Technology Diffusion and Firm Networks
Paper Session
Monday, Jan. 5, 2026 1:00 PM - 3:00 PM (EST)
- Chair: Timothy DeStefano, Georgetown University
Technology Spillovers, Diffusion and Rivalry in Firm Networks
Abstract
We show how the benefits of new technology adoption diffuse along a supply chain, improving the performance of firms that do not adopt new technology themselves, and also increasing their likelihood of adopting. We measure new technology adoption in terms of robotization, and firm performance in terms of TFP, wages, employment and adoption probability, with firm level VAT data matched with custom and employer-employee data from Turkey. We find sizable positive causal impacts of upstream robotization on downstream productivity, even beyond direct input-output linkages. We do not find similar impact for upstream capital deepening and for firms without direct, or indirect, input-output linkages with robot adopters in the supply chain. Spillovers are instead negative if the connected firms share the same customer base, due to rivalry. We finally investigate the heterogeneity of the impacts with respect to robots' quality, as proxied by their origin country and import price. Higher quality robotization upstream has larger positive impact downstream. Taken together, these results suggest that new technology adoption via robotization improves the quality of inputs along the supply chain with productivity enhancing effects, that build up from the firm to the aggregate level.Barriers to Knowledge Creation in the Multinational Firm
Abstract
This paper shows that knowledge creation, as measured by patents, is increasingly conducted in cross-border collaborative teams of inventors. It documents the importance of cross-border communication costs by showing that a higher overlap in business hours is associated with increased cross-border collaboration. This effect is distinct from the effect of physical distance, which matters as well. It is stronger for technology classes where lab experiments are involved and thus more frequent interactions may be required. The data further suggest that episodes of telecommunications liberalization (and the resulting decline in the cost of international calls) lead to an increase in cross-border collaboration, particularly when the business hour overlap between the headquarters and a subsidiary is larger. This effect is stronger for experiment-based technology classes. Less successful inventors respond more than their most successful peers.AI Diffusion Within Multinationals
Abstract
Artificial intelligence (AI) is disproportionately adopted by multinationals (MNEs), however, the mechanisms behind technology diffusion within MNEs are largely unknown. This paper examines the role of communication costs in explaining AI diffusion from the MNE headquarters (HQ) to their affiliates. To do so it uses a novel dataset capturing establishment-level technology adoption and ownership linkages for MNEs across 30 countries from 2019 to 2022. It employs both geographic proxies of communication costs between the HQ and each affiliate (distance or time zones) and technological proxies (using similar pre-existing production technologies, digital technologies or having the same technology provider). HQ AI adoption is instrumented with proximity to local AI university research. It finds that affiliates are more likely to adopt AI after the HQ does so, especially those nearby or in similar time zones. However, technological proxies matter more than geography, such as employing complementary technologies, particularly cloud computing, and having the same cloud computing vendor.Discussant(s)
Katherine Stapleton
,
World Bank
Ester Faia
,
Goethe University Frankfurt
Stefan Pauly
,
Sciences Po
Jonathan Timmis
,
World Bank
JEL Classifications
- O3 - Innovation; Research and Development; Technological Change; Intellectual Property Rights
- F2 - International Factor Movements and International Business