We investigate how provisioning models interact with bank regulation to affect banks' risk-taking behavior. We study an accuracy versus timeliness trade-off between an incurred loss model (IL) and an expected loss model (EL) such as current expected credit loss model or International Financial Reporting Standards 9. Relative to IL, even though EL improves efficiency by prompting earlier corrective action in bad times, it induces banks to originate either safer or riskier loans. Trading off ex post benefits versus ex ante real effects, we show that more timely information under EL enhances efficiency either when banks are insufficiently capitalized or when regulatory intervention is likely to be effective. Conversely, when banks are moderately capitalized and regulatory intervention is sufficiently costly, switching to EL impairs efficiency. From a policy perspective, our analysis highlights the roles that regulatory capital and the effectiveness of regulatory intervention play in determining the economic consequences of provisioning models. EL spurs credit supply and improves financial stability in economies where intervening in banks' operations is relatively frictionless and/or regulators can tailor regulatory capital to incorporate information about credit losses.