This paper has reviewed recent firm-level evidence on the drivers of upgrading in manufacturing firms in developing countries. From a measurement perspective, the literature faces a number of challenges. TFP measures have the conceptual advantage that they aim directly at capturing firm capabilities, but they have a number of well-known shortcomings. I have argued that focusing on directly observable measures of upgrading - technology use, quality ratings, expansions of product scope, and productivity measured under controlled conditions - is a particularly promising way forward. At the same time, such measures are often available only for particular sectors, and increases in these measures are not necessarily optimal either for firms or for the broader economies in which they are embedded. It seems clear that there is value in improving measurement both of indirect measures such as TFP and of more direct measures of upgrading.
Despite the difficulties of measurement, several empirical patterns emerge. Selling to developed country consumers, either directly or indirectly through value chains with richer-country end-consumers, appears to be robustly associated with upgrading. Increased availability of high-quality inputs also appears to promote upgrading. It is not clear that developing-country firms are making mistakes by not upgrading but there is growing evidence that tailored, intensive consulting interventions can improve firm performance. A broader conclusion is that developing-country firms appear to be constrained by a lack of know-how - both of internal capabilities and knowledge of products and techniques. A key challenge, perhaps the key challenge, in promoting upgrading is to promote learning by firms.
A number of important questions remain open. One is the strength of the link between the products that a firm specializes in and the rate of learning. Does producing higher-quality products, for instance, generate a greater accumulation of know-how? The hypothesis that there is a link between the pattern of specialization and upgrading was central to the thinking of an early generation of development economists (e.g. Prebisch (1950)). In recent years, it has been advanced by Dani Rodrik, Ricardo Hausman and others (see e.g. Hausmann et al. (2007), Hausmann et al. (2014)) and investigated largely at the sectoral level. Now that firm-product-level datasets are increasingly available, the time seems ripe for investigating the link at the firm level.
Another important question is to what extent behavioral biases of entrepreneurs lead them not to maximize profits. I have argued that what appears to be non-optimizing behavior by firms can often be explained by firms’ lack of know-how, organizational dynamics, or other constraints, and that we need to think carefully about these possibilities before concluding that individuals are failing to optimize. But firm owners and managers are human, and they may procrastinate, put more weight on losses than gains, ignore evidence that does not comport with their priors, and display all the other foibles that other humans do. There is a need for research designs that can measure these propensities separately from other factors.
Also important is to what extent firm capabilities can be acquired on markets or must be homegrown. In principle, one would expect firms to be able to hire consultants to acquire the know-how needed to upgrade. One puzzling fact, worthy of further investigation, is that in many developingcountry settings the consulting market is either extremely thin or non-existent. But even where consulting markets exist, it requires time and effort for firms to incorporate new knowledge or practices into the everyday functioning of an organization. A related question is to what extent firms can improve their performance by hiring highly skilled managers, even if their homegrown capabilities are low. In many developing countries, the supply of highly skilled managers is limited. But it also appears that top managers cannot just parachute in and impose new practices; the capabilities to implement practices effectively must be developed internally as well. More research on these issues is much needed.
Finally, it is natural to ask about the implications of the recent literature on upgrading for industrial policy. This review has focused on the determinants of upgrading behavior by firms, with the idea that such an understanding will eventually be useful in policy design. But policymakers must face a number of additional constraints not considered here, among them pressures from different interest groups and the limited knowledge of government officials. More research is needed on what works and what does not work in industrial policy, especially given the limited capacity of many developingcountry governments. If policies are to be implemented at scale, designers will also need to confront the general-equilibrium effects of industrial-policy interventions, which have not been the focus here. Analyzing these issues will likely require more guidance from economic theory than the primarily reduced-form studies discussed in this review have relied on.
Although much work obviously remains to be done, there are many reasons for optimism about the prospects for future research on firm-level upgrading in developing countries. The data frontier has been expanding quickly, with information on customs transactions, firm-to-firm trade, quantities and prices at the product level, banking relationships, and other sorts of contracts becoming increasingly available. Appreciation is growing in a number of fields - macroeconomics, industrial organization, and international trade, as well as development - for careful firm-level empirical work on the determinants of innovative behavior. And policymakers in many countries are hungry for rigorous, evidence-based advice about how to promote upgrading. It is an exciting time for the field.
