Part 5 of 7
Input-Output Networks and Economic Activity
Domar aggregation, Hulten's theorem, propagation, and distortions in networks.
The previous sections studied aggregation and misallocation in economies with heterogeneous firms but without explicit intermediate-input linkages. The next step is to allow sectors to use each other as inputs. Once we do that, a productivity shock in one part of the economy affects aggregate activity not only because that sector becomes more productive, but also because all of its customers, and the customers of its customers, face different costs. This is the key reason why input-output networks matter both for propagation and for accounting for economic activity.
This perspective goes back to Leontief (1936) and to the growth-accounting tradition of Domar (1961); Hulten (1978). In modern macroeconomics, the same ideas reappear in multisector business cycle theory à la Long and Plosser (1983), in the network approach of Acemoglu et al. (2012), and in the recent work surveyed by Carvalho and Tahbaz-Salehi (2019); Baqaee and Rubbo (2023) and developed further by Baqaee and Farhi (2019); Baqaee and Farhi (2020); Baqaee and Burstein (2026); Baqaee and Burstein (2026). These notes follow especially closely the exposition in these papers, while keeping notation close to the rest of these notes.
I cannot recommend enough Emmanuel Farhi's Plennary Talk at the 2019 SED. Watch it! You can find it on Youtube: https://youtu.be/92_-kIyVC1Y?si=SYHCuURa3Bkn2CZ_. You can also watch Emmanuel's presentation in VMACS of his paper with David Baqaee on the effects of COVID 19: https://www.youtube.com/live/Yek0x91ZE-0?si=SSySMysFlWNLhmGz.