We reveal evidence that the US aggregate bank dividends exercise a causal impact on the US real GDP growth during the period from the introduction of the Prompt Corrective Action framework in 1992 until the outburst of the subprime mortgage market crisis in 2007. Over this period, the positive signalling effects of bank dividends outperform the negative effect of dividends on default risk. During the pre‐Prompt Corrective Action and the recent post‐2007 periods, bank dividends do not affect GDP growth. This regime‐dependent relation is due to an asymmetric role of bank default risk. These findings are of interest to bank regulators in reassessing the role of bank dividends within Basel III and carry important policy implications as bank dividends constitute an important tool available to policy makers for strengthening real activity. Copyright © 2014 John Wiley & Sons, Ltd.