Revisiting asset co-movement: Does network topology really matter?

Asset co-movement has garnered increasing attention from researchers in both traditional finance and emerging interdisciplinary disciplines. However, there is limited knowledge re-garding the consistency of market-wide co-movement proxies constructed based on these two perspectives. Employing Fama-French industry portfolios as samples, we construct market -wide co-movement proxies in terms of R2-based and network-based approaches. We further examine if and how market-wide co-movement is priced using supervised principal analysis (SPCA) proposed by Giglio et al. (2021). Our findings include: First, most topological properties are highly correlated with the R2-based proxies. Second, the risk-premium estimates by the SPCA range from 90 bps to 130 bps per month, depending on the size of the latent factors considered. Furthermore, most co-movement proxies are linked to the characteristics regarding asset fundamentals. However, the associated R2 values remain around only 15%, highlighting the challenges in hedging the risks associated with market-wide co-movement in equity markets.

Automated summary

Asset co-movement has gained increasing attention from researchers in traditional finance and interdisciplinary fields. However, there is limited knowledge regarding the consistency of market-wide co-movement proxies constructed based on these two perspectives. This study constructs market-wide co-movement proxies using R2-based and network-based approaches and examines their pricing using supervised principal analysis. The findings show that most topological properties are highly correlated with the R2-based proxies, and the risk-premium estimates range from 90 bps to 130 bps per month. Furthermore, the co-movement proxies are linked to asset fundamental characteristics, but the associated R2 values remain low, highlighting the challenges in hedging the risks associated with market-wide co-movement in equity markets.