A test for counting sequences of integer-valued autoregressive models

Abstract

Non-negative integer-valued time series have been intensively studied for decades. Integer autoregressive (INAR) models are one of the most popular models in this field. To define the INAR model, the binomial thinning operator or, more generally, the generalized Steutel and van Harn operator guarantees the integer-valued nature. However, the distributions of the counting sequences used in the operators have been determined by the preference of analysts without statistical verification so far. The aim is to propose a test for counting sequences of INAR models and show that the test has an asymptotically correct and consistent size.

This talk is a joint work with Kou Fujimori (Shinshu University).

Please click here for participating the talk online.

Download