Abstract
This study considers the problem of testing a parameter change in general nonlinear integer-valued time series models where the conditional distribution of current observations is assumed to follow a one-parameter exponential family. We consider score-, (standardized) residual-, and estimate-based CUSUM tests and show that their limiting null distributions take the form of the functions of Brownian bridges. Based on the obtained results, we then conduct a comparison study of the performance of CUSUM tests through the use of Monte Carlo simulations. Our findings demonstrate that the standardized residual-based CUSUM test largely outperforms the others.
Similar content being viewed by others
References
Al-Osh, M. A., Aly, E.-E. A. (1992). First order autoregressive time series with negative binomial and geometric marginals. Communications in Statistics-Theory and Methods, 21(9), 2483–2492.
Alzaid, A., Al-Osh, M. (1990). An integer-valued pth-order autoregressive structure (INAR(p)) process. Journal of Applied Probability, 27, 314–324.
Berkes, I., Horvth, L., Kokoszka, P. (2004). Testing for parameter constancy in GARCH(p, q) models. Statistics & Probability Letters, 70(4), 263–273.
Chen, C. W., Lee, S. (2016). Generalized Poisson autoregressive models for time series of counts. Computational Statistics & Data Analysis, 99, 51–67.
Christou, V., Fokianos, K. (2014). Quasi-likelihood inference for negative binomial time series models. Journal of Time Series Analysis, 35(1), 55–78.
Csörgö, M., Horváth, L. (1997). Limit theorems in change-point analysis (Vol. 18). New York: Wiley.
Davis, R. A., Liu, H. (2016). Theory and inference for a class of observation-driven models with application to time series of counts. Statistica Sinica, 26(4), 1673–1707.
Davis, R. A., Wu, R. (2009). A negative binomial model for time series of counts. Biometrika, 96(3), 735–749.
Diop, M. L., Kengne, W. (2017). Testing parameter change in general integer-valued time series. Journal of Time Series Analysis, 38(6), 880–894.
Doukhan, P., Fokianos, K., Tjøstheim, D. (2012). On weak dependence conditions for Poisson autoregressions. Statistics & Probability Letters, 82(5), 942–948.
Doukhan, P., Fokianos, K., Tjøstheim, D. (2013). Correction to “On weak dependence conditions for Poisson autoregressions”. Statistics & Probability Letters, 83(8), 1926–1927.
Ferland, R., Latour, A., Oraichi, D. (2006). Integer-valued GARCH process. Journal of Time Series Analysis, 27(6), 923–942.
Fokianos, K., Fried, R. (2010). Interventions in INGARCH processes. Journal of Time Series Analysis, 31(3), 210–225.
Fokianos, K., Fried, R. (2012). Interventions in log-linear Poisson autoregression. Statistical Modelling, 12(4), 299–322.
Fokianos, K., Rahbek, A., Tjøstheim, D. (2009). Poisson autoregression. Journal of American Statistical Association, 104(488), 1430–1439.
Fokianos, K., Gombay, E., Hussein, A. (2014). Retrospective change detection for binary time series models. Journal of Statistical Planning & Inferences, 145, 102–112.
Francq, C., Zakoïan, J.-M. (2004). Maximum likelihood estimation of pure GARCH and ARMA-GARCH processes. Bernoulli, 10(4), 605–637.
Franke, J., Kirch, C., Kamgaing, J. T. (2012). Changepoints in times series of counts. Journal of Time Series Analysis, 33(5), 757–770.
Heinen, A. (2003). Modelling time series count data: an autoregressive conditional Poisson model. Available at SSRN 1117187.
Hudecová, Š. (2013). Structural changes in autoregressive models for binary time series. Journal of Statistical Planning & Inferences, 143(10), 1744–1752.
Hudecová, Š., Hudecová, M., Meintanis, S. (2016). Change detection in INARCH time series of counts. In R. Cao, W. Gonzalez Manteiga, J. Romo (Ed.), Nonparametric Statistics (Vol. 175, pp. 47–58).
Huh, J., Kim, H., Lee, S. (2017). Monitoring parameter shift with Poisson integer-valued GARCH models. Journal of Statistical Computation & Simulation, 87(9), 1754–1766.
Kang, J., Lee, S. (2009). Parameter change test for random coefficient integer-valued autoregressive processes with application to polio data analysis. Journal of Time Series Analysis, 30(2), 239–258.
Kang, J., Lee, S. (2014). Parameter change test for Poisson autoregressive models. Scandinavian Journal of Statistics, 41(4), 1136–1152.
Kim, H., Lee, S. (2017). On first-order integer-valued autoregressive process with Katz family innovations. Journal of Statistical Computation & Simulation, 87(3), 546–562.
Lee, S., Lee, J. (2015). Parameter change test for nonlinear time series models with GARCH type errors. Journal of Korean Mathematical Society, 52(23), 503–553.
Lee, S., Ha, J., Na, O., Na, S. (2003). The cusum test for parameter change in time series models. Scandinavian Journal of Statistics, 30(4), 781–796.
Lee, S., Tokutsu, Y., Maekawa, K. (2004). The cusum test for parameter change in regression models with ARCH errors. Journal of Japan Statistical Society, 34(2), 173–188.
Lee, S., Lee, Y., Chen, C. W. (2016). Parameter change test for zero-inflated generalized Poisson autoregressive models. Statistics, 50(3), 1–18.
Lee, Y., Lee, S., Tjøstheim, D. (2018). Asymptotic normality and parameter change test for bivariate Poisson INGARCH models. TEST, 27(1), 52–69.
McKenzie, E. (1985). Some simple models for discrete variate time series1. Journal of American Water Resource Association, 21(4), 645–650.
McKenzie, E. (2003). Ch. 16. Discrete variate time series. Handbook of statistics, 21, 573–606.
Neumann, M. H. (2011). Absolute regularity and ergodicity of Poisson count processes. Bernoulli, 17(4), 1268–1284.
Oh, H., Lee, S. (2017). On score vector- and residual-based CUSUM tests in ARMA-GARCH models. Statistical Methods & Applications: Online published.
Oh, H., Lee, S. (2018). Modified residual CUSUM test for location-scale time series models with heteroscedasticity. Accepted.
Weiß, C. H. (2008). Thinning operations for modeling time series of counts-a survey. AStA-Advances Statistical Analysis, 92(3), 319–341.
Weiß, C. H., Testik, M. C. (2009). Cusum monitoring of first-order integer-valued autoregressive processes of Poisson counts. Journal of Quality Technology, 41(4), 389–400.
Weiß, C. H., Testik, M. C. (2011). The Poisson INAR(1) CUSUM chart under overdispersion and estimation error. IIE Transactions, 43(11), 805–818.
Zhu, F. (2011). A negative binomial integer-valued GARCH model. Journal of Time Series Analysis, 32(1), 54–67.
Zhu, F. (2012a). Modeling overdispersed or underdispersed count data with generalized Poisson integer-valued GARCH models. Journal of Mathematical Analysis & Applications, 389(1), 58–71.
Zhu, F. (2012b). Zero-inflated Poisson and negative binomial integer-valued GARCH models. Journal of Statistical Planning & Inferences, 142(4), 826–839.
Acknowledgements
We thank the Editor, an AE, and two referees for their careful reading and valuable comments. This research is supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (No. 2018R1A2A2A05019433).
Author information
Authors and Affiliations
Corresponding author
Appendix : Proofs of Theorems 1 and 2
Appendix : Proofs of Theorems 1 and 2
Before proving Theorems 1 and 2, we prepare some lemmas. In what follows, we use notation \(\eta _t^0 = \eta _t(\theta _0)\) and \( \eta _t^n= \eta _t(\hat{\theta }_n)\).
Lemma 1
Suppose that conditions (A0), (A6), and (A10) hold. Then, we have
Proof
Note that
Then, using the mean value theorem and (A10), we have
where \(\eta _t^* = B^{-1}(X_t^*)\) and \(X_t^*\) is an intermediate point between \(\tilde{X}_t(\theta )\) and \(X_t(\theta )\). Hence, using (A6), we establish the lemma. \(\square \)
Lemma 2
Suppose that (A0)–(A13) hold. Then, under \(H_0\), as \(n\rightarrow \infty \),
where \(\theta _n^*\) is the intermediate point \(\hat{\theta }_n\) and \(\theta _0\).
Proof
(i) It suffices to show that
Note that, by the mean value theorem, (A10), and Lemma 1,
for some intermediate points \(X_t^*\) between \(X_t(\theta )\) and \(\tilde{X}_t(\theta )\) and \(\eta _t^* = B^{-1}(X_t^*)\). Since
according to (A6), we can show that (8) holds.
(ii) Note that
Hence, we have
Using (A6), (A7), (A9)–(A11), Lemma 1, and the following facts: \(|\tilde{U}_t (\theta )|\le |U_t(\theta )| + V\rho ^t\) and
we can see that the first term of (9) is \(o_P(1)\).
On the other hand, by Lemma 1, the second term of (9) is bounded by
which is \(o_P (1)\) due to (A7) and (A10).
(iii) Note that
Therefore, we have
Since \(B^{\prime }(\eta _t)\partial \eta _t / \partial \theta = \partial X_t(\theta )/\partial \theta \), the first and second terms of the RHS of (10) converge to 0 a.s. as \(t \rightarrow \infty \) because of (A7), (A9), and (A10). On the other hand, the fourth and fifth terms converge to 0 a.s. as \(t \rightarrow \infty \) owing to (A6), (A7), (A9), and Lemma 1. Due to (A11), we have
Henceforth, the third term converges to 0 a.s. as \(t \rightarrow \infty \) owing to (A7) and (A10).
(iv) This can be proven similarly to the proof of Proposition 5 of Lee et al. (2016). \(\square \)
Lemma 3
Suppose that conditions (A0)–(A13) hold. Then, under \(H_0\), as \(n\rightarrow \infty \),
where \(\tilde{S}_k (\theta ) = \sum _{t=1}^k \partial \tilde{\ell }_t (\theta )/\partial \theta .\)
Proof
As \(\hat{\theta }_n\) is the CMLE of \(\theta _0\), it suffices to show that
By Taylor’s theorem, we have
where \(\theta _n^*\) is an intermediate point between \(\theta _0\) and \(\hat{\theta }_n\). Thus, we have
Note that, by Proposition 1 and (iv) in Lemma 2,
Meanwhile, due to (A7), for some \(0<\gamma <1/2\),
Furthermore, since
owing to (iv) in Lemma 2, we can show that \(I_n = o_P(1)\). This asserts (11), and the lemma is validated. \(\square \)
Proof of Theorem 1
Since \(\{\partial \ell _t (\theta _0)/\partial \theta , {\mathcal {F}}_t \}\) forms a sequence of stationary ergodic martingale differences, using a functional central limit theorem, we have
where \(S_k (\theta ) = \sum _{t=1}^k {\partial \ell _t (\theta )}/{\partial \theta }\) and \(\{\mathbf B_d (s), 0<s<1\}\) is a d-dimensional standard Brownian motion. Further, from (ii) in Lemma 2, we have
Then, using Lemma 3, we obtain
This establishes the theorem. \(\square \)
Proof of Theorem 2
We consider \(T_n^{\mathrm{res},2}\) only. (The proof of \(T_n^{\mathrm{res},1} {\mathop {\longrightarrow }\limits ^{w}} \sup _{0\le s \le 1} | \mathbf {B}_1^{\circ } (s) |\) is similar to that of Kang and Lee (2014).)
We write
Noting (3), it suffices to show that
i.e., for \(i=1,2,3\),
Firstly, we express
where
Using the mean value theorem with intermediate points \(\theta _{n,1}^*\) and \(\theta _{n,2}^*\) between \(\hat{\theta }_n\) and \(\theta _0\), we have
where we have used Proposition 1 and (A7), (A10), and (A12). Since \(\hat{\eta }_t\) can be represented as \(\tilde{\eta }_t(\hat{\theta }_n)=B^{-1}(\tilde{X}_t(\hat{\theta }_n))\) with \(\tilde{X}_1 (\hat{\theta }_n)=\hat{X}_1\), we have
with intermediate point \(\theta _{n,1}^*\) between \(\hat{\theta }_n\) and \(\theta _0\), due to (A7), (A10), and (A11). Furthermore, note that \(|\hat{X}_t - X_t (\hat{\theta }_n)| \le V\rho ^t\) a.s. since owing to (A0),
Then, by using this and (A10),
Thus, (13) holds for \(i=1\).
Secondly, we express
where
Similarly to the case of \(I_{n,2}\), we have
Using Taylor’s theorem, we have
where \(\theta _n^*\) is an intermediate point between \(\hat{\theta }_n\) and \(\theta _0\), and \(\zeta _t(\theta )=B''(\eta _t)B'(\eta _t)^{-3/2} \frac{\partial \eta _t}{\partial \theta } \), so that
where
Since \(\{\epsilon _{t,1} \zeta _t(\theta _0)\}\) is ergodic, \(\sqrt{n} \Vert \hat{\theta }_n - \theta _0\Vert = O_P(1)\), and \(E|\epsilon _{t,1}| \Vert \zeta _t(\theta _0)\Vert < \infty \) owing to (A6), (A7), (A10), and (A12), we have \(II_{n,2}^{\prime }=o_P(1)\). Moreover, because
and \(E \sup _{\theta \in \Theta } \Vert \zeta _t(\theta )\Vert ^2 < \infty \) owing to (A7), (A10), and (A12), using (A3), (A6), and the dominated convergence theorem, we can have \(II_{n,2}^{\prime \prime }=o_P(1)\) (cf., Proposition 5 in Lee et al. (2016)). Thus, (13) holds for \(i=2\).
Finally, using Taylor’s theorem, we have
for some \(\theta _n^*\) between \(\hat{\theta }_n\) and \(\theta _0\). Then, similarly to the case of \(II_{n,2}\), we can show that (13) holds for \(i=3\). Hence, (12) is verified. \(\square \)
About this article
Cite this article
Lee, Y., Lee, S. CUSUM test for general nonlinear integer-valued GARCH models: comparison study. Ann Inst Stat Math 71, 1033–1057 (2019). https://doi.org/10.1007/s10463-018-0676-7
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10463-018-0676-7