Even when the water is gone, Houston will not be as it was.

Residents who fled their homes to escape Hurricane Harvey will return find their cars junked, their houses full of mold, their furniture destroyed. And they’ll have a visceral, first-hand experience of just how bad it can get when Houston floods.

Natural disasters are many things, and they can be an impetus for people to move. Leah Boustan, an economist at Princeton—along with her colleagues Matthew Kahn, Paul Rhode, and Maria Lucia Yanguas—have tracked migration after 5,000 natural disasters in the United States between 1920 and 2010. I spoke to her about what to expect after Houston recovers from Hurricane Harvey.

A transcript of our conversation, condensed and edited for clarity, is below.

Sarah Zhang: Harvey is a serious disaster. Your data set captures 5,000 other natural disasters, ranging from a bad winter storm to Katrina. What kind of response do you see to the most severe natural disasters like Katrina and perhaps Harvey?

Leah Boustan: The typical disaster includes everything from floods, tornadoes, earthquakes, hurricanes, and major winter storms. We do find a migration response to an event like that. But for a very severe disaster—and Harvey looks like it’s going to be in that category—the response is twice as large.

Part of that has to do with people learning about the risk factors. Maybe they didn’t know the area they’re living in is so susceptible to storms.

Part of it is watching whether the existing infrastructure really works.There’s discussions now about Houston not really having enough of a drainage system. People might have known, yes, there are tropical storms, but they may not have understood the tropical storm is going to be such a devastating effect.

Zhang: Like when levees broke during Katrina.

Boustan: Yes, does the bayou hold or does the floodwall to seawall hold? You can observe another city going through it but then it may not be salient until it actually happens to you.

Zhang: The Gulf Coast is obviously pretty prone to hurricanes, so it’s not surprising that a big one would eventually hit Houston. Do people in disaster-prone areas respond differently than those in places that don’t get hit as often?

Boustan: The results there are really striking. We use some of the geography for different areas to say what the characteristics are of an area that make it susceptible—if you’re on the coast, if you’re in the Gulf of Mexico, if you’re along a major river, if you’re at a low elevation. And it’s precisely in the places in the path of regular disasters that, when there’s a really big disaster, people tend to start to pack it in and leave.

That was sort of puzzling to us as economists—an economist being the type of person who’s like, “Shouldn’t people be able to gather information and assess risk and make these decisions, like looking a spreadsheet?” There’s new information coming when it’s a surprise: You’re not expecting a disaster at all, and you haven’t had the experience of stocking up your house and getting the candles and getting the food ready. [You might think that] people who are in the wake of disasters might know how to respond and it’s not really that surprising. But that doesn't seem to be the case.

Zhang: You usually study migration in a more general context. How did you get interested in specifically looking at natural disasters?

Boustan: I’ve worked on the black migration from the rural South to industrial cities during and after World War II, and one of the theories in that literature is that the migration out of the South began in part in response to a devastating flood in 1927 of the Mississippi River.

Zhang: The years you studied, 1920 until 2010, cover a pretty long period, including a lot of changes in how the government responds to disaster. How has the government stepping in affected how people react to, say, a hurricane?

Boustan: In the 1920s and 1930s, the federal government kind of contracted out the disaster response to the Red Cross. It’s not quite fair to say that it was a completely private response to disasters at the time, dependent on only private donations, but there wasn’t the same kind of government infrastructure like FEMA [the Federal Emergency Management Agency] that we’re used to today.

FEMA started in the early ’70s, and it gets its own independent status as an agency in ’78. We looked at before and after the ’70s, there was a hypothesis, well—and I’ve heard a lot of this post-Harvey—that when you have centralized disaster response, there’s not really an incentive to move out.

Zhang: This is the idea of “moral hazard”: When you’re protected from the consequences of your actions, you take more risks.

Boustan: Right, like there’s going to be big government payout, and that encourages people to stay put in places that are risky. You know you’ll get your FEMA payout. We actually didn’t have any difference of course in the migration response before and after FEMA.

But of course, this is really just a before and after, and there’s a lot things about the ’40s, ’50s, and ’60s, that could be different about the ’80s, ’90s, and 2000s. In particular, you can really see the number of disasters and severity of disasters increasing. There are two things going on that could be kind of confounding. On one hand, there’s government response. On the other hand, disaster activity is getting worse. We can’t really separate those two things, but it looks like because disasters are getting worse, there’s just as much of a migration response more recently than there was in the ’30s, ’40s, and ’50s.

Zhang: With climate change, a lot of coastal cities may actually become uninhabitable. Could we start to see migration away from these cities because of natural disasters?

Boustan: That’s one of the things we were thinking that explains why a severe disaster in a prone area leads to more migration. If it is the case that disasters are getting more frequent and more severe with climate change, and you’re in the path of disaster activity, when you see that kind of thing happening, you say, all right, now is a good time for me to get out.