The State and Industrial Transformation: Insights from Peter Evans () Embedded Autonomy: States and Industrial Transformation

Reviewing ‘Embedded Autonomy’ for EDUC 306B

Reading Notes
Courses
Political Economy
State Theory
Author

Tales Mançano

Published

January 3, 2026

Full citation: Evans, P. B. (1995). Embedded autonomy: States and industrial transformation. Princeton University Press.

What are the central research questions or problems raised by the authors?

What core concepts, evidence, and research methods are utilized?

As you do the readings, think about what the author did right as well as wrong.

What are the interesting ideas in the paper?

If you disagree with an argument, what would it require to persuade you?

Can these differences be adjudicated through further empirical study?

Personal Notes and highlights

Uma coisa que me marcou nesse livro é a metafora inicial dos leões que comem burocratas, me parece uma otima metáfora, mas não é assim que começaria o livro.

As Evans puts it: “Identification with the escaped lion is natural, but until less hierarchical ways of avoiding a Hobbesian world are discovered, the state lies at the center of solutions to the problem of order. Without the state, markets, the other master institution of modern society, cannot function. We do not spend our valuable time standing in lines in front of the counters of bureaucrats because we are masochists. We stand there because we need what the state provides. We need predictable rules, and these in turn must have a concrete organizational structure behind them. Fervent calls for the dismantling of the state by late-twentieth-century capitalist free-marketeers served to derail the state’s ability to act as an instrument of distributive justice, but not to reduce its overall importance.”

How this text makes me feel:

Eu não soumuito fã desse estudo de livro, é meio confuso para mim, As metáforas são legais, mas a forma como constroi o problema teórico, bem, não é tanto a minha praia. Fico pensando nas confounders e em perder o ponto, é um modelo de livro bastante aberto, me da um pouco de frio na Barriga.

— Central questions (Mix Me and LLM)

What are the central research questions or problems raised by the authors?

Evans central problem is how state structures and state-society relations shape possibilities for industrial transformation in newly industrializing countries (NICs). The core puzzle is: why do some states successfully promote economic development while others impede it? More specifically, he asks whether states can deliberately change their position in the international division of labor, especially in technologically demanding sectors like information technology. Theoretically, his contribution is a shift from debating “how much” states intervene to what kind” of state involvement matters, and why states differs in those perspectives. He challenges the neo-utilitarian (or liberal) orthodoxy dominant in the 1980s that portrayed state intervention as inevitably creating rent-seeking behavior and economic distortions.

Derivative questions (in portuguese):

  1. Não “quanto” o Estado deve intervir, mas “que tipo” de intervenção e estrutura estatal promove o desenvolvimento?.
  2. Por que alguns estados (como a Coreia do Sul) conseguem promover a transformação industrial e outros (como o Zaire) se tornam predadores que destroem a economia?.
  3. Como as estruturas internas do Estado e suas relações com a sociedade (especificamente a classe industrial) variam e impactam o desenvolvimento?.
  4. É possível para países de industrialização tardia (NICs) alterarem deliberadamente sua posição na divisão internacional do trabalho, construindo vantagens comparativas em setores de alta tecnologia (como a informática)?

Core Concepts

  1. Embedded Autonomy:

The book’s central theoretical concept is “embedded autonomy”—an apparently contradictory combination of bureaucratic insulation and deep social ties. Evans argues that developmental states require both: - 1.1 Autonomy: Weberian bureaucratic coherence through meritocratic recruitment, long-term career rewards, and corporate identity that prevents individual incumbents from pursuing purely personal goals

  • 1.2 Embeddedness: Dense concrete ties binding the state to society through institutionalized channels for continual negotiation of goals and policies

  • 1.3 Neither element works alone. Pure autonomy would lack intelligence and implementation capacity, while embeddedness without internal coherence would prevent the state from transcending individual private interests.

  1. State Roles

Evans develops four analytical categories to describe state involvement patterns:

  • Custodian: Regulatory role privileging policing over promotion, aimed at preventing or restricting private initiatives
  • Demiurge: Direct state production assuming private capital incapable of sustaining necessary commodity production
  • Midwifery: Assisting emergence of new entrepreneurial groups through protection, subsidies, and signaling
  • Husbandry: Ongoing support for established firms facing global challenges through R&D support and complementary state activities

Evans argues midwifery-husbandry combinations work better than custodian-demiurge approaches.

Evidence and Research Methods

Evans employs a comparative institutional approach combining multiple evidence types:

  1. Case Selection He examines three NICs—Brazil, India, and Korea that were chosen because they represent countries desperate enough to attempt transformation but less constrained than raw material exporters (?). Within each, he focuses on the information technology sector during the 1970s-1980s as an exceptionally strong test case.

  2. Data Sources

  • Secondary literature: Scholarly accounts and World Bank analyses
  • Government documents and statistics: Regulatory agency data, industry association publications
  • Key informant interviews: Dozens of current and former government officials and private executives provided crucial insights into state structures, bureaucratic dynamics, and firm strategies
  • Business press: Day-to-day reporting on firm and product evolution
  • Firm-level data: Annual reports and corporate publications

Evans emphasizes avoiding state reification by examining specific agencies, organizations, and individual actors. He acknowledges that informants offer biased accounts, but treats these biases as evidence themselves, juxtaposing theories from different bureaucratic positions.

Strengths of the Analysis

Theoretical Innovation

Evans successfully moves beyond sterile debates about market versus state by demonstrating that variation in state structures matters more than degree of intervention. His embedded autonomy concept elegantly synthesizes Weber’s bureaucratic theory, Gerschenkron’s developmental state insights, and contemporary analysis of East Asian success.

The framework effectively explains why Korea outperformed Brazil and India: Korea approximated embedded autonomy while the others exhibited “partial and imperfect approximations”. His sectoral analysis validated these general patterns—Korea’s Ministry of Communications-ETRI-chaebol networks epitomized embedded autonomy, while India’s Department of Electronics showed the costs of insufficient embeddedness.

Empirical Richness

Evans provides compelling evidence that developing countries can construct comparative advantage in high-technology sectors, contradicting conventional economic wisdom. By the late 1980s, Korea had become one of the world’s largest semiconductor producers, Brazil had created a multibillion-dollar locally-owned computer industry, and India developed significant software export capacity.

His analysis reveals surprising patterns: comparative advantage lay not in low-wage assembly but in skilled intellectual labor and design-intensive production. India excelled in software engineering despite negative value-added in routine assembly. Brazil achieved international competitiveness in financial automation systems.

Dynamic Analysis

Perhaps most impressive is Evans’s attention to how successful transformation changes the political foundations of state involvement. He documents a contradictory dynamic: nationalist greenhouse (Protectionism) policies created local entrepreneurial groups that eventually allied with transnational corporations, undercutting political support for the very state agencies that helped create them.

This insight contradicts neo-utilitarian predictions of static rent-seeking symbiosis between officials and firms. Instead, Evans finds that “state involvement was associated with economic dynamism, and the result was political contestation, not symbiosis”.

Weaknesses and Limitations

Incomplete Political Dynamics Theory

Evans acknowledges a “serious lacuna” in his framework. While the comparative institutional approach facilitated uncovering contradictory political dynamics, it did not predict them. His initial discussion offered ideas about how states affect transformation but “had relatively little to say about how this transformation would change the basis of subsequent state involvement”.

The book documents how successful midwifery produces “gravediggers” that undermine developmental state coalitions, but doesn’t provide a systematic theory of this process. When do transformative projects strengthen versus undermine their own foundations?

Measurement Challenges

Evans’s concepts, while analytically powerful, face operationalization difficulties. How precisely do we measure “embeddedness”? When does a state cross from “partial approximation” to genuine embedded autonomy? The framework sometimes appears post-hoc—we know Korea succeeded, therefore it must have had embedded autonomy.

The reliance on interview data, while providing rich texture, raises questions about selection bias and generalizability. How representative were Evans’s informants? Did survivorship bias shape which officials and entrepreneurs he interviewed?

Sectoral Specificity

The information technology focus provides analytical leverage but raises questions about external validity. IT is characterized by rapid technological change, relatively low capital intensity compared to steel or chemicals, and possibilities for modular production. Do Evans’s findings apply to other sectors with different characteristics?

He acknowledges this concern but argues IT provides an “exceptionally strong test” precisely because conventional wisdom would predict state involvement should fail. Still, the scope conditions of embedded autonomy remain underspecified.

Interesting and Provocative Ideas

Bureaucracy as Scarce Resource

Evans inverts conventional wisdom: “It is the scarcity of bureaucracy that undermines development, not its prevalence”. While bureaucracy remains pejorative for citizens and policymakers, Evans shows that approximating Weberian bureaucratic ideals is exceptionally difficult in most Third World states, and its absence cripples developmental efforts.

User-Producer Linkages

The analysis of Brazil’s financial automation sector reveals how tight user-producer linkages can generate international competitiveness despite high component costs. When major banks were simultaneously producers and users, intimate knowledge of local idiosyncrasies translated into custom-designed systems that were 30% cheaper than international alternatives.

Learning State Agencies

Evans documents remarkable institutional evolution: India’s Department of Electronics, initially dedicated to protecting state-owned ECIL, transformed itself toward promotion rather than policing. Despite neo-utilitarian predictions that bureaucrats would defend their regulatory power, DOE officials embraced liberalization because they were “immersed in a project of transformation that was of greater interest than minor individual perquisites”.

This suggests that technocratic commitment to sectoral success can override bureaucratic self-interest—a finding with profound implications for institutional reform.

Points of Disagreement and Evidence Requirements

The Inevitability of Internationalization

Evans argues the “new internationalization” of the late 1980s—growing alliances between local firms and TNCs—was “in part the product of successful midwifery” rather than simply “the empire strikes back”. This interpretation seems optimistic. What evidence would distinguish successful developmental partnerships from recolonization?

To be persuasive, we would need: - Longitudinal data on value capture: Do local firms in joint ventures retain growing shares of profits, or do returns accrue primarily to TNC partners? - Technology transfer metrics: Are local capabilities deepening, or are firms becoming dependent on proprietary TNC technology? - Trajectory analysis: Do alliance firms eventually compete independently, or do they devolve into permanent subsidiaries?

Evans provides some evidence on these questions but acknowledges “the specter of getting trapped in the high-tech equivalent of trading cheap cotton for expensive cloth remained real”. More systematic comparison of alliance outcomes versus autonomous development paths would strengthen the argument.

State Capacity Thresholds

Evans argues that even Brazil and India, with “partial and imperfect approximations” of embedded autonomy, achieved significant results. But how much state capacity is minimally necessary? What happens in cases below Brazil and India’s level?

To adjudicate this, we would need: - Analysis of failed cases: Countries that attempted similar policies but achieved worse outcomes - Threshold identification: Specific capacity indicators (meritocratic recruitment rates, bureaucratic tenure patterns, institutionalized state-business consultation mechanisms) that distinguish effective from ineffective states - Decomposition studies: Which elements of embedded autonomy are necessary versus sufficient?

Counterfactual Scenarios

Evans’s strongest claim is that state involvement caused industrial transformation. But what would have happened without these policies? Korea’s chaebol were already powerful before specific IT policies. Would they have entered the sector anyway?

Persuasive evidence would include: - Within-country variation: Comparing sectors with and without state support in the same country - Timing analysis: Demonstrating that firm entry followed rather than preceded policy changes - Firm-level studies: Showing that state support influenced specific investment decisions

Evans provides some such evidence (e.g., the timing of Korean chaebol entry into semiconductors following state initiatives) but systematic analysis remains limited.

Empirical Adjudication Possibilities

Many of Evans’s arguments can be adjudicated through further empirical study:

State Structure Measurements

Comparative surveys could measure Weberian characteristics across countries: recruitment procedures, career patterns, salary structures, professionalization indicators. Correlating these with developmental outcomes would test the embedded autonomy hypothesis more rigorously.

Network Analysis

Contemporary social network analysis tools could map state-business ties, measuring embeddedness density, reciprocity, and centralization. Comparing network structures across successful and unsuccessful cases would validate or challenge Evans’s framework.

Long-term Performance Tracking

Following the IT sectors Evans studied into the 21st century reveals whether his predictions held. Did Korean firms move up the value chain? Did Brazilian and Indian alliances lead to capability deepening or hollowing out? Such longitudinal analysis tests the theory’s predictive power.

Mechanisms Testing

Evans proposes specific mechanisms (e.g., midwifery induces entrepreneurial entry, husbandry sustains competitiveness). Case studies examining variation in these mechanisms and their effects would identify which are essential and under what conditions.

Conclusion

Evans provides a theoretically sophisticated and empirically rich analysis that fundamentally advances our understanding of state roles in industrial transformation. His embedded autonomy concept elegantly captures the paradoxical requirements of developmental states, while his sectoral analysis demonstrates that even technologically demanding industries are amenable to strategic state action.

The book’s greatest contribution may be its dynamic perspective: recognizing that successful transformation reshapes the political foundations of future state action. This insight opens crucial questions about the sustainability of developmental coalitions and the reconstruction of state-society relations—questions that remain central to development studies today.

Countries Classification

Com base na obra Embedded Autonomy de Peter Evans, a classificação dos países nos conceitos de estrutura estatal, bem como a justificativa para a escolha dos estudos de caso, pode ser detalhada da seguinte forma:

  1. Classificação dos Países e Conceitos Evans utiliza três categorias principais para classificar os estados com base em suas estruturas internas (burocracia) e suas relações externas com a sociedade (inserção):
  • Estado Predatório (Exemplo: Zaire sob Mobutu):
    • Características: Falta de burocracia competente e ausência de regras previsíveis. Os cargos públicos são tratados como propriedade privada para extração de riqueza pessoal, onde “tudo está à venda”. O Estado atua contra o desenvolvimento, extraindo excedentes da sociedade sem fornecer bens coletivos em troca.
    • Justificativa: O Zaire (atual República Democrática do Congo) é usado como o arquétipo do “pesadelo neoutilitário”, onde a lógica de mercado domina o comportamento administrativo sem qualquer coerência corporativa.
  • Estado Desenvolvimentista (Exemplos: Japão, Coreia do Sul, Taiwan):
    • Características: Possuem Autonomia Inserida (Embedded Autonomy). Combinam uma burocracia weberiana meritocrática e coesa (autonomia) com redes densas de conexão com o setor industrial privado (inserção).
    • Justificativa: O Japão é o modelo original, mas a Coreia do Sul é o foco principal da análise empírica de Evans. O Estado coreano conseguiu disciplinar e apoiar os grandes conglomerados (chaebol) para promover a transformação industrial e exportações.
  • Estado Intermediário (Exemplos: Brasil e Índia):
    • Características: Exibem aproximações parciais da autonomia inserida. Não são puramente predatórios, mas carecem da coerência total dos estados desenvolvimentistas.
      • Brasil: Caracterizado por uma burocracia fragmentada, modernizada por “adição” e não por transformação, resultando em “bolsões de eficiência” (como o BNDE) cercados por clientelismo. As relações com a sociedade são muitas vezes individualizadas (“anéis burocráticos”) em vez de institucionais.
      • Índia: Possui uma longa tradição burocrática (o IAS), mas sofre de uma inserção ineficaz. O Estado manteve-se historicamente isolado e ambivalente em relação ao capital privado, focando em regulação restritiva (o “raj de licenças”) em vez de promoção conjunta.
  1. Por que esses países foram escolhidos? Peter Evans selecionou explicitamente o Brasil, a Índia e a Coreia do Sul como o trio central para sua análise comparativa. As razões para essa escolha são:
  • Novos Países Industrializados (NICs): A escolha focou em países em desenvolvimento que fossem grandes ou avançados o suficiente para suportar uma gama completa de produção industrial. Eles representam o grupo para o qual o desafio da transformação industrial é mais saliente.
  • Diversidade de Abordagens: À primeira vista, é um trio improvável, o que enriquece a comparação:
    • O Brasil era o arquétipo do “desenvolvimento dependente”, impulsionado por multinacionais e desigualdade social.
    • A Índia era conhecida por sua tendência à autarquia e desconfiança do capital privado.
    • A Coreia do Sul era o modelo de crescimento liderado por exportações e grandes conglomerados locais.
  • Fator Comum (Envolvimento Estatal): Apesar das diferenças, em todos os três casos, o envolvimento do Estado na industrialização é inegável. Isso permitiu a Evans testar sua tese de que a questão não é “quanto” o Estado intervém, mas “como”.
  1. Qual a justificativa para o foco no setor de Informática? Evans escolheu analisar a indústria de Tecnologia da Informação (TI) nesses três países como uma lente para testar suas teorias por razões estratégicas:
  • O Teste Difícil: A indústria de informática é um setor de alta tecnologia e rápida mudança, onde a sabedoria convencional (neoliberal) diria que o Estado tem menos chances de sucesso e deveria intervir menos. Se o Estado pode desempenhar um papel transformador aqui, o argumento a favor da intervenção estatal é fortalecido.
  • Conspiração Multidimensional: A TI é vista como o setor com maior probabilidade de gerar externalidades positivas e impulsionar o desenvolvimento no final do século XX.
  • Controlar Variáveis: Ao olhar para o mesmo setor em três países diferentes, Evans pôde controlar as características tecnológicas e de mercado, focando nas variações causadas pelas diferentes estruturas estatais e papéis adotados (Custodiante, Demiurgo, Parteira, Husbandry).

LLM Generated content

NotebookLM

Perplexity

NotebookLM : #### Qual a tese central do texto?

  1. ideas, concepts, arguments that you found stimulating, worth remembering and building on,
  2. questions, concerns, disagreements with ideas encountered,
  3. connections, linkages, contradictions between one idea or approach and another.

Short memos are due by 5pm Monday, the day before class. A third of your course grade will be based on the short memos. Please submit them through Canvas.

  1. Long memos: For four of the topics, students will prepare a more detailed memo (5 pages) assessing the weekly readings. You choose which week’s readings you wish to analyze, but you must complete this assignment before the date the topic is discussed in class. The purpose of the longer memos is to help you grapple with the readings and respond with questions, criticisms, and new ideas. The main thrust of the longer memos should be on developing promising research ideas suggested by the readings. We strongly encourage you use the long memos as a vehicle for advancing your own work. When you choose to write a longer memo about a particular topic, you should consult the additional readings for that week. Longer memos are due by 9am on Tuesdays. A third of your course grade is based on the long memos.

  2. All students should arrive at class with questions, topics, and issues to be vetted and debated. You should come to class prepared to answer: What are the central research questions or problems raised by the authors? What core concepts, evidence, and research methods are utilized? As you do the readings, think about what the author did right as well as wrong. What are the interesting ideas in the paper? If you disagree with an argument, what would it require to persuade you? Can these differences be adjudicated through further empirical study? A good seminar should have active dialog and debate. If someone proposes an idea that is contrary to your view, speak up. I will often be intentionally provocative, so be prepared to push back. Your task is to engage one another in an assessment of the readings.

Note

This post was first written as a reading exercise for the course EDUC 306B, this was my second reading of the text.

Overview

In EDUC 306B: Global Education Policy & Organization we recently delved into the work of Peter Evans, specifically his 1995 most famous book, **.

Evans provides a critical framework for understanding why some states succeed at industrialization while others fail, focusing on the internal structure of the state and its relationship with society.

🏗️ What is “Embedded Autonomy”?

The central thesis of the book is that the state’s ability to transform its economy depends on a unique structural combination:

  • Autonomy (Corporate Coherence): A professional, meritocratic bureaucracy that can act independently of narrow interest groups.
  • Embeddedness (Social Ties): A dense network of connections to the private sector that allows the state to gather information and implement policy.

“The state must be sufficiently autonomous to avoid being captured, but sufficiently embedded to be effective.”

State Archetypes

Evans contrasts different types of states based on their developmental outcomes:

State Type Characteristics Examples
Developmental High Embedded Autonomy; fosters growth. South Korea
Intermediate Mixed capacity; inconsistent results. Brazil, India
Predatory Lacks autonomy; extracts wealth from society. Zaire (DRC)

🛠️ The Four Roles of the State

Evans identifies how states engage with the industrial sector, particularly in high-tech fields like informatics during the 1970s and 80s:

  1. Custodian: The regulator. Focuses on “policing” and setting legal frameworks.
  2. Demiurge: The producer. The state acts as a player by creating State-Owned Enterprises (SOEs).
  3. Midwifery: The promoter. Fostering the birth of new private sector firms through protection or incentives.
  4. Husbandry: The nurturer. Supporting existing firms to help them navigate global competition and R&D.

🌎 Case Studies: Brazil, India, and Korea

Evans uses an institutional comparative approach to analyze the IT and computer industries. He argues that:

  • South Korea succeeded through a combination of Midwifery and Husbandry, moving quickly into global markets.
  • Brazil and India initially focused on the Demiurge and Custodian roles, which provided mixed results in the high-tech sector.

The lesson for us in EDUC 306B is clear: industrial and educational outcomes are not just about “good policy”—they depend on the character of the state structures that implement them.

💡 Personal Reflection

Reading Evans helps bridge the gap between “politics” and “economics.” It suggests that for educational reforms to be successful, the state must not only have the will to change but also the institutional capacity and social connections to see that change through.

Chapter 3

References Evans, P. B. (1995). Embedded Autonomy: States and Industrial Transformation. Princeton, NJ: Princeton University Press.

Next Steps

Would you like me to help you draft a specific section comparing the “Husbandry” role in Brazil versus South Korea based on Evans’ analysis?