In the wake of the financial crisis, the IMF promoted levies on the risky part of bank funding as a tool to increase revenue collection from the financial sector while at the same time contributing to financial stability by creating incentives for banks to adopt less risky capital structures. Such bank levies have been adopted in 14 European countries and are still under consideration in others. This research project studies how banks responded to the levies, with the ultimate aim of assessing whether they have been successful in reducing risk in the financial sector. We use data from the financial reports of more than 5,000 European banks.
Our first main finding is that the levies did have a statistically and economically significant effect on banks’ funding choices – our results suggest that banks raised their equity-asset ratio by 1 to 1.5 percentage points on average in response to the levies. This suggests that the levies can be successful in reducing banks’ funding risk. However, we also find that the levies had an effect on banks’ portfolio choices. Specifically, we find that banks that had a low initial level of capital (ie. that were more risky on the funding side) changed their portfolio of assets in response to the levies so as to increase the average portfolio risk. This seems likely to be the result of an unintended interaction with financial regulation, which imposes a minimum ratio of bank capital to risk-weighted assets. Hence, to the extent that banks raise more capital because of the levies, they are also able to increase the risk of their assets while still complying with the regulatory capital requirements.
These results suggest that the intended behavioural response, the decrease in funding risk, was largest in relatively safe banks whereas the unintended response, the increase in portfolio risk, occurred only in relatively risky banks. This seems to imply that the levies were more successful in reducing total risk in initially safe banks than in initially risky banks.
Michael Devereux and John Vella, CBT Working Paper 13/25