The Growth Lab Research Seminar series is a weekly seminar that brings together researchers from across the academic spectrum who share an interest in growth and development.
Speaker: François Lafond
Abstract: Firm-level datasets on production networks are often confidential and arise from different data collection methods, making it difficult to determine stylized facts. Are standard network properties similar across all available datasets, and if not, why? We provide benchmark results from two administrative data sets (Ecuador and Hungary) which are exceptional in that there are no reporting thresholds. We compare these networks to a leading commercial data set (FactSet), and a systematic synthesis of published results. The administrative data sets with no reporting thresholds have remarkably similar properties, but differ substantially from non-administrative sources, or from administrative datasets with higher reporting thresholds. Our results provide a roadmap for a distributed micro-data project, offer insights into the direction of biases on key metrics when using partial datasets, and have wide implications for reconstructing the global firm-level production network and for modelling heterogeneity in macro. (Joint work with Andrea Bacilieri, András Borsos, and Pablo Astudillo-Estevez).
The seminar will be on Zoom. The link to the registration page is: https://harvard.zoom.us/j/640145716?pwd=bTUxOXRad0o2ZTJtbWRxTGpta1BlQT09
About the Speaker: François Lafond is deputy director of the Complexity Economics group at the Institute for New Economic Thinking, University of Oxford, and an associate member of Nuffield college, Oxford. His main areas of research are in the economics of innovation and productivity, environmental economics, networks and complex systems, applied econometrics and forecasting.