This thesis consists of two self-contained studies on the implications of lump sum costs for empirical design in the context of corporate finance. The first study analyzes the dynamics of cash holding management and empirically determines the optimal policy for cash reserves. The framework stipulates that cash management is associated with the following actions: 1) allow cash level to freely float within the range bounded by two barriers; 2) refinance back to the target value immediately when cash level hits either barrier. The endogenous pattern identified under this framework facilitates understanding of the dynamics of cash holdings since it allows to estimate both the triggers of cash adjustments, as well as the target of each component of the policy. Further, this empirical application emphasizes the importance of the adjustment cost setting which refers to the interpretation of cost types for cash refinancing (fixed or fractional). In our model, we allow both fixed and fractional cash adjustment cost, which allows to identify a number of dynamic aspects such as target and thresholds in liquidity management. Also, this study enriches the existing studies of determinants of cash holdings by demonstrating novel effects of covariates on target cash holdings such as the negative impact of cash flow (profitability) much larger than previously estimated effects of industry risk. These findings differ from those in existing studies, either in signs or in magnitude, but are fully consistent with the underlying theory. Overall, the presented research quantifies cash holding management in a dynamic double-barrier model, allows to estimate the trigger of cash refinancing, and hence enhances our understanding of determinants of cash holding policy.
The second study investigates the stickiness in credit rating. The existing literature on credit ratings typically assumes on accurate match between credit quality and agency ratings. This assumption ignores the agencies’ trade-off between the reputation among investors and the revenue from issuers when updating credit ratings. Our model controls for the adjustment cost for rating agencies, and hence explains the stickiness embedded in rating assignments. Presented tests empirically demonstrate the existence of the stickiness and its significant impact. This is the first study to explicitly model decision (partial) irreversibility in credit rating research. This paper offers therefore a different explanation of the observed rating deterioration to the upgrades becoming increasingly difficult. Lastly, this study personalizes the standard ordered-probit estimation to allow for stickiness (path-dependence). Our estimation identifies upper and lower threshold groups in which the credit quality does not match assigned ratings, and calculates the likelihood specifically based on their features.