This paper examines analytically and empirically whether the Basel III leverage ratio (LR) and output floor (OF) requirements act as substitutes for or complements to each other and how they interact with risk‑based regulatory requirements. These requirements determine banks’ overall capital requirements and therefore their capacity to support credit and growth in the economy as well as their resilience to negative shocks.
A simple analytical framework maps the three requirements to a common metric, the risk‑weighted asset (RWA) density, and defines regions in which either the risk‑based requirements, the LR or the OF determine a bank’s required capital. The framework identifies the conditions under which one backstop can substitute for the other and derives the potential change in capital from removing a binding backstop.
Publicly available data for 29 global systemically important banks (G‑SIBs) over 2014–25 show that the proportion of banks for which the LR is the highest requirement increased materially alongside a broad decline in unfloored RWA densities. Yet the most constraining requirement is not necessarily stable: banks switched between LR and risk‑based constraints 55 times in the period, based on annual frequency data. More generally, and in line with the conceptual analysis, the highest requirement varies significantly across banks and jurisdictions due to differences in the prevailing requirements and RWA densities. For the 13 G-SIBs currently disclosing OF data (ie those in Canada, the European Union, Japan and Switzerland), none was bound by the OF at end‑2025 owing to transitional arrangements. However, under a fully phased‑in 72.5% OF, six banks would be bound by the OF. Interestingly, this includes cases where the LR would not even partially reproduce the outcome of the OF constraint in its absence, highlighting the complementary nature of the two backstops.
Although both backstops constrain required capital when risk weight densities are undesirably low, they are not substitutes. Each addresses a failure that the other cannot: the LR constrains excessive leverage that risk measures might miss, while the OF constrains excessive RWA variability due to internal models by linking effective RWA to standardised approaches, preserving buffer usability and some risk sensitivity.