contributed by Otto Seiskari) to the package. Like fread(), fwrite() is also written in C and is surprisingly efficient. The fwrite() function has truly come through for me, allowing me to write terrabytes of weather data without the complications I have run into with other packages. ### Installing fwrite() functionality. As of March 30, 2017, fwrite() function hasn't been added to current, core CRAN release of the package. To install the development version of the data.table library with the fwrite() command, you'll have to uninstall your current standard version of the library and install the development version from the github repository: ```R remove.packages("data.table") install.packages("data.table", type = "source", repos = "http://Rdatatable.github.io/data.table" ) ``` See [the data.table wiki for more](https://github.com/Rdatatable/data.table/wiki/Installation).s ```R installing to /home/XXX/R/x86_64-pc-linux-gnu-library/3.3/data.table/libs ** R ** inst ** byte-compile and prepare package for lazy loading ** help *** installing help indices ** building package indices ** installing vignettes ** testing if installed package can be loaded * DONE (data.table) ``` There wont be any documentation; don't be alarmed when nothing shows up when you type ?(?)fwrite: ```R >?fwrite No documentation for 'fwrite' in specified packages and libraries: you could try '??fwrite' ``` For my current project I have to repeatedly process a hundred files, each with over 40 million lines (around 4 gigs each). Writing these data.frames to a.csv takes under a minute on a pretty lowly desktop: ![Image](https://nathanlane.info/images/blog/assets/fwritespeed.png) Much faster than other methods. Also: fairly recently, the prolific Wes McKinney and Hadley Wickham developed the feather package ([See: http://github.com/wesm/feather](http://github.com/wesm/feather).) to address some of the memory issues that plague folks trying to edit and save large data sets in R (See a pretty good discussion [here](https://blog.rstudio.org/2016/03/29/feather/).). ## About https://nathanlane.info/about/ **Nathan Lane, PhD**, is an Assistant Professor of Economic Development at the London School of Economics, in the [Department of International Development](https://www.lse.ac.uk/international-development). He is an affiliate of CESifo (Munich), and LSE’s [Centre for Economic Performance](https://cep.lse.ac.uk/) (CEP) and [Programme on Innovation and Diffusion](https://poid.lse.ac.uk/) (POID). He co-founded the empirical economics research lab, the [Industrial Policy Group](https://www.industrialpolicygroup.com/). His research covers applied machine learning, economic development, political economy, and industrial economics. Before joining the LSE, Dr. Lane was an Assistant Professor at the University of Oxford, a Senior Lecturer at Monash University, and co-founded [SodaLabs.io](https://sodalabs.io), a lab focused on addressing social science questions using big data and computational tools. He also worked as a postdoctoral researcher at the Massachusetts Institute of Technology under Professor Daron Acemoglu. Dr. Lane’s research agenda focuses on establishing fundamental facts about economic change, with particular emphasis on understanding the efficacy, determinants, and patterns of industrial policies. His work has been featured in _The Economist_, _The Financial Times_, _The Wall Street Journal_, _Trade Talks_, and other major outlets. Through the Industrial Policy Group, Dr. Lane aims to establish industrial strategy as a rigorous object of inquiry within economics. His recent projects quantify global policy patterns using large-scale textual data on commercial policy and examine the role of policy in the global semiconductor sector. Dr. Lane is passionate about computational methods, machine learning, and unstructured data. He has been a Linux user for 30 years. Outside academia, he enjoys comedy, hardcore punk music, and exercise science. ## Media https://nathanlane.info/media/