The Long Earthquake Swarm
The human cost of the earthquakes that hit the Apennines in August and October is now known. However, it will take more time to assess the overall economic damage of the earthquake on the Italian territory. While the Italian Civil Protection Department has released first estimates for the cost of the destructions in the affected areas, the economic impact of the earthquake is unfortunately unlikely to be limited to the areas surrounding its epicenter. In particular, a propagation channel to other Italian regions rests on the fact that firms often interact through business-to-business relationships, making them sensitive to disruptions hitting their suppliers or customers.
In recent work using data on the occurrence of major natural disasters over the last 30 years in the U.S. (including the 1989 Loma Prieta earthquake in northern California, but also several hurricanes and floods), I and Jean-Noel Barrot (MIT Sloan School of Management) have shed light on how disruptions to individual firms spread to other firms in production networks (Input Specificity and the Propagation of Idiosyncratic Shocks in Production Networks, in Quarterly Journal of Economics, doi: 10.1093/qje/qjw018).
At the firm-level, we might think that shocks should be easily absorbed in production networks. For instance, it is plausible that firms organize their operations to avoid being affected by shocks to their supplies. Even when they face such disruptions, firms are supposedly flexible enough to recompose their production mix, or to switch to other suppliers. The development of online business should make it even easier for firms to adjust their sourcing.
Our research shows instead that when a supplier is hit by a natural disaster, there are important disruptive effects along the supply chain. Specifically, we find that the sales growth of supplier firms directly hit by a natural disaster drops by around five percentage points over two or three quarters. Importantly, the customers of these suppliers are also significantly affected, as their sales growth drops on average by two percentage points when one of their suppliers is hit by a natural disaster. This is a strikingly large effect, which is not immediate, but rather takes a few quarters to materialize. Interestingly, the effect is concentrated among customers with lower inventories, and for natural disasters that hit specific suppliers, meaning firms producing differentiated goods, generating high R&D expenses, or holding patents.
Finally, we look at whether the shock originating from one supplier propagate further in production networks, to other suppliers of the same customer, which were not directly affected by the natural disaster. You might expect that firms would continue to buy from other suppliers outside of the natural disaster zone, or that the other suppliers would find alternative buyers. However, our research shows large negative spillovers of the initial shock to other suppliers as well.
Surprisingly, we found that other suppliers of a main customer see a drop in sales growth by roughly three percentage points.
Overall, our findings highlight the presence of strong interdependencies in production networks, and are likely to apply to other contexts as well, such as for instance strikes and financial crises. As for the economic impact of the earthquake this summer, we might expect that the economic effect will not be limited to the disaster zone, but will be propagated to other parts of the economy through the relationships that firms in the earthquake zone have with customers in the rest of the country.