Panel Smooth Transition Model with Covariate-Dependent Thresholds and its Application to the Nexus between Investment and Cash Flow

Abstract This paper introduces a panel smooth transition model with a covariate-dependent threshold (PSTCT). To overcome the difficulty in estimating threshold parameters of covariate-dependent thresholds, we develop a new algorithm based on the within-group transformation and simulated annealing (SA) technique. We also suggest test statistics for threshold effect, threshold constancy and cross-section dependency in the proposed model. A variable selection procedure is suggested to choose the covariates that affect the threshold. An extension to the setting with multiple regimes is also discussed. Monte Carlo simulations are conducted to investigate the finite sample properties of the proposed estimation and testing procedures. The model is illustrated by revisiting the relationship between firms’ investment and cash flow using the dataset of González, A., T. Teräsvirta, D. van Dijk, and Y. Yang. 2017. “Panel Smooth Transition Regression Models.” CREATES Research Paper . Comparing the proposed model with PSTR model of González, A., T. Teräsvirta, D. van Dijk, and Y. Yang. 2017. “Panel Smooth Transition Regression Models.” CREATES Research Paper , we find different empirical results after incorporating the covariate-dependent thresholds setting, highlighting the importance of splitting the sample based on a number of indicator variables in the cash flow/investment relationship.