Reduced-rank matrix integer-valued autoregressive model

Integer-valued time series are widely present in many fields, such as finance, economics, disease transmission, and traffic flow. With data dimensions surging, the traditional multivariate generalized integer autoregressive (MGINAR) model faces parameter overload, poor interpretability, and structural information loss. Matrix integer-valued autoregression (MINAR) model captures row-column cross-correlations and reduces the number of parameters to be estimated. However, further growth in dimensionality causes data redundancy, which degrades the MINAR model’s performance and increases the number of parameters. To solve the limitations of the MINAR model described above, this paper proposes the reduced-rank matrix integer-valued autoregression (RRMINAR) model. Reducing rank is achieved by adding low-rank constraints to the coefficient matrices in the MINAR model, leading to RRMINAR reducing parameter quantity while incorporating matrix structure information. We develop an iterative conditional least squares estimation and analyze its asymptotic properties. Simulation results demonstrate that the proposed RRMINAR model exhibits more robust parameter estimation and higher prediction accuracy than MGINAR and MINAR models when the data structure is low-rank. Empirical analysis using criminal data validates the proposed RRMINAR model’s effectiveness and uncovers structural temporal–spatial information in criminal behavior.