“…In this connection, we note that for both the INAR and PAR models, a typical example of distributional law involves the Poisson distribution. Recently, however, there have been proposed several alternatives or more general specifications; see, for instance, Jazi et al (), Bouzar & Jayakumar (), Bourguinon & Vasconcellos (), Goncalves et al (), Schweer & Weiss (), Zhu () and Fokianos (). One major aspect that might lead to non‐Poisson specifications is (marginal or conditional) overdispersion or underdispersion of the data at hand, and it is conceivable to expect that such a feature might manifest itself only beyond a certain point in time, thus signalling a structural change in the observations that cannot be accommodated by a potential change in the parameters as long as the (equidispersed) Poisson law remains in place.…”