NewDataMetrics: Econometrics for New Data: Theory, Methods, and Applications
Lead Research Organisation:
University College London
Department Name: Economics
Abstract
Major advances in quantitative economic modeling over the past half-century have often come about through the development of new types of data, such as panels and networks. With increased digitalization and the proliferation of online platforms, there is now a vast array of new unstructured types of data, principally text, but also images, browsing histories, and purchase histories, to name a few, that exist alongside conventional quantitative economic data.
This project sets forth an innovative and wide-reaching agenda for harnessing new types of data to improve quantitative economic modeling. Contributions will be theoretical, methodological, and applications based.
Recent years have seen a wealth of applications of new unstructured types of data in quantitative economic modeling. But even state-of-the-art empirical methods are largely ad hoc and prone to several inference problems. This project brings much needed discipline to quantitative economic modeling with unstructured data. It will first set out a coherent framework for modeling unstructured and quantitative data together. The framework will be leveraged to develop novel, innovative, and theoretically sound econometric methods for quantitative modeling with unstructured data. These methods will not be prone to inference problems that have plagued the literature to date. The project therefore has great potential to enhance the credibility and reliability of empirical work. It will also push the frontier of econometric theory and practice. To rigorously justify the methods, the project will develop new statistical theory for combined quantitative and unstructured data. Further, it will develop a novel efficiency theory to guide best practice for quantitative economic modeling with unstructured data. The power of the methods will be illustrated with several important empirical applications to macroeconomic forecasting, policy analysis, and demand estimation, to name a few.
This project sets forth an innovative and wide-reaching agenda for harnessing new types of data to improve quantitative economic modeling. Contributions will be theoretical, methodological, and applications based.
Recent years have seen a wealth of applications of new unstructured types of data in quantitative economic modeling. But even state-of-the-art empirical methods are largely ad hoc and prone to several inference problems. This project brings much needed discipline to quantitative economic modeling with unstructured data. It will first set out a coherent framework for modeling unstructured and quantitative data together. The framework will be leveraged to develop novel, innovative, and theoretically sound econometric methods for quantitative modeling with unstructured data. These methods will not be prone to inference problems that have plagued the literature to date. The project therefore has great potential to enhance the credibility and reliability of empirical work. It will also push the frontier of econometric theory and practice. To rigorously justify the methods, the project will develop new statistical theory for combined quantitative and unstructured data. Further, it will develop a novel efficiency theory to guide best practice for quantitative economic modeling with unstructured data. The power of the methods will be illustrated with several important empirical applications to macroeconomic forecasting, policy analysis, and demand estimation, to name a few.
