Your cart is currently empty!
Fannie Mae and the Mortgage Bankers Association (MBA) publish economic & housing finance forecasts showing quarterly projections for key indicators for the forthcoming 24-48 months. This includes projections for total (agency and non-agency) first-lien, closed-end 1-to-4 single family mortgage originations, by original loan balance and loan count. These forecasts are updated monthly, including adjusting historical actuals once HMDA data is released the following year.
This article compares the Fannie Mae and MBA forecasts with the 2023 actual volumes from the agency capital markets data. This data from Fannie Mae, Freddie Mac and Ginnie Mae is available monthly, and represents 71.5% of all first-lien, closed-end 1-to-4 single family mortgage originations.
A second article will be published after the banks and credit unions release their financial disclosures which will provide actual volume estimates for a further 26.4% of first-lien, closed-end 1-to-4 single family mortgage originations. The remaining originations of 2.5% is from private label securitization.
Mortgage technology companies and service providers rely on the Fannie Mae and MBA housing finance forecasts as a baseline for annual budgeting, and planning the appropriate resourcing and investment, but struggle to compare these forecasts with actual loan volumes in a timely and consistent manner. To increase complexity, the baseline for historic actuals and future projections frequently changes.
Fannie Mae and MBA Forecasts, through 2023
Below is a summary of the published Fannie Mae and MBA housing finance forecasts for first-lien, closed-end 1-to-4 single family mortgage originations throughout 2023 – showing January 2023, April 2023, July 2023, October 2023 – and concluding in January 2024 with the recently published forecast. This shows original loan balance in $BN, as published and updated by Fannie Mae and MBA.
The chart below shows MBA and Fannie Mae reduced their projections for Q4 2023 by 26% to $399BN and 25% to $328BN respectively, with Fannie Mae’s estimate 18% lower than MBA for Q4 2023, and 8.6% lower across 2023 in the latest January 2024 publications.
When HMDA data is released in early Q2 2024, Fannie Mae and MBA will reconcile actuals against this data source, so this is still an estimate of the total originations for 2023. More information is available in this recent article on HMDA data.
MtgeFi Mortgage Originations Model
The newly released MtgeFi mortgage originations model allows customers to reconcile these industry housing finance forecasts with aggregate industry actuals on a quarterly basis, and to review progress against business plans and adjust annual budgets as needed.
The chart below shows a high-level comparison of loan count in (‘000s), published by MBA and estimated for Fannie Mae (from average purchase and refinance loan amounts), compared with known agency issuance and portfolio lending by banks and credit unions for 2023.
Based on the variance between housing finance forecasts versus actuals for 2023, the model provides the ability to adjust projected loan volumes through 2024 and 2025. Adjustments are then made throughout the year to reflect quarterly loan volumes as they are recorded from agency issuance and bank and credit union disclosures.
MtgeFi provides these actuals at an aggregate industry level, and also by individual lender, securitization month, origination month, agency, institution type, loan purpose, origination channel and property valuation method.
Further measures available upon request include property type, original and unpaid balance, loan product type, amortization schedule, agency specific programs, State, servicer, first-time buyer, mortgage insurance (MI), remote online notary (RON), jumbo, ARM and various borrower credit and security performance characteristics.
When the HMDA data is released later this year, the Total Reconciled Loans will be updated and released. The HMDA 2022 data has been fully reconciled and validated against agency issuance and bank and credit unions actuals.