Seeing the Future: Improving Macroeconomic Forecasts with Spatial Data and Recurrent Convolutional Neural Networks

Jonathan Leslie ()
Additional contact information
Jonathan Leslie: Indiana University, Department of Economics

CAEPR Working Papers from Center for Applied Economics and Policy Research, Department of Economics, Indiana University Bloomington

Abstract: This paper presents a method of leveraging the information content of high dimensional, spatially-distributed economic data through borrowing techniques common in visual recognition artificial intelligence. Specifically, I cast spatially-disaggregated U.S. economic data as a sequence of quarterly geographic images in a deep learning computer vision setting to evaluate whether leveraging the spatio-temporal distribution of predictors can improve macroeconomic forecasts. This spatial forecasting model produces highly-accurate out-of-sample forecasts of GDP (0.136 percentage point average mean absolute error (MAE)), inflation (0.066 percentage point average MAE), and industrial production (0.368 percentage point average MAE) across a four-quarter horizon. The model substantially outperforms both more traditional linear methods as well as deep learning methods that do not leverage the spatial distribution of the data.

Keywords: Macroeconomic Forecasting; Machine Learning; Deep Learning; Computer Vision; Economic Geography (search for similar items in EconPapers)
Pages: 26 pages
Date: 2023-03
New Economics Papers: this item is included in nep-big, nep-cmp, nep-for, nep-geo and nep-ure
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://caepr.indiana.edu/RePEc/inu/caeprp/caepr2023-003.pdf (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:inu:caeprp:2023003

Access Statistics for this paper

More papers in CAEPR Working Papers from Center for Applied Economics and Policy Research, Department of Economics, Indiana University Bloomington Contact information at EDIRC.
Bibliographic data for series maintained by Center for Applied Economics and Policy Research ().