In the first chapter, we focus on explaining the U.S. commercial banking failures
during the recent financial crisis. We employ the semi-parametric mixture
hazard model (MHM) with both continuous and discrete time specifications to
first, distinguish between troubled and healthy banks and second, to estimate the
probability and the timing of their failure. We combine the MHM with the stochastic
frontier model (SFM) to explore the role of managerial inefficiency on a
bank's longer term viability. We find that the discrete-time MHM which takes
the managerial inefficiencies into account fits well and dominates other competing
specifications by accurately predicting the timing of failures both in and out of the
The second chapter explores a new class of flexible cross-sectional parametric
SFMs that impose an unobservable bound on the inefficiency term. We consider
doubly truncated normal, truncated half-normal, and truncated exponential distributions
to model the inefficiencies. We extend the models to the panel data setting
and specify a time-varying inefficiency bound. We apply these models to analyze
the performance of the U.S. commercial banking industry during 1984-2009.
In the third chapter, we address the issue of the "wrong" skewness of the least
squares residuals that often arises in applied studies using the traditional SFM.
Findings of "wrong" skewness imply that the SFM is misspecified and all firms
are fully efficient. Based on doubly truncated normal distribution that displays
both positive and negative skewness, we prove that "wrong" skewness does not
necessarily imply that the SFM model is misspecified.
The fourth chapter investigates the existence of heterogeneous technologies in
the U.S. commercial banking industry through the threshold effects estimation
techniques, modified to allow for time-varying effects. We employ the total assets
as a threshold variable and determine seven distinct technology-groups.
In the fifth chapter, we describe the commercial banking data that are extracted
from the quarterly Consolidated Reports of Condition and Income (Call Reports).
We detail the construction of the key variables used in this thesis, which mainly
contain output quantities, input quantities and prices, bank-specific structural and
geographical characteristics, as well as a number of measures of risk.