The above estimate reduces to 0.1 if the scaling relation of van der Marel (1999), which agrees better with the XRB intensity, is used, but can increase towards unity if stellar mass loss is recycled into new stars, so that a ~ 1. A mass-to-energy efficiency of 0.1 has been used but it can be 0.06 if the black hole is not spinning. or 0.42 if it becomes a maximally spinning, Kerr, black hole.
An even more extreme possibility which defines an upper limit on the efficiency relative to the final (dead) black hole mass is to assume that the black hole was maximally spinning during the accretion phase and then spun down by, say, the Blandford-Znajek (1977) mechanism. The total energy released relative to the final black hole mass allows for an order of magnitude uncertainty in and thus EAGN. Of course a high value here, which maximises EAGN / E*, overpredicts the XRB intensity unless most of the growing phase of black holes is Compton thick. It is also possible that a significant fraction of the power from an AGN is in the form of a wind and not directly in radiation. As discussed later, growing black holes may be both Compton thick and powering winds. If this is correct, then EAGN / E* may be significantly higher than the estimate in the last section.