Estimating Large-Dimensional Connectedness Tables: The Great Moderation Through the Lens of Sectoral Spillovers

Sectoral crises (for example, the 1973 oil crisis) often spill over to related sectors. These spillovers expose links in a network of sectors. Understanding the size of sectoral links is crucial to predicting the impact of a sectoral crisis on the whole economy.

We investigate a class of techniques to estimate a network of spillover links from sectoral time-series data. Networks of sectoral links typically have many parameters but relatively few observations. We therefore apply statistical learning techniques that avoid estimating links between sectors that are truly independent to focus on the links that are most important. We first compare the performance of these techniques in a simulation study. We then apply them to study how links across industries changed before and with the onset of the Great Moderation in the mid-1980s.

Our results illustrate the importance of employing statistical learning techniques when measuring large networks of links. These approaches substantially outperform traditional estimation techniques. Empirically, we find that the importance of some sectors decreased drastically with the onset of the Great Moderation. This suggests that this period had a structural aspect and may not have been driven entirely by policy interventions.