Skip to main content

Mortgage Loan Securitization and Relative Loan Performance


We compare the ex ante observable risk characteristics, the default performance, and the pricing of securitized mortgage loans to mortgage loans retained by the original lender. In our sample of loans originated between 2000 and 2007, we find that privately securitized fixed and adjustable-rate mortgages were riskier ex ante than lender-retained loans or loans securitized through the government sponsored agencies. We do not find any evidence of differential loan performance for privately securitized fixed-rate mortgages. We find evidence that privately securitized adjustable-rate mortgages performed worse than retained mortgages, although other observable factors appear to be more economically important determinants of mortgage default. We do not find any evidence of a compensating premium in the loan rates for privately securitized adjustable-rate mortgages.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3


  1. See also Jones (2004; Dionne and Harchaoui (2003; Minton et al. (2004; Nolan (2005), and Ambrose et al. (2005).

  2. See Bubb and Kaufman (2009) for a contrasting interpretation of this cutoff rule.

  3. According to the California Association of Realtors the median single-family house price in California in 2007 was $476,000, above the prevailing conforming loan limit of $417,000 at that time.

  4. Jiang et al. (Securitization and loan performance: a contrast of ex ante and ex post relations in the mortgage market, unpublished), also link the securitization decision to eventual loan performance.

  5. We exclude mortgages securitized through Ginnie Mae from our data set.

  6. For example, a “plain-vanilla” security might stipulate that an investor is entitled to a pro-rata share of the cash flows generated by the pool of mortgages owned by the trust. Alternatively, the claims on the assets may be structured, in which case investors in different securities have different priority to the cash flows generated by the underlying mortgages.

  7. Losses due to borrower default may not necessarily be borne by investors in the private-label securitizations. The arrangers may have “over-collateralized” the structure by setting the par value of the securities below the face value of the underlying mortgages, thus building in some buffer for losses. Alternatively, the arrangers may have purchased some limited insurance to cover losses. Either way, the focus here is on the different information produced by the originator relative to the arrangers and, in turn, to the investors.

  8. In LPS’s marketing literature, they claim that their participating servicers account for about 60 % of the entire mortgage market.

  9. We measure loan-to-value and house price appreciation as ratios rather than percents.

  10. We estimated the models separately for purchases and refinances, and the results were very similar.

  11. We constructed the equity ratio and nonperforming loan ratio variables as follows. We calculated the means of each of these variables for each lender, with the mean taken over that lender’s closing dates. The low, low-mid, and high-mid equity ratio and nonperforming loan ratio variables in the regressions are dummies indicating the positions in the sample distributions of these means of the actual values of the lender’s equity ratio and nonperforming loan ratio at the time of loan closing–below the 25th percentile, between the 25th and 50th percentiles, and between the 50th and 75th percentiles, respectively. Thus, the coefficients on these variables indicate the marginal effect on the probability of securitization of moving from a value of the corresponding variable that is above the 75th percentile to a lower value. Note that it is not true that, for example, 25 % of the loans in the sample have a value of 1 for “low equity ratio.” In fact, only about 11 % do. This is because banks with higher mean equity ratios tend to originate more mortgages than banks with lower mean equity ratios and/or banks tend to originate more mortgages when their equity ratios are high relative to their own equity ratios at other times.

  12. The appearance of equal marginal effects and standard errors for low documentation in columns (vi) and (vii), accompanied by differing significance levels, is due to rounding.

  13. For example, in California at this time, many alt-A loans were labeled thus because the loan amounts exceeded the GSE’s conforming loan limits, or sometimes the borrower debt-to-income or LTV ratios were considered too high.

  14. The canonical competing risks proportional hazards model is estimated by maximizing the partial likelihood function

    $$ L(\beta) = \prod\limits_{j=1}^m \prod\limits_{i=1}^{k_j} \frac{{\rm exp}[x_{ji}(t_{ji}) \beta_j ]}{\sum_{l \in R(t_{ji})} exp[ x_l (t_{ji}) \beta_j ] }, $$

    where i denotes a history, j = 1,..,m denotes the types of termination, k j denotes the number of subjects in the data with termination type j, and R(t ji ) denotes the set of observations exposed to risk j after t periods of history. The likelihood function is “partial” in the sense that the method produces consistent estimates of the βs without a simultaneous estimation of the baseline hazard. See Cameron and Trivedi (2005) for details.

  15. Note that our measure of the current LTV is for the first lien. We do not observe combined loan to value which would include the loan principal from any second liens on the mortgaged property.

  16. There is a potential for measurement error in constructing the spread. Borrowers may choose to pay upfront “points” in order to pay down the loan rate, thus giving rise to unobserved heterogeneity in spreads.

  17. Market concentration also may affect the probability of securitization. However, we did not include market concentration in the securitization regressions because it is too correlated with MSA and year for the multinomial regressions to converge when it is included.

  18. Consistent with this result, though, Jiang et al. (Securitization and loan performance: a contrast of ex ante and ex post relations in the mortgage market, unpublished), find that differences in delinquency rates between full- and low-documentation mortgages did not appear until 2007, the end of our sample period.

  19. See also Elul (2011) for a similar result.

  20. Note that we do not posit that the act of securitization may cause a change in pricing. We only compare the pricing of loans that end up being securitized with the pricing of loans that end up being retained.

  21. Note that unlike the FRM spread, the spread on an ARM may be relative to a rate that is not risk free (e.g., LIBOR). Recall that we used the same margin rate in our construction of the current mortgage rate variable that went into the default risk modeling.

  22. The Board of Governors of the Federal Reserve System finds that the market for mortgage lending is national (Bank of America Corporation/Countrywide (order dated June 5, 2008)).


  • Ambrose BW, LaCour-Little M, Sanders AB (2005) Does regulatory capital arbitrage, reputation, or asymmetric information drive securitization? J Financ Serv Res 28(1–3):113–133

    Article  Google Scholar 

  • Ashcraft AB, Schuermann T (2008) Understanding the securitization of subprime mortgage credit. Federal Reserve Bank of New York Staff Reports (318)

  • Bank of America Corporation/Countrywide (order dated June 5, 2008) (2008) Fed Reserve Bull 94(2nd Quar):C81–C92

  • Bruskin E, Sanders AB, Sykes D (2000) The nonagency mortgage market: background and overview. In: Fabozzi F, Ramsey C, Marz M (eds) The handbook of nonagency mortgage-backed securities, 2nd edn. John Wiley & Sons

  • Bubb R, Kaufman A (2009) Securitization and moral hazard: evidence from a credit score cutoff rule. Federal Reserve Bank of Boston Public Policy Paper (5)

  • Calomiris C, Mason J (2004) Credit card securitization and regulatory arbitrage. J Financ Serv Res 26(1):5–27

    Article  Google Scholar 

  • Cameron AC, Trivedi PK (2005) Microeconometrics: methods and applications. Cambridge University Press

  • Carey M (1998) Credit risk in private debt portfolios. J Finance 53(4):1363–1387

    Article  Google Scholar 

  • Dionne G, Harchaoui TM (2003) Banks’ capital, securitization and credit risk: an empirical evidence for Canada. Working paper, HEC Working Paper No 03-01

  • Elul R (2011) Securitization and mortgage default. Federal Reserve Bank of Philadelphia working paper

  • Jones D (2004) Emerging problems with the Basel capital accord: regulatory capital arbitrage and related issues. J Bank Financ 24:35–58

    Article  Google Scholar 

  • Karaoglu NE (2005) Regulatory capital and earnings management in banks: the case of loan sales and securitizations. Working paper 2005-05, FDIC Center for Financial Research

  • Keshk OMG (2003) CDSIMEQ: a program to implement two-stage probit least squares. Stata J 3(2):1–11

    Google Scholar 

  • Keys BJT, Mukherjee K, Seru A, Vig V (2010) Did securitization lead to lax screening? Evidence from subprime loans. Q J Econ 125(1):307–362

    Article  Google Scholar 

  • Maddala GS (1983) Limited-dependent and qualitative variables in econometrics. Cambridge University Press

  • Minton B, Sanders A, Strahan PE (2004) Securitization by banks and finance companies: efficient financial contracting or regulatory arbitrage. Working paper No 2004-25, Ohio State University

  • Nolan A (2005) Sweet Basel II: the regulatory capital treatment of securitization and credit derivatives activities of banking institutions. Goodwin Proctor LLP

  • Pavel C, Phillis D (1987) Why commercial banks sell loans: an empirical analysis. Econ Perspect II(3):3–14

    Google Scholar 

Download references

Author information

Authors and Affiliations


Corresponding author

Correspondence to John Krainer.

Additional information

Both authors are from the Federal Reserve Bank of San Francisco. The views expressed are those of the authors and not necessarily those of the Federal Reserve System. We thank Larry Cordell, Diana Hancock, Christopher James, Michael Koetter, Michael LaCour-Little, Shane Sherlund, and seminar participants at the Research Subcommittee of the Basel Committee on Banking Supervision, the Federal Reserve Bank of Atlanta, and the International AREUEA meetings. We would like to thank James Gillan and Armando Franco for excellent research assistance.



Table 10 Competing risks models: Prepayment hazard
Table 11 Competing risks models: FRM prepayment hazard
Table 12 Competing risks models: ARM prepayment hazard
Table 13 First-stage securitization regressions

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Krainer, J., Laderman, E. Mortgage Loan Securitization and Relative Loan Performance. J Financ Serv Res 45, 39–66 (2014).

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


  • Mortgage lending
  • Securitization
  • Loan quality
  • Asymmetric information

JEL Classifications

  • G21
  • L11
  • D82