Date of Degree
Wim P. Vijverberg
This thesis proposal consists of three essays on the estimation methods and applications of spatial econometric models and one essay on the generalized autoregressive conditionally heteroskedastic (GARCH)-type models in financial time series. The first essay discusses the heteroskedasticity robust generalized method of moments estimator (RGMME) for the spatial models that allow for spatial dependence in both the dependent variable and the disturbance term (SARAR(1,1)). First, we show that the maximum likelihood estimator (MLE) is generally inconsistent in the presence of unknown heteroskedasticity. Then, we extend robust GMM approach in Lin and Lee (2010) to SARAR(1,1). The large sample properties are rigorously studied and presented for the RGMME. Through a comprehensive Monte Carlo study, we compare the finite sample properties of the RGMME with some other estimators proposed in the literature.
The second essay focuses on the GMM estimation of the spatial autoregressive models which impose a moving average process for the disturbance term (SARMA). We extend the best GMM estimator (BGMME) of Liu et al. (2010) to the SARMA models and provide the best set of instruments for the SARMA(1,1) and the SARMA(0,1) specifications. The large sample properties are rigorously studied and presented for the BGMME. The finite sample properties are investigated through an extensive Monte Carlo study. To confirm our results from the Monte Carlo study, we replicate the results for the SARMA(1,1) specification in Behrens et al. (2012) in an empirical illustration.
The third essay investigates the effect of foreign direct investment (FDI) on economic growth through a spatially augmented Solow growth model. The current literature on the relationship between FDI and economic growth uses canonical cross-country growth regression specifications that are derived from the textbook Solow growth model for closed economies. We claim that these specifications cannot reflect the relationship between economic growth and FDI, because they model each country as an isolated island that does not interact with the rest of the world. On the other hand, a spatially augmented Solow growth model allows
for technological interdependence among countries through spatial externalities. The
modified growth model yields regression specifications that properly account for spatial autocorrelations. We construct a panel of 85 countries for the period 1980-2010 and estimate the modified specifications with the tools from spatial econometrics. Our findings indicate that FDI inflows have a significant positive effect on the growth rate of host countries.
The final essay proposes a flexible distribution for the maximum likelihood estimation
of the GARCH-type time series models. The new distribution can better account for the potential skewness and leptokurticity in the driving noise sequence. We study the large sample properties of the new estimator following the methodology presented in Francq and Zakoïan (2004). To investigate the finite sample properties of the new estimator, we first conduct a Monte Carlo study. Furthermore, to test the relative out-of-sample predictive power of the new estimator, we test for its prediction power on two data sets using the methods described in White (2000) and Hansen et al. (2003).
Taspinar, Suleyman, "Essays On Robust Estimators For Non-Identically Distributed Observations In Spatial Econometric And Time Series Models" (2014). CUNY Academic Works.