Although the source and duration of the destruction differ, wars and natural disasters disrupt economies in similar ways. This column examines the effects of natural disasters and the impact of government rebuilding policies by focusing on the impacts of Hurricane Maria, which struck Puerto Rico in 2017. The findings underscore how the scale and design of post-disaster assistance shape both reconstruction and return migration.
Three years into Russia’s full-scale invasion, the estimated direct damage to Ukraine’s buildings and infrastructure now exceeds $176 billion, equivalent to nearly 100% of Ukrainian GDP in 2024 (World Bank et al. 2024). Although the source and duration of destruction differ, wars and natural disasters disrupt economies in similar ways. In addition to the tragic loss of human life, these shocks damage physical capital including infrastructure, private capital, and housing, and often prompt large-scale temporary and permanent migration. Evidence from post-disaster recoveries, such as those following major hurricanes, can offer potentially useful insights for planning Ukraine’s reconstruction (see Becker et al. 2023 for a comprehensive macroeconomic framework).
Building on a long tradition of work that studies the impacts of aggregate shocks to capital and subsequent rebuilding after WWII (King and Rubelo 1993, Gilchrist and Williams 2004, Auray et al. 2014), my research (Aron-Dine 2025) seeks to understand the effects of natural disasters and the impact of government rebuilding policies. The study focuses on the heterogeneous impacts of Hurricane Maria, a Category 4 hurricane that struck Puerto Rico, an island territory of the US, in 2017.
Destruction induces migration for many reasons
The initial destruction shock to local capital stocks lowers the value of staying in the impacted location in several ways. First, damage to housing directly reduces the utility that households derive from living in their house and, in aggregate, reduces the effective supply of housing in a location, pushing up prices and rents. Second, damage to infrastructure such as utilities, roads, and schools diminishes the quality of life in a community, even for those with intact homes. Third, firm closures caused by physical damage or infrastructure disruptions reduce local labour demand. Faced with lower earnings, diminished services, and more expensive housing, households that can afford to migrate to a better and safer location are likely to do so, at least temporarily until the conflict ends and rebuilding can occur.
Measuring migration in the aftermath of large destructive shocks is difficult. Conventional sources of demographic statistics, such as household surveys, are often only available with very long lags and at annual frequency. In addition, the shocks themselves disrupt such data collection, making it even harder to track population movements in real time. In my work on the aftermath of Hurricane Maria, I therefore relied on alternative statistics such as commercial flight records to infer high-frequency population movements. A comparable approach has been used by the United Nations during the war in Ukraine, where data on border crossing provide measures of population flows (UNHCR 2025). These alternative data sources, while providing timely aggregates, reveal nothing about the characteristics or motives of migrants and miss internal displacement altogether.
Although the motives for leaving an area after a natural disaster may differ from those during an armed conflict, the aggregate migration profiles that follow these shocks are remarkably similar. Figure 1 plots monthly net passenger outflows from Puerto Rico after Hurricane Maria (in 2017) and monthly net border crossing out of Ukraine after Russia’s full-scale invasion (in 2022). To make the outflows comparable, each series is scaled by the relevant pre-shock population (roughly 3.2 million for Puerto Rico and 41 million for Ukraine). In both cases, out-migration peaks in the immediate months and then declines. There is even return migration within the first year of the shocks. The return migration is much larger in the case of Puerto Rico, where the hurricane was a one-time event and the long process of rebuilding began within a few months. In contrast, return migration to Ukraine has been much slower because of the ongoing conflict.
Figure 1 Net-migration outflows after the Russian Invasion of Ukraine and Hurricane Maria (percent of pre-shock population)
There are many key differences between wars and natural disasters that could explain the differences in return migration. Natural disasters are usually short-lived events and reconstruction can begin as soon as the storm has passed and debris has been cleared. In contrast, it remains unclear if and when the war in Ukraine will have a defined end. As the fighting continues, Ukraine faces continuing Russian missile and drone attacks. Each new strike not only destroys more capital and adds to the ultimate cost of reconstruction, but also makes it impossible to begin the work of rebuilding. After a natural disaster in the US, the main uncertainty is around the generosity of aid from the federal government, while there are many sources of uncertainty about the future of Ukraine. The timing and terms of the end of the conflict depends on many uncertain geopolitical forces both within Ukraine but also in Europe, Russia, the US, and the rest of the world. Without a credible ceasefire and timetable for rebuilding, those who have fled have little incentive to return.
Who leaves and who returns?
Identifying who leaves and who eventually returns is difficult in Ukraine’s still-evolving crisis, but Hurricane Maria offers a useful benchmark. Using direct survey evidence and census data, I find that homeowners who suffer direct property damage tended to migrate only temporarily. These households returned to Puerto Rico after a few months to repair their homes. In contrast, it appears that young renters were more likely to move away permanently after the hurricane. These households were not locked in place through homeownership and had the most to gain from relatively higher wages in the mainland US.
Early indicators suggest that similar mechanisms may be playing out in the decision to return to Ukraine. The top destinations for Ukrainian refugees are Germany, Poland, and Czechia. Once established outside Ukraine, it may be challenging to induce young workers to return. However, one crucial difference between these two settings is that Puerto Ricans are US citizens and face no legal barriers to employment anywhere in the US, whereas Ukrainian refugees’ labour market access depends on each host country’s immigration policies.
Standard cost-benefit models of migration apply imperfectly in war zones, where decisions are often driven by threats of physical violence rather than economic calculus (Becker and Ferrara 2019). For those who choose to remain in Ukraine, they face a non-trivial risk of attack from Russia. The conflict also imposes very large intergenerational costs. Ukrainian children face prolonged school disruptions and war-related trauma and thousands of children have been forcibly transferred to Russia. The cumulative effect of these psychological scars and disruption to human capital formation will likely far outlast any physical destruction of buildings and infrastructure (Akbulut-Yuksel 2022, Égert and de La Maisonneuve 2023).
How many people return will depend on government policy
Evidence from natural disasters in the US underscores how the scale and design of post-disaster assistance shape both reconstruction and return migration. More research is needed to better understand whether these findings generalise to other contexts. Over the last 40 years, US policy has been to provide very large federal grants to local governments and homeowners hit by severe natural disasters. A recent paper (Roth Tran and Wilson 2025) shows that on average during this period of generous government rebuilding policy, areas of the US that experienced a natural disaster were able to recover to pre-disaster levels of household income, property values, and population within a decade. In contrast, during the era before structured federal post-disaster assistance in the US, Boustan et al. (2020) document net out-migration and lower property values after disasters. Taken together, these two studies show that it is possible, but expensive, to fully offset the long-run economic damage of capital destruction. Well-structured aid can also generate large positive externalities by coordinating communities to rebuild (Fu and Gregory 2019).
However, financing rebuilding poses a substantial challenge for Ukraine. Disasters in the US typically affect only a very small share of the population, so spreading relief costs across the federal tax base is manageable. For example, the US spent roughly $900 per tax filer in 2004 over two decades to finance rebuilding after Hurricane Katrina: $120 billion spending on Katrina (source: CRS) and 132.4 million tax filers in 2004 (source: IRS). The US Treasury has historically been able to borrow at low interest rates, given the US dollar’s status as the world’s reserve currency. Ukraine, by contrast, faces an estimated $486 billion reconstruction bill, or about $13,000 per resident, as well as much higher borrowing costs. Figure 2 plots the yield on a one-year Ukrainian government bond (a higher bond yield implies higher borrowing costs). In the immediate aftermath of the full-scale Russian invasion, Ukrainian government bonds traded at prices typically seen for counties about to default. Since then, prices have recovered somewhat, but borrowing costs remain elevated. Successfully rebuilding Ukraine will depend on large-scale, low-interest lending and grants from international financial institutions and, where legally feasible, on redirecting seized Russian assets (Gorodnichenko and Becker 2024).
Figure 2 Ukrainian one-year government bond yield (%)
Data source: Trading View
Whether displaced Ukrainians return will depend not only on security conditions, but also on the timing and scale of reconstruction. The order of rebuilding is important: rapid reconstruction of infrastructure, such as the power grid, raises the value of all subsequent investments in housing and firm capital. Aid directed to homeowners may also potentially increase return migration. Such programmes often face a trade-off between better targeting and more flexibility. Finally, if current host countries allow for long-term employment in higher-wage EU labour markets, the opportunity cost of returning to Ukraine will rise, resulting in fewer Ukrainians returning. The scale of rebuilding in Ukraine and immigration policy in the EU will likely determine the scale of return migration at the end of the conflict.
Source : VOXeu
