Modeling CMBX Prices Using Commercial Mortgage Delinquency Rates

This paper explores the relationship between the performance of commercial mortgages and the returns of Commercial Mortgage-Backed Securities (CMBS) and CMBX indices. Leveraging extensive CMBS deal data and CMBX pricing data from January 2012 to March 2024, the paper constructs a pricing model for CMBX using loan delinquencies, interest rates, and other market variables. Focusing on CMBX 6 AAA and CMBX 6 BB, the paper tests three model specifications: predicting prices with delinquency rates only, with both delinquency rates and interest rates, and with delinquency rates, interest rates, and deal and market characteristics. For each specification, three models are run: predicting prices using the previous month’s delinquencies, predicting prices using the next month’s delinquencies, and predicting prices using the current month’s delinquencies. The study’s findings suggest that while some delinquency metrics and market conditions can help predict variability in CMBX prices to a certain extent, the predictive power of the individual variables is relatively unstable across different model specifications and data samples. Adding interest rates alongside delinquencies provides statistically significantly more predictive power for the BB index, but it does not provide any benefit when predicting the AAA index. The research also reveals that AAA index prices are best predicted by delinquencies in the preceding month, while BB index prices are best predicted using delinquencies from the succeeding month. Even when isolating pre-COVID data, individual variables are not particularly helpful in forecasting prices despite an increase in model fit. The paper concludes that while some relationships between commercial mortgage delinquency and CMBX pricing are observable, they are not consistently reliable for predictive modeling.