Does the Financial Reporting Transparency of Securitization Affect Bank Lending Decisions?

Abstract

This thesis examines the effect of financial reporting transparency for securitization on banks' mortgage lending decisions. Prior research (e.g., Mian & Sufi 2009, Keys, Mukherjee, Seru, & Vig 2010} shows that securitization reduces banks' incentives to screen and monitor borrowers. I posit that transparency plays a significant role by affecting bank stakeholders' ability to monitor and discipline bank lending decisions. I identify three specific channels for monitoring and discipline, namely corporate governance, regulatory oversight, and market discipline by uninsured depositors. I hypothesize that transparency affects bank risk-taking in mortgage lending, and that monitoring and discipline from external stakeholders moderates this effect.

I test my hypotheses using difference-in-differences tests around five FASB pronouncements relating to securitization, issued since 1996, of which one decreases and the others increase transparency. I obtain loan-level data to construct new measures of bank risk-taking in mortgage lending based on the borrower income, loan amount and property location. I validate the proposed measures using bank-level future mortgage delinquencies and charge-offs.

The main results generally do not support my hypotheses. I find that, in most cases, the effect of transparency on risk-taking is either insignificant or in the opposite direction of the prediction. These findings are robust to multiple sensitivity tests. However, I find some evidence supporting my hypotheses when comparing bank lending decisions during the least transparent period to those during the most transparent period within my sample.

As a whole, my findings support the null hypothesis that transparency does not affect banks' risk-taking in their mortgage lending decisions. This conclusion is counter-intuitive and contrary to the commonly held view that transparency promotes better stakeholder monitoring of bank risk-taking. I identify alternative explanations for the null results, including: (i) whether the accounting pronouncements affect transparency as expected, (ii) the complexity of the setting, and (iii) potential noise in the data sources and the development of my measures.

This thesis contributes to the literature in multiple ways. The proposed risk-measures might prove useful to future researchers examining risk-taking in mortgage lending. My findings are also relevant to the branches of literature examining the effect of securitization on bank lending decisions, the effect of transparency on bank risk-taking, and the real effects of accounting standards. This thesis might also be useful to standard setters and regulators in their attempt to improve financial reporting quality and to promote better decision making.