This is how we can reduce risk and improve stability within the financial system

Following the experience of the Global Crisis, countries have significantly increased their efforts to cooperate in the supervision of their banks.1 Most notably, the euro area has now the Single Supervisory Mechanism, which places the supervision of large banks in the hands of the ECB. However, little is known about whether such cooperation ‘works’, that is, does it help improve the stability of the financial system? And even if it works, what is the channel? And does the effectiveness of cooperation depend on the environment within which supervisors operate, or the type banks to which it applied?

While many prominent cooperation arrangements are of a recent nature, for decades, it has been common for countries to cooperate with each other. For example, among a group of 93 countries, almost 200 bilateral cooperation agreements were signed between 1995 and 2013, in addition to numerous multilateral agreements. These agreements are of varied nature, ranging from a common supervisor to more limited types of cooperation, such as agreements on information sharing or joint exercises on crisis prevention and resolution. The agreements can arise because of a direct bilateral agreement or because countries are part of a multilateral agreement.

The 2008 crisis that saw the failure of several large cross-border banks provided the impetus for an additional wave of (more intensive) cross-border agreements between supervisors to cooperate.

How to measure the impact of cooperation?There is a rich history in countries forming cooperation agreements… While many prominent cooperation arrangements are of a recent nature, for decades, it has been common for countries to cooperate with each other. For example, among a group of 93 countries, almost 200 bilateral cooperation agreements were signed between 1995 and 2013, in addition to numerous multilateral agreements. These agreements are of varied nature, ranging from a common supervisor to more limited types of cooperation, such as agreements on information sharing or joint exercises on crisis prevention and resolution. The agreements can arise because of a direct bilateral agreement or because countries are part of a multilateral agreement. The 2008 crisis that saw the failure of several large cross-border banks provided the impetus for an additional wave of (more intensive) cross-border agreements between supervisors to cooperate. … which offers a way to study their effectiveness In a recent paper, we examine the effectiveness of supervisory cooperation agreements (Beck et al. 2018a). A large theoretical literature has argued for the potential benefits of international banking supervision (see for example Acharya 2003, Calzolari and Loranth 2011, Carletti et al. 2016, Dell’Ariccia and Marquez 2006, and Goodhart and Schoenmaker 2009).2 Cooperation should lead to higher supervisory stringency as supervisors now take into account the cost of bank failures in other countries. In addition, cooperation provides supervisors with new information that should result in better decision-making. On the other hand, some suggest that more cooperation does not necessarily result in higher banking stability (Dell’Ariccia and Marquez 2006, Beck et al. 2013, and Calzolari et al. 2018). For example, the presence of foreign ownership of banks may make domestic regulators excessively strict, in which case cooperation is expected to reduce bank stability (Beck et al. 2013). Also, supervisors face many constraints in practice: they have limited legal powers, are subject to regulatory capture, have imperfect information and/or face political pressure. Many of these constraints are likely to be compounded in an international setting. Cooperation agreements—even if well-intended—may hence not result in higher stability.

We investigate supervisory cooperation at the bank-level. We calculate bank-specific supervisory cooperation indices that measure the degree to which a global bank’s parent-subsidiary structure is covered by cross-border cooperation between home and host country supervisors. Specifically, the index gauges the share of a parent bank’s foreign assets in subsidiaries whose host country supervisors have an agreement with the parent-bank supervisor.

These bank-level (supervisory) cooperation indices vary within a country, as different parent banks will have different geographic dispersion of their subsidiaries. This facilitates identification of the relationship between cooperation and stability, as it allows for other home-host country factors.

Using panel analysis for a large sample of cross-border banks, we study whether a higher incidence of supervisory cooperation is associated with higher bank stability, as measured by the Z-score, an accounting-based measure of distance to default.

We find an economically large effect of supervisory cooperation. For example, a standard-deviation increase in the supervisory cooperation intensity at the bank level improves a bank's Z-score by 24%. The large result is robust to an instrumental variable approach that alleviates concern about the endogeneity of supervisory cooperation.3

Interestingly, we find the association to be concentrated at the smaller institutions in our sample of cross-border banks.4 This may be explained by the fact that large banks are more complex, and hence more difficult to supervise. Consistent with this, we show that banking supervision is less effective for banks that have a larger number of subsidiaries. We also find evidence for a convex relationship—the effect of cross-border cooperation on stability increases in the share of a parent bank’s foreign asset being covered by cross-border agreements.

Focusing on the sample of smaller banks, we show that the link between cooperation and bank stability runs through asset risk. This is consistent with the notion that asset risk is difficult to observe and control at arms-length; intensive cooperation and information exchange should hence have a pronounced effect. By contrast, bank leverage (which also affects the Z-score) is not improved through cooperation. This can be explained by leverage already being well covered by existing (international) regulations, such as capital adequacy standards, and hence being less affected by cooperation.

We also find that the characteristics of a country's supervisory and financial system influence the effectiveness of supervision. First, supervisory effectiveness is higher when both home and host supervisor are more stringent. Second, supervisory effectiveness is also higher when the home supervisor has access to higher quality information. Third, cooperation is more effective when there are fewer limits to foreign entry. This likely reflects that there are then a lower number of other foreign banks, leading to less dilution in supervisory attention.

A potential concern is that cooperation is effective in normal times but breaks down during crises, and thus when it is most needed. Our empirical results, however, suggest that cooperation remains effective in reducing bank risk during the Global Crisis, and—if at all—the influence of cooperation may even get larger. We also find cooperation to be effective conditional on a systemic event, as we show that it improves a bank’s marginal expected shortfall.

Our analysis offers several important lessons for policy. First and foremost, more cooperation is in principle to be welcomed as it is effective in improving banking stability. However, in order for it to work, supervisors need to have the right institutional conditions, such as having sufficient powers and access to quality information. Also, the effectiveness of cooperation declines with bank size, possibly reflecting the reality that supervision of more complex institutions is more difficult. This points to a significant downside of cooperation, as the very large institutions are the ones that pose the highest risk to financial stability.