Error rates depend on many factors including scanner model, population, number of fingerprints per user, enrollment strategy (1 sample vs. more samples of the same finger, quality check, etc), quality of fingerprints (wet, dry, dirty fingers).

We do statistical evaluation on datasets from optical scanners (touch, large area). IEngine_CharacterizeScore (not available in IDKit version 3.0.6 and higher) contains results (correspondences between similarity threshold, FRR and FAR) from such statistical evaluation.
However, as error rates may depend on various factors, sometimes non-related to fingerprint algorithm, we always recommend software integrators to conduct tests on their own data.