Unpacking the U.S. Census Data on 'Natural Disasters' Part I: Displacement

Over 3.3 million adults displaced by disasters in 2022

I had a delightfully nerdy mentor in graduate school who would rub her hands together at the end of a study, smile and say: “And now…we get to play with DATA!”

Playing with data is all-too-rare in the disaster world. More on this in a future newsletter, but the lack of standardized, longitudinal data has been a thorn in the side of U.S. disaster research and policy-making for as long as anyone can remember. It is cause to celebrate, then, when the U.S. Census Bureau releases new data.

The Household Pulse Survey is a new (since mid-2020), experimental data product developed by the U.S. Census Bureau and other federal agencies in the wake of the Coronaivrus pandemic. The survey is a rapidly deployable, online survey that studies “how the coronavirus pandemic and other emergent issues are impacting households across the country from a social and economic perspective.” Lucky for us one of the emergent issues included on the latest phase of the Pulse Survey questionare (3.7) is “the impact of living through natural disasters.” Specifically, this latest questionaire includes the following questions:

1. In the past year, were you displaced from your home because of a natural disaster?

2. What type of natural disaster?

3. How long were you displaced from your home?

4. Altogether, how much damage to your property or possessions did you experience as a result of natural disasters in the last year? Would you say no damage, some damage, a moderate amount of damage, or a lot of damage?

5. In the first month after the natural disaster, to what extent did you experience any of the following:

   1. A shortage of food?

   2. A shortage of drinking water?

   3. Loss of electricity?

   4. Unsanitary conditions?

   5. Feeling isolated?

   6. Fear of crime?

   7. Offers that seemed like a scam?

We have data. Now lets play with it.

The data tables for the survey helpfully break these questions down by a number of characteristics including age, education, household size, income and a host of others. The tables also include breakdowns by state and metro regions. As the Census Bureau makes clear, however, these are experimental data with potentially small sample sizes and large standard errors (the displacement data by metro area, for example, is largely unusable). For that reason I am mostly sticking to examinig national level trends, but may jump into state-level findings in future newsletters.

Disaster and Displacement

Mike Schneider at the Associated Press covers some of the headline numbers from the survey, namely that 1.3% of the adult population (3.33 million) was displaced by a disaster at some point in 2022, with the greatest impact coming from hurricanes (1.9 million), other (685k), flood (665k) and fire (669k).

• A brief comment - why did the questionnaire writers list ‘fire’ but not differentiate between wildfire and house fires, which average about 350k per year? Why is the ‘other’ category so large and the responses not presented in the data tables?

The state-level numbers are interesting because they show that Louisiana had the highest rate of displacement (15.2% of respondents, with a +/- of 2.6%) despite having a relatively quiet hurricane season. The next highest state, not suprisingly, was Florida followed by Nevada and Texas.

Interestingly, household income tracks strongly with top-level displacement numbers, with those earning less than $25,000 being displaced at the highest rate (2.8% of respondents) and steadily declining from there until you reach high-income households (making $200,000 or more annually) when the share jumps up again:

This certainly tracks with a lot of research showing that hazard-prone land is relatively less valuable than less hazard-prone land, especially for floods, and therefore is often home to households making less income. It also tracks with a hurricane season that saw lots of damage to high-value Florida coastal real estate, which may account for the jump for high-income households.

When we look at top-level displacement data we also see some interesting and not-too-suprising trends by Hispanic origin and Race:

Respondents who were White alone, not Hispanic had the overall lowest rate of displacement at 1.2%, followed by Hispanic or Latino (1.4%), Asian alone (1.5%), Black alone (1.7%) and two or more races + Other races, not Hispanic (1.8%). I’d be interested to see some analysis of these numbers by state and cross-tabbing income and hazard data, but the sample sizes might be getting too small at that point to ensure robust results.

One last finding on these top-level numbers - households with children under 18 reported being displaced at a significantly higher rate (1.6%) than those without (1.1%).

How about the length of displacement? Overall we see that the majority of households are back in their home in less than a week (39%), while 24% were displaced more than six months or never returned home.

• Another methodological note: I find these lengths of time curious. Specifically, the gap between one month and six months is quite large, especially when you consider the implications for transitional housing needs (i.e. housing someone for 5 weeks is much different than 5 months). I would like to see another choice of 1-3 months. Second, ‘never returned home’ is misleading because the survey only asks about the past year. That is, in many disasters it takes longer than 1 year for households to rebuild and return. In-fact, my colleagues at the Urban Institute recently published a paper showing that the average CDBG-DR recovery housing project from FY2005-2015 took 3.8 years from declaration to completion. So I would rather see "‘still displaced after 1 year” as the choice, or perhaps an option of “not planning to return home.”

Finally, lets look at the length of time of displacement by income:

These results are more mixed and it is tough to see any firm patterns - I think I will need to dig more into the state- and hazard-specific data to really test some assumptions we might have. One interesting finding that ‘pops’ is that households making over $200k per year are much more likely to report never returning home after a disaster…and that makes sense, if you consider they likely have the most resource flexibility to pickup and move to a new home, or have careers that are highly mobile. Similarly we see relatively high ‘never returned home’ among the households with the lowest incomes, and I suspect that these may be renters who move on to different housing situations. It is hard to know, though, because the survey does not ask about housing tenure, a major ommission that I would love to see addressed in future iterations.

Some concluding thoughts

I hope that the Census Bureau will continue this effort and even expand the effort to include other key questions related to disaster and climate change and especially issues like migration and long-term disaster recovery. I would love to see a group of disaster scholars propose updated questions - maybe a roundtable topic for the upcoming Natural Hazards Workshop?

I’m looking forward to hearing from you in the comments below or on Twitter or Linkedin. What questions do you wish the Census Bureau and its partners (including HUD but not FEMA or DHS) would include on future phases of the Household Pulse Survey? What trends do you see in the data that I did not discuss?

Stay tuned for Part II of this post, where I will look at the Household Pulse Survey data on damage and experience of environmental hazards and stressors.