Shock Propagations in Granular Networks

Abstract

This paper studies a number of features of transaction networks, firm sales growth, and buyer-supplier comovements of sales using a large scale dataset on the Japanese interfirm transaction network. Larger firms have higher sales growth rates and smaller growth dispersion. Well-connected firms also exhibit higher growth rates, but there is no systematic relationship between the number of partners (degree) and sales growth dispersion. Using a statistical test for spatial interdependence, it is confirmed that there exists a significant network interdependence of sales growth. By employing spatial autoregressive models, various propagation factors are estimated. In the baseline specification, the elasticity of average sales growth of suppliers is estimated to be 0.153 while that of customers is 0.257 for 2012. In all years, the upstream propagation factor is larger than the downstream factor, implying difficulty in replacing an existing customer or adjusting to a demand shock. The manufacturing sector is characterized by a large degree of propagation. For both downstream and upstream propagations, manufacturing and wholesale sectors exhibit higher propagations factors while retail and service sectors exhibit lower ones. The interdependence of intermediate physical inputs produced by other firms may generate an additional margin for the buyer-supplier comovements. It was also found that larger firms have higher propagation factors. Larger firms have more partners, and their degree of propagation is also higher. This result stresses an even larger impact of big firms for aggregate fluctuations in a granular production network.