Paper: arXiv 2610.23901

Authors: Arka P. Bandyopadhyay

Abstract

I use a deep neural network (DNN) on a proprietary panel of commercial mortgages spanning 1998–2021 to separate liquidity-constrained default from strategic default – two competing, observationally similar explanations for delinquency that are difficult to disentangle with standard parametric models. Conditional on delinquency, property income (net operating income, or NOI) is a more important determinant of the default rate than leverage (loan-to-value, or LTV), despite LTV’s status as the workhorse variable in the commercial mortgage default literature. I recover borrower motive from otherwise observationally equivalent delinquency states through a joint analysis of default rates across NOI and LTV buckets, and show that the resulting thresholds are stable across the 2008 financial crisis and the 2020–2021 COVID-19 shock to commercial real estate. Because commercial borrowers are firms rather than households, and because responsibility for a missed payment is diffused across an organization rather than borne by a single decision-maker, the moral and reputational frictions that discourage strategic default in the residential setting are attenuated – making commercial mortgages a setting in which strategic behavior should be comparatively easier to detect, and comparatively costly to ignore for investors, servicers, and regulators.

Complexity vs Empirical Score

  • Math Complexity: 6.5/10
  • Empirical Rigor: 8.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper demonstrates strong empirical rigor through its use of a proprietary dataset, DNN, and robust stress tests across financial crises. Its novelty lies in applying deep learning to disentangle default motives in commercial mortgages, challenging existing literature. The mathematical complexity is present in the DNN application, though the paper focuses more on empirical results than theoretical derivations.

Research Flowchart

  flowchart TD
    A[Research Goal: Disentangle Liquidity-Constrained vs. Strategic Default] --> B{Methodology: Deep Neural Network (DNN)};
    B --> C[Data: Proprietary Commercial Mortgage Panel (1998-2021)];
    C --> D[Computational Process: Joint Analysis of Default Rates across NOI & LTV Buckets];
    D --> E[Finding 1: NOI is More Important than LTV for Default Rate (Conditional on Delinquency)];
    D --> F[Finding 2: Recovered Borrower Motive Thresholds Stable Across 2008 & COVID-19 Crises];
    D --> G[Implication: Commercial Mortgages Facilitate Strategic Default Detection Due to Attenuated Frictions];

Cited References (90)

  1. Bank Delays in the Resolution of Delinquent Mortgages: The Problem of Limbo Loans (2015) — link
  2. Identification Is Not Causality, and Vice Versa (2017) — link
  3. An Option-Based Model of Mortgage Default (1984)
  4. Equilibrium Mortgage Default Pricing with Non-Optimal Borrower Behavior (1991)
  5. Bartik Instruments: What, When, Why, and How (2020) — link
  6. The Effect of Board Structure on Firm Disclosure and Behavior: A Case Study of Korea and a Comparison of Research Designs (2012) — link
  7. Moral Hazard versus Liquidity in Household Bankruptcy (2020) — link
  8. Measuring Expectations (2004)
and 82 more
  1. Five Facts about Beliefs and Portfolios (2019) — link
  2. Biased By Choice: How Financial Constraints Can Reduce Financial Mistakes (2018) — link
  3. Should retail investors’ leverage be limited? (2019) — link
  4. Strategic Actions and Credit Spreads: An Empirical Investigation (2007) — link
  5. Strategic Default and Equity Risk Across Countries (2012) — link
  6. The Leverage Cycle (2010) — link
  7. The Adjustable Balance Mortgage: Reducing the Value of the Put (2012) — link
  8. Optimal Put Exercise: An Empirical Examination of Conditions for Mortgage Foreclosure (2001)
  9. The Determinants of Attitudes toward Strategic Default on Mortgages (2013) — link
  10. Frailty Correlated Default (2009) — link
  11. The Determinants of Attitudes toward Strategic Default on Mortgages (2013)
  12. Classification and Regression Trees (1984)
  13. An Introduction to Statistical Learning: with Applications in R (2013) — link
  14. Multivariate Adaptive Regression Splines (1991)
  15. Interpreting Neural-network Connection Weights (1991) — link
  16. Back-propagation neural networks for modeling complex systems (1995)
  17. Data Mining of Inputs: Analysing Magnitude and Functional Measures (1997)
  18. Robust Locally Weighted Regression and Smoothing Scatterplots (1979)
  19. Applied predictive modeling (2013) — link
  20. Commercial Mortgage-Backed Securities (CMBS) and Market Efficiency with Respect to Costly Information (2008)
  21. On the Number of Linear Regions of Deep Neural Networks (2014)
  22. Multilayer feedforward networks are universal approximators (1989)
  23. Uniqueness of the weights for minimal feedforward nets with a given input-output map (1992)
  24. For neural networks, function determines form (1993)
  25. Dropout: A Simple Way to Prevent Neural Networks from Overfitting (2014)
  26. Approximation capabilities of multilayer feedforward networks (1991)
  27. The contagion effect of foreclosed properties (2009)
  28. Spillover Effects of Foreclosures on Neighborhood Property Values (2009)
  29. The Contagion Effect of Neighboring Foreclosures (2013)
  30. The Subprime Virus (2015) — link
  31. Commercial Mortgage-Backed Securities (CMBS) and Market Efficiency with Respect to Costly Information (2008) — link
  32. Adverse selection in mortgage securitization (2012)
  33. Estimating Default Probabilities Implicit in Commercial Mortgage Backed Securities (CMBS) (2009)
  34. Commercial Mortgage-Backed Securities: Prepayment and Default (2003)
  35. The Pricing of Multiclass Commercial Mortgage-Backed Securities (1996)
  36. Estimating Default Probabilities of CMBS Loans with Clustering and Heavy Censoring (2008)
  37. Determinants of Credit Spreads in Commercial Mortgages (2005)
  38. Volatility, Mortgage Default, and CMBS Subordination (2008)
  39. Differences Across Originators in CMBS Loan Underwriting (2012)
  40. Differences Across Originators in CMBS Loan Underwriting (2012)
  41. The Yield on Insured Residential Mortgages (1974)
  42. Default Risk on FHA-Insured Home Mortgages as a Function of the Terms of Financing: A Quantitative Analysis (1969)
  43. The Determinants of Default on Insured Conventional Residential Mortgage Loans (1983)
  44. The Relative Termination Experience of Adjustable to Fixed-Rate Mortgages (1990)
  45. Mortgage Terminations, Heterogeneity and the Exercise of Mortgage Options (2000) — link
  46. What “Triggers” Mortgage Default? (2010)
  47. Reducing Foreclosures: No Easy Answers (2009)
  48. Negative Equity and Foreclosure: Theory and Evidence (2008) — link
  49. The Dow Theory: William Peter Hamilton’s Track Record Reconsidered (1998)
  50. A Nonparametric Approach to Pricing and Hedging Derivative Securities via Learning Networks (1994)
  51. No Arbitrage and Arbitrage Pricing: A New Approach (1993)
  52. A Model Selection Approach To Real-Time Macroeconomic Forecasting Using Linear Models And Artificial Neural Networks (1997)
  53. Testing for neglected nonlinearity in time series models: A comparison of neural network methods and alternative tests (1993)
  54. Modelling Nonlinear Relationships between Extended-Memory Variables (1995)
  55. Adaptive Learning with Nonlinear Dynamics Driven by Dependent Processes (1994)
  56. Consumer credit-risk models via machine-learning algorithms (2010)
  57. Risk and risk management in the credit card industry (2016)
  58. Subadditive approaches in integer programming (2007)
  59. Deep learning for finance: deep portfolios (2017)
  60. Logistic Regression in Rare Events Data (2001) — link
  61. Machine Learning and Knowledge Discovery in Databases, European Conference, ECML/PKDD 2008, Antwerp, Belgium, September 15-19, 2008, Proceedings, Part I (2008) — link
  62. SMOTE: Synthetic Minority Over-sampling Technique (2002) — link
  63. Back-propagation neural networks for modeling complex systems. (1995) — link
  64. An accurate comparison of methods for quantifying variable importance in artificial neural networks using simulated data (2004)
  65. Data Mining of Inputs: Analysing Magnitude and Functional Measures (1997) — link
  66. Residuals and Diagnostics for Binary and Ordinal Regression Models: An Introduction to the sure Package (2018) — link
  67. Hybrid Gradient Boosting Trees and Neural Networks for Forecasting Operating Room Data (2018) — link
  68. XGBoost: A Scalable Tree Boosting System (2016) — link
  69. Greedy Function Approximation: A Gradient Boosting Machine (2000)
  70. Hedonic housing prices and the demand for clean air (1978) — link
  71. Consumer Credit-Risk Models via Machine-Learning Algorithms (2010) — link
  72. Measures of Diversity in Classifier Ensembles and Their Relationship with the Ensemble Accuracy (2003) — link
  73. Predicting mortgage default using convolutional neural networks (2018)
  74. A Unified Approach to Interpreting Model Predictions (2017) — link
  75. Deep Learning for Mortgage Risk (2016) — link
  76. Explaining the Housing Bubble (2012)
  77. An Empirical Model of Subprime Mortgage Default From 2000 to 2007 (2008) — link
  78. Predicting Consumer Default: A Deep Learning Approach (2019)
  79. Why Do Borrowers Default on Mortgages? (2023) — link
  80. Is COVID Revealing a Virus in CMBS 2.0? (2023) — link
  81. Work from Home and the Office Real Estate Apocalypse (2026) — link
  82. HPImaster.csv — link

0 of 90 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.