256
G. Marti et al.
10.5.2 Based on Supply Chain Data
• Reference [153] explores “the interactions from the perspective of an economywide supply-chain network, and propose a list of network centrality measures
to capture the relative importance of each company within this network.” Using
FactSet Supply Chain Relationships database, the author estimates dynamic supplier networks and compute centrality measures for each company. From there,
the author constructs a supplier central portfolio of stocks based on the top ten
companies with the highest centrality in the network. The author finds that the
stock performance of supplier central portfolios tends to predict the movements
of the overall stock market.
• In [24], authors study the network structure implied by firms’ cash flows on the
cross-section of expected returns. They find that firms which are more central have
lower P/D ratios and higher expected returns.
• Reference [1] find that there is significant credit risk propagation along supply
chains by analyzing credit default swap spreads.
10.5.3 Based on Transaction Data
• In [83], authors study a large proprietary dataset which consists of a network built
from transactional data of the payment platform of a major European bank. These
transactions link around 2.4 million Italian firms whose credit risk rating is known
for a large fraction of them. Authors find that this network, like many financial
networks, is sparse but made of a single component, scale free and verifies the
small world property. The main contribution of the authors is to document significant correlations between local topological properties of a node (firm) and its risk.
They also employ machine learning techniques to build classifiers for predicting
the risk rating using as inputs network properties (e.g. degree, community) alone,
i.e. no balance sheet information or any other features. Their classification method
outperforms significantly its benchmark, i.e. random assignments. Another contribution is to show the existence of an homophily of risk, i.e. the tendency of
firms with similar risk profile to be statistically more connected among themselves through payments. Risk is therefore not spread uniformly on the network,
but rather concentrated in specific areas. This implies that an idiosyncratic shock
on a single firm can propagate more or less quickly depending on the local network
structure and the community the node belongs to.
G. Marti et al.
10.5.2 Based on Supply Chain Data
• Reference [153] explores “the interactions from the perspective of an economywide supply-chain network, and propose a list of network centrality measures
to capture the relative importance of each company within this network.” Using
FactSet Supply Chain Relationships database, the author estimates dynamic supplier networks and compute centrality measures for each company. From there,
the author constructs a supplier central portfolio of stocks based on the top ten
companies with the highest centrality in the network. The author finds that the
stock performance of supplier central portfolios tends to predict the movements
of the overall stock market.
• In [24], authors study the network structure implied by firms’ cash flows on the
cross-section of expected returns. They find that firms which are more central have
lower P/D ratios and higher expected returns.
• Reference [1] find that there is significant credit risk propagation along supply
chains by analyzing credit default swap spreads.
10.5.3 Based on Transaction Data
• In [83], authors study a large proprietary dataset which consists of a network built
from transactional data of the payment platform of a major European bank. These
transactions link around 2.4 million Italian firms whose credit risk rating is known
for a large fraction of them. Authors find that this network, like many financial
networks, is sparse but made of a single component, scale free and verifies the
small world property. The main contribution of the authors is to document significant correlations between local topological properties of a node (firm) and its risk.
They also employ machine learning techniques to build classifiers for predicting
the risk rating using as inputs network properties (e.g. degree, community) alone,
i.e. no balance sheet information or any other features. Their classification method
outperforms significantly its benchmark, i.e. random assignments. Another contribution is to show the existence of an homophily of risk, i.e. the tendency of
firms with similar risk profile to be statistically more connected among themselves through payments. Risk is therefore not spread uniformly on the network,
but rather concentrated in specific areas. This implies that an idiosyncratic shock
on a single firm can propagate more or less quickly depending on the local network
structure and the community the node belongs to.