GSE Multifamily Loan Performance: National vs. NY Delinquency Trends

Executive Summary

This report analyzes multifamily loan performance across the GSEs (Fannie Mae and Freddie Mac) to answer three core questions:

1. What is the delinquency trend for the GSEs’ multifamily loans over time? Nationally, delinquencies remain relatively low (currently at 0.5%), but New York is experiencing an elevated rate in distress (currently at 1.3%) that has diverged from the national data. Furthermore, removing New York from the dataset results in a visible drop in the national average.
2. What is the trend for loans in NY? The GSEs’ exposure in New York is structurally different from the rest of the country. The active New York portfolio is skewed toward small-balance loans (median size of $6,000,000), whereas the national portfolio is concentrated in larger-scale projects (median size of $9,700,000). However, due to property values, New York carries significantly more debt per individual unit.
3. Is NY driving the national trends? Yes. New York holds a disproportionate share of the nation’s multifamily distress. Despite only accounting for 6.6% of the total active portfolio by balance, New York currently holds 16.2% of the nation’s delinquent UPB.

(Note: Data Standardization, Methodology, and further exploratory hypotheses are located in the Appendix at the end of this report).

---

What is the delinquency trend over time? (National vs. NY)

Visualizing the Trend: Full Context

Historically, multifamily GSE delinquencies generally trend together, but New York has proven vulnerable to localized shocks. The massive spike in New York delinquencies in the early 2000s (peaking in 2002-Q1) was caused by the 9/11 attacks. The attacks disrupted the city’s economy, causing immediate job losses, closing businesses, and triggering an exodus from certain NYC neighborhoods. This localized plunge in revenues resulted in a wave of delinquencies.

Visualizing the Trend: Modern Divergence (Zoomed to 4.0%)

By zooming in, we can see the scale of the current trend relative to the initial 2020 Covid shock. During the initial COVID-19 lockdowns, both National and NY rates spiked, with NY rates peaking in 2020-Q3 at nearly a full percentage point higher than the National in both Loan Count and UPB. However, as the National rates have begun to settle back down, NY’s rates have remained elevated, with Loan Count having surpassed its initial COVID-era peak.

Comparing the National Average to the Rest of the Country

Now that we’ve compared the nation directly to New York, let’s look at how the national average compares to the other 49 states when New York is removed from the equation. This helps us clearly spell out exactly how much New York’s specific distress is pulling up the overall numbers for the rest of the country.

For most of the past 25 years, including the Great Recession and the initial 2020 peak, the National and Rest of US trends overlap almost perfectly, indicating that distress was relatively uniform across the country. The widening gap on the far right side suggests that New York’s current elevated delinquencies are heavy enough to inflate the entire national average today.

Overall vs. NY Delinquency Rates

Note: The first ten observations are shown by default. To see other information use the filters/sliders.

---

What is the trend for loans in NY? (Distribution)

To understand why New York is behaving differently, we must look at the physical makeup of the loans themselves. These density plots show the distribution of loan sizes, revealing that the New York portfolio is structurally different from the rest of the country.

Total Loan Size (UPB)

Understanding the Y-Axis: The y-axis does not represent the raw percentage of total loans. Instead, it measures probability density; a heat map for clustering. A higher peak means that a larger volume of the portfolio’s loans are packed together at that specific dollar amount on the x-axis.

New York’s curve peaks to the left of the National curve. This shows that the GSE’s New York exposure is driven by a higher volume of smaller-balance loans, whereas the rest of the nation is more evenly distributed towards larger-scale projects.

Loan Size Per Unit (UPB per Unit)

When normalized on a per-unit basis, the scene flips. The New York curve sits noticeably to the right of the National curve. This shows that while New York properties account for fewer units and lower total UPB, they carry more debt per individual apartment unit than the national average. This would make New York’s housing stock more sensitive to potential increases in operating costs or other cash-flow disruptions.

---

Disproportionate Share: New York’s Footprint in the National Portfolio

Small Loans: Notice that New York’s Share of Delinquent Loans is higher than its Share of Delinquent UPB. This corroborates our earlier findings that the distress in New York is being driven by a high volume of smaller-balance loans. Rather than a few massive luxury high-rises defaulting; the defaults are concentrated among smaller projects.

Q4 2024 Drop: The data shows a drop in NY’s Share of Delinquent UPB in 2024-Q4. During late 2024, lenders exposed to NY distress – notably New York Community Bank (Flagstar) – executed loan portfolio sales, took charge-offs, and offloaded non-performing assets to Mr. Cooper, a major national mortgage servicing company, and other buyers to raise capital and reduce their risks¹. Similarly, the FDIC continued to resolve the impaired assets left over from the Signature Bank collapse². When large tranches of bad NY loans are sold to private equity or fully charged off, they are removed from the active loan pool, which shrinks the visible delinquency numerator for the quarter.

The Loan Size Factor: While New York has a high concentration of small-balance loans, further exploratory analysis (detailed in the Appendix) confirms that small loan size alone does not explain this trend. As demonstrated in Hypothesis 1, a small loan in New York defaults at a significantly higher rate than a similarly sized loan anywhere else in the country, pointing directly to region-specific conditions.

---

Appendix

Exploratory Analysis & Hypotheses

On review, several hypotheses emerged regarding the underlying drivers of the data.

Hypothesis 1: Are these trends driven specifically by Loan Size or by New York conditions?

To test if New York’s distress is solely a byproduct of having smaller loans, we binned the entire national portfolio into four equal-sized quartiles.

By comparing New York to the Rest of the US within each specific size quartile, we effectively isolate and control for the size of the loan.

If small loans were simply riskier by nature, then small loans everywhere would default at a similar rate. Because New York has a large concentration of small loans, that alone could potentially inflate its overall delinquency average. However, the data reveals this is not the case.

Even when we isolate the analysis to just the smallest loans (Q1), New York properties default at a higher rate than similarly sized properties elsewhere in the country. This elevated distress persists across Q2 and Q3 as well. This implies that region-specific conditions (such as rent stabilization laws, operating cost pressures, etc.) are impacting NY’s small to mid-market projects, rather than just the intrinsic risk of small-balance lending.

Hypothesis 2: How does the “Age of the Loan” affect these portfolios?

To test whether New York’s distress is driven by an aging, legacy portfolio, we calculated the “Loan Age” (in years) for every active loan based on its original funding date.

The data shows that New York properties are not burdened with significantly older, legacy debt. In fact, the New York distribution closely mirrors the rest of the country, with a large concentration of originations occurring in the past 5 to 7 years. The age of the loan itself does not explain the region’s disproportionate delinquency rate.

Hypothesis 3: Is the NYC Metro Area driving the State’s trend?

To identify if the NYC Metro Area specifically is driving the distress, we mapped the relevant regional identifiers and isolated the datasets.

First, we look at the combined GSE master dataset to compare the rate of distress alongside the sheer volume of active loans in each region.

As of mid-2024, the Rest of NY area delinquency rate approached 4.0%. In contrast, the portfolios for the NYC Metro and the Rest of US maintained delinquency rates closer to 1.0% or below. This shows that the Rest of NY is defaulting a higher percentage than the metro area, potentially making it the primary driver of the broader state and, effectively, national trends.

The spike in the Rest of NY region, while significant, does not represent the majority of the loans within the state. By inspecting closer the counts of the loans in both regions, we see that the metro area holds 2x the amount of loans found elsewhere in the rest of the state. The higher counts in the metro will hold more weight in the equation of which region drives the state’s trend. To explain, though, the increased rate of delinquencies within the Rest of NY, we can assume that the increased concentration of rental support programs within the NYC Metro area is helping keep more projects from falling into delinquencies.

Potential geographic discrepancy (Fannie vs. Freddie) While the combined data shows a clear trend, it does not account for the discrepancy in how the two agencies report location data. Fannie Mae assigns loans to standardized Metropolitan Statistical Areas (MSAs). Freddie Mac, however, often inputs the localized city name.

Because our aggregation logic filters for the string "NEW YORK|NYC", Freddie Mac’s downstate suburban loans (e.g., properties listed in “White Plains” or “Yonkers”) bypass the Metro filter and fall into the “Rest of NY” bucket. By splitting the graph below, we can see how the different methodologies affect tracking.

Distress Rates by Origination Date (Pre/Post Oct 2020)

This plot isolates the active portfolio based on origination dates, testing whether older, pre-pandemic debt performs differently than debt originated during or after the pandemic rate environment.

GSE Portfolio Comparison: Fannie Mae vs. Freddie Mac

This chart plots the aggregate delinquency rates over time for the Fannie Mae and Freddie Mac portfolios independently, allowing for an evaluation of how the two datasets perform relative to one another.

Delinquency Trends by Interest Rate Type: Fixed vs. Floating

This chart compares the delinquency performance of fixed-rate and floating-rate loans over time, measuring each cohort’s delinquency rate as a percentage of its own total UPB.

Floating-rate loans have historically shown greater sensitivity to economic disruptions — most visibly during the post-2020 rate environment, where rising SOFR benchmarks increased debt service obligations on variable-rate loans originated during the low-rate era. Fixed-rate loans, we see, offer borrowers payment certainty and tend to show lower and more stable delinquency rates.

NYC MSA vs. Other Blue State MSAs (High-Exodus Markets)

This chart compares the delinquency performance of the NYC MSA against other major metropolitan areas in high-outmigration blue states (California, Illinois, Washington, Oregon, and the DC metro). Each MSA’s delinquency rate is measured as a percentage of its own total UPB.

Peer markets such as Los Angeles, San Francisco, and Chicago have experienced varying degrees of multifamily stress tied to comparable dynamics: post-2020 rent regulation tightening, elevated operating costs, and outmigration-driven demand pressure.

Fannie Mae Localized Distress by Zip Code

Because Freddie Mac datasets omit zip code identifiers, this table calculates distress distributions using just the active Fannie Mae portfolio.

GSE Exposure & Delinquency in Rent-Stabilized NYC Stock

This interactive map details GSE multifamily delinquency rates with rent stabilization concentration by zip code in New York City.

Note: only Fannie Mae data is shown in this map, given that it is the only dataset that provides zip code level information

NY State Foreclosures and REO Counts (2010 - 2025)

Composition of Serious Delinquencies Over Time (National)

This stacked bar chart breaks down the underlying makeup of the serious delinquencies within the master dataset over time, utilizing Unpaid Principal Balance (UPB) to measure the proportional severity of distress across the different delinquency stages. Counts for the number of loans in each category are overlaid.

Data Dictionary: Shared GSE Variables

Core Concept	Fannie Mae Column	Freddie Mac Column	Definition
Loan Identifier	Loan.Number	lnno	Unique ID for the loan.
Reporting Period	Reporting.Period.Date	quarter	The timeframe of the current performance snapshot.
Origination Date	Issue.Date	dt_fund	When the loan was funded/issued.
Maturity Date	Maturity.Date...Current	dt_mty	The date the final loan payment is due.
Original Balance	Original.UPB	amt_upb_pch	The loan amount at origination.
Current Balance	UPB...Current	amt_upb_endg	The unpaid principal balance at the reporting period.
Property City	Property.City	geographical_region	City where the asset is located.
Property State	Property.State	code_st	State where the asset is located.
Property Zip Code	Property.Zip.Code	N/A	5-Digit zip code. Freddie Mac data omits this level of granularity.
Unit Count	Property.Acquisition.Total.Unit.Count	cnt_rsdntl_unit	Total number of apartment units.
Interest Rate	Note.Rate	rate_int	The current interest rate on the loan.
LTV	Loan.Acquisition.LTV	rate_ltv	Loan-to-Value ratio at underwriting.
DSCR	Underwritten.DSCR	rate_dcr	Ratio of net operating income to debt obligations.
Amortization Term	Amortization.Term	cnt_amtn_per	Total months the loan amortizes over.
Interest-Only Term	Original.I.O.Term	cnt_io_per	Number of months where only interest is paid.
Delinquency Status	Loan.Payment.Status / SDQ.Indicator	mrtg_status	Current payment performance or default code.
Liquidation Date	Liquidation.Prepayment.Date	liq_dte	Date the loan was paid off, sold, or liquidated.
Credit Loss	Lifetime.Net.Credit.Loss.Amount	credit_loss	Financial loss recognized by the GSE.
Defeasance	Defeasance.Date	flag_defeased	Indicates if the loan collateral was replaced with securities.
Lien Position	Lien.Position	lien_number	Seniority of the mortgage debt.

---

1. New York Community Bancorp (Flagstar) executed significant residential mortgage servicing rights (MSR) sales and loan portfolio offloads to Mr. Cooper Group in Q4 2024 to raise capital. Source: PR Newswire, “Flagstar Bank Closes on the Sale of its Mortgage Servicing and Third-Party Origination Business to Mr. Cooper,” November 1, 2024. Link: https://www.prnewswire.com/news-releases/flagstar-bank-closes-on-the-sale-of-its-mortgage-servicing-and-third-party-origination-business-to-mr-cooper-302294447.html↩︎

2. The FDIC continuously marketed and resolved joint venture stakes in Signature Bank’s rent-stabilized commercial real estate loan portfolio throughout 2023 and 2024. Source: Commercial Observer, “Signature Sale Hangover Still Felt in 2024,” January 8, 2024. Link: https://commercialobserver.com/2024/01/signature-sale-hangover-still-felt-2024/↩︎