2016ApJ...830..150P

We consider a sample of 31 exoplanetary systems detected by gravitational microlensing and investigate whether or not the estimated distances to these systems conform to the Galactic distribution of planets expected from models. We derive the expected distribution of distances and relative proper motions from a simulated microlensing survey, correcting for the dominant selection effects that affect the sensitivity of planet detection as a function of distance, and compare it to the observed distribution using Anderson-Darling (AD) hypothesis testing. Taking the relative abundance of planets in the bulge to that in the disk, f_bulge, as a model parameter, we find that our model is consistent with the observed distribution only for f_bulge<0.54 (for a p-value threshold of 0.01) implying that the bulge may be devoid of planets relative to the disk. Allowing for a dependence of planet abundance on metallicity and host mass, or an additional dependence of planet sensitivity on event timescale, does not restore consistency for f_bulge=1. We examine the distance estimates of some events in detail, and conclude that some parallax-based estimates could be significantly in error. Only by combining the removal of one problematic event from our sample and the inclusion of strong dependences of planet abundance or detection sensitivity on host mass, metallicity, and event timescale are we able to find consistency with the hypothesis that the bulge and disk have equal planet abundance.