Many-to-one contagion of economic growth rate across trade credit network of firms

TL;DR

This study empirically tests the hypothesis of 'many-to-one' contagion of growth rates in a trade credit network of firms, finding limited evidence of contagion but noting a strong macroeconomic impact during the crisis.

Contribution

It introduces a novel empirical approach to measure direct contagion effects in trade credit networks and applies it to real data from Italian firms during 2007.

Findings

Limited contagion observed in the network

Strong macroeconomic effects during the crisis

Correlation between predicted and actual growth is weak

Abstract

We propose a novel approach and an empirical procedure to test direct contagion of growth rate in a trade credit network of firms. Our hypotheses are that the use of trade credit contributes to contagion (from many customers to a single supplier - "many to one" contagion) and amplification (through their interaction with the macrocopic variables, such as interest rate) of growth rate. In this paper we test the contagion hypothesis, measuring empirically the mesoscopic "many-to-one" effect. The effect of amplification has been dealt with in another paper. Our empirical analysis is based on the delayed payments between trading partners across many different industrial sectors, intermediated by a large Italian bank during the year 2007. The data is used to create a weighted and directed trade credit network. Assuming that the linkages are static, we look at the dynamics of the nodes/firms. On the ratio of the 2007 trade credit in Sales and Purchases items on the profit and loss statements, we estimate the trade credit in 2006 and 2008. Applying the "many to one" approach we compare such predicted growth of trade (demand) aggregated per supplier, and compare it with the real growth of Sales of the supplier. We analyze the correlation of these two growth rates over two yearly periods, 2007/2006 and 2008/2007, and in this way we test our contagion hypotheses. We could not find strong correlations between the predicted and the actual growth rates. We provide an evidence of contagion only in restricted sub-groups of our network, and not in the whole network. We do find a strong macroscopic effect of the crisis, indicated by a coincident negative drift in the growth of sales of nearly all the firms in our sample.