Non-agency Mortgage-backed Securities Research Articles

Does Securitization Weaken Screening Incentives?

AbstractWe test whether lenders’ screening incentives weaken when faced with the possibility of loan sales. We adopt a new measure of lending standards, the processing time for mortgage applications at the loan level, and use the collapse of the nonagency mortgage-backed securities issuance market as a natural experiment. Secondary market liquidity for nonconforming loans decreased significantly at the end of 2007, but the market for securitizing conforming loans did not experience the same breakdown. Following this event, lenders spent significantly more time screening applications for loans larger than the conforming loan limits than for those below the limits. The processing-time gap widened more for banks with lower capital, greater involvement in the originate-to-distribute model, and larger assets.

The Mortgage Meltdown and the Failure of Investor Protection

The aftermath of the US mortgage meltdown of the late 2000s revealed the biggest failure of investor protection since the Great Depression: the securities laws did not allow investors in non-agency mortgage-backed securities (MBS) to recover losses on the securities even though there had been material misstatements and omissions when the securities were sold. The key issue was the short time limit for starting a lawsuit under those laws. Many investors did not become aware of the nature of their claims until after the time limit had expired. A lot of investors pursued alternative legal theories, such as contract claims, but short time limits were also an issue there. The mortgage meltdown produced losses on the order of $1 trillion (±20%), most of which were borne by non-agency MBS investors. The underlying cause of those losses was an industry-wide breakdown of mortgage lending and securitization practices from around 2005 through 2007. The nature of that breakdown was not fully revealed until many years later through investigations by the US Department of Justice, which revealed the nature and extent of the problems in a series of major settlements starting in 2013. <b>TOPICS:</b>MBS and residential mortgage loans, financial crises and financial market history, CMBS and commercial mortgage loans <b>Key Findings</b> • The mortgage meltdown produced $1 trillion (±20%) of losses from 2007 through 2016. The losses were borne primarily by investors in non-agency mortgage-backed securities (MBS). • The cause of the mortgage meltdown was an industry-wide breakdown of loan origination and securitization practices. The breakdown of practices was not disclosed to non-agency MBS investors when they made their investments. • Laws intended to protect investors from misstatements or omissions in the sale of securities did not work effectively. The maximum time limit for starting lawsuits to recover losses was very short. In most cases, the time limit expired before investors even learned the relevant facts about how they had been misled.

Securitization and Screening Incentives: Evidence from Mortgage Processing Time

We test whether lenders' screening incentives weaken when faced with the possibility of loan sales. We adopt a new measure of lending standards, mortgage application processing time at the loan level, and use the collapse of the non-agency mortgage-backed securities issuance market as a natural experiment. The event significantly reduced liquidity for non-conforming loans, but had little impact on conforming loans. Following the collapse, lenders spent significantly more time screening applications for loans larger than the conforming loan limits than those below. The processing time gap widened more for banks with lower capital, greater involvement in the originate-to-distribute model, and larger assets.

Decisions on Bond Issuances by Banks

Next to deposits, European banks have, unlike their US counterparties which heavily rely on securitization to fund mortgages, historically largely used bank obligations such as covered bonds and senior unsecured bonds. We assess the decision of 402 European banks between 2006 and 2014 to issue three types of bonds – (private label) mortgage backed securities (MBS), covered bonds (CB) and senior unsecured bonds (SUB) – to account for balance-sheet substitution effects and their underlying channels. We estimate conditional probit and tobit models to simultaneously account for the choice between the three securities implying dependence between the decisions to issue either instrument. We show evidence for substitution through a funding liquidity channel; banks with more liquid balance sheets would choose to issue less CB and MBS whereas banks with high balance sheet maturity mismatch would choose to issue more MBS. There are substitution effects also for banks which more heavily rely on deposits issuing less CB and SUB. Overall, there is a higher probability to issue non-agency MBS in Europe by banks with high reliance on deposits, less risky loan portfolios, and high maturity mismatch prompting to a different role of this instrument in European bank funding as compared to the US. Most of the above effects vanish for the crisis period suggesting that banks’ funding decisions may have been driven by unobserved factors.

The effects of SFAS 157 disclosures on investment decisions

This paper examines whether public bank managers change both the composition and classification of their investment portfolios after SFAS 157. We first show that non-agency mortgage-backed securities (MBSNA) are the asset class most likely to be measured using level 3 inputs, which are based on unobservable information. We then find that relative to a control sample of private banks, public banks altered their investment portfolios in a manner that reduced the percentage of MBSNA holdings for which SFAS 157 disclosures are required. Taken together, this evidence is consistent with public banks attempting to avoid disclosure of level 3 assets through changes in both asset composition and classification.

Evaluation of Mortgage Credit Risk

This article describes a stochastic home price appreciation (HPA) model that can be used to obtain random paths of simulated home prices that are consistent with historical behavior and with market-implied HPA. The stochastic HPA model has been integrated with Citi’s prepayment, default, term structure, and MOATS models and can be used to calculate credit-adjusted option-adjusted spread (OAS) for non-agency mortgage-backed securities and loans. The credit-adjusted OAS model is available to users via the Yield Book. <b>TOPICS:</b>MBS and residential mortgage loans, credit risk management, factor-based models

Lending sociodynamics and economic instability

We show how the dynamics of economic instability and financial crises articulated by Keynes in the General Theory and developed by Minsky as the Financial Instability Hypothesis can be formalized using Weidlich’s sociodynamics of opinion formation. The model addresses both the lending sentiment of a lender in isolation as well as the impact on that lending sentiment of the behavior of other lenders. The risk associated with lending is incorporated through a stochastic treatment of loan dynamics that treats prepayment and default as competing risks. With this model we are able to generate endogenously the rapid changes in lending opinion that attend slow changes in lending profitability and find these dynamics to be consistent with the rise and collapse of the non-Agency mortgage-backed securities market in 2007/2008. As the parameters of this model correspond to well-known phenomena in cognitive and social psychology, we can both explain why economic instability has proved robust to advances in risk measurement and suggest how policy for reducing economic instability might be formulated in an experimentally sound manner.

Completing the Picture of Loan Modifications

One of the major lessons learned from the liquidity crisis in bond structures backed by non-conforming home loans is that analysis and credit surveillance need to be performed at the loan level. To more accurately forecast returns on existing investments, analysts must perform analytics on individual loans to predict the non-scheduled cash flows derived from the mortgages backing the trusts, rather than rely on summary statistics. Because of current economic conditions and the state of the housing markets, projections on borrower defaults, payment delinquencies, and subsequent loss severities are now the major components of non-scheduled cash flows. Loan modifications and resulting recidivism rates have obvious effects on these cash flows. This article examines the mechanics of identifying loan modifications in large data sets, understanding that there are significant issues in the timing and comprehensiveness of reporting loan modifications on the parts of servicers and trustees. The authors have developed a process for identifying modifications not indicated in trustee reports and providing better insight into all modifications that have happened in non-agency mortgagebacked securities going back to 1999. Initial findings from this project offer valuable insight into the non-conforming RMBS market and offer researchers a path to new insights, underscoring the need for better loan-level data as mortgage markets re-emerge in 2010. <b>TOPICS:</b>MBS and residential mortgage loans, real estate

Valuation of Mortgage-Backed Securities: A Portfolio Credit Derivatives Approach

Non-agency mortgage-backed securities (MBS) are typically priced and traded on discounted cashflow basis where a cashflow projection is made under a prepayment and default scenario and discounted with a discount margin (DM) that supposedly measures credit risk. Whilest simple and intuitive to traditional buy-and-hold investor base, it quickly became clear that DM was not effective for pricing and risk management, when mortgage credit rapidly grew out of usually mild credit environment and more sophisticated participants entered into the market. In particular, synthetic players' push for standardization of CDS on ABS and index products (ABX and CMBX) started to demand pricing consistency across cash and synthetics. In this paper we advocate application of portfolio credit derivative no arbitrage pricing framework to mortgage securitization products and their derivatives. To demonstrate we focus on the first toxic out of the Pandora's box into the financial system -- subprime MBS. We develop an amortizing mortgage loan model with competing arrival of random prepayment and default events following reduced form modeling of corporate default. Loan prepayment correlation, default correlation, and cross correlation can be built drawing into vast literature of corporate default correlation modeling. As both prepayment and default depend on house price appreciation and interest rate, we introduce a prepayment and default factor copula (Gaussian) to statically model these two factors. Analysis of recent vintage subprime deal structure reveals a near sequential pay waterfall which allows semi-analytical pricing of MBS. Calibration to ABX.HE benchmark is satisfactory and pricing of bespoke ABS off the index not only offers fair value data points when there is no market but also an innovative way to assess relative values. ABS CDO and TABX can be consistently priced with underlying ABS and the model naturally extends to CMBS/CMBX/CRE CDO and CLO/LCDX.