Date of Award


Degree Name

Doctor of Philosophy



First Advisor

Dr. C. James Hueng

Second Advisor

Dr. Matthew Higgins

Third Advisor

Dr. Joseph W. McKean


This dissertation collects three empirical essays on inflation and economic growth. The first essay examines the impact of inflation uncertainty on the level of inflation. Uncertainty is measured by the conditional variance of inflation, and inflation is modeled as a GARCH-in-mean process with a two-regime Markov-switching coefficient on uncertainty. Using a Bayesian estimator with the Markov Chain Monte Carlo approach, we find that the impacts of uncertainty on inflation are statistically significant and have different signs in different regimes for the U.S. postwar data. The regime switching nature of the inflation process can explain the contradictory theoretical predictions and empirical evidence shown in the existing literature.

In the second essay, we develop a time-varying parameter model with survey information to forecast future inflation rates. To capture the inflation dynamics, we first specify quarterly U.S. inflation as an AR(2) process with time-varying unobservable parameters. The model is estimated by the Kalman filter algorithm. We then examine survey data information by combining Survey of Professional Forecasters (SPF) forecasts into the model. Compared to the survey data and the classical ARIMA models, the time-varying parameter models significantly reduce out-of-sample prediction errors. We find that the improvement lies over the time-varying feature of the models and that including survey data does not significantly improve predictive ability, indicating that the survey data do not contain too much information beyond realized inflation rates.

The third essay re-examines the convergence hypothesis by revising a four-step procedure of the panel unit root test suggested by Evans and Karras (1996). We use data on output for 24 OECD countries over 40 years. We incorporate the spatial autoregressive error structure into a fixed-effect panel model to account for spatial dependence that may induce significant size distortion of the conventional panel unit root tests. Bootstrap procedures are employed to find related critical values. In contrast to the results obtained from the test that does not consider the spatial error structure, our results indicate that there is no output convergence among the OECD countries.

Access Setting

Dissertation-Open Access

Included in

Economics Commons