BLOQUE III: ELEMENTOS CONCEPTUALES DEL ESTADO DE LA CUESTIÓN:
1. Innovación
1.2.7. Pedagogía de las múltiples inteligencias
The empirical analyses presented in this chapter provide little evidence supporting my hypotheses. Below, I discuss several possible reasons for the lack of evidence supporting my hypotheses. I plan to explore these further in future work.
First, it is possible that transparency as defined in this thesis does not affect bank lend- ing decisions, or that it affects them differently from my predictions. Transparency refers to the financial reporting transparency of securitization. I hypothesized that transparency affects lending decisions though its effect on bank stakeholders’ (shareholders’, creditors’, and regulators’) ability to monitor bank risk-taking. I did not consider how investors in the securitization vehicles may have affected bank lending decisions, because I expect securiti- zation accounting standards to have little effect on their decision to invest in asset-backed securities and the expected returns they demand to do so. I conjecture that securitization investors are more likely to rely on information about the composition of the securitized assets and the structure of the transaction, which are not included in great detail in annual reports or FR Y-9Cs. However, it is conceivable that the securitization investors and their investment and pricing decisions play a primary role in affecting bank lending decisions, resulting in little or no room for the other stakeholders considered in this thesis.
Second, it is possible that the research design lacks power to identify the hypothesized effects. For example, the propensity score matching may not lead to otherwise identical treatment-control pairs. In section 5.5, I performed additional analyses to test the sen- sitivity of my results to alternative assumptions for the propensity score model, to an alternative matching technique, and to an alternative research design that does not rely on any matching method. The results, while different in some cases, lead to similar con- clusions about my hypotheses, suggesting that the lack of results is likely not driven by the specific research design.
Third, the accounting pronouncements may not affect transparency, or not in the direc- tion I expected, and the transparency implications of the pronouncements may vary across stakeholders. For example, contrary to my expectation, if FAS 125 introduced consistency to securitization accounting, the benefits of that consistency may outweigh the negative implications that arise from the financial components approach and QSPEs, resulting in enhanced transparency in the post-FAS 125 period. For FAS 140, the negative effects of the mandatory non-consolidation of QSPEs, which can allow easier off-balance-sheet treat- ment, may overshadow the positive effects of enhanced transparency. Consistent with this explanation, I find results supporting H1 when I limit my analyses to comparing the least transparent period (pre-FAS 125 or pre-FAS 140) to the most transparent (post-FAS 166 & 167) period.
Last, the risk measures may not capture the desired underlying construct of bank risk- taking in mortgage lending. In section 5.5, I consider an alternative definition of the risk measure, which excludes components with weak associations with BHC-level mortgage performance measures in the validity tests. The results and conclusions remain similar when using the alternative risk measure. However, I believe that this does not rule out the possibility that the proposed risk measures lack power to identify annual changes in bank lending decisions.
In summary, I find little evidence supporting my hypotheses and I identify several explanations for this apparent lack of results. I believe that further research is needed: (i) to develop better mortgage lending risk measures, (ii) on the effect of the securitization accounting pronouncements on transparency, and (iii) on the mechanisms through which transparency can affect bank lending decision. I find some evidence consistent with (ii).
Chapter 6
Large Securitizing BHCs
6.1
Introduction
This chapter examines the lending behavior of the five large securitizing BHCs, hereafter top-five BHCs, that are excluded from the main sample, focusing on potential differences from the BHCs included in the sample. The top-five BHCs are among the largest and most complex BHCs, ranking in the top 20 banks by total assets throughout my sample period 1993-2015. These BHCs also dominate the mortgage securitization market in the United States, with their market share ranging from 40% in 1998 to 91% in 2008. I excluded the top-five securitizing banks from the main sample because they are too different from the other banks to find matching non-securitizing banks within a reasonable tolerance. For instance, I fail to find matching control observations for most of the top-five BHC-years using propensity score matching with a caliper width of 25%, which is considerably wider than the 10% caliper I use in the main analyses in chapter 5, and the 1-3% caliper typically used in the literature. In addition, the large jump in securitization activity in 2001, most of which can be traced to these five banks, coincides with the effective date of FAS 140, and with the new regulatory disclosure requirements. As a result including these BHCs in the test sample may affect my ability to make inference around FAS 140.
I dedicate this chapter to examining these BHCs separately because they have a signifi- cant market share in the securitization market and in the banking industry. Moreover, the top-five BHCs are those most likely to benefit from explicit and implicit guarantees from regulators and the government that are afforded to too-big-to-fail banks. These guarantees can increase their appetite for risk-taking, making these banks likely to take advantage of the opacity of securitization to increase their risk-taking in mortgage lending.
The remainder of this chapter is organized as follows. Section6.2presents a background discussion about the top-five BHCs, and a brief review of the literature on too-big-to-fail banks. Section6.3 contains empirical analyses. Section 6.4 concludes the chapter.