Hurricane Ian, Florida, 2022 – Post-Crisis Resilience Benchmark

Hurricane Ian made landfall in Florida on September 28, 2022, as a powerful Category 4 hurricane with maximum sustained winds of 115 mph. The storm produced extreme rainfall, with Placida receiving 15 inches in just 12 hours—a 1-in-1,000-year rainfall event. Combined with catastrophic flooding and storm surge of 12 to 18 feet above ground level, Ian became the costliest hurricane in Florida’s history and the third-costliest in U.S. history overall. Despite this historic economic impact, mortgage losses remained remarkably contained.

Risk-adjusted analysis reveals that default rates in Ian’s path were 1.5 times hurricane-free/crisis-free levels, approximately one-third of crisis-period rates and significantly lower than Hurricane Irma’s 4.5x increase just five years earlier. When controlling for portfolio risk using the Spero Portfolio Risk Measure, Loss Given Default rates during Ian were uniformly lower than hurricane-free periods—approximately half the normal LGD rates. This pattern likely reflects insurance proceeds, FEMA assistance, forbearance programs, and strategic defaults where borrowers delayed payments until assistance arrived.

The net result was striking: overall net charge-off rates were 0.10 times hurricane-free/crisis-free averages (11x smaller), demonstrating that multiplying elevated defaults by suppressed LGDs yields loss rates approaching zero. This analysis confirms the mature post-crisis regulatory and support framework, including federally-backed forbearance programs, carries customers through until National Flood Insurance Program and FEMA assistance becomes available. For CECL and loss forecasting purposes, models must account for both the PD spike and LGD drop to maintain accuracy.

Hurricane Harvey, Texas, 2017 – Uninsured Exposure and Post-Crisis Framework

Hurricane Harvey struck Houston on August 25, 2017, during the city’s recovery from mid-decade oil price declines. The storm’s defining feature was unprecedented rainfall—over 50 inches in parts of southeast Texas—that stalled for approximately four days, causing catastrophic flooding far outside designated floodplains. Over 270,000 homes were impacted, with nearly 80,000 sustaining at least 18 inches of floodwater and 23,000 experiencing more than 5 feet of flooding. The scale of uninsured exposure was staggering: 80% of households impacted by Harvey lacked flood insurance, leaving most homeowners reliant on FEMA assistance capped at $33,300.

Risk-adjusted default analysis shows quarterly default rates within Harvey’s path at 97 basis points—7 times higher than hurricane-free/crisis-free times (13.5 bps) and 4 times crisis rates. This 7x increase matched Hurricane Irma’s magnitude but diverged in crisis comparison due to baseline conditions: Florida’s crisis was substantially more severe (5x hurricane-free/crisis-free rates) than Texas context (1.7x hurricane-free/crisis-free rates). Despite these elevated defaults, LGDs during Harvey were uniformly lower than hurricane-free periods, with storm-period LGDs approximately half of normal-weather LGDs.

Net charge-off analysis reveals hurricane path losses of approximately 0.62 basis points—modestly higher than Irma (0.19 bps) and Ian (0.06 bps) where flooding was contained to traditional flood zones. However, critical context shows path charge-offs were actually 0.08 bps lower than hurricane-free/crisis-free periods and dramatically lower than crisis-period rates of 2.28 bps. The combination of high uninsured exposure and limited federal assistance (FEMA explicitly stated assistance was not intended to return homes to pre-disaster condition) explains why Harvey losses were slightly elevated compared to other post-crisis storms, though still minimal in absolute terms. The rebuilding process catalyzed significant employment growth, with payroll employment expanding by 1.9% in Q4 2017.

Hurricane Irma, Florida, 2017 – Multi-Hazard Event with Zero Net Losses

Striking Florida within days of Harvey hitting Texas, Hurricane Irma made landfall in a state with strong post-recession economy and healthy housing market characterized by significant homeowner equity. Unlike Harvey’s concentrated rainfall, Irma’s impact featured widespread severe wind damage alongside flooding, creating a complex insurance claim environment. The storm’s powerful winds caused substantial structural damage from roof failures to shattered windows, which subsequently allowed rainwater penetration. This multi-hazard damage pattern created a critical insurance coverage puzzle, where properties lacking flood insurance but sustaining both wind damage and subsequent interior water intrusion faced significant coverage gaps.

Risk-adjusted default analysis reveals hurricane path default rates uniformly exceeded non-path rates across the risk spectrum. Average default rates for the hurricane path were 7 times hurricane-free/crisis-free defaults—matching Harvey’s magnitude—but only 1.2x larger than crisis period default rates compared to Harvey’s 4x. This difference reflects Florida’s substantially more severe crisis experience (5.7x hurricane-free/crisis-free defaults) versus Texas. The data shows a clear increasing convex function between the Spero Portfolio Risk Measure and default rates during hurricane-free times, validating the proprietary risk metric’s effectiveness.

Loss Given Default analysis during Irma shows rates essentially at zero when controlled for risk, with hurricane-period LGDs dramatically suppressed compared to normal times. This combination of insurance complexity (multiple pathways for financial recovery through flood insurance, homeowners policies for wind damage, FEMA assistance, and forbearance programs) alongside more effective post-crisis forbearance created the offsetting mechanism. The result: hurricane NCOs at 0.17x hurricane-free/crisis-free times and 0.01x crisis times (whereas crisis NCOs were nearly 25x hurricane-free/crisis-free periods). Overall loss rates across path, non-path, and hurricane-free observations are statistically indistinguishable from each other when risk-adjusted, providing little justification for any hurricane-related CECL overlay despite the dramatic default spike.

Hurricane Sandy, Long Island, New York, 2012 – Transitional End-of-Crisis Event

Hurricane Sandy struck Long Island, New York in late October 2012, arriving at a critical moment in the nation’s housing recovery. The broader economy remained in early, fragile stages of recovery from the 2008 Financial Crisis, with homeowner equity still depressed across the region. Unlike the robust housing markets characterizing 2017 and 2022 hurricane seasons in the South and Southeast, the New York housing market lacked momentum, pricing strength, and significant equity buffers that would later protect borrowers through subsequent disasters. Sandy’s primary destructive mechanism was extreme storm surge pushing unprecedented ocean water levels into densely populated coastal communities, with surge heights of 8 to 11 feet above normal tidal levels in most areas and reaching 13-14 feet in Staten Island.

This surge-driven saltwater flooding caused cascading damage to electrical systems, heating infrastructure, and structural foundations. In a high-cost urban environment where property values were substantial but equity constrained, this created particularly challenging recovery scenarios. Damage extended beyond individual homes to critical infrastructure including subway systems, power generation facilities, and transportation networks, amplifying economic dislocation. Risk-adjusted default analysis focusing on Long Island—where storm effects were most concentrated—shows default rates from Sandy approximately 2x hurricane-free/crisis-free times, demonstrating patterns consistent with post-crisis hurricane events rather than pre-crisis catastrophes.

When adjusted for risk using the Spero Portfolio Risk Measure, quarterly default rates within Sandy’s path show elevation compared to non-path areas, with expected convex relationship between risk measure and defaults during hurricane-free times. Consistent with other post-crisis hurricanes, LGDs during Sandy show lower levels in the storm path compared to hurricane-free periods, likely reflecting insurance payouts, FEMA assistance, and forbearance programs. The net charge-off impact appears minimal when properly risk-adjusted, with overall losses at 1.2x hurricane-free/crisis-free periods—substantially lower than Katrina’s 6-6.5x impact. This transitional event demonstrates emerging post-crisis resilience patterns, though occurring before the full maturation of regulatory frameworks evident in 2017 and 2022 events.

Hurricane Katrina, Louisiana, 2005 – Pre-Crisis Benchmark and Regulatory Failure

Hurricane Katrina, striking Louisiana in August 2005, serves as the critical pre-crisis benchmark for understanding how regulatory frameworks and loss mitigation transformed mortgage resilience. The Category 3 hurricane with 125 mph winds caused catastrophic flooding through widespread levee failures, displacing over 1 million residents and resulting in approximately 1,400 deaths. The convergence of three factors created conditions producing mortgage losses an order of magnitude higher than any post-crisis hurricane: a catastrophic natural disaster, weak underlying loan quality from the origination era, and primitive, inconsistently-applied loss mitigation tools. The regulatory environment in 2005 was fundamentally different, with weak loan quality standards, widespread low-documentation originations, and minimal equity requirements.

When defaults occurred, loss mitigation options were primitive and inconsistently applied—no standardized, federally-mandated forbearance programs existed. The government and insurance response proved wholly inadequate: the National Flood Insurance Program was overwhelmed, private insurers faced insolvency, and FEMA assistance was slow and insufficient. Homeowners facing total loss had limited recovery pathways—servicers offered little flexibility, insurance wouldn’t pay, and government aid came months or years later if at all. Risk-adjusted default analysis reveals mortgages in Katrina’s path experienced defaults 43 times higher than hurricane-free/crisis-free periods and approximately 23x higher than crisis periods, though Louisiana’s relatively modest crisis impact (3x increase from hurricane-free to crisis) reflects both storm damage and substantial population decline (New Orleans city population fell 50%, metro area fell 25%).

Loss Given Default analysis shows that even during Katrina, LGDs in hurricane path areas were lower than hurricane-free/crisis-free, crisis, and non-path periods—demonstrating the suppression pattern existed even in the pre-crisis environment. However, Katrina’s extraordinarily high default rate meant that even with lower LGDs, net charge-offs remained dramatically elevated. Hurricane NCOs exceeded hurricane-free/crisis-free NCOs by approximately 6.5x and crisis NCOs by about 2x. This compares starkly to post-crisis events: Hurricane Sandy produced 1.2x hurricane-free/crisis-free impact, Louisiana 2016 flooding tracked near hurricane-free/crisis-free levels, Harvey came in at 0.9x, while Irma (0.17x) and Ian (0.10x) registered well below normal levels. The housing market impact was severe and permanent, with entire neighborhoods experiencing wholesale property loss, collapsed values across affected parishes, and permanent displacement creating conditions where mortgage defaults inevitably converted to foreclosures and charge-offs at dramatically higher rates.

Hurricane Katrina, Mississippi, 2005 – Record Storm Surge and Pre-Crisis Losses

While Hurricane Katrina is often remembered for Louisiana’s levee failures, the Mississippi Gulf Coast experienced equally catastrophic devastation from a massive storm surge. The same Category 3 hurricane generated surge heights of 24 to 29 feet along the Mississippi coast, with a record high water mark of 27.8 feet recorded at Pass Christian—the highest storm surge ever recorded on the U.S. coast at that time. This surge-driven destruction obliterated communities from Biloxi to Pass Christian, with entire residential blocks and commercial districts simply erased and surge penetrating as much as 6 miles inland in some locations. The economic and regulatory context was identical to Louisiana: a pre-crisis environment with looser underwriting standards, lower equity requirements, and inadequate loss mitigation tools.

When disaster struck, response mechanisms failed in the same ways as Louisiana—overwhelmed insurance, slow and insufficient government assistance, and servicers with no standardized protocols for helping distressed borrowers. Homeowners in Mississippi faced identical limited recovery pathways: insufficient FEMA aid, inadequate insurance payouts, and minimal servicer flexibility. The housing market impact mirrored Louisiana’s collapse, with coastal properties experiencing total or near-total loss, values plummeting across affected counties, and permanent displacement fragmenting communities. Risk-adjusted default analysis shows hurricane-impacted areas in Mississippi experienced default rates 23.5x higher than hurricane-free/crisis-free periods and 12.5x higher than crisis period alone.

The pattern of defaults as a function of risk for mortgages in the path shows fairly strong increasing relationship, though what’s striking compared to Texas or Florida is that default rates for hurricane-free/crisis-free times don’t spike as dramatically as risk measure increases. This is partly a scaling effect but also reflects that Katrina hit before the crisis, and Mississippi didn’t see the frothy, bubbly housing market that other Sunbelt states experienced. Loss Given Default analysis confirms that LGDs in hurricane path areas were lower than in hurricane-free/crisis-free, crisis, and non-path periods. However, Katrina’s unique characteristic was extraordinarily high default rate, meaning even with lower LGDs, net charge-offs remained elevated. Hurricane NCOs exceeded hurricane-free/crisis-free NCOs by approximately 7x and crisis NCOs by 1.5x—producing losses comparable to Louisiana’s outcome and demonstrating the uniform pre-crisis regulatory failure across the region.

Louisiana Flooding, 2016 – Same Region, Zero Losses Under Modern Regulations

In August 2016, widespread flooding in Louisiana caused by an intense, slow-moving weather system provides the most compelling evidence of post-crisis regulatory transformation. The unnamed event produced extraordinary rainfall across South Louisiana: some areas received 10-20 inches over 48 hours, with hardest-hit areas experiencing 20-30 inches, and Livingston Parish recording 32 inches—designated by the National Weather Service as a “1,000-year” rainfall occurrence. This historic flooding damaged over 90,000 homes and required rescue of over 28,000 people from floodwaters, affecting many of the same parishes devastated by Katrina a decade earlier. By 2016, the post-crisis framework was fully established: mortgages originated since 2010 were subject to much stricter underwriting (Ability-to-Repay/QM rules), and the disaster response playbook including immediate forbearance and coordinated federal aid was standard operating procedure.

This event provides an ideal “apples-to-apples” comparison of major flooding in the same geographic region under vastly different regulatory regimes. Risk-adjusted default analysis shows mortgages in the flood zone experienced defaults greatly exceeding non-path areas—approximately 3-4x higher in basis points when visually inspected on risk-adjusted scatter plots. Average default rates for path areas were 4.3x hurricane-free/crisis-free levels. The flood zone default spike was actually more significant than Financial Crisis peaks in Louisiana, with the worst quarter for 2016 nearly twice the peak crisis-related default rates from 2010. This demonstrates the 2016 flood was genuinely damaging and not merely a “blip” when viewed on proper scale rather than Katrina’s overwhelming magnitude.

Loss Given Default analysis shows the critical difference: LGDs during the flood were approximately one-third of normal-weather levels when adjusted for risk. This suppression pattern, identical to other post-crisis events, created the offsetting mechanism. The 2016 flood caused severe default spike (3x normal), but the combination of higher-quality underlying mortgages and effective loss mitigation completely offset the rise in defaults. LGDs were suppressed so effectively that final net loss impact was zero—flood NCOs were one-third of rates during the crisis in Louisiana, contrasting starkly with Katrina’s impact of over double crisis rates. This same-location comparison definitively demonstrates that post-crisis regulatory framework (strict underwriting, mandatory insurance requirements, standardized forbearance, coordinated FEMA response) eliminated the catastrophic losses that characterized the pre-crisis era, even when facing similar natural disaster severity.

Methodology and Risk Adjustment Framework

Our analytical methodology employed the proprietary Spero Portfolio Risk Measure to ensure valid comparisons across different portfolios and time periods. For each event, we defined the hurricane’s geographic path and temporal duration (measured in quarters), then segmented contemporaneous loans into three comparison categories: path (directly affected areas), non-path (geographically distant areas in same state), and loans from hurricane-free periods. The analysis compared four key dimensions: (A) default rates over time to establish temporal patterns, (B) risk-adjusted default rates across categories, (C) risk-adjusted Loss Given Default rates, and (D) overall loss rates calculated as net charge-offs (PD × LGD).

We defined default consistently with standard banking practices as loans ≥90 days past due. This definition is critical for timing considerations, since major events occurring in late August or early September would not show elevated defaults or losses in third quarter data. The Spero Portfolio Risk Measure—constructed on a loan-by-loan basis and updated each quarter—provides robust, intuitive, and theoretically sound risk adjustment. While risk can be measured across multiple dimensions such as FICO, LTV, and DTI, updated values aren’t readily available for Fannie loans, necessitating our proprietary measure. We aggregated similar observations to make risk-adjusted comparisons, ensuring comprehensive portfolio or sub-portfolio assessments. Our methodology is transparent and explainable—never a black box solution—and we’ve applied similar analytical frameworks to develop effective and intuitive loss forecasting (CECL and stress testing) models using both internal bank data and external data sources.

Data source and scope: We used Fannie Mae’s publicly available mortgage loan data for all analyses. Our findings reflect the Fannie Mae Resi dataset and should not be directly applied to bank-held mortgages, portfolio loans, or other loan types without appropriate validation and adjustment for portfolio-specific characteristics. Observation counts vary across analyses, and ranges with fewer than 10,000 observations are filtered to ensure statistical robustness. This research demonstrates that for CECL and loss forecasting purposes, models must explicitly control for hurricane periods to maintain accuracy—accounting for BOTH the PD spike AND the LGD drop. Weather-related default spikes and LGD drops can distort the historical relationship between economic indicators and credit losses in models, requiring controls for (1) higher, weather-related defaults to avoid over-estimating probabilities of default and (2) lower LGDs during extreme events to avoid under-estimating prospective losses.

Key Findings and Implications for Risk Management

The transformation in hurricane-related mortgage losses between pre-crisis and post-crisis periods is fundamental and dramatic. Pre-crisis events (Katrina 2005 in both Louisiana and Mississippi) produced net charge-off rates approximately 6-7 times hurricane-free/crisis-free levels, while post-crisis events show dramatically lower impacts: Sandy reached 1.2x, Louisiana 2016 flooding tracked near hurricane-free levels, Harvey came in at 0.9x, and Irma (0.17x) and Ian (0.10x) registered well below normal levels. This represents a transformation from 6 basis points in losses to 0-2 basis points despite defaults still spiking 3-6 times during hurricane events. The offsetting mechanism—elevated defaults paired with suppressed LGDs—creates near-zero net losses in the post-crisis era.

For modern hurricanes in well-regulated portfolios with appropriate insurance requirements, the evidence shows zero difference in loss rates between storm path and non-path observations. This analysis relies on Fannie Mae data with stricter insurance requirements; lenders with more lenient standards may see higher losses, especially regarding flood insurance. The individual components still matter for CECL and loss forecasting purposes: mortgages in storm-affected areas consistently exhibit higher default rates (requiring incorporation into default modeling frameworks), while LGD rates during storm events decline by roughly the same magnitude (requiring corresponding adjustments to LGD estimation models). Getting both components right ensures accurate component-level forecasts even when net loss effect is small.

Critical modeling implications: Despite federal banking regulators withdrawing the “Principles for Climate-Related Financial Risk Management for Large Financial Institutions,” knowing your financial institution’s exposure in historical and future weather events remains critical for sound risk management. Ignoring these factors introduces material distortions into credit risk models. Models should account for both shifts in PDs and LGDs after hurricanes. While our analysis did not focus on macroeconomic aspects, hurricanes significantly impact CECL variables—Hurricane Katrina, for example, caused state-level unemployment shocks that diverged sharply from national trends, leaving lasting imprints on historical data. The evidence demonstrates that post-crisis regulatory framework (strict underwriting, mandatory insurance, standardized forbearance, coordinated federal response) has effectively eliminated catastrophic mortgage losses during major weather events, fundamentally changing the risk profile for hurricane-exposed portfolios.