Biases in stellar masses of JWST high-z quasar host galaxies caused by quasar subtraction

Abstract: JWST has enabled a new era of understanding high-z galaxy and black hole evolution with more than 30 high-z quasar host galaxy detections. Many of these observations imply galaxies with black holes that are overmassive compared to their low-z counterparts. However, the bright quasar point source removal may cause significant biases in these stellar mass measurements. We develop a simulation-based inference method to disentangle the quasar host galaxy stellar mass measurements from observational biases during the point source removal. We use the BlueTides simulation to generate mock images and perform point source removal on thousands of simulated high-z quasar host galaxies, constructing corrected host magnitude posteriors. We find that JWST photometry tends to either correctly recover or modestly misestimate host magnitudes, with a maximum magnitude underestimate of 0.21 mag. With our corrected magnitude posteriors, we perform SED fitting on each quasar host galaxy and compare the stellar mass measurement before and after the correction. We find that stellar mass estimates are generally robust, or overestimated by $\leq0.3$ dex. We also find that the stellar masses of a subset of hosts (J1120+0641, J0844-0132, J0911+0152, and J1146-0005) remain unconstrained, as key photometric bands provide only flux upper limits. Understanding these biases is essential to uncovering the evolutionary pathways of high-z quasars with their hosts.