Nonstationary INAR(1) Process with 𝑞th-Order Autocorrelation Innovation

Table 1

Bias and MSE results of 𝛼 for NSINARMA(1,1) model.

CLS
	𝜆 = 0.3			𝜆 = 5
Sample size	100	300	800	100	300	800
𝜃1=0.1
Bias(α)	−2.3411𝑒−03	−1.3253𝑒−04	−2.4503𝑒−05	−9.7709𝑒−05	1.2879𝑒−05	9.7185𝑒−07
MSE(α)	1.6443𝑒−03	5.2694𝑒−06	1.8012𝑒−07	2.8641𝑒−06	4.9761𝑒−08	2.8335𝑒−10
𝜃1=0.4
Bias(α)	−1.8837𝑒−03	−6.2751𝑒−05	−1.9073𝑒−05	−9.3204𝑒−05	1.1088𝑒−05	2.1430𝑒−06
MSE(α)	1.0645𝑒−03	1.1813𝑒−06	1.0913𝑒−07	2.6061𝑒−06	3.6884𝑒−08	1.3778𝑒−09
𝜃1=0.7
Bias(α)	−1.3353𝑒−03	−8.5133𝑒−05	−1.3719𝑒−05	−7.6709𝑒−05	2.1019𝑒−05	2.5092𝑒−06
MSE(α)	5.3488𝑒−04	2.1743𝑒−06	5.6460𝑒−08	1.7653𝑒−06	1.3254𝑒−07	1.8889𝑒−09
𝜃1=0.9
Bias(α)	−4.6077𝑒−04	−6.1094𝑒−05	6.5116𝑒−06	−2.1260𝐸−04	1.8841𝐸−05	1.3932𝐸−06
MSE(α)	6.3694𝑒−05	1.1197𝑒−06	1.2720𝑒−08	1.3559𝐸−05	1.0649𝐸−07	5.8233𝐸−10