Familiar ground: why wildfire evacuees gravitate toward places they already know

When the Marshall Fire tore through suburban Boulder County on 30 December 2021, Colorado residents had hours, not days, to decide where to go. Wind gusts above 100 mph drove flames through neighborhoods in Superior and Louisville that had not previously been considered high wildfire risk, destroying more than 1,000 homes and burning about 6,000 acres in what became the most destructive fire in state history at the time. Two years later, the patterns of where those evacuees actually went have begun to come into focus, and they cut against the assumptions embedded in most municipal shelter plans.

New work on the Marshall Fire suggests that, given the choice, displaced households overwhelmingly relocated to addresses they already recognized: a relative's house across town, a friend's apartment in a nearby city, a neighborhood they had passed through on the way to work. The instinct is small in scale but large in implication. The places most disaster-ready on paper, dedicated shelters and purpose-built reception facilities, were often the places evacuees avoided. The places that absorbed most of the displaced population were the networks of ordinary relationships that planners cannot easily map.

Where the displaced actually went

Researchers tracked Marshall Fire households in the weeks and months after the fire and found a consistent preference for staying close to home in social terms even when the physical distance was significant. Evacuees frequently moved to the homes of family or friends, then drifted through a sequence of temporary addresses before either returning to a rebuilt house or relocating somewhere permanent. The pattern held across income groups, though households with fewer local ties tended to settle further afield sooner.

The finding complicates the logic of regional shelter systems, which are typically built around a small number of large-capacity sites run by the Red Cross, county emergency management, or municipal agencies. Those facilities did operate during the Marshall Fire; they were not overrun. They were simply not where most people went. The mass displacement was absorbed by the informal economy of couches, spare bedrooms, and rental listings that nobody had pre-registered as a shelter.

The counter-narrative

The standard emergency-management literature treats that pattern as a coordination failure. If evacuees will not go to designated shelters, the argument runs, planners need better signage, better communication, and better incentives to push them toward facilities that have medical staff, food, and a documented headcount. The implicit assumption is that the formal system is the desired destination and the informal one is a deviation.

The Marshall data point the other way. People knew, often accurately, what they would find at a shelter: cots, strangers, a hard check-in time, and a long bus ride back to a neighborhood they could not enter. They also knew what they would find at their cousin's house: a door, a dog, and someone who would not ask for paperwork. Treating that as a problem to be engineered away assumes the formal option is the better one. The evidence here suggests the evacuees, given the choice, did not.

Why familiarity wins

The structural point is that disasters do not reset social geography. They interrupt it. A household that has lived in a suburban cul-de-sac for fifteen years has built up a web of routine contacts: the pediatrician, the auto mechanic, the school pickup line, the neighbor with the spare key. None of that is in a county emergency operations plan, and none of it shows up on a shelter map. When the order comes to leave, the web is what people reach for first.

That has measurable consequences. Households with denser local ties tend to return to their pre-disaster neighborhoods more often, which matters because returning residents are the constituency for rebuilding permits, school re-enrollment, and the slow restoration of property tax bases. Households with thinner ties are likelier to relocate permanently, sometimes to the same county, sometimes across state lines. The Marshall Fire, like Hurricane Katrina before it and the Camp Fire after it, accelerated a sorting process that was already underway: people with anchors stayed; people without them drifted.

What planners can and cannot fix

There is a real case for better formal capacity. The Marshall Fire displaced thousands of people in winter, in the middle of a pandemic, in a state whose shelter infrastructure is sized for a different threat profile. The fact that informal networks absorbed most of the load does not mean the formal system was unnecessary. It means the formal system was needed by the people who had no informal network to fall back on, and the data suggest that group is smaller than planners assumed, but not zero.

The harder policy question is whether to try to redirect evacuees toward the formal system or to recognize the informal one and build around it. The first option runs into the finding that evacuees are not making a mistake; they are making a preference, often with good information. The second option would mean treating households, not shelters, as the unit of disaster response: pre-registering extended family networks, funding small-dollar direct assistance that follows people to wherever they actually go, and accepting that a successful evacuation may look like a thousand private arrangements rather than a single gymnasium with cots.

The Marshall Fire was, by the standards of recent western disasters, a tractable event. The next one in a fast-growing Sun Belt county may not be. How emergency managers respond to the data on where evacuees actually go will shape whether the next mass displacement is absorbed by the social fabric or breaks it.

The desk note: this article is built on a single peer-reviewed behavioral study of the 2021 Marshall Fire; figures and patterns above trace to that work. Wire coverage of the fire itself focused on damage counts and cause, not evacuee destination choice. Monexus treats the shelter-versus-informal-network tension as a policy question in its own right.