<p>This paper investigates the statistical properties of estimators for a partially linear time-varying coefficient spatial autoregressive panel data model with fixed effects. As a semiparametric extension of classical spatial panel models, this framework incorporates both time-varying and time-invariant regressors while accounting for spatial dependence and unobserved individual heterogeneity. A profile quasi-maximum likelihood estimation based on the local linear method has been proposed to estimate the constant coefficients and nonparametric time-varying coefficients. However, the theoretical properties of these estimators remain understudied. This paper establishes the consistency and asymptotic normality of both the parametric and nonparametric estimators. Monte Carlo simulations demonstrate that the proposed estimation method performs robustly under various distributions of the error term, different spatial adjacency matrices and varying values of the autoregressive parameter. Finally, an empirical application to provincial carbon emissions in China illustrates the practicality of the model and the proposed estimation method.</p>