Research conference at the SSM
Early May saw the first SSM research conference, where I had the honour to serve as chair of the programme committee. Lots of great submissions and an excellent conference programme. While there is a rich research culture in the ECB and national central banks in Europe, such a research culture is now starting to be built up in the SSM. It is important to note that there was an active participation of senior management and board members of the SSM so clearly the programme also reflected SSM priorities in the area of bank supervision and financial stability.
I will not be able to do justice to all the excellent papers that were presented, so here a quick rundown of some that caught my attention.
Several papers have gauged the impact of the extraordinary macroprudential measure of asking banks to refrain from paying dividends during the pandemic. One further contribution is by Ernest Dautovic and co-authors. Using data on planned but non-distributed dividends (thus exploiting cross-bank variation) and credit registry data they can disentangle loan demand and supply effects by focusing on firms borrowing from several banks. Their result suggests that the dividend restrictions helped support financially constrained firms, especially small and mid-sized enterprises and firms operating inCovid-19 vulnerable sectors. At the same time and reassuringly, there seems no evidence of a significant increase in lending to riskier borrowers and zombie firms.
In a theory paper, Gyöngyi Lóránth and co-authors model supervisory interventions in cross-border banks under different institutional architectures where a bank may provide voluntary support to an impaired subsidiary in a different country using resources from a healthy subsidiary. With supervisory authority on the national level, there is the risk of ring-fencing, thus authorities restricting the parent bank from supporting an impaired subsidiary in another country. This risk increases the more correlated the returns of subsidiary and parent bank are as national authorities will be worried about the costs for the national deposit insurer. Such ringfencing would be eliminated under supranational supervision (i.e., SSM). However, there are also incentive effects, with weaker cross-border banks possibly taking more risks (or exerting less effort) under supranational rather than national supervision. An important contribution to the debate on cross-border supervisory cooperation (and directly related to my own research with Wolf Wagner).
Following the Global Financial Crisis, there was pressure to move from backward-looking to forward-looking provisions to make them less cyclical (‘provisioning too little, too late’). Expected Credit Loss (ECL) approaches such as the International Financial Reporting Standard 9 (IFRS 9) were introduced (although partly suspended in March 2020 given the high uncertainty). IFRS 9 requires provisions to be based on estimated future credit losses, expected to be varied over time as credit risk evolves and usually obtained from dedicated provisioning models operated by banks themselves. Using loan-level data from the ECB’s credit registry, Anacredit, Markus Behn and co-authors gauge how the implementation of IFRS 9 has affected euro area banks’ provisioning behaviour. They find that, overall, provisioning is generally higher under IFRS 9, but the bulk of provisioning is still done at time of default. They also find evidence that IFRS 9 has increased variation in provisioning practices across banks, with banks with higher capital headroom more likely to provision to the same borrower than banks with less capital head room. Thus, accounting discretion and capital management motives seem to affect not only the overall level of provisioning, but also the distribution of provisions across a borrower’s banks.
Finally, Barry Eichengreen and Orkun Saka have a fascinating paper (that I will discuss myself later this week in a conference in Tilburg) on how cultural stereotypes influence cross-border banks’ investment decisions in sovereign bonds. They use hand collected bi-annual data on banks’ investments in European sovereign debt and show that when residents of the country or countries where a bank operates have a high level of trust in residents of another country, the bank is more likely to hold claims on that other country. Specifically, they create bank-specific measure of trust in a specific country, by calculating a weighted average of bilateral trust between two countries, where weights are the share of host-country branches in the network of the bank. They show that this bank-level measure of trust predicts banks’ entry/exit decisions vis-a-vis sovereign debt of a country and that the effect is less strong for more diversified banks, while – not surprising – it is stronger for countries that have gone through a sovereign debt crisis.