BLOG POSTS ARE PRELIMINARY MATERIALS CIRCULATED TO STIMULATE DISCUSSION AND CRITICAL COMMENT. THE VIEWS EXPRESSED ARE THOSE OF THE INDIVIDUAL AUTHORS. WHILE BLOG POSTS BENEFIT FROM ACTIVE UHERO DISCUSSION, THEY HAVE NOT UNDERGONE FORMAL ACADEMIC PEER REVIEW.
By Dylan Moore and Baybars Karacaovali (Office of Tax Research and Planning, Hawai’i Department of Taxation)*
Executive Summary:
An analysis of state income tax filing addresses indicates that the 2023 Maui wildfires reduced Maui’s population by at least 1,000 residents, due to both increased out-migration and reduced in-migration. At least 430–510 Maui residents have moved out of state because of the fire. These migration impacts will cost the state’s economy at least $50m of annual income. These preliminary estimates likely underestimate the magnitude of the effect, but nonetheless help to shed light on the evolving impact of the fires.
More than a year after the devastating Maui wildfires, questions remain about the scale of the population loss that may result. Reports from wildfire survivors and other Maui residents suggest they face challenges that have led to displacement: a lack of affordable housing, job losses, financial uncertainty, emotional strain, and uncertainty about rebuilding their lives. A survey by the Hawai‘i State Rural Health Association found that about one in five fire-affected households are seriously considering moving away from Maui.
However, putting precise numbers on the migration effects of the fires has proven difficult, due to a lack of quickly available reliable data. In this brief, we use information from state income tax returns to answer two questions:
- Where are fire-displaced households living now?
- How did the fires impact migration patterns in the state?
Filing addresses are required on state tax returns to allow the Department of Taxation to correspond with taxpayers and, where necessary, to issue refund checks by mail. While these records are not intended for tracking population movements, filing addresses can be used to track the movements of most Hawaii residents over time. This includes not only movements within the state, but also outside the state, as households that leave Hawaiʻi typically file at least one or two final state tax returns from their new location. Although filing address is not a perfect measure of location – as we discuss below – our estimates provide the first comprehensive, statewide assessment of the fires’ migration effects.[1]
Below, we present a summary of our findings about both of these questions, alongside two data visualizations. The details of how we reached our conclusions are presented at the end of the post.
Where Did Displaced Households Go?
We confine our analysis to a group of individuals who are highly likely to have been displaced by the fires. This “displaced” group consists of households which filed a resident state tax return at some point between January 1, 2022 and August 7, 2023 (i.e. before the fires), and whose most recent tax return in that period was filed from an address which was so heavily damaged by the fires that it was deemed unsafe by inspectors.[2],[3] There are about 5,089 people listed on such returns (either as a tax filer or the dependent of a filer).
The figure below shows where previously displaced residents—those whose pre-fire filing address was rendered unsafe by the fires —filed their state tax returns after the fires.[4]
Of the displaced residents, we can see that 1,420 relocated elsewhere within the Lahaina zip code (i.e. West Maui), and 1,058 moved elsewhere on Maui. However, some 369 displaced residents have left the County: 242 moving out of state, and 127 relocating to other Hawaiian counties. These, and other statistics are presented in the data visualization tool below. Encouragingly, the share of displaced residents moving to various locations is fairly similar to the migration patterns of respondents to UHERO’s Maui Recovery survey.
Nonetheless, there are three important limitations of these results. First, the sample of households we have labeled as displaced is not comprehensive, in large part due to the fact that some filing addresses could not be confidently linked to county records. As well, some households whose properties were not heavily damaged were still displaced from their homes by the fires.
Second, a number of other households are excluded from our analysis because they do not appear in the state tax return data. Over 600 displaced households that filed before the fires have not yet filed after the fires, and some households never filed for tax years 2021 through 2023 to begin with. Finally, we note that a filing address is not always a good guide to a current residential address. This is quite clear if you look at our data, as the second most commonly reported type of filing address for displaced residents is an unsafe address: that is, they filed from an address where they almost certainly did not reside at the time. For more details on these, and other caveats, see this section.
How Did the Fires Change Migration Patterns?
While the relocation of displaced households is an interesting question, it does not tell us how the fires impacted migration patterns in the State. Some displaced households who moved away from Maui likely would have done so regardless of the fires. One might reasonably ask, how many of these moves happened because of the fires. Additionally, the fires may have influenced migration decisions for households beyond the immediately displaced group, including those in unburned parts of Lahaina and elsewhere in Maui.
In this section, we present estimates of the impact of the fire on migration in each of three regions within Maui: (i) the burned parts of Lahaina; (ii) unburned parts of Lahaina; and, (iii) the rest of Maui County. Importantly, living in the so-called “burn area” of Lahaina is not the definition of displacement we used in the section above: not all of those living in the burn area have properties that meet the definition of “unsafe.” However, this group was quite heavily impacted by the fires: a UHERO survey suggests 90% were forced to leave their homes.
The wildfire migration impact visualization tool presented below displays our impact estimates. As the tool’s tutorial explains, our impact estimates are based on comparing observed post-fire migration rates in each of the three affected areas in Maui against estimates of the rates we would have expected to see absent the fires. In each case, we obtain estimates of what would have happened by assuming migration rate trends in other parts of Hawai‘i during the same period are indicative of what would have happened in these three fire affected areas. For more detailed information about our estimates, see here.
Our estimates reveal that relocations from Lahaina to other parts of Maui are the largest migration effect of the fires. However, we also find evidence that many people have left Maui: the total population has fallen by at least 1000 people. About 63-64% of this results from increased out-migration[5] from Maui, with some residents leaving for other counties in Hawai‘i (at least 160-200 people), and others leaving the state (at least 430-510 people). Reduced in-migration[6] makes up the remainder of the effect: we estimate that 370 fewer people moved to Maui from outside the state than would have been expected if the fires had not occurred.
It is important to note that our impact estimates suffer from some of the same limitations as our tracking of displaced residents. The impact estimation tool shows this: we estimate that the fires have caused around 1300 residents to “disappear” from our data. That is to say, that’s how many more people failed to file a timely 2023 state tax return than we would have expected absent the fires. And, as with tracking displaced residents, our results are undermined to some extent by lack of knowledge about the true location of households who still file taxes from unsafe addresses. For further discussion of these limitations, see here.
Economic Implications
The observed out-migration may have significant implications for the economic conditions and revenues of both the County of Maui and the State as a whole. Our estimates suggest that nearly $60 million of annual income were lost in Maui due to changes in out-migration and in-migration caused by the fires.[7] This represents a little over 1% of total annual income Maui residents reported on state tax returns in the year before the fires. Much of this money would have eventually been spent in Maui and thus generated GET revenue for the County, while also supporting jobs for local residents. Similarly, over $50 million of income lost to the State would have generated spending that supported state jobs and state GET revenues. Finally, the migration effects of the fires cost the State over $3 million in income tax revenues. As in other cases, these numbers understate the true impact of fire-induced migration, not only because true out-migration effects will be larger than what we observe in the tax data, but also because of the spillover effects the spending of these absent residents would have had on local economic activity.
More importantly, these dramatically underestimate the economic and revenue impacts of the fires, as they do not account for the impact of the fires on the incomes of the people who remained in Maui. Those impacts are quite likely to be much greater in magnitude than the ones reported here. We intend to revisit that subject in future work.
Appendices
Appendix A: Impact Estimation Method
To understand how the Maui wildfires influenced migration patterns, we need to compare observed post-fire migration rates to the rates we would have expected if the fires had never occurred. Thus, estimating the fires’ impact on migration requires estimating unobserved, counterfactual migration rates.
To be more specific, the impact of the fires on migration rates in a given area can be written as the difference between what we observe and what would have happened without the fires:
Impact in Area = Observed Rate in Area – Rate in Area if Fires Hadn’t Happened
There is no perfect way to guess what would have happened to migration rates in a fire affected area if the fires hadn’t happened. We adopt a commonly used approach by assuming that, absent the disaster, migration in the fire-affected areas would have followed similar trends to what was observed in parts of Hawai‘i that were unaffected by the fires.
To be more concrete, consider the case of folks moving from the burned areas of Lahaina to someplace out-of-state. Before the fires, 1.92% of residents living in this area in early 2022 had moved out of state by early 2023: a pre-fire baseline out-of-state out-migration rate of 1.92%. By contrast, looking at the residents filing from this area in early 2023, we find that 4.31% of them had moved out of state by 2024: a post-fire out-of-state out-migration rate of 4.31%. This observed increase suggests a fire-induced effect, but this is not enough. We still need an estimate of the counterfactual scenario: what would the post-fire rate have been if the fires had not occurred?
To find this counterfactual, we examine migration trends in the rest of Hawai‘i and use them to infer the expected post-fire migration rate absent the fires. Returning to our example, outside Maui, the out-of-state out-migration rate changed from 2.59% before the fires to 2.26% after the fires. We consider two possible ways to estimate a counterfactual rate based on this trend:
- The Absolute Changes Approach
Under this approach, we assume that the affected area’s out-migration rate would have moved by the same absolute number of percentage points as it did elsewhere. Since the rest of Hawai‘i’s rate fell by 0.33 points, the Lahaina burn area’s expected post-fire rate (if no fires had occurred) would be 1.92% – 0.33% = 1.59%. Instead, we see 4.31%. This difference of 2.73 percentage points translates to an estimated 146 additional out-of-state movers attributable to the fires. - The Proportional Changes Approach
This method assumes that, absent the fires, migration rates in an affected area would have changed by the same proportion as rates in unaffected areas. The 0.33 percentage point drop in the out-of-state out-migration rate in the rest of Hawai‘i represented a 12.7% change from the of the baseline rate of 2.59% (0.33 / 2.59 x 100% = 12.7%). If this same decline had occurred in the burned parts of Lahaina, we should have expected a post-fire rate of 1.67% (1.92% – 0.127 x 1.92% = 1.67%). Given the observed out-of-state out-migration rate of 4.31%, we can estimate that the fires added roughly 2.64 percentage points to the rate, implying about 141 extra out-of-state movers.
Although these two methods produce slightly different numbers, both indicate a substantial fire-induced increase in out-migration. Our wildfire impact tool displays results from both approaches, allowing readers to see how sensitive our estimates are to the method chosen.
These estimates rely on two key assumptions:
- We assume that, had the fires not occurred, migration rates in our three regions of Maui would have evolved similarly to those in the rest of Hawai‘i.
- We also assume that the fires did not significantly alter migration patterns outside Maui.
These assumptions cannot be directly tested for the post-fire period, but they provide a reasonable starting point for preliminary estimates. We anticipate updating these estimates using more refined methods as more complete data becomes available.
Finally, we do not present estimates in situations where we believe these assumptions clearly fail. For example, we do not estimate the effect of the fires on migration flows into Maui from other Hawaiian counties, because our out-migration estimates show that the fires causes increases flows from Maui to other parts of the state, which implies that out-of-county, within-state in-migration in other parts of the state was impacted by the fires. By contrast, we have no strong reasons to expect that in-migration from the mainland to areas outside Maui was significantly affected by the fires. Consequently, we do present estimates of the fires’ impact on in-migration from out-of-state.
Appendix B: Underestimation of Migration
These numbers are quite likely underestimates of the true effects. There are two main reasons for this. First, tax filing addresses are not always the same as residential addresses. Consequently, using this data source may lead us to misidentify the location of some households that appear in the tax data. For example, a household might set up mail forwarding after moving and still file their tax return from their old address. Indeed, the USPS explicitly encouraged many wildfire-affected households to do so, and also allowed residents whose homes were damaged, destroyed, or inaccessible to collect their mail at a post office. As our first visualization shows, a large number of households are still filing state tax returns from addresses that were rendered unsafe for habitation by the fires, suggesting that this issue is quite pervasive. Using the tax data, we cannot say with any confidence that these households do or do not currently reside in Maui.
The second reason to think our estimates are too low is that some households do not file tax returns, or file their returns quite late. We exclude both types of households from our analysis. We also exclude non-resident tax filers. This means that certain types of immigrants, including undocumented immigrants, as well as nonresident aliens are not included in our results. To get some sense of how important this issue is, note that Census estimates of the Maui population just before the fires (July 1, 2023) are 164,183 people, whereas our dataset includes 128,122 people who appear on a state income tax return filed from Maui between January 1, 2023 and August 7, 2023.[8]
This second problem is exacerbated by the fact that the fires appear to have increased the share of households that don’t file and/or that file late. About 700 more tax returns from Maui were filed late or unfiled than would have been expected absent the fires. About 1300 people are associated with these returns, and we make no assumptions here about where these people are currently located.
* The views expressed here are solely those of the author and should not be attributed to the Office of Tax Research and Planning or Hawai‘i Department of Taxation.
[1] Analysis of resident tax return data as conducted via a partnership between UHERO and the Hawai’i Department of Taxation (DOTA) Office of Tax Research and Planning. At no time did UHERO researchers have direct access to any personally identifiable information about taxpayers.
[2] If a household filed more than one state tax return during this period, we take the address from the most recently filed return. This can happen because filing dates vary widely across households, and within households from one year to the next.
[3] Data on the results of post-fire building inspections were obtained from Maui County via Matt Jachowski, who cleaned the dataset to allow merging with other property records.
[4] In a minority of cases, a household may file two state tax returns after the fires (due to late filing of a 2022 tax return). In such instances, our analysis is based on the address listed on the return filed most recently.
[5] Out-migration from an area refers to folks leaving that area for another.
[6] In-migration to an area refers to folks arriving in that area from another.
[7] Our measure of income is federal annual gross income (FAGI) which includes many types of income, but also allows for certain types of deductions (for tax purposes). Thus, our income measure is likely to slightly understate income for the purpose of estimating economic and tax revenue effects.
[8] Thus, if migration effects of the fires are similar amongst the groups excluded from our dataset, the population changes would be 28% higher than we reported above. However, we have no good reason to expect such similarity, as these groups are quite different from one another.