How natural disasters affect banking

Natural disasters affect firm activities in a variety of ways. They disrupt business operations directly through the loss of lives and the destruction of buildings and equipment, and indirectly through the impact on firms’ transaction partners such as their customers, suppliers, and providers of finance.

The indirect damage through transaction partners can be substantial. For example, a survey conducted in July 2012 following the Tohoku earthquake, which wreaked havoc in the northern part of Japan on 11 March 2011, suggests that 62.5% of firms in the Tohoku area suffered direct damage from the earthquake. In addition, 36.5% of firms also indicated that they were negatively affected by damage to their suppliers, and 44% stated that they were negatively affected by damage to their customers. 11.4% of firms indicated that, following the earthquake, the bank that was their most important source of lending could not operate the branch with which they transacted, and 4.8% of firms replied that they were adversely affected by the fact that the bank with which they transacted suffered damage as a result of the earthquake.

Despite these potentially substantial indirect effects, much of the extant economic literature has focused solely on the direct impact of natural disasters and the recovery from them (see, e.g., Leiter et al. 2009, De Mel et al. 2012). As a result, our understanding of the indirect economic effects of natural disasters is limited; a better understanding of such effects is indispensable for designing policies for recovery and the appropriate allocation of limited resources.

Among firms’ transaction partners, suppliers of funding – especially banks – play a particularly important role in the process of economic recovery from natural disasters. When banks suffer physical or capital damage as a result of a natural disaster, they may not be able to provide sufficient funds at the same interest rates as before. In frictionless financial markets, firms should be able to offset such a negative shock to their funding by switching from a damaged to an undamaged bank, or raising funds from the financial markets. In practice, however, switching may involve substantial costs or time due to the role played by private information on firms’ creditworthiness, potentially making it difficult for firms to obtain funds from alternative sources in a timely and cost-efficient manner. Thus, whether damage to banks affects firm activities or not is an empirical issue.

While there is a considerable body of literature examining banks’ role in providing funding during normal times and times of financial crisis, there are relatively few studies that examine how impairment of banks’ functioning in the wake of a natural disaster affects firm activities. Against this background, the aim of this column is to show how damage to banks affects firm activities, such as capital investment and exports, based on the results of studies we conducted on Japan’s 1995 Kobe earthquake (Hosono et al. 2012, Miyakawa et al. 2014).

Empirical strategy to study financial constraints

Damage to banks is likely to undermine their lending capacity and to tighten the financial constraints that firms face when they undertake capital investment (Kahle and Stulz 2013) or start exports, i.e., the extensive margin of exports (Amiti and Weinstein 2011, Paravisini et al. 2011). Due to such financial constraints, firms may not be able to finance capital investment or the fixed costs associated with entering a foreign market (e.g., the costs of building a sales network abroad and adapting products to local tastes and regulations). Financial constraints may also restrict the volume of exports (i.e., the intensive margin of exports) of firms that have already started exporting due to a reduction in the amount of trade finance.

Although the theoretical predictions on the relationship between financial constraints and firm activities are straightforward, examining this relationship empirically is fraught with difficulties due to identification problems. That is, not only may lending by banks affect the performance of borrowing firms, but also borrowing firms’ performance may have a significant impact on lenders’ financial health and lending. Furthermore, there may be assortative matching between firms and banks (e.g., Sorensen 2007). Specifically, better-performing firms may be more likely to transact with better-performing banks. Both of these issues mean that it is difficult to clearly identify the direction of causality between bank lending and firm performance.

Two studies by a team of researchers of which we are members (Hosono et al. 2012, Miyakawa et al. 2014) investigated the effects of bank damage due to a natural disaster on borrowing firms’ capital investment and export behaviour. In order to circumvent the identification problems indicated above, the studies took advantage of the natural experiment provided by the 1995 Kobe earthquake.

Devastating natural disasters such as the Kobe earthquake are likely to impose significant financial constraints on firms through the exogenous financial shock caused by the reduced lending capacity of banks from which firms borrow. For example, a natural disaster may obliterate information on borrowers’ creditworthiness accumulated at the disaster-hit banks and thus destroy their managerial capacity to originate loans, including the ability to screen and process loan applications. Natural disasters may also cause damage to borrowing firms located in the disaster-hit areas, which may lead to a deterioration in banks’ loan portfolios and hence weaken their risk-taking capacity. A reduction in credit supply from banks whose lending capacity has declined may have a negative impact on borrowing firms’ activities through the mechanisms described above.

In the two studies mentioned above, in order to separate out the purely exogenous shock to firms’ financing stemming from the reduced lending capacity of banks, we specifically focused only on firms that did not suffer any direct damage of their own as a result of the disaster, but that were borrowing from damaged banks. If we focused on firms that suffered any direct damage of their own, our analysis would suffer from the identification problem described above. Any reduction in such firms’ capital investment or exports might reflect their loss of assets or lack of production of goods to export rather than the effect of financial constraints due to damage to their banks.

Thus, by focusing on non-damaged firms, we are able to extract the impact of the purely exogenous financial shock resulting from the earthquake. Employing this approach also circumvents any potential problem caused by positive assortative matching, since it would be implausible to assume that better-performing firms choose banks that are less likely to be hit by a natural disaster.

Related literature

As mentioned above, there are very few studies that examine the adverse impact of a natural disaster through damage to banks, and there are only a limited number of studies that successfully overcome the aforementioned identification problems.

One notable exception is the study by Amiti and Weinstein (2013), who used matched bank-firm loan data to identify idiosyncratic bank shocks and to examine the impact of loan supply shocks on firms’ capital investment. Using Japanese data, they showed that idiosyncratic bank shocks have a large impact on firms’ investment.

Another notable study, this time focusing on exports, is that by Paravisini et al. (2011) who showed that a decline in credit due to the reversal of capital flows during the 2008 crisis reduced the intensive margin of exports, but had no effect on the extensive margin. Yet another example is the study by Amiti and Weinstein (2011) who employed Japanese firm-level data with information on each firm’s main bank to examine the impact of the financial crisis in Japan during the 1990s on firm exports. They found that banks’ financial health, measured by their market-to-book ratio, has a significant impact on the intensive margin of firms’ exports, which suggests that bank financial health affects firm exports through the availability of trade finance. They also show that their results are not contaminated by a reverse causality.

None of these studies, however, examined the role of financial constraints on firms’ capital investment and export behaviour in the context of a natural disaster.

Evidence on the effects of bank damage on firm investment

The first of the two studies that we are introducing here, Hosono et al. (2012), focused on the effects of damage to banks on firms’ investment activity. In the study, two alternative definitions of bank damage were used:

• Damage to a bank’s headquarters, which is a dummy that takes value of one if the bank was headquartered in the earthquake-hit area, and aims to capture the decline in the bank’s managerial capacity to process loan applications at the back office; and
• Damage to a bank’s branch network, which is the ratio of the number of branch offices located inside the earthquake-affected area to the total number of branches and aims to capture the decline in the bank’s financial health and risk-taking capacity.

The results of the study can be summarised as follows.

• First, firms located outside the earthquake-affected area, but associated with a main bank located inside the area, have a significantly lower investment ratio than firms outside the affected area associated with a main bank that was also located outside the area, indicating that bank damage has a negative impact on firm investment.

Moreover, this negative impact is not only statistically significant, but also economically significant. For example, in the case that bank damage is defined as damage to banks’ headquarters, the investment ratio of undamaged firms whose main bank was damaged is 8.2 percentage points lower than that of undamaged firms whose main bank was not damaged. This impact is economically significant, given that the average investment ratio for undamaged firms in FY1995 was 13.1%. This result clearly implies that the exogenous damage to banks’ lending capacity has a substantial adverse effect on firm investment.

• Second, the finding above is robust to the two alternative definitions of bank damage.

We interpret these results as evidence that financial constraints play an important role in the observed relationship between bank damage and the investment of firms outside the unaffected areas.

• Finally, bank damage, either to headquarters or branch networks, does not last for very long – three years after the earthquake the effect had dissipated.

Evidence on the effects of bank damage on firm exports

The second study, Miyakawa et al. (2014), examined the effects of damage to banks on firm exports. Specifically, like Hosono et al. (2012), the study focused on firms that were located outside the area affected by the 1995 Kobe earthquake and that were hence undamaged, and divided these firms into those whose main bank was located within the affected area and was damaged, and those whose main bank was not located in the affect area. The results of the study can be summarised as follows.

• First, as for the extensive margin of exports, we found that firms whose main bank is damaged are less likely to start exporting or expand the regions to which they export than firms whose main bank is unaffected.

Moreover, the impact of bank damage on undamaged firms is not only statistically but also economically significant. For example, the probability of starting exporting is 4.5 percentage points lower for firms whose main bank is damaged than for firms whose main bank is unaffected. This impact is substantial given that the average probability of starting exports of firms in unaffected areas in FY1995 was 4.4%. The negative impact increases to 6.7 percentage points for FY1996 before declining to 3.3 percentage points for FY1997. This finding suggests that the exogenous damage to banks’ lending capacity has a substantial adverse effect on the extensive margin of firm exports.

• Furthermore, the finding holds regardless of which of the two alternative measures of bank damage, i.e., damage to a bank’s headquarters or damage to a bank’s branch network, is used.

Thus, our results imply that deterioration in banks’ lending capacity has an important adverse impact on the extensive margin of exports. However, it is also interesting to note that although both measures of bank damage yielded significant results, there are some differences. Specifically, we find that whereas damage to a bank’s headquarters affected firm exports in FY1995, the financial year immediately following the earthquake (i.e., in the financial year typically from April 1995 to March 1996; the earthquake occurred in January 1995), damage to a bank’s branch network affected firm exports only a year later, in FY1996. This implies that the impairment of banks’ managerial capacity affects firms’ exports within a relatively short period of time, while the impairment of banks’ financial health and risk-taking capacity only has a delayed effect.

• Second, as for the intensive margin, we found that a firm whose main bank is damaged has a lower export-to-sales ratio than a firm whose main bank is unaffected.

This finding is consistent with the prediction that the exogenous damage to banks’ lending capacity has an adverse effect on the intensive margin of firm exports through a smaller provision of trade finance. Specifically, we found that in FY1995 the export-to-sales ratio of firms whose main bank was damaged was 6.5 percentage points lower than that of firms whose main bank was unaffected, while in FY1996 and FY1997 it was 7.4 and 6.5 percentage points lower, respectively. This impact is economically significant, given that the average export-to-sales ratio for firms in the unaffected areas was 9.1% in FY1995, 9.2% in FY1996, and 9.7% in FY1997.

Lessons from the empirical results

The results above imply that for economic recovery in the wake of a natural disaster, it is important to ensure that the functioning of financial intermediation is maintained so as to mitigate any potential financial constraints and allow firms to carry on with their capital spending and export plans. To this end, government support, including capital injection into banks affected by a natural disaster, as provided in the aftermath of the Tohoku earthquake, may be justified. In other words, in addition to speeding up recovery through the reconstruction of infrastructure and support for damaged firms, it may be necessary for the government to introduce policies for the purpose of mitigating a potential credit crunch triggered by the disaster.

In this context, it should be noted that the analysis in the two studies focused solely on damage to a firm’s main bank. Our results thus imply that firms are unable to fully offset the shock to their main lender bank by finding alternative sources of funding, providing further justification for government intervention in loan markets in the case of a massive natural disaster. At the same time, however, it is also important to note that our findings indicate that the negative effects of bank damage on firm activities do not last for a long time. This means that government measures to intervene in the supply of credit should be timely and can be terminated within a short period.

Published in collaboration with Vox

Authors: Kaoru Hosono, Professor of Economics, Gakushuin University and Daisuke Miyakawa, Associate Professor of Economics, Nihon University

Image: Vehicles make their way through high water as they drive after hurricane Arthur passed through in Nags Head, North Carolina July 4, 2014. REUTERS/Chris Keane