Previously, I took a look at the JER for a variety of R's:

https://drupal.star.bnl.gov/STAR/blog/dmawxc/update-08072018-run-9-embedding-jet-energy-resolution-r-02-03-04-and-05

And in the process, realized that the matching efficiency was abnormally low for R = 0.4:

So as a first-order check, I looked into the qT and dR distributions I use to select the best match for a given particle jet.  The details of those quantities can be found in these posts:

https://drupal.star.bnl.gov/STAR/blog/dmawxc/update-06052018-run-9-embedding-new-matching-algorithm
https://drupal.star.bnl.gov/STAR/blog/dmawxc/update-06272018-run-9-embedding-applying-max-qt-cut

And here are the distributions:

All four look reasonable.  Here's the percentage of jets lying in the region I cut on when matching particle to detector jets (i.e. lying in qT = 0.15 - 1.85 and dR < Rjet):

So the qT, dR cut doesn't seem to be the issue...