| This thesis explores the relationship between statistical performance and simplicity in financial econometrics. The work draws upon two financially important areas: dynamic latent variables models and event study econometrics. It develops a computationally simple and promising method called the restricted efficient method of moments (REMM) to estimate latent variable models. REMM addresses situations when the distribution of latent variables conditional on the observed data is intractable, which prevents direct use of the EM algorithm and erodes its simplicity. However, the REMM framework retains the simplicity of the EM setting even in the presence of the intractability by following a very direct estimation approach that I term true quasi-maximum likelihood filtering. The theoretical foundation of REMM is formalized. It is shown that under certain, less than restrictive, conditions that characterize the filtered density REMM estimators converge to maximum likelihood estimates. Simulation results indicate that REMM performs at least as well as, if not better than, EMM, GMM, MCMC, MCL, QML, and SEM with minimal computational burden in both a finite sample MA(1) and stochastic volatility setting.; Next, the standard FFJR portfolio approach event study methodology is reviewed and compared to a more robust generalized panel regression. The appeal of the panel regression is hampered due to its difficult design. This problem is documented by its limited use in event studies. However, it is shown that after a simple variance adjustment, the FFJR approach is equivalent to a generalized fixed-effects panel regression when identical explanatory variables are shared across all units, such as a β-pricing model. So a great deal of efficiency over the standard FFJR portfolio method is achieved with a marginal addition of implementation difficulty.; Overall, the trade-off between robustness and simplicity is an important issue. The econometrician has to be aware of this delicate balance because too much of one and too little of the other creates a situation of negative marginal returns. Fortunately, this thesis demonstrates that creative solutions do exist that trade-off one ingredient for another to reposition an unfavorable balance towards optimality. |
