Three Essays on Cross-Sectionally Dependent Panel Data Models

This thesis develops the panel data models that are designed to capture and explain observed comovements among macroeconomic/finance variables. In Chapter 1, we develop a unifying econometric framework for analysing the heterogeneous spatial panel data models with common factors. In particular, a CCEX-IV estimation procedure is developed to tackle the challenging issues of endogeneity due to the spatial lagged term and the correlation between the regressors and factors. Asymptotic properties of the proposed estimators are established and Monte Carlo simulations confirm their satisfactory finite sample performances. The proposed method is then applied to analyse the growth of UK house prices over 1997Q1-2016Q4. Chapter 2 extends the previous analysis to a dynamic framework and proposes a spatial-temporal autoregressive model with unobserved factors. An iterative procedure is developed for the consistent estimation of parameters. The properties of the proposed estimators are investigated both theoretically and via extensive Monte Carlo simulations. Moreover, we develop network connectedness measures that can track the evolving influence of any node on others at both individual and regional levels through using the diffusion FEVDs and multipliers. We finally employ the method to analyse the synchronisation of international business cycles using the data for 79 countries over 1970-2019. While the first two chapters study the conditional mean effects, we investigate the conditional distributional effects in Chapter 3. Specifically, we develop a two-step procedure for estimating the dynamic quantile panel data model with unobserved common factors. The proposed estimator is shown to be consistent and follow an asymptotic normal distribution, but it is subject to asymptotic bias due to the incidental parameters. We then apply the split-panel jackknife approach to correct the bias and confirm its satisfactory performance by Monte Carlo simulations. Finally, the proposed method is applied to an analysis of bilateral trade flows for 380 country pairs over 1960-2018.