Liang Xu

Abstract

This dissertation consists of three essays studying the impact of accounting quality on the design of syndicated loan contract terms and structures. The first essay examines the relation between accounting quality, as proxied by accruals quality (AQ), and the interest spreads of syndicated loans. Further, I investigate the underlying mechanisms of this relation. Specifically, I focus on two streams of theoretical works that potentially explain AQ’s impact on interest spreads. The first one is agency theory which suggests that AQ’s pricing effect is induced by the agency problem between borrowers and lenders; and the other is Yee’s (2006) theoretical conclusion that AQ’s pricing effect depends on fundamental risk (i.e., the unresolved uncertainty regarding a firm’s ability to pay dividends /interests in the future). I empirically test whether these two theories provide valid explanations for why AQ affects interest spreads by studying whether AQ’s pricing effect varies systematically with the magnitude of the agency problem and fundamental risk. I proxy the magnitude of the agency problem using the firm’s default risk and the reputation of the loan arranger, while fundamental risk is estimated using firm age, firm size and macroeconomic factors. The results show that AQ has a significant impact on interest spreads and this impact is more pronounced when the borrowing firm is subject to a higher default risk, the lead arranger of the loan less reputable and in expansion periods, but does not change with either firm age or firm size. Overall, my findings support agency theory as a reasonable explanation for AQ’s pricing effect, but cast doubt on Yee’s (2006) arguments.

The second essay is a comprehensive study of how agency problems affect the design of various contract terms and the ownership structures of syndicated loans. I focus on two potential determinants of agency costs in loan syndication: (i) the borrowing firm’s information risk (proxied by the borrowing firm’s accruals quality, audit quality and previous relationship with the lead bank) and (ii) the lead bank’s reputation (proxied by the lead bank’s market share and previous relationship with the participant banks). The findings indicate that loans issued to borrowers with higher information risk are subject to higher interest spreads, shorter maturity and a higher likelihood of collateral requirements. Conversely, loans originated by less reputable lead banks not only incur higher interest spreads, shorter maturity and a higher likelihood of collateral requirements but also have more covenants, presumably to substitute for the lead bank’s monitoring duties. Moreover, the less reputable lead banks are forced to retain larger proportions of the loan in their own portfolios and to form more concentrated syndicates.

In the third essay, I examine whether lenders are able to detect and how they respond to the borrowing firm’s engagement with real earnings management. My evidence is consistent with lenders detecting certain real earnings management activities, such as sales manipulation and overproduction. These activities are likely to result in an increase in both the borrowing firm’s information risk and default risk. Accordingly, banks charge higher interest rates and enhance security requirement to address this incremental risk. However, I do not find any statistically significant relations between the loan contract terms and the borrowing firm’s abnormal discretionary expenditures, suggesting that lenders may not be able to detect a firm’s manipulation of earnings numbers by decreasing discretionary expenditures.