At least since Gerschenkron (1962), the “advantages of backwardness” - above all, the accumulation of advanced technologies and products in industrialized countries that developing-country firms can then adopt - have been well appreciated. Since Gerschenkron’s landmark study, a number of developing countries, disproportionately in East Asia, have been able to industrialize, and to do so more quickly than earlier industrializers. But for many other countries, the purported advantages of backwardness have remained elusive. Something seems to be getting in the way of the adoption of advanced technologies and products, a process often referred to as industrial upgrading. What are these barriers? Since to identify a barrier is implicitly to identify a factor that promotes upgrading (if only by removing or mitigating the barrier), the question can be restated in a positive way: What are the drivers of industrial upgrading in developing countries?
This paper reviews recent empirical research on firms that sheds light on this question. I focus primarily on studies of larger firms (with more than a handful of employees) in manufacturing. This choice reflects a number of judgments: that such firms, although they make up a small share of the total firm population in most countries (Tybout, 2000; Hsieh and Olken, 2014), are crucial for growth; that the issues facing them are distinct from those facing very small firms, agricultural producers, and service-sector firms; and that the literatures on small firms (including entrepreneurship) and agricultural producers have been well covered by other recent reviews.[1] To keep the review to a manageable size, I concentrate primarily (but not exclusively) on studies employing quasi-experimental and experimental methods to isolate causal relationships of interest. I also focus on studies that consider upgrading (in one of the senses discussed below) as an outcome.
The first part of the review (Section 2) discusses what is meant by the term upgrading, conceptually and empirically. The term encompasses innovation as commonly defined, but also reflects the fact that innovative activities among developing-country firms are often oriented toward catching up to the world frontier, rather than pushing it forward. I set out a simple organizing framework, which helps to clarify the four related but distinct ways in which the term upgrading has typically been used: learning, the accumulation of knowledge about products or techniques or about how to implement those techniques; quality upgrading, an increase in the weighted-average quality of goods produced by a firm; technology adoption, the adoption of a technique not previously used by a firm; and product innovation, the enlargement of the set of varieties produced by a firm. With these conceptual definitions in hand, I then review the various ways that researchers have sought to measure upgrading. As we will see, the mapping between the conceptual definitions and the empirical measures is less than perfect and existing measures have different strengths and weaknesses.
The second part of the review (Section 3) considers the drivers of upgrading. I classify them into three categories: drivers on the output side, including consumer preferences and the degree of competition in export and domestic markets (Section 3.1); drivers on the input side, including conditions in credit, labor, and intermediate-input markets (Section 3.2); and drivers of firm capabilities, including mechanisms that affect the entrepreneurial ability or knowledge possessed by firms (Section 3.3). The categorization is necessarily somewhat loose, because some mechanisms span more than one category.
A number of themes emerge from the review. First, a methodological point: there is great benefit to using directly observable information on upgrading outcomes - technology use (including management practices), quality ratings, product scope, and productivity under controlled conditions. These measures are often available only for specific sectors, and questions naturally arise about external validity, but the approach of building up from direct observation of particular sectors seems particularly promising. Second, there is accumulating evidence that the demand side matters: selling to richer buyers, or supplying inputs in value chains destined for richer buyers, seems to be robustly associated with upgrading. Third, evidence has also accumulated of causal links between the cost, quality, and variety of inputs and upgrading outcomes. Increased access to imported inputs, for instance, appears to stimulate upgrading. Fourth, it is clear that developing-country firms are often constrained by a lack of know-how. Several types of informational interventions have been successful in improving firm performance. At the same time, organizational dynamics are complex and learning is costly, and a lack of upgrading should not simply be attributed to a failure of individuals to optimize. A number of other insights will be highlighted as we proceed.
This review is related to a number of existing reviews, beyond those cited above. In its focus on firms in developing countries, it is similar in spirit to an older review by Tybout (2000), but with different topical emphases. Several reviews from the perspective of international trade have covered work in developing countries, including Tybout (2003), De Loecker and Goldberg (2014), Goldberg and Pavcnik (2016), Shu and Steinwender (2019), and Atkin and Khandelwal (2019); the current review is broader in considering drivers of upgrading unrelated to trade, but also narrower in focusing on firmlevel empirical work on upgrading outcomes using quasi-experimental and experimental strategies. Also related are the handbook chapter of Harrison and Rodríguez-Clare (2010) on the theory and practice of industrial policy in developing countries, and recent policy-oriented overviews by Crespi et al. (2014), Cirera and Maloney (2017), and Cusolito and Maloney (2018).[2] The current review is focused on evaluating what we know about how firms behave, which is relevant to policy design, but not specifically on the practical issues of what works or does not work in industrial policy.
[1] See McKenzie and Woodruff (2013), Woodruff (2018), Quinn and Woodruff (forthcoming), Foster and Rosenzweig (2010), Jack (2013), de Janvry et al. (2017), and Magruder (2018).
[2] See also Lane (forthcoming).
