Thumbnail Image


dissertation.pdf (1.1 MB)
No. of downloads: 1535

Publication or External Link






The first part of the dissertation is a Monte-Carlo study of the small sample

properties of various estimators of the parameters of single equation model with

a spatially lagged dependent variable and a spatially lagged disturbance term.

We focus on the small sample behavior of the maximum likelihood estimator

(MLE) and spatial instrumental variable (IV) estimators. These IV estimators

are feasible spatial two-stage least squares (FGS2SLS) and series estimators which

were suggested by Kelejian and Prucha (1998, 2001), the best GS2SLS estimator

which was suggested by Lee (2000).

The findings indicate that the finite sample properties of the IV estimators

are almost identical. Furthermore, the advantage of the ML estimator over the

spatial IV estimators is very limited or nonexistent in most of the cases considered.

These results have important implications in terms of e.ciency and

computational feasibility of these estimators.

The second part analyses the importance of alternative channels of contagion

during the Asian, Russian and Brazilian crisis episodes. We consider four contagion

channels relating to the extent of trade, financial links through common

lenders (bank lending channel), similarity in risk, and neighborhood e.ects.

In order to assess the significance of each we apply a spatial modeling technique

to weekly stock market returns of a cross-section of countries. The paper

improves upon previous contagion studies with similar methodology in two aspects.

First, the parameters of the model are estimated by a consistent procedure.

This clearly leads to more reliable inferences. Second, we use a data set involving

a larger sample of countries. This should alleviate some of the potential sample

selection biases inherent in previous studies.

The results indicate that (a) the bank lending channel was important in all

three crisis episodes, (b) the trade channel was relevant in the Russian and Brazilian

crisis episodes, but not in the Asian crisis, (c) there is some evidence of the

risk similarity channel during the Asian crises, but not in the Russian and Brazilian

crises, (d) neighborhood e.ects were important in all three crisis episodes.

Furthermore, there is an evidence of negative trade spillovers from Japan during

the Asian crisis.