AGN must be very efficient at powering outflows

Galaxy evolution is affected by competing feedback processes. Stellar feedback dominates in low-mass galaxies, while AGN feedback predominantly affects massive ones. Recent observational results reveal the dependence of black hole accretion rate (BHAR) and star formation rate (SFR) on galaxy stellar mass, and give information on the galaxy mass at which the changeover between dominant feedback mechanisms occurs. I use this information to derive an empirical estimate of the coupling efficiency, $f_{\rm AGN}$, between AGN luminous energy output and AGN-driven galactic outflows, and the momentum loading factor $f_{\rm p,AGN}$ between the momentum of AGN radiation field and the outflow. The results are independent of any particular model of AGN feedback and show that AGN feedback must be very efficient and/or have very large momentum loading in order to explain current observations. I discuss possible ways of reaching the required efficiency and loading factor, and the selection effects that might result in only weak outflows being observed, while the most powerful ones may be generally obscured. There are significant uncertainties involved in the derivation of the result; I suggest ways of reducing them. In the near future, better estimates of coupling efficiency can help distinguish among AGN feedback models, investigate the redshift evolution and mass dependence of feedback efficiency.