Varieties of deindustrialization and patterns of diversification: why microchips are not potato chips

https://doi.org/10.1016/j.strueco.2021.01.009Get rights and content

Highlights

  • There is no evidence supporting invariant tendencies in deindustrialisation.

  • Sectoral technological opportunities and learning regimes matter.

  • A diversified employment structure fosters catching-up.

  • Globalization accelerates deindustrialization and hinders development opportunities for most.

Abstract

Contrary to the notion of a natural tendency in deindustrialization, this paper, performing a cross-country long-term analysis, documents the existence of a variety of patterns of deindustrialization. Looking at industrial sectors and their technological characteristics, categorised on the ground of the Pavitt (1984) taxonomy, we do find a markedly uneven process of deindustrialization with Science Based and Specialised Suppliers not presenting any inverted U-shaped pattern, neither in employment nor in value added. The heterogeneity holds both for the four Pavitt aggregates and under further disaggregation at industry level. We then study whether the uneven sectoral composition had an impact on the timing of deindustrialization. Overall, our analysis supports the notion that “microchips” are not “potato chips” in their influence on the patterns of long-term economic development of different countries. Moreover, during the phase of globalization the probability for low-income countries to be stuck to produce “potato chips” has increased and that of transition toward the production of “microchips” has been reducing.

Introduction

A process of generalised deindustrialization represents a widespread feature of the current phase of capitalist development. From the peak of 40% share of the overall workforce employed in the manufacturing sector in the sixties, nowadays the overall manufacturing employment share in advanced economies ranges between 10%–25%.

Deindustrialization beyond a certain income threshold is often thought of a “natural tendency” of capitalism. In such a perspective, deindustrialization should occur evenly across countries and industrial sectors. However, the historical evidence is at odds with this view. Countries deeply differ both in their process of industrialization and eventually deindustrialization, and with that, in the consequences that deindustrialization bears thereafter upon the pattern and “quality” of development. Consider two “archetypical” alternative examples like Brazil and South Korea. Both countries experienced a rapid process of catching-up with the former starting indeed in the fifties from much higher levels of GDP per capita. However their development path followed completely different trajectories, with Brazil falling behind since the eighties and South Korea nowadays included among the most prosperous economies. And now they differ even in the response to major exogenous shocks. South Korea has recently demonstrated a superb capability in managing the spreading of the COVID-19 pandemic. The pandemic is teaching us that, other things being equal, more industrialized countries are better equipped to manage it.

Contrary to the belief on any “natural tendency”, this paper, examining both deindustrialization and premature deindustrialization for developing countries, performs a cross-country, long-term analysis, documenting the existence of a variety of patterns. Looking at industrial sectors and their technological characteristics, categorised on the ground of the Pavitt (1984) taxonomy, we do find a markedly uneven process of deindustrialization with Science Based and Specialised Suppliers sectors not presenting any clear inverted U-shaped pattern, neither in employment nor in value added. The heterogeneity holds both for the four aggregates of the Pavitt taxonomy and under further disaggregation at industry level.

We then study whether the uneven sectoral composition might have exerted an impact on the timing of deindustrialization. We find, first, that Scale Intensive and Specialised Suppliers industries have reduced their employment shares more dramatically than Supplier Dominated ones. Countries stuck into the latter seem to have missed major opportunities of catching-up. Second, after performing a cluster analysis, we do find that the higher the degree of diversification in industrial composition the higher the level of GDP per capita. In this respect, four different clusters in terms of composition of manufacturing employment shares do emerge, distinct in terms of their degrees of diversification. Finally, we do find evidence that the post-1990 time period is characterised by increasing probabilities of falling behind even for industrialized rich countries. That is, globalization has contributed in general to freeze technological upgrading and opportunities of catching-up from traditional toward innovative manufacturing sectors. With China, of course, standing out as a major exception.

Overall, our analysis brings support to the notion that “microchips” are not equivalent to “potato chips1”: the industrial composition of manufacturing highly influences the patterns of long-term economic development of the countries. During the phase of globalization the probability for low-income countries to produce “potato chips” has increased while the transition probability toward the production of “microchips” has been reducing.

The paper is organised as follows: Section 2 discusses the main theoretical background, Section 3 presents the data and performs the analysis of deindustrialization patterns disaggregating by the Pavitt taxonomy. Section 4 analyses the possible processes of premature deindustrialization by technological classes while in Section 5 we perform a cluster analysis to detect the underlying patterns of country diversification. Finally, our conclusions and policy implications are sketched in Section 6.

Section snippets

Deindustrialization, sectors and quality of specialization

The very first question one needs to address is whether it is still relevant to discuss about the role of the manufacturing sector for the process of economic development. Back to the structuralist perspective but even earlier to List (1841), what a country produces, does matter. According to the “Hirschman-Prebisch” approach, the manufacturing sector is the engine of growth for two specific reasons namely, first, the rate of productivity growth occurring in the manufacturing sector is

Varieties of deindustrialization: sectoral and technological heterogeneity

In this section, we first present the data set and then document the patterns of sectoral and technological heterogeneity in industrial structure and deindustrialization. We consider both employment and nominal valued added shares.2

Timing heterogeneity and premature deindustrialization

What about heterogeneity in the timing of deindustrialization? Let us consider it by disaggregating into the four Pavitt classes and focusing on the sectoral contribution of each class to the aggregate premature deindustrialization process, splitting the analysis between pre- and post-1990 periods.

Fig. 8 offers an overview of the time patterns of deindustrialization using year dummies.10 All

Patterns of diversification

The process of economic development historically can be seen as an upgrading path over technological capabilities and learning regimes. If the textile sector has typically represented the opportunity to discipline and organise a previously unstructured labour force, steel and heavy metal industries often have been the chance to accumulate massive productive capacity and capital equipment. Finally, information and communication technologies, fine chemistry and drugs entailed a tighter connection

Discussion and conclusions

The process of development is highly uneven. Cross-country evidence highlights a variety of industrialization/deindustrialization patterns. The institutional shock associated with globalization discussed by Rodrik (2016) has yielded an average acceleration in the timing and pace of deindustrialization. However, such average tendency hides ample heterogeneity driven by underlying diversities in technological capabilities and learning regimes. We try to account for them employing the Pavitt

Credit Author Statement

All authors have equally contributed to the paper.

References (50)

  • K.J. Arrow

    The economic implications of learning by doing

    Rev. Econ. Stud.

    (1962)
  • D. Baker et al.

    Employment Multipliers in the US Economy

    Technical Report

    (1993)
  • W. Baumol

    Macroeconomics of unbalanced growth: the anatomy of urban crisis

    Am. Econ. Rev.

    (1967)
  • G. Berlingieri et al.

    The Great Divergence(s)

    Technical Report

    (2017)
  • J. Bivens

    Updated Employment Multipliers for the US Economy

    Technical Report

    (2003)
  • H. Chenery et al.

    Patterns of Development 1950-1970

    (1975)
  • H. Chenery et al.

    Patterns of Development 1950-1983

    World Bank Discussion Papers, No: 41, Washington OC

    (1989)
  • H. Chenery et al.

    Development patterns among countries and over time

    Rev. Econ. Stat.

    (1968)
  • M. Cimoli et al.

    Technological paradigms, patterns of learning and development: an introductory roadmap

    J. Evol. Econ.

    (1995)
  • M. Cimoli et al.

    Intellectual Property Rights: Legal and Economic Challenges for Development

    (2014)
  • M. Cimoli et al.

    Industrial Policy and Development: The Political Economy of Capabilities Accumulation

    (2009)
  • C. Clark

    The Conditions of Economic Progress

    (1957)
  • G. Dosi et al.

    The Economics of Technical Change and International trade

    (1990)
  • G. Dosi et al.

    The Role of Comparative Advantage and Endowments in Structural Transformation

    LEM Working Paper Series, n. 33/2018

    (2018)
  • G. Dosi et al.

    Whither the evolution of the contemporary social fabric? New technologies and old socio-economic trends

    Int. Labour Rev.

    (2019)
  • Cited by (19)

    • The North-South divide: Sources of divergence, policies for convergence

      2023, Journal of Policy Modeling
      Citation Excerpt :

      Alternative theoretical perspectives have addressed the problem of geographical divergence. The structuralist approach highlights the importance of asymmetric productive compositions, patterns of technological learning, and specialisation between advanced and developing economies (Cimoli, 1988; Cimoli & Dosi, 1995; Cimoli & Porcile, 2014; Dosi et al., 2009; Dosi, Riccio et al., 2021). The core-periphery approach studied asymmetries among member states within the European Monetary Union (EMU) (Celi et al., 2019; Landesmann et al., 2015; Storm & Naastepad, 2015), and uneven development along international value chains in general (Pavlínek, 2018), drawing from the dependency theory (Prebisch, 1950).

    • Premature deindustrialisation and growth slowdowns in middle-income countries

      2022, Structural Change and Economic Dynamics
      Citation Excerpt :

      Policies that focus on productivity enhancement in most competitive sectors can foster economic growth (Lin, 2012). In this regard, a crucial directive for countries is to achieve a higher degree of diversification in industrial composition, as it is associated with higher economic growth (Dosi et al., 2021, Foster-McGregor et al., 2015). Similarly, developing countries can take advantage of international production fragmentation by specialising in those stages where they have a comparative advantage (Baldwin, 2018).

    • Quantity or quality: who makes manufacturing stronger?

      2024, Journal of the Asia Pacific Economy
    View all citing articles on Scopus

    The authors wish to thank two anonymous reviewers, participants to the 31st EAEPE, Warsaw (2019), to the 8th WIPE, Reus (2020) for their useful comments and suggestions. The authors acknowledge support from European Union’s Horizon 2020 research and innovation programme under grant agreement No. 822781 GROWINPRO – Growth Welfare Innovation Productivity.

    View full text