No structural surprises are presented in this paper, which contains main results, a relative application, and proofs. In general, however, the author lets his work do the talking. The author proves rates of strong convergence of M-estimators in stationary autoregressive models, with an autoregression function that is not necessarily smooth, but is Lipschitz-continuous. The properties presented are proven for the first time in connection with autoregressive models.

Liebscher promotes his contribution by placing it in the context of earlier works in the field, in a precise literature review. His references to earlier work are found in all sections of the paper, pointing out extensions, contributions, and differences. It is very interesting to see a variational principle from stochastic optimization theory, proved by Shapiro [1], employed in the proof process. Surely, there are several problems in estimation theory whose proofs can also follow this approach. As if all this were not enough, the author then treats us to an additional section where continuous threshold models provide a sound paradigm for the applicability of the results.

This paper could draw potential readers from a variety of disciplines. They will find in it common points of interest, new ways to approach problems, and a sound tool.