Diagnostic tests for homoskedasticity in spatial cross-sectional or panel models

• Main Authors: BALTAGI, Badi H., PIROTTE, Alain, YANG, Zhenlin
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2018
• Subjects: Adjusted quasi-scores; Dynamics; Fixed effects; Heteroskedasticity; Non-normality; Martingale difference; Score tests; Short panels; Spatial effects; Econometrics
• Online Access: https://ink.library.smu.edu.sg/soe_research/2179; https://ink.library.smu.edu.sg/context/soe_research/article/3179/viewcontent/Tests_Homo_BPY_June2018F_.pdf
• Institution: Singapore Management University

We propose tests for homoskedasticity in spatial econometric models, based on joint or concentrated score functions and an Outer-Product-of-Martingale-Difference (OPMD) estimate of the variance of the joint or concentrated score functions. Versions of these tests robust against non-normality are also given. Asymptotic properties of the proposed tests are formally examined using a cross-section model and a panel model with fixed effects. Monte Carlo results show that the proposed tests based on the concentrated score function have good finite sample properties. Finally, the generality of the proposed approach in constructing tests for homoskedasticity is further demonstrated using a spatial dynamic panel data model with short panels.