This paper examines the impact of natural disasters on a wide range of bank performance measures, namely; credit risk, default risk, liquidity and profitability. Further, it explores the moderating effect of financial integration (measured by the ratio of foreign bank presence and foreign claims of international banks to GDP) on the relationship between natural disasters and bank performance. The East Asia Pacific region provides a good context to examine the impact of natural disasters on bank performance and the moderating role of financial integration for two main reasons.
This article focuses on the impact of natural disasters on country-level financial stability (as measured by the Z-score of the national banking sector) of 81 countries over the period 1997–2010. The literature provides conflicting predictions about the moderating role of financial integration on the relationship between natural disasters and bank performance. Finally, regardless of the type of indicators used (foreign claims of international banks relative to GDP and foreign bank penetration), financial integration worsens the impact of disasters on bank performance.
Unfortunately, there has been no empirical evidence on the impact of disasters on bank liquidity. Noth and Schumer (2018) provide empirical evidence on the negative impact of disasters on US bank profitability. In fact, there was only one evidence for the moderating role of capital account opening in the impact of natural disasters on economic growth.
Measures of financial integration
To reduce endogeneity concerns, some papers (such as Klomp, 2014 and Noy, 2009) construct a count variable to proxy for accident frequency. However, EM-DAT does not include the entire universe of events, so there is no measure of the number of accidents from which to calibrate the frequency estimate (Felbermayr and Gröschl, 2014).
Results and discussion 1. Descriptive analysis
The impact of natural disasters on bank performance
- The impact of country disaster index on bank performance
- The impact of various type of disasters on bank performance
- The impact of disasters on bank performance- by the economic development of a country
We start the discussion with the aggregate impact of disasters measured by the country-year index ("DISINDEX"). 1) which includes only the lagged dependent variable and the disaster index. This may explain the marginal significance of the coefficient, since the ultimate effect of disasters on bank defaults also depends on whether banks could increase or have sufficient equity buffers. As reported in column 8, the impact of disasters on ROA is positive (with the coefficient of 0.002) but insignificant.
On the one hand, bank profitability could directly benefit from the increase in lending following post-disaster reconstruction.6 The increase in total lending after a disaster in the affected areas has been found in the literature (Chavaz, 2014; Cortés and Strahan 2015; Koetter et al. 2016). Therefore, the ultimate impact of disasters on profitability depends on whether the benefit of additional credit activity outweighs the cost of lower credit quality. We observed a positive but insignificant impact of the disaster intensity index (“DISINDEX” on the loan to total assets ratio).
To assess the impact of specific types of disasters on the bank's performance, we separately include a measure of intensity for earthquakes, volcanic eruptions, storms, floods, droughts and extreme temperatures (instead of the overall disaster index "DISINDEX" as in section 4.2.1 ) in Eq. The total impact of several disasters (proxied by the disaster index "DISINDEX") is more relevant to explain the reduction of banks' stability. Overall, the difference in the impact of different types of disasters on banks' profitability partly helps to explain the insignificant coefficient of the aggregate disaster index (DISINDEX) found in Section 4.2.1.
There is little evidence of the gradual and incremental impact of slow events such as droughts, floods or extreme temperatures. In order to better understand how the level of economic development of a country moderates the impact of disasters on bank performance, the sample is separated into developed and developing countries.8 Table 5 reports the impact of the Disaster Intensity Index (DISINDEX) on different levels of banks. performance measures for the two subsamples. Accordingly, the adverse impact of disasters is reflected in the deteriorating quality of loans on bank balance sheets.
Together with these results, we observe a consistent negative impact of natural disasters on the liquidity of banks, regardless of the level of economic development of the country. This again highlights the importance of studying the impact of disasters on bank liquidity, which has not been explored in the existing literature.
The moderating role of financial integration
These results help to explain the vulnerability of banks in developing countries in the presence of natural disasters. The evidence can be attributed to the recovery lending opportunities after catastrophic events in developing and developed countries. Regarding the interaction term between DISINDEX and CLAIM, the significant positive (0.012) and negative (-0.001) coefficients in columns 1 and 2, respectively, suggest that disasters increase credit risk and fragility to a greater extent in the case of banks operating in more financially integrated countries.
A reduction in foreign assets could slow the recovery of capital of countries, tighten capital constraints of companies regardless of the efforts of commercial banks in expanding loans for rehabilitation. This capital constraint would have slowed down the rebuilding process, making the impact of the disaster even more dramatic in the years leading up to it. Regarding ROA, the interaction term between DISINDEX and CLAIM is negative (-0.001) and significant at the 1% level.
Similarly, the interaction term between DISINDEX and FOR shows that the presence of foreign banks significantly exacerbates the impact of disasters on banks' risks and performance. 10 In Appendix 4, we report the models that estimate the impact of disasters on bank performance by bank ownership. This result indicates that most of the significance reported in Table 3 can be attributed to the impact of disasters on domestic banks.
The core business of domestic banks is concentrated within each country and therefore they are more exposed to the adverse impact of a natural disaster. Finally, in countries with a more open capital account, the adverse impact of disasters on bank performance is greater. The size of these coefficients is also larger than that of the coefficients obtained for the interaction terms between EIS (or VIR) and DISINDEX.
The evidence is consistent with Noy (2009) who finds that countries with less capital account openness appear to experience less output losses due to natural disasters. Taken together, natural disasters disproportionately affect bank performance in countries with different levels of financial integration, strengthening our support for Hypothesis 3.
In other words, if domestic banks are playing an important role in covering the lack of liquidity of firms facing a sudden disaster shock, a higher proportion of foreign banks will hinder the recovery process. This shows that in case of severe disasters, international banks withdraw their capital from the affected countries. The response is more pronounced in countries with higher capital account openness, which allows for more flexible movement of international capital across borders.
In short, the evidence from KAOPEN strengthens our results when DEMAND and FOR are used to measure financial integration.
Conclusion and implications
In this sense, the methodology underestimates the impact of disasters, as some banks may be significantly affected, but the negative impact will not be detected unless all banks (or a large number of them) are affected. Finally, there are some limitations in Ifo-GAME measurement for floods and droughts; therefore, our paper could not provide a complete assessment of the impact of such events on the bank's performance. This calls for further studies to find appropriate datasets and empirical techniques to measure the impact of these events, which are becoming more frequent due to climate change.
There are 8,299 bank-year observations from eleven countries in the East Asia-Pacific region (comprising data from China, Japan, Korea, Indonesia, Malaysia, Philippines, Singapore, Thailand, Vietnam, Australia and New Zealand) covering the period 1992- 2010. The table reports the descriptive statistics (mean and standard deviation) for the disaster intensity index (DISINDEX) and financial inclusion indicators (CLAIM, FOR and KAOPEN) for each country in the sample. The table presents the empirical result for equation (1) using Fixed Effect-OLS regression with robust standard errors clustered around banks.
The dependent variables (Y) are ROA (bank profitability), CRERISK (credit risk), LIQ (liquidity) and ZSCORE (natural logarithm of z-scores to approximate default risk). Control variables including CON (market concentration), GDP (GDP growth rate), IFL (inflation) and PRICRE (private sector loans to GDP) are introduced with a one-year lag. Specific intensity measures for each disaster type come from the Ifo-GAME dataset; specifically, the Richter scale for earthquakes, the VEI volcanic explosivity index for volcanic eruptions, wind speed for storms, temperature for extreme temperatures, millimeter precipitation for floods, and an indicator variable for droughts.
The table presents the empirical result for equation 1 using Fixed Effect-OLS regression with robust standard errors clustered around banks. The table presents the empirical result for equation 2 using Fixed Effect-OLS regression with robust standard errors clustered around banks. Financial integration is measured by CLAIM (foreign claims of international banks to GDP), FOR (ratio of foreign banks to the total number of banks), KAOPEN (capital account opening index).
The interaction terms between DISINDEX and various indicators of financial integration (PRETI, PER, KAOPEN) are created, then included in the model separately. The table presents the empirical result for Equations 1 and 2 using Fixed Effect-OLS regression with robust standard errors clustered around banks. Additional bank-level variables such as SIZE (natural logarithm of total assets), EQUITY (equity to total assets), INDIV (income diversification), COST (overall cost) are included.
The country-level covariates such as LNGDP (natural logarithm of GDP per capital), POLITY (Polity IV index), INTEREST (annual real interest rate).
Definition and specification of variables
33 EQUITY Share of equity = total capital/ balance sheet total (%) Bankscope INDIV Income dispersion = (share of non-interest income/ . total income) (%) Bankscope.
Number of banks and observations for each country and year
The moderating role of financial integration- An interaction term between DISINDEX and an indicator variable of CLAIM
The impact of disasters on bank performance- by type of bank ownership
Bank regulation, institutional quality and banking risk in emerging and developing countries: an empirical analysis.