Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

R. Carter Hill


The dissertation addresses the issues of small sample properties of estimators and predictors. Economic analysis usually relies on the asymptotic properties of estimators and predictors which may not be the same as their asymptotic counterparts. Furthermore, some biased estimators and predictors used in economic studies have certain asymptotic properties which are not fully understood. Consequently, sampling techniques are used to explore the small sample properties and construct confidence intervals for predictors and estimators. In the dissertation, first, Monte Carlo experiments are used to find an appropriate estimation procedure for a system of simultaneous equations which involves a latent endogenous variable. Second, Monte Carlo experiments are used to explore the small sample property of the 'equity estimator' and compare it to the small sample properties of the 'traditional' estimators. Third, bootstrap sampling techniques is utilized to construct confidence intervals for the out-of-sample forecasts obtained via biased predictors which cannot be constructed in the usual way. The findings are (1) an instrumental variables approach is an appropriate alternative estimation technique of the system of simultaneous equation involving a latent endogenous variable; (2) the small sample of the equity estimator is dependent on the vector lengths and the conditioning of the data; and (3) bootstrap method produces reasonable confidence intervals for out-of-sample forecasts.