Job Market Paper
The Adoption and Diffusion of New Technologies
Abstract: The diffusion of new general purpose technologies transforms innovation and growth. I study the drivers and consequences of technology diffusion using an endogenous growth model of firm-level innovation and technology adoption. The central mechanism in the model is that knowledge embodied in firms is more complementary to innovations by other firms using the same technology. Firms produce and innovate using either an old or new technology. Firms hire researchers to improve products or to adopt by learning to apply the new technology to existing products. Adoption of the new technology benefits firms by directly increasing productivity and reducing competition from other firms. Using patent citation links, I construct a new measure of information-communication technology (ICT) related patents that accounts for the application of ICT to new innovations. Empirically, I document that firm-level patent quality and productivity growth increase after firms start to issue ICT-related patents. Using this new measure, I calibrate the model to match empirical features of ICT diffusion. There are three key results. First, diffusion consists of an early phase where firms adopt the new technology directly and a late phase where previous adopters expand relative to non-adopters. Second, productivity growth declines by 10-15% following the introduction of the new technology, despite an increase in long-run growth. This is driven by low complementarities between embodied knowledge and innovations. Third, policies designed to aid diffusion can unintentionally misallocate resources, prolonging the period of low growth.
Idiosyncratic Distortions and Technology Adoption
Abstract: Recent work shows that resource misallocation has a substantial negative effect on aggregate productivity in developing countries. In this paper, I show that the same underlying institutions are also important for explaining cross-country technology differences. I study a model of heterogeneous firms that choose both labour and technology inputs. Firms differ in underlying productivity and an idiosyncratic distortion - modeled as a wedge on revenues. Distortions delay adoption by lowering the marginal benefits firms receive from adopting new technologies. At the aggregate level, distortions that tend to target, otherwise, highly productive firms delay the initial adoption of new technologies. In the calibrated model, distortions account for a large portion of the observed cross-country technology differences. Moving from the distortions of the bottom decile economy to the United States level increases productivity by 89% and explains just under half of the observed adoption lag. Furthermore, over half of the change is attributed to a dynamic channel in which firms adjust their technology.
Market Constraints, Misallocation, and Productivity in Vietnam Agriculture (with L. Brandt and D. Restuccia)
Abstract: We examine important changes in agriculture in Vietnam in the context of ongoing structural changes in the economy. We use a household-level panel dataset and a quantitative framework to document the extent and consequences of factor misallocation in agriculture during the period between 2006 and 2016. Despite rapid growth in agricultural productivity and a reallocation of factor inputs to more productive farmers, we find that misallocation across farmers remains high and increased during the period. Reallocation of factor inputs has not been strong enough to accommodate substantial changes in farm productivity over time. Our analysis also reveals important differences between the north and south regions.
Work in Progress
Trade and Diffusion of Embodied Technology (with F. Ibrahim, G. MacKenzie and S. Rachapalli)
Abstract: We examine international knowledge spillovers through the trade of technology embodied in products. We use patent data to construct measures of embodied technology and knowledge linkages across sectors. Empirically, we find that an increase in the trade of embodied knowledge increases spillovers to downstream innovative sectors. Next, we construct a tractable dynamic model of trade by multi-product firms. Firms add products to their portfolio either by inventing new products or innovating on existing products produced by other firms. Firms exit the economy when they either can no longer produce any frontier products or are unwilling to pay the associated fixed cost of operation. Opening to trade causes low productivity firms to exit the market, which reallocates resources to high productivity firms. Additionally, opening to trade increases knowledge spillovers, and this increases local innovation and causes firms to exit more frequently. Finally, we use the empirical measures of embodied knowledge to quantify the model. The quantitative model predicts that knowledge spillovers increase the aggregate gains from trade, and the dispersion of these gains across firms. This is driven by higher firm turnover relative to a trade model without knowledge spillovers.
Learning and Diffusion (with G. MacKenzie)
Abstract: How do best practices diffuse through the economy? We explore the role of employees learning on-the-job as a source of diffusion. We study a model in which researchers learn to manage new firms by imitating their current employers. New firms inherit the originating firm’s innovative type (i.e. best practices) and steal market share in the product market. Imitation contributes to aggregate growth by (1) improving the quality of innovations by the originating firm; and (2) improving the distribution of firm types. However, the possibility of losing market share to future imitators decreases incentives for firms to innovate, lowering growth. We calibrate the model to match empirical evidence on researcher mobility and firm types using US patent data. We use the quantitative model to assess ownership rights to incumbent firms on growth. Increasing protection lowers entry and worsens the distribution of firm types but increases investment by incumbent firms.