A disaster under-(re)insurance puzzle: Home bias in disaster risk-bearing

BIS Working Papers  |  No 808  | 
30 August 2019


One basic role of international financial markets is to share risks across economies. For instance, investors swap the equity of firms at home for those abroad in order to diversify away from sources of loss that are unique to the domestic economy. In an ideal world of complete risk-sharing, domestic investors would bear such a loss only to the extent of their share of world wealth. Thus, a small economy would lay off on the rest of the world almost all risk that is unique to it. Natural disasters pose such a risk. In theory, financial markets might spread their risk around the world to approach the ideal. But how well do they work in practice?


We start with data compiled by Munich Re, a major reinsurer, on the largest natural catastrophes. These give us estimates on the monetary value of the direct losses and the fraction covered by insurance. We then search balance of payments data to identify related receipts of reinsurance payments from the rest of the world. These data let us measure, for the first time, the fraction of the cost that economies share internationally for dozens of disasters. We then decompose the departure from the ideal of perfect international risk-sharing into two parts. The first arises from a lack of insurance in the first place. The second arises because insurers do not reinsure risks to the extent that, in principle, they should.


This paper finds that individual economies share the cost of natural disasters to a surprisingly small extent. Two big earthquakes hit in 2011. Japan absorbed over nine-tenths of the losses in Japan. New Zealand laid off about half the losses in New Zealand. Japan turns out to be the more typical case. If governments pick up the bill for natural disasters ex post, then they may convert natural risks into fiscal risks.


The losses from the 2011 earthquakes in Japan remained in Japan, while reinsurance spread the losses from that year's New Zealand earthquake to the rest of the world.This paper finds that the Japanese case is more typical: losses from natural disasters are shared internationally to a generally very limited extent. This finding of home bias in disaster risk-bearing poses a puzzle of international risk-sharing. We decompose international risk-sharing into the portion of losses insured and the portion ofinsurance that is internationally re-insured. We find that the failure of international risk-sharing begins at home with low participation in insurance. Regression analysis points to economic development and institutional/legal quality as important determinants of insurance participation. We propose a new method to measure international reinsurance payments with balance of payments data. This method identifies for the first time the cross-border flow of reinsurance payments to 88 economies that experienced insured disasters in the 1985-2017 period. Regression analysis of these data points to small size and de facto financial integration as positively related to the reinsurance share, as one might expect. However, we also find that more internationally wealthy economies reinsure less, suggesting that net foreign assets substitute for international sharing of disaster risk. For advanced economies, a lack of international risk-sharing is correlated with a lack of fiscal space. Thus, the governments under more pressure to provide ex post government insurance through the budget have less room to manoeuvre to do so. At high levels of public debt, a lack of ex ante insurance can turn disaster risk into financial risk.

JEL codes: F32, G15, G22, Q54

Keywords: international risk-sharing, earthquake insurance, reinsurance