Modeling Interest-Only Adjustable-Rate Mortgages with Rate and Property Scenarios
Summary
The document considers quantitative modeling of interest-only adjustable-rate mortgages. It proposes using index forward curves to project how mortgage rates may change after an initial fixed-rate period, then examining effects on borrower costs and payment obligations. The questions also include estimating loan margins from borrower and property characteristics such as credit scores and loan-to-value ratios, and stressing rates against an amortization profile.
The response recommends a quantitative model that explicitly represents loan characteristics and refers to a research paper whose model includes an interest-only ARM. Simulations from such a model can explore how internal rates of return change under different assumptions for property prices and policy rates. This is a direction for analysis, not a worked implementation: the document supplies no calibration data, inferred underwriting standards, forward-curve results, or numerical IRRs. Reliable estimates would depend on loan-level assumptions and scenario design.
Key ideas
- A quantitative mortgage model can represent borrower characteristics such as credit score and loan-to-value ratio.
- Forward index curves can support scenarios for rate resets after an ARM’s fixed-rate period.
- Rate stress scenarios can be applied to payment and amortization profiles.
- Simulations can examine how property-price and interest-rate assumptions affect mortgage investment returns.
- The suggested approach requires suitable model assumptions and data, which the document does not provide.
Tags
Full text
# Modeling Interest-only Mortgages # Modeling Interest-only Mortgages Can we infer a range of future all-in costs for I/O ARMs with current index forward curves? Essentially, just taking a worksheet like this and adding some type of ramping capability after the fixed rate period ends. I'm interested in how quickly ARM rates adjust to changes in indices, and implications for borrowers given market expectations about the future path of monetary policy decisions. On a less quantitative note, loan characteristics are hard to find; all of the premium sources at my disposal focus on capital markets. Margin is likely some function of FICO, LTV etc. Can we reliably infer some of these from publicly available data? Every underwriter has slightly different standards, but I'm sure there are white papers out there somewhere. Advanced applications would include rate stressing, i.e. putting Fitch's forward LIBOR curve into a given amortization profile and seeing how much P&I we end up owing. Changes to property values over time would be very interesting to model as well - if I take out an I/O loan, LIBOR rises in lockstep with Fed Funds, and the underlying property appreciates or depreciates ten percent, what does the IRR look like? First post on this forum - happy to be here. Please give feedback if this is off-topic so I can more meaningfully contribute moving forward. Please let me know if anything is unclear. Thanks! ## Answer by phdstudent (score 1, accepted) https://quant.stackexchange.com/a/38858 You probably should try to model it properly in a quantitative model that allows you to account for FICOs, LTVs, etc. Take a look at this paper. The authors in section 3.3 solve a model with an interest only ARM. You can use the simulations of such a model to understand how IRRs change if you change assumptions regarding how property prices change, federal funds rate, etc.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.