银行的监管风险承受能力

Banks’ regulatory risk toleranceMikael Juselius, Aurea Ponte Marques,Nikola Tarashev Non-Technical Summary and performance.We contribute to the literature on bank capital management in four main ways. First, we provide evidenceof explicit capital management strategies.Second, we introduce a new metric – regulatory risk tolerance(RRT) – which decreases with the buffer target and speed of reversion but increases with the buffer volatility.Third, we study whether RRT varies systematically with bank characteristics and with the degree of bank- GFC across almost all banks in our sample, largely reflecting reductions in management buffer targets thatcushioned the impact of higher regulatory requirements on capital ratios.To avoid standing out in termsof RRT, banks with more volatile management buffer shocks set higher management buffer targets andhigher speeds of reversion. That said, some systematic differences across banks do emerge: for instance, lessprofitable banks exhibit higher RRT, consistent with a lower ability to retain earnings. In addition, banks implications of capital management choices. requirements and balance-sheet ratios. These data cover several regulatory regimes and both US and euro- area banks – including the latter’s non-public requirements.Second, as a new summary metric of capitalmanagement, we introduce “regulatory risk tolerance” (RRT), which combines estimates of asteady-stateMB target,2 the speed of reversion to that target3and the volatility of MB shocks.And we study theextent to which RRT differs across individual banks and between groups of banks based on nationality,size, business model, riskiness and performance.Third, we study whether the overhaul of international We design the RRT metric so that it relate directly to banks’ trade-off between (i) the cost of breachingregulatory requirements and (ii) the cost of maintaining a robust capital position. The first cost is a mani-1BCBS (2021) notes similar effects of increases in MBs that resulted from policy decisions to (i) reduce regulatory require-ments and/or (ii) limit or ban distributions. Findings in the same vein are reported by Jim´enez et al. (2017), Sivec et al. (2019),and Behn et al. (2020), with the latter stressing the importance of policy communication. By contrast, Andreeva et al. (2020)highlight that financial market pressure may constrain banks’ willingness to use released buffers.2In line with the notion of a time-invariant target, Gropp and Heider (2010) find notable statistical and economic significanceof bank fixed effects as explanatory variables of leverage.3The MB target and speed of MB reversion are in the spirit of the “partial adjustment” model in Flannery and Rangan a partial adjustment model of the MB. The first period covers 34 pre-GFC quarters – corresponding to theBasel I and II regulatory standards, which did not differ in terms of the level of capital requirements andthe attendant definition of capital. The second period covers 34 post-GFC quarters that are associated withBasel III, which overhauled the previous international standards, and saw a new set of jurisdiction-specificrequirements (see below).5We estimate regime-specific MB target, speed of reversion to the MB target andvolatility of MB shocks – at the level of all banks, groups of banks and individual banks – thus allowing for thepossibility that each period features a distinct steady state. We use local projection to obtain estimates thatare robust to non-linearities in the evolution of MB, controlling also for bank-specific and macroeconomic ECB Working Paper Series No 3161 steeper increases in their capital requirements under Basel III. This suggests that RRT counteracts theregulatory overhaul’s impact on capital ratios.Second, in the post-GFC period – though not in the pre-GFC period – banks experiencing greater volatility in MB shocks tend to maintain higher buffer targets andexhibit faster speeds of reversion towards those targets. This configuration of the three components – which we also observe at the level of bank groups – dampensthe RRT cross-sectional dispersion, consistent with a strategy to avoid standing out in terms of capitalmanagement post-GFC. based on size, business model, performance or asset riskiness.9In each case, we effectively seek to uncoverwhether the banks in one group face a different RRT-related trade-off relative to those in the other group.Despite banks’ tendency to cluster around a common RRT level, we identify systematic, albeit eco-nomically modest, differences between bank groups in the post-GFC period. For instance, US banks havesignificantly lower RRT than euro-area banks, mainly because of lower volatility of MB shocks. Conversely, ECB Working Paper Series No 3161 Concretely, a one-standard deviation negative shock to the MB leads high-RRT banks to reduce their netlending by 0.75% over a two-quarter horizon.Given high-RRT banks’ greater conservatism, this findingsuggests that cutting