Not directly implemented, but there are a couple of thing you can try.
1. If you are interested in specific donor-acceptor pairs you can use CORRel to extract the timeseries of the distance between the atoms, and the process that with your own program.
2. The MRD option of coor anal (corman.doc) does calculate the intermittent correlation function (ie, no requirement that the "donor"/"acceptor" are within cutoff distance at all times). Something like this (not tested):
COOR ANAL SELE first-atom(s) END SITE SELE second-atom END -
FIRSTU 51 NUNIT 2 NCORS 100 IMRD 21 RRES 3.0
This should output the correlation function in the fourth column for atom(s) in the first selection around the atom in the second selection.
Note that correlation times extracted from these correlation functions are not the same thing (and possiblby more approximate) as the (exact) averages calculated by COOR HBONd. You can also get lifetime distributions from COOR HBONd, which should be related.