Working Paper Series no. 935. This paper provides empirical evidence on the compositional effect of weather-related disasters on consumer prices. We combine data on monthly granular inflation for 12 CPI product categories with data on extreme weather events for four French overseas territories sporadically hit by large weather-related disasters. We find that disasters lead to a maximum rise in consumer prices of 0.5 percent with substantial heterogeneity in the price response. An immediate strong surge in the prices of food, and notably of fresh products, is partially offset by a decline in the prices of manufactured products and services. The effects of weather-related disasters dissipate after four months and differ along the income distribution, notably raising inflation for low-income households by more. Price controls dampen the price response on impact, but lead to similar adjustments in the price level after six months.
How do weather-related disasters affect consumer prices? In times when central banks consider climate risks in their operational frameworks, this becomes a relevant monetary policy question. The existing empirical literature has focused mostly on the aggregate price effect. However, natural disasters are a complex mix of supply disruptions driving up prices in the short run, combined with a shock to the composition of aggregate demand, with differing effects in sign and magnitude across types of goods and services. Aggregate price effects of natural disasters will depend on the relative strength of these supply and demand effects over time.
This paper documents how prices of granular CPI product categories respond to weather-related natural disasters. We focus our empirical analysis on prices in four French overseas territories (DCOM, Départements et Collectivités d’Outre-Mer), which are regularly exposed to significant weather-related disasters and located in different places of the world. For each of these small territories, the French statistical office (Insee) produces harmonized price indices at a disaggregate product level and available at a monthly frequency between 1999 and 2018.
We identify weather-related disasters by combining administrative and meteorological data sets. While administrative databases detect disasters with significant economic consequences, they may suffer from different reporting biases: for instance, the probability to declare the state of emergency is positively correlated with the insurance coverage in the community, and the intensity of natural disasters reported in news-driven and insurance-based data is correlated with GDP per capita. Conversely, an approach only relying on the intensity of meteorological data imperfectly takes into account the heterogeneity in regional vulnerabilities. Overall, these data limitations to measure economic consequences of natural disasters are likely to generate both attenuation biases and omitted variable biases. To reduce these biases, we rely on an instrumental variable approach where we use meteorological records of wind speed and rainfall to predict the occurrence of extreme weather events that imply significant economic damage as reported by administrative data. We then measure the impact of these disasters on the evolution of prices for up to six months after the shock, using a local projection method. We compute the price responses at the product and aggregate level.
We find that weather-related disasters induce a temporary but statistically significant rise in headline consumer prices, with a peak at 0.5 percent two months after the disaster occurrence. The overall observed effect is driven by an immediate strong surge in the prices of fresh food products of 11 % after two months, which vanishes after four months. Prices of other food products also increase, but more moderately and in a more sustained manner (+0.3 %). By contrast, the prices of services and manufactured products decline moderately (and in a less statistically significant way), by about -0.2 %. The positive effects on food prices are likely to reflect negative supply shocks, as we observe a simultaneous decrease in agricultural employment. To the contrary, the negative effects on the prices of manufactured products and services are likely to reflect negative demand shocks.Our results point to small and temporary effects on headline inflation, mainly related to distortions of relative prices.
These effects translate into distributional effects through the heterogeneity in household consumption structure. Overall, the weather-related disasters increase temporarily inflation inequality, with a difference of up to 0.2 pp between the bottom and upper quintiles of the household income distribution. The rise in inflation inequality is primarily due to the fact that the weight of food in the consumption basket is higher for low income households.We also analyse the effects of the introduction of price cap policies, namely the Bouclier Qualité-Prix introduced in 2013. Price caps lower the impact response of price reactions to weather-related disasters in the sample period. However, cumulated over six months, price reactions are not significantly affected by the introduction of price cap policies, implying that the adjustment in the price level is just spread over the horizon of six months.
Finally, this paper documents the importance of controlling for region-specific seasonality for the unbiased estimation of inflationary effects, which affects both the occurrence of disasters and price variations. This matters in particular, but not only, for fresh-food prices.