Download Download Preview (PDF, 866.68 KB) PDF

TitleDownload Preview (PDF, 866.68 KB)
LanguageEnglish
File Size866.7 KB
Total Pages51
Table of Contents
                            Related Literature
Bankruptcy Treatment of Qualified Financial Contracts
Hypotheses
Data and Summary Statistics
	Bank Holding Company Data
	Hand collected data on repo contracts
	Mortgage Pricing data
	Structured Mortgage Issuance Data
Research Design
	Identifying Assumption
Empirical Results
	Banks hold more securitized products after expansion of safe harbor
	Effect of safe harbor expansion on MBS yields
	Banks increase securitization activity after expansion of safe harbor
	Real effects of safe harbor expansion
		Mortgage origination
		Mortgage loan pricing
The mix of collateral in repo contracts
Robustness
	Was the increase in mortgage securitization part of an overall trend?
	Are results sensitive to quartile-based cut-off points?
	Is mortgage securitization driven by systematic variation in loan demand between treated and control banks?
	Did BAPCPA increase mortgage securitization independent of the repo-collateral-demand channel?
	Does treated group assignment matter for the observed difference in mortgage securitization?
	Are results driven by a size effect?
Conclusion
                        
Document Text Contents
Page 1

The Securitization Flash Flood

Kandarp Srinivasan∗

November 20, 2016

Abstract

What caused the
ood of securitized products in the years immediately preceding the

crisis? This paper presents evidence that demand for safe collateral in repo markets

made it attractive for �nancial institutions to issue securitized products. Using the 2005

Bankruptcy Abuse Prevention and Consumer Protection Act (BAPCPA) as a natural

experiment that shocked the demand for collateral, this paper establishes collateralized

borrowing in short-term debt markets as a contributing factor to the rise of mortgage

securitization. Hand-collected data on underlying collateral of over 900 repurchase

contracts reveals underwriters of securitized products increased use of mortgage-based

repos in the months following the law change. Highlighting an important connection

between repo markets and securitization activity, this paper draws attention to an

unintended consequence of bankruptcy law which has important policy implications.

∗Olin Business School, Washington University in St.Louis. Author can be reached at kan-
[email protected] I thank my advisor Radhakrishnan Gopalan, members of my dissertation com-
mittee Anjan Thakor, Mark Leary, Jennifer Dlugosz, Xiumin Martin, seminar participants at the Midwest
Finance PhD Symposium, Phil Dybvig, Taylor Begley, Janis Skrastins, Todd Gormley, Walter Theus - Se-
nior Chapter 11 Attorney - Office of the General Counsel, PhD students in Finance and Accounting at Olin
business school for helpful comments and suggestions. All errors are my own.

1

Page 2

Introduction

The traditional view of securitization is that of transferring risks out of a bank’s balance

sheet. But events during the �nancial crisis of 2007-2009 were puzzling because risks ended

up back on banks’ balance sheets: U.S banks faced large (>$40bn) write-downs on their

exposures to AAA-rated mortgage-backed securities (He et al. [2010], Vyas [2011], Beltran

et al. [2013]). Why did banks hold large quantities of AAA-rated MBS on their balance

sheets in the run up to the crisis?

Existing evidence (Erel et al. [2014]) suggest a strong correlation between the holdings

of AAA-rated tranches and the securitization activity of banks. Banks that were active in

securitization before the crisis also held AAA-rated securitized products on their balance

sheet. While Erel et al. [2014] take an important step in documenting this correlation, their

evidence leaves a fundamental question unanswered: What explains thedramatic rise in

mortgage securitization activity (and holdings of MBS products), speci�cally in the years

immediately preceding the crisis?1

The main �nding of this paper is that changes to repo collateral demand in the pre-crisis

period was a contributing factor to the securitization
ash
ood. Securitized products are

used as collateral in short term debt (repurchase) markets. A typical repo contract involves

the sale and repurchase of safe, liquid collateral. Senior tranches of securitized products

are attractive as collateral because of their safety and liquidity properties. This paper �nds

banks respond to a shock to repo collateral demand by a) increasing their holdings of AAA-

rated securitized products and b) increasing supply of these products via securitization

of mortgages. The interesting relationship between repo markets and securitization has

been hypothesized before (Acharya et al. [2010], Gorton and Metrick [2012a], Nadauld and

Sherlund [2013]), however systematic evidence linking the two is surprisingly missing.

I address this gap by exploiting a natural experiment (the Bankruptcy Act of 2005)

which introduced a shock to repo collateral demand for a speci�c asset class - mortgage-

backed securities. For a subset of banks, this shock led to a di�erential increase in repo

collateral demand. For these banks, I document greater holdings of highly rated tranches

as well as greater mortgage securitization activity in a di�erence-in-di�erences setting.

1Mortgage-backed issuance increased over 250% between 2004Q4 and 2007Q1. See Figure 1.

1

Page 25

Appendix

Variable definitions

• Treated: A dummy variable that takes a value 1 if a bank’s trading liabilities (as a

percentage of total liabilities) lies in the top 25% of the distribution.

• Trading Liabilities: Item 3548 (BHCK series) de�ned as \Trading Liabilities, Total".

This item includes liabilities for short positions (Equity, Debt and other securities)

and derivatives with a negative fair value.

• Total Liabilities: Item 2948 (BHCK series) de�ned as \Total Liabilities and Minority

Interest". This item includes the sum of all liability items including subordinated

notes and debentures.

• Repo Exposure: Item B995 (BHCK Series) de�ned as \Securities Sold Under Agree-

ments to Repurchase".

• Tier 1 Capital : Item 8274 (BHCK series) de�ned as \Tier 1 Capital Allowable under

the Risk-Based Capital Guidelines".

• Non-Interest Income: Item 4079 (BHCK series) de�ned as \Total Noninterest In-

come".

• Highly Rated Residual : Replicates the construction in Erel et al. [2014], reproduced

in Table A.1 for convenience.

• Mortgage Securitization Activity: Item B705 (BHCK series) de�ned as \Outstand-

ing Principal Balance of Assets Sold and Securitized with Recourse or Other Seller-

provided Credit Enhancements - 1-4 Family Residential Loans"

• Mortgage Lending Activity : Ratio of total loans secured by 1-4 family residential

properties over total loans. Sum of Items BHDM1797 (revolving), BHDM5367 (�rst

liens) and BHDM5368 (junior liens) over BHCK2122 (total loans and leases, net of

unearned income).

24

Page 26

Table A.1: Highly Rated Residual construction as per Erel et al. [2014]

Variable Definition

+ BHC21754 Held-to-maturity securities, total

+ BHC51754 Held-to-maturity securities, total

+ BHC21773 Available-for-sale securities, total

+ BHC51773 Available-for-sale securities, total

- BHCK1294

Amortized cost of held-to-maturity u.s. government agency and corporation

obligations issued by u.s. government- sponsored agencies (excluding

mortgage-backed securities)

- BHCK1297

Amortized cost of available-for-sale u.s. government agency and corporation

obligations issued by u.s. government- sponsored agencies (excluding

mortgage-backed securities)

- BHCK1703
Amortized cost of held-to-maturity mortgage pass-through securities issued by

FNMA and FHLMC

- BHCK1706
Amortized cost of available-for-sale mortgage pass-through securities issued by

FNMA and FHLMC

- BHCK1714
Amortized cost of other held-to-maturity mortgage-backed securities (include cmos,

remics, and stripped mbs) issued or guaranteed by fnma, fhlmc, or gnma

- BHCK1716
Amortized cost of other available-for-sale mortgage-backed securities (include cmos,

remics, and stripped mbs) issued or guaranteed by fnma, fhlmc, or gnma

- BHCK1718

Amortized cost of other held-to-maturity mortgage-backed securities (include cmos,

remics and stripped mbs) collateralized by mbs issued or guaranteed by fnma,

fhlmc, or gnma

- BHCK1731

Amortized cost of other available-for-sale mortgage-backed securities (include cmos,

remics and stripped mbs) collateralized by mbs issued or guaranteed by fnma,

fhlmc, or gnma

- BHCK8496
Amortized cost of held-to-maturity securities issued by states and political

subdivisions in the u.s.

- BHCK8498
Amortized cost of available-for-sale securities issued by states and political

subdivisions in the u.s.

+ BHCK3536 Trading assets - all other mortgage-backed securities

25

Page 50

Figure 11: Block bootstrap assignment of treated banks

This figure plots bootstrapped t-statistics from randomly assigning a block of banks to treatment. To plot

this distribution, the difference-in-differences estimate in Table 5 (Column (1)) is run 1,000 times. The

t-statistic corresponds to the coefficient on Treated X 2006Q1 . The figure confirms the assignment to

treatment matters for the observed difference in securitization activity.

49

Page 51

Figure 12: Overlap of control branch network with homestead exemption states

This figure picks a sample control bank (SVB bank), and plots the overlap between states the passed

homestead exemption laws in the sample period and the branch network of that bank. The plot shows

significant overlap between the branch network and homestead exemption states. Since treated banks such

as Bank of America have presence in all homestead exemption states, this overlap suggests there are no

systematic differences between treated and control banks with respect to homestead exemptions.

50

Similer Documents