Progress and Vulnerability Two Years After the Wildfires: A Data-Driven Update from the Maui Recovery Survey

Executive Summary

Two years after the devastating August 2023 wildfires, thousands of fire-affected people continue to navigate a long and complex path to recovery. While many are still facing hardship — including displacement, financial uncertainty, and unmet needs — others have regained stability in many aspects of their lives. This report presents findings from the Maui Recovery Survey, an ongoing monthly survey that has collected near real-time data from over 950 individuals who lived, worked, or owned a business in West Maui or Kula at the time of the fires.

Displacement continues. Most fire-affected people express a desire to remain on Maui, and many plan to return to West Maui if possible. As of mid-2025, however, about 40% of those who lived in West Maui at the time of the disaster have not returned from elsewhere on Maui or out of state. Displacement also extended beyond West Maui and Kula. Among fire-affected people who lived elsewhere on the island but worked or owned businesses in fire-affected areas, almost one-third were displaced from their homes and remain displaced to this day.

Housing recovery is uneven but moving forward. More than half of fire-affected households now live in permanent housing, a significant milestone. In addition, nearly half of temporary residents have lived in the same home for more than a year, providing them some stability to recover past community ties and form new ones. The share of people facing precarious housing conditions – such as living with family or friends or being unhoused – has also halved over the last year.

Rent burdens vary widely. About a third of fire-affected renters who now live in a different property than before the fires still receive full rental support, such that they pay little or no rent. Almost three-quarters of this group pay less rent in the second quarter of 2025 than before the wildfires. But others, especially those without formal assistance, pay rents well above pre-fire market rates, which were already the highest in the state. Another milestone in the path to recovery are signs that the rental market is stabilizing: rents for studios, one-bedroom and two-bedroom units have fallen over the last 12 months. However, rents for larger units have been more persistent, fetching nearly double their pre-fire rent. As temporary programs near their expiration dates in 2026, ensuring stable, affordable housing for all fire-affected households will remain a major challenge.

Employment and income recovery remain limited. Full-time employment is still well below pre-disaster levels for fire-affected people, with many individuals unemployed, working part-time involuntarily, or out of the labor force altogether. Income losses are widespread and the share of households living in poverty has risen substantially. However, a growing number of individuals report less need for financial assistance, pointing to greater economic stability for parts of the fire-affected population.

Support systems are shrinking, even as needs continue. Assistance from FEMA, other government entities, and community organizations has declined substantially over the past year. In the second quarter of 2025, nearly half of fire-affected individuals reported receiving no assistance at all. While this may indicate less need for assistance, as urgent needs for support have begun to decline in some areas, the reduction in assistance risks leaving the most vulnerable behind.

Overall, the findings point to an ongoing, yet uneven recovery process. While some fire-affected people continue to experience serious challenges, others are beginning to regain stability in housing, work, and daily life. The Maui Recovery Survey will continue to offer timely, data-driven insights into how recovery is progressing and where gaps remain. The data supports long-term planning efforts that center the experiences of those most impacted.

Introduction

On August 8, 2023, the Maui wildfires devastated Lahaina and some areas of Kula. The fires claimed many lives; destroyed homes, businesses, public infrastructure and cultural resources, and displaced thousands of residents. The disaster triggered a broad emergency response and set in motion a long-term recovery process involving federal, state, local, and community-based efforts.

Two years later, the recovery is still ongoing. While some residents have been able to find stable housing and regain employment, others continue to face challenges. Some remain displaced or live in temporary housing; others have experienced long-term income loss and are still seeking employment or support. The rebuilding of permanent housing is expected to take several years, and economic recovery remains uneven across communities. In this context, understanding the evolving needs and conditions of fire-affected residents remains essential to guiding effective long-term recovery strategies.

Amid this evolving landscape, reliable and timely data on how people are faring has remained limited. Most official statistics have long time lags or do not reflect the specific experiences of fire-impacted individuals. To help fill this gap, the University of Hawai‘i Economic Research Organization (UHERO), with support from the Hawai‘i Community Foundation, launched the Maui Recovery Survey: Housing & Jobs in August 2024, one year after the wildfires. This ongoing, monthly survey captures frequent information on housing status, employment, income, and other socioeconomic indicators from people who lived, worked, or owned a business in West Maui or Kula at the time of the fires.

The survey is designed to track change over time and provide insight into the lived experience of recovery. Participants can join at any point and are not required to respond every month, making the survey accessible to a broad population, including those still navigating instability. It is available in multiple languages and outreach has been facilitated by local community organizations. So far, over 950 fire-affected individuals have participated, with an average of around 430 responses per month. To help ensure that the findings have relevance beyond the survey sample, the data has been weighted to reflect the broader population of fire-affected individuals.

This report presents key findings from the first year of the Maui Recovery Survey. Using data collected between August 2024 and June 2025, it provides an updated picture of housing stability, income and employment outcomes, unmet needs, and assistance. We used statistical methods to analyze key indicators over time and identify determinants of recovery trends. The findings are intended to inform ongoing recovery efforts and ensure that the experiences of fire-impacted residents are reflected in public discussions and decisions. A detailed summary of our survey and methodology is provided in the appendix along with an overview of sample demographics.

Displacement

The August 2023 wildfires triggered widespread displacement across West Maui. Two years later, a significant share of fire-affected households remain unable to return to their previous homes or communities. Among those who lived in West Maui at the time of the fires, nearly 80% were still displaced from either their home or West Maui in the second quarter of 2025 (April to June). Of those who had to leave West Maui, 31% still reside in other parts of the island and 9% have relocated elsewhere. This represents a substantial level of ongoing geographic displacement, especially when considering that people who left the island altogether are likely underrepresented in the survey due to outreach centered on Maui-based networks. Many of the displaced households did not live in the immediate burn area.

In addition, displacement also extended beyond West Maui and Kula. Among fire-affected people who lived elsewhere on the island but worked or owned businesses in West Maui or Kula, 18% were displaced from their homes but remained in the same area, while an additional 14% were displaced from both their home and the broader area where they had previously worked or owned a business, as shown in Figure 1. Overall, almost one-third were displaced from their homes and remain displaced to this day. This group faces a distinct set of challenges: although their residences were not directly impacted by the fires, the loss of employment or business activity in West Maui or Kula disrupted their financial stability, the pre-existing and long-standing housing shortage was exacerbated by the disaster, and assistance programs are limited for those whose homes were not destroyed. All of these challenges may force people to relocate. The continued displacement of this group underscores the far-reaching ripple effects of the disaster, affecting not only those in the direct burn areas but also people whose lives were closely connected to them through work and local economic activity.

Looking ahead, most fire-affected people express a desire to remain on Maui, and many plan to return to West Maui if possible. Among those currently living in West Maui, more than 90% expect to still be there in a year. Among households currently living in other parts of Maui, two-thirds plan to stay in their current region, while about a quarter hope to return to West Maui within the next year. Less than 10% expect to leave Maui entirely.

These expectations indicate that many fire-affected individuals remain connected to West Maui as their home and wish to return. However, this anticipated movement stands in contrast to the slow pace of rebuilding and the limited number of temporary housing units currently available. As of mid-2025, 275 homes are under (re)construction, 44 have been completed, and an additional 422 permits have been issued. An additional 750 temporary housing units are completed or nearing completion (County of Maui, 2025), but this supply may not be sufficient to accommodate all those planning or hoping to return. The high proportion of individuals expressing a desire to remain in West Maui or return, especially from other parts of Maui, underscores the need for long-term planning that includes a sustained increase in housing supply. Recovery efforts must consider not only the logistical and financial barriers to return but also the expectations and hopes of displaced residents for whom West Maui is home.

Housing Stability

Two years after the wildfires, housing instability remains a central challenge for fire-affected households. The disasters disrupted existing housing arrangements for many people and significantly altered the landscape of homeownership and renting on Maui. Yet even amid the upheaval, the data show important differences in how people have been able to recover, depending on where they started before the fires and the support available to them since.

The most common housing transition since the fires has been from owning to renting. Among fire-affected individuals who owned their homes before the wildfires, in the second quarter of 2025 (April to June), only about two-thirds remain homeowners, while 30% have become renters and 3% have lost stable housing altogether and continue to live with family, friends, or they have become unhoused (see Figure 2). This shift reflects both the destruction of owner-occupied homes and the difficulty of reentering the ownership market after displacement, especially given the extremely limited supply of housing and the need to continue mortgage payments on uninsured or underinsured properties. The increase in former homeowners now renting also contributes to higher demand for rental properties, placing additional pressure on an already tight market.

The vast majority of pre-fire renters remain renters, but similar to owners, about 3% have lost stable housing and now live with family or friends or are unhoused. Only a small fraction have transitioned into homeownership. At the same time, a notable minority of new renters were previously living with family or friends or were unhoused, suggesting that access to disaster assistance may have improved housing stability for some households and provided much needed support for existing, pre-disaster disparities. Still about 30% remain without stable housing, a reminder that housing precarity continues for many of the most vulnerable groups.

These shifts in tenure are closely tied to the type of housing fire-affected people currently occupy. A significant proportion – almost 40% – still reside in temporary housing in the second quarter of 2025 (see Figure 3). However, it is encouraging that as of the end of last year, the share of temporary housing has started to decline. And, there has been a gradual but steady rise in the share of people living in permanent new homes over the last year, now accounting for more than 20% of fire-affected households. In addition, the share of people facing precarious housing conditions (living with family/friends or unhoused) has halved over the last year. Together, these figures point to important progress in the transition from temporary to permanent housing, although a large share of the population remains in interim arrangements.

While the prevalence of temporary housing remains high, one emerging sign of stabilization is a significant increase in duration of stays. Among those still in temporary housing, the share of people who have lived at their current address for more than one year has increased from under 5% a year ago to almost 50% now (see Figure 4). Although many fire survivors are still without permanent housing, their housing has become much more stable over the past year.

Rent

The fires disrupted the rental housing market on Maui overnight, leading to widespread displacement and fierce competition for an even more limited supply. Two years later, the rental situation for fire-affected households remains shaped by a complex mix of temporary assistance, market strain, and uneven recovery trajectories. While some renters today remain insulated from higher costs by assistance programs, others bear large rent burdens on their own, often for smaller units than they had before the disaster.

To better understand access to rental assistance over time, a statistical model was used to estimate the probability of receiving full rental assistance for fire-affected households still living on Maui. Figure 5 shows that since August 2024, the probability of paying zero or near-zero rent—defined as less than $100 per month—has grown noticeably, reaching a peak at the end of 2024. This is consistent with the opening of multiple temporary housing projects in the second half of 2024, in particular Hale ‘O Lā‘ie and Ka La‘i Ola where rent is free for the time being. Since the end of last year, the probability of receiving full rental assistance has declined such that over the last three months only about a third of fire-affected households have received full assistance. Although FEMA began requiring rent contributions in March 2025, most households were approved to pay a minimal $50 monthly fee—still classified as “near-zero” rent in this analysis. Therefore, the observed decline in assistance rates likely reflects factors beyond federal policy shifts and warrants continued monitoring.

Our analysis also reveals which groups are most likely to receive full rental assistance (see Table 1 in the Appendix). People not in the labor force – often due to retirement, disability, or caregiving – have been substantially more likely to benefit, with significantly higher odds of paying minimal rent than employed people. Renters who are unemployed also tend to receive more assistance, though the relationship is somewhat weaker and not statistically significant. As expected, those in temporary rentals without government assistance and also people living with family or friends are substantially less likely to pay zero or near zero rent. In addition, the results show that for the same household, every additional $1,000 in monthly income (compared to their usual average) is linked to a 13% lower chance of paying zero or near-zero rent. Even though this difference was not statistically significant, it suggests that higher-income households are likely to receive fewer rental subsidies. This is consistent with support frameworks that target assistance to households with greater financial need.

To track housing affordability among renters without access to full rental assistance, we developed models of median rent by unit size over time (see Figure 6). For the majority of renters who pay more than $100 per month, rent levels remain elevated, particularly for those in larger units or without assistance. Median predicted rents vary widely by unit size, with studios and one-bedroom apartments remaining the most affordable at about $1,800 in June 2025 (see Figure 6A). Over the last year, median out-of-pocket rents in these smaller units have declined steadily, likely due to increased availability and targeted assistance. However, even for small units, median rents remain about 50% higher compared to pre-fire levels of $1,200.

Two-bedroom and especially three or more-bedroom units continue to carry even higher costs and rents have remained more persistent over time (see Figure 6B and 6C). While there has been a drop in median rent for two-bedroom units in the second half of last year, median rents have subsequently remained stable at about $2,100 to $2,200. For three or more bedroom housing, median rents have shown limited movement and remain at about $3,200 to $3,400, almost double of what they were before the wildfires.

Household income and housing type are among the most consistent predictors of rent levels across unit sizes (see Table 2 in the Appendix). As expected, renters living in informal or shared housing arrangements, such as staying with family, tend to pay much less than those renting independently. But these arrangements may not be sustainable in the longer term.

Other factors also matter. Households with higher incomes and older adults tend to be able to afford higher rents. Meanwhile, identifying as Filipino was associated with notably lower rents for zero- to two-bedroom housing, which may reflect different housing arrangements, multigenerational living patterns and lower disposable income in this group.

The impact of higher market rents on affordability depends strongly on the receipt of comprehensive rental assistance. Looking at households on lower or similar incomes as before the wildfires, it appears that rental affordability has improved in the last two years. Almost three-quarters pay less rent now than before the fires, offsetting some of their financial constraints from lower incomes. But this does not hold true for households without full or near full rental assistance. More than half of this subgroup are paying more rent now than before the fires, even when their income is similar or lower than before the fires (see Figure 7).

Taken together, our rent analyses highlight both progress and vulnerability. While assistance has significantly reduced rent burdens for some, many continue to face extraordinarily high housing costs. For those without access to formal rental support, current rents far exceed pre-fire levels. As temporary programs near their expiration dates in 2026, ensuring stable, affordable housing for all fire-affected households will remain a major challenge.

Employment

Two years after the Maui wildfires, many fire-affected people remain in a different employment situation than they were before the disaster. As shown in Figure 8, before the fires, 66% were working full-time. In the second quarter of 2025, that share remains at only 45%. While there has been little change in part-time work by choice, the share of people who cannot get as many hours as they want has more than doubled. This points to sustained underemployment. In addition, many people continue to struggle finding any work or they have dropped out of the labor force altogether. The share of fire-affected people who are unemployed remains elevated at 8%, compared to a county-wide unemployment rate of 3%. More than a quarter of fire-affected people have retired or are otherwise not seeking employment (up from 19% before the fires). These changes suggest that for many individuals, the path to economic recovery has been slow and uneven. The drop in full-time employment and increase in part-time work, unemployment and labor force exit are likely a consequence of the closure or downsizing of businesses, ongoing displacement, caregiving responsibilities, or health and mental health challenges that have emerged or persisted in the wake of the disaster.

A major driving force of these changes in the labor market is the tourism sector which has long been central to Maui’s economy. The tourism industry shows signs of long-term disruption. Even almost two years after the disaster, real (inflation adjusted) visitor spending remains more than 20% below pre-fire levels (Hawaii Tourism Authority, 2025). This has had clear consequences for employment, as shown in Figure 9. Among those employed in tourism before the wildfires, only 66% remain in the sector in 2025, and only 45% of those who previously held full-time positions continue to work full-time in tourism jobs. Almost a quarter of former tourism workers have shifted into other industries, and almost one in five are currently not working because they are unemployed or have left the labor force. This realignment suggests that many former tourism workers have had to adapt, either by retraining or relocating within the labor market, and two years after the fires, many remain out of work altogether.

Income

Two years after the wildfires, many affected households continue to face financial hardship (Figure 10). More than half report having reduced income, including 22% with much less (i.e. less than half of pre-fire income) and 35% with less (i.e. 50-90% of pre-fire income). These results highlight the lasting economic impact of the disaster, with job losses, business closures, and displacement continuing to hinder recovery for many families. Still, it is important to note that not all of the economic hardship can be directly attributed to the wildfire; more can still be done to encourage the return of visitors to Maui to support local businesses and their employees.

Income losses following the wildfires have pushed a substantial number of households into poverty. As shown in Figure 11, 12% of households who were above the poverty line before the fires have fallen below it by mid-2025. An additional 9% who were already in poverty remained there. Only 3% moved out of poverty during this period. By the second quarter of 2025, more than one in five households were living below the poverty line— up significantly from 12% before the fires.

To understand changes in economic well-being over time, a regression model was used to estimate the median income-to-poverty ratio for fire-affected households still living on Maui (see Table 3 in the Appendix). This ratio compares household income to the federal poverty threshold, adjusted for household size. A ratio of 1 indicates a household is at the poverty line; below 1 means the household is in poverty, while a ratio of 2 indicates income double the poverty threshold.

As shown in Figure 12, the median income-to-poverty ratio has changed little over the past year, holding relatively stable at around 1.7 which is well below the pre-fire baseline of 2.2. The model also highlights characteristics associated with lower or higher ratios. Households with unemployed members, part-time workers, or those not in the labor force had significantly lower income-to-poverty ratios than those with full-time workers. For example, unemployment and part-time employment was associated with a nearly 0.8-point drop in the ratio—a substantial decline in financial stability.

In addition, living in government-supported housing is associated with lower income-to-poverty ratios, suggesting these programs are reaching some of the most economically vulnerable households. Somewhat smaller effects are observed for people staying with family or friends or in rentals without government support. Finally, identifying as Black or Native American or having a disability is negatively associated with financial stability suggesting continued gaps in support for these populations. These findings reflect the persistent and uneven economic impacts of the disaster across communities.

Needs and assistance

Two years after the Maui wildfires, many fire-affected residents continue to report substantial unmet needs. As shown in Figure 13, a significant share of households rate their needs as  urgent or serious (4 or 5 on a 5-point Likert scale), particularly in areas such as financial assistance, housing, and food. However, it is encouraging that these unmet needs have gradually declined over time. An exception for this trend is the share of people needing help navigating available resources which has remained unchanged, suggesting persistent barriers to accessing available support.

At the same time, overall assistance has declined. While many continue to receive some form of help, fewer households are getting the support they once had (Figure 14). For example, only 28% of respondents received FEMA assistance in the second quarter of 2025, compared to 51% at some point earlier in the recovery. Use of SNAP (Supplemental Nutrition Assistance Program, “food stamps”), WIC (Special Supplemental Nutrition Program for Women, Infants and Children), or TANF (Temporary Assistance for Needy Families) also fell, with fewer than one in five receiving these benefits recently, down from 35%.

Support from nonprofit and community sources have followed a similar pattern. About a quarter of respondents received support from the American Red Cross in the second quarter of 2025, compared to over half earlier in the recovery. Support from other community organizations dropped from 53% to 22%, and reliance on family, friends, or individual donors has fallen from almost 60% to just over one quarter.

Perhaps most notable, 44% of respondents reported receiving no assistance at all in the second quarter of 2025, despite only 4% saying they had received no help during the entire recovery period. This may reflect a shift in needs as some households regain stability and economic security, especially given the decline in reported unmet needs. However, it also signals potential gaps emerging as emergency programs phase out, philanthropic support wanes, or  barriers to access persist.

Two years after the disaster, a central challenge remains: transitioning from short-term emergency aid to long-term recovery support that is accessible, sustained and responsive to evolving needs.

Conclusion

Two years after the Maui wildfires, the recovery is ongoing, marked by both progress and persistent challenges. The findings from the Maui Recovery Survey reveal the disaster’s deep and lasting disruptions to housing, livelihoods, income, and daily life. Many fire-affected households remain displaced or in temporary housing, often facing rent burdens that exceed pre-disaster norms. Job opportunities, particularly in tourism, have not fully returned, and income losses have pushed a sizable share of households into poverty. While urgent needs have declined in some areas, some residents still face barriers to accessing the programs and services critical to recovery. The decline in assistance over the past year raises concerns that some households are being left behind as temporary support phases out.

Amid these difficulties, there are also meaningful signs of progress. More fire-affected residents have transitioned into permanent housing, and precarious housing conditions have  decreased. Some have found new employment or returned to full-time employment in different sectors. Although income remains lower for many, it has stabilized. The drop in assistance may, in part, reflect improved financial stability for some households. These developments suggest that recovery is moving forward, even if uneven and incomplete, and Maui communities continue to demonstrate remarkable resilience.

Sustained investment and targeted support will be essential to maintain this progress and address ongoing gaps. Rebuilding in West Maui and other affected areas will take years, and it is important that long-term planning reflects both the scale of the disaster and the hopes of those seeking to return. To monitor and support this effort, the Maui Recovery Survey will continue for at least another year, offering timely insights into household well-being and helping to ensure recovery efforts remain rooted in the lived experiences of those most directly impacted.

Appendix

Methodology

Survey methodology

The Maui Recovery Survey focuses on people who lived, worked, or owned a business in West Maui or Kula during the Maui wildfires in August 2023. Participant recruitment is facilitated by community and government partners. Participants receive a $20 gift card for the first survey and a $10 gift card for each monthly follow-up survey that they take. It is not a closed cohort, i.e. eligible people can join anytime. The survey form is available online in six languages: English, Spanish, Tagalog, Ilocano, Tongan, and Vietnamese. Recruitment primarily relies on multilingual email outreach but is supplemented by physical flyers available in community spaces and at local events. The Council for Native Hawaiian Advancement provided critical support for community outreach and initial project planning.

The survey questionnaire is designed to follow key indicators such as income, employment, housing type, rent, residential and job location over time. Participants are asked about their situation before the wildfires, followed by monthly updates on the same set of questions starting in August 2024.

We have used a wide range of data verification checks and individual participant follow-ups to eliminate fraudulent responses from the data. We are confident that this data represents real fire-impacted people. So far more than 950 people have participated in the survey, with an average monthly sample of about 430 responses. All housing and income-related data is reported on a household basis whereas employment and needs data is based on individual-level records.

The survey data has been weighted to maximize representativeness relative to the fire-affected population as a whole. The person-level data has been weighted by income-to-poverty ratio by age group, employment status and tenure by age group. Because of substantial monthly sample churn, we modeled a synthetic cohort that mimics typical sample characteristics across months as the empirical basis for developing survey weights. Iterative proportional fitting (raking) has been used to implement survey weighting. This method continues to adjust the weights until the sample distribution approximately matches the population distributions for the selected weighting variables. Reference data has been obtained from the American Community Survey 5-year sample for 2022 to avoid bias from 2023 data that could have been collected after the fires. Because most of the fire-affected people lived in Lahaina prior to the wildfires, we used data for the Lahaina Census Designated Place as the reference population. The design effect is 1.3 with a margin-of-error of 1.1.

The weighted survey sample closely aligns with the reference population for the weighting variable age, employment, poverty and tenure (see Figures A1 to A8 in the Appendix). In addition, race/ethnicity and place of birth have similar distributions. The survey does include more women than men and due to lack of reference data, we have been unable to compare non-binary people. The share of disabled people appears higher in the survey, but this is likely largely related to differences in the definition of disability.

Analysis methods

Our analysis combines descriptive statistics with econometric modeling to track change over time and identify key drivers of outcomes such as housing, rent, income, and unmet needs. All reported percentages and cross-tabulations are weighted descriptive proportions. For point-in-time analyses including pre- and post-fire comparisons, we pool data from respondents who completed at least one survey during the second quarter of 2025 (April to June 2025) and apply survey weights to calculate population-representative estimates. For time series analyses, we include only individuals who have responded to three or more monthly surveys. This helps reduce month-to-month noise due to changes in sample composition.

To visualize trends in key outcomes over time – such as housing type, displacement, and self-reported needs – we use LOESS (Locally Estimated Scatterplot Smoothing), a non-parametric method that fits a line through the data by averaging local subsets of points. This technique does not assume a specific functional form and is useful for identifying gradual changes or inflection points in the data by filtering out short-term fluctuations.

To model the probability of paying zero or near-zero rent (defined as paying less than $100 per month), we use a Bayesian generalized linear model with a logistic link function and include individual-level random effects to account for repeated observations within respondents. To address the potential endogeneity of the random effects, we incorporate Mundlak adjustments, which include the person-specific means of time-varying covariates as additional predictors. This approach helps separate the between-person variation from within-person changes over time and yields more robust estimates of the time-varying effects. The model includes polynomials of time (survey month) up to the third degree, allowing for curved patterns in how rent assistance has changed over time. The plotted time series of the predicted probability has been LOESS-smoothed.

To assess rent levels for households who pay more than $100 per month, we estimate median rent over time using quantile regression models. Natural splines of time with internal knots placed at key months were included if they meaningfully improved the explanatory power of the model. Three model specifications are used for zero or one-bedroom, two-bedroom and three or more-bedroom housing respectively. These models help capture differences in rent burden for different types of renters, including by unit size, income level, housing type, and employment status, while allowing rent to change over time in a flexible way. The plotted time series of the predicted median rent has been LOESS-smoothed.

To examine changes in economic stability, we estimate median income-to-poverty ratios over time using quantile regression. The income-to-poverty ratio compares a household’s income to the federal poverty threshold for their household size. A ratio of 1 means the household is right at the poverty line, while a ratio below 1 means they are living in poverty. At a ratio of 2 a household has twice as much income as a comparable household at the poverty line. Our model estimates how this ratio has changed since August 2024 and which factors are associated with better or worse outcomes. Key predictors include employment status, housing type, prior location (e.g., living in the Lahaina burn area), household size, race/ethnicity, and disability status. The plotted time series of the predicted median income-to-poverty ratio has been LOESS-smoothed.

All models are weighted using the survey weights described in the Survey Methodology section, ensuring that results reflect the broader fire-affected population. Models are estimated in R. Model fit is evaluated using standard diagnostics, including pseudo R², Bayesian R², Brier scores (for binary models), and changes in average absolute deviation for quantile models.

Regression results

Table 1: Estimated probability of paying zero or near zero rent for fire-affected households living on Maui

	Probability of zero or near zero rent
Time (Linear)	1.127 (0.638, 1.951)
Time (Quadratic)	0.957 (0.872, 1.052)
Time (Cubic)	1.001 (0.995, 1.006)
Employment status: Not in labor force	6.286 (0.968, 49.74)
Employment status: Unemployed	1.168 (0.314, 4.335)
Housing type: Rental without government support	0.000 (0.000, 0.000)
Housing type: Staying with family/friends	0.004 (0.000, 0.082)
Income (in thousands)	0.87 (0.715, 1.052)
Mean of employment status: Not in labor force	0.223 (0.013, 3.836)
Mean of employment status: Unemployed	6.324 (0.341, 125.384)
Mean of income (in thousands)	0.84 (0.572, 1.21)
Mean of housing type: Staying with family/friends	1.093 (0.022, 52.88)
Mean of housing type: Rental without government support	0.006 (0.000, 0.084)
Bayesian R2	0.864
Brier Score	0.027
Mean Residual	<0.001
AUC	0.996
Odds ratios, 95% confidence intervals in parentheses. Reference category for housing type: Same rental as before wildfires. Reference category for employment status: Employed.

Table 2: Estimated median rent by unit size for fire-affected households currently living on Maui

	(1) Median rent: 0-1 bedrooms	(2) Median rent: 2 bedrooms	(3) Median rent: 3+ bedrooms
Intercept	806.417** (287.5)	807.358** (288.25)	2238.39** (790.662)
Spline Segment 1 (Month, knots at 4, 8)	-15.034 (110.965)
Spline Segment 2 (Month, knots at 4, 8)	-220.743 (230.884)
Spline Segment 3 (Month, knots at 4, 8)	-176.232* (99.181)
Spline Segment 1 (Month, knots at 5, 7, 9)		0.000 (212.353)
Spline Segment 2 (Month, knots at 5, 7, 9)		-75.963 (195.667)
Spline Segment 3 (Month, knots at 5, 7, 9)		-27.58 (487.954)
Spline Segment 4 (Month, knots at 5, 7, 9)		-3.388 (168.881)
Housing type: FEMA supported housing	-402.534* (239.064.7)	-40.316 (283.694)	450.356 (417.162)
Housing type: New rental without assistance	-13.905 (94.89)	576.686*** (125.24)	700* (365.786)
Housing type: State supported housing		945.36 (1023.485)	4074.634* (1775.805)
Housing type: Temporary rental with other assistance	125.701 (129.165)	1088.851 (658.763)	834.634* (422.302)
Housing type: Temporary with family/friends	-715.345*** (201.925)	-354.166* (140.773)	-511.766 (424.429)
Income	148.408*** (32.006)		185.366*** (37.444)
Number of bedrooms			-608.378** (194.617)
One bedroom (binary)	297.573*** (88.968)
Employment status: Not in labor force			1218.022** (382.739)
Employment status: Unemployed			119.083 (597.031)
Age	10.172** (3.029)	23.921*** (4.158)
Disability		-857.385*** (245.001)
Gender: Female			834.549* (362.303)
Born in Hawaii		306.336** (111.905)	740** (234.373)
Race: Filipino	-369.144*** (109.768)	-707.998*** (104.138)
Race: Other Asian		358.495* (163.73)
Race: White	158.944* (88.465)		542.439* (227.184)
Pseudo R2	0.158	0.14	0.193
Change in average absolute deviation from the median relative to null model	17.425	25.943	28.167
Significance levels: * 10%, ** 5 %, *** 1%. Standard errors in parentheses. Reference category for housing type: Same rental as before wildfires. Reference category for employment status: Employed. Race variables are binary and allow for multi-racial identification, e.g. Filipino (alone or in combination with other races) with reference category: Not Filipino.

Table 3: Estimated median income-to-poverty ratio for fire-affected households currently living on Maui

	Median income-to-poverty ratio
Intercept	1.822*** (0.151)
Spline Segment 1 (Month, knots at 2, 5)	-0.083 (0.093)
Spline Segment 2 (Month, knots at 2, 5)	-0.062 (0.152)
Spline Segment 3 (Month, knots at 2, 5)	0.09 (0.059)
Employment status: Unemployed	-0.789*** (0.078)
Employment status: Part-time	-0.769*** (0.063)
Employment status: Retired or not seeking employment	-0.619*** (0.065)
Household size	-0.076*** (0.011)
Housing type: Government supported housing	-0.384*** (0.07)
Housing type: Rental without government support	-0.171 (0.078)
Housing type: Staying with family/friends	-0.195 (0.107)
Pre-fire income-to-poverty ratio	0.167*** (0.028)
Age	0.006*** (0.002)
Race/ethnicity: Other (Black or Native American)	-0.199* (0.084)
Disability	-0.2*** (0.053)
Pseudo R2	0.216
Change in average absolute deviation from the median relative to null model	18.951
Significance levels: * 10%, ** 5 %, *** 1%. Standard errors in parentheses. Reference category for housing type: Same rental as before wildfires. Reference category for employment status: Employed. Reference category for housing type: Same rental as before wildfires. Race variables are binary and allow for multi-racial identification, e.g. Filipino (alone or in combination with other races) with reference category: Not Filipino.

Demographics of the survey cohort

Age

Race/ethnicity

Gender

Figure A3: Gender distribution of the survey participants

Disability before the wildfires

Poverty before the wildfires

Employment before the wildfires

Figure A6: Distribution of the employment status of the survey participants before the wildfires

Tenure before the wildfires

Place of birth

Bibliography

County of Maui. (2025). Rebuild Dashboard [Dataset]. https://www.mauirecovers.org/recoverydashboard

UHERO Data Portal. (2025). Total Real Visitor Expenditures [Dataset]. https://data.uhero.hawaii.edu/#/series?id=164727&data_list_id=41&geo=MAU&freq=Q&start=2019-10-01&end=2024-10-01