Floods are the costliest type of natural disaster in Europe, causing more than €12 billion in damages each year on average (European Environment Agency 2020, Fatica et al. 2024). And we are increasingly witnessing flooding events with a much larger damage bill. The 2021 floods in Belgium, Germany, and the Netherlands resulted in estimated damages of €44 billion, while the 2023 floods in Slovenia caused direct damages of around €10 billion, equivalent to 16% of the country’s GDP. Increased flooding results from more frequent and intense heavy precipitation events, which hydrological models project will intensify in the coming decades (IPCC 2021).
A growing body of research has documented the macroeconomic impact of floods and other climate-related disasters, showing their significant effects on GDP and inflation (Parker 2018, Kabundi et al. 2022, Sudo and Hashimoto 2022). At the aggregate level, flooding disrupts economic activity, yet its impact on prices remains ambiguous, raising questions about whether floods act more as supply or demand shocks (Krebel et al. 2025). Even less is known about the role of adaptation investments, such as flood defences, in shielding the economy from flood damage. While flood defences do not address the root cause of floods, they are the most accessible tool for governments to mitigate their frequency and impact (Fried 2022).
In recent research (Ficarra and Mari 2025), we study how floods affect output and prices at the sectoral level in England, and we provide a novel empirical assessment of the role of adaptation investments. England is a highly flood-prone country (Figure 1) and has seen growing debate over sufficient levels of flood defence investment (Financial Times 2024).
Figure 1 Number of floods and flooded area by year
Note: We treat each flood event as a single flood, assign it to every ITL3 area hit, and compute the flooded area accordingly.
Source: EA Recorded Flood Outlines and authors’ calculations.
The impact of floods is heterogeneous across sectors
To estimate the sector-level impact of floods, we use data on gross value added and prices from the UK Office for National Statistics for 1998–2021, along with county-level data for each of the 309 local authorities (ITL3 regions) in England.
We find that floods affect sectors differently, not just in terms of magnitude, but also in timing and sign. Figure 2 shows that in some sectors (manufacturing and trade in particular), output dampens immediately, suggesting supply-side disruptions – for instance, related to lack of accessibility of business premises and supply chain effects. In other sectors (such as construction and food and beverage services), it takes longer to see an impact, in line with this being a consequence of demand-side effects. Accommodation services and civil engineering even see temporary gains due to the increasing demand in the aftermath of these events, as a result of forced displacement from flooded residences and urgent remedial works to key infrastructures.
Figure 2 Gross value added response to floods by sector
Note: Dynamic impulse response functions of gross value added (GVA) to a one-standard-deviation increase in the number of floods. Estimates are based on LP-IV. All specifications include ITL3 and year fixed effects. Controls include population size and one lag of GVA. Standard errors are clustered at the ITL3 level. Shaded areas denote 68% and 90% confidence bands.
Inflation exhibits similar sectoral heterogeneity (Figure 3). Except for textile manufacturing, floods generally reduce inflation. Most sectors experience immediate but temporary price effects, while wholesale and retail trade show delayed and persistent responses. These findings reconcile opposing views in the literature by demonstrating that floods can act as both supply and demand shocks, depending on the sector. The aggregate effect likely depends on the economy’s sectoral composition. Moreover, our results challenge the assumption that climate change only impacts headline inflation through food and energy prices, showing that floods also influence core inflation sectors.
Figure 3 Inflation response to floods by sector
Note: Dynamic impulse response functions of gross value added to a one-standard-deviation increase in the number of floods. Estimates are based on LP-IV. All specifications include ITL3 and year fixed effects. Controls include population size and one lag of inflation. Standard errors are clustered at the ITL3 level. Shaded areas denote 68% and 90% confidence bands.
Can flood defences help?
The effectiveness of adaptation investments in mitigating flood damage is a critical policy question, especially as governments face mounting pressure to act. During the UK’s 2023 flooding season, poor flood defences were blamed for the rising number of affected households (Horton 2024). The National Audit Office found that the number of properties expected to receive better protection from flooding by 2027 has been cut by 40%, and 500 of 2,000 new flood-defence projects have been abandoned (see Haliday 2023, Horton 2023). Maintenance work is also needed, as over 4,000 of England’s vital flood defences are in very poor conditions (Halliday 2023).
To study the role of flood defences, we use counties’ balance sheet data on revenue expenditure for flood defence, land drainage, and coast protection for the fiscal years 2008–2009 to 2023–2024, and cumulate expenditure over time to build a proxy for adaptation capital. We study the role of adaptation policy along both the extensive and the intensive margin. While we expect flood defences to be effective in reducing flood risk (the extensive margin), whether they can help once a flood hits (the intensive margin) is less obvious.
We first estimate how increases in adaptation expenditure and capital affect flood frequency in counties that are prone to flooding (Figure 4).
Figure 4 Adaptation investments: Extensive margin
Note: The bars represent point estimates of a regression where the dependent variable is the number of floods in county i at time t+h. A county is flood prone when it is flooded more times than the national average. The skewers represent 99% confidence intervals. Black bars report the point estimate when the independent variable is flood-defences expenditure over GDP, while red bars report the point estimate when the independent variable is flood-defences capital over GDP. We include three lags of the dependent variable and population size, and 1 lag of GDP. All regressions include ITL3 and year fixed effects, and standard errors are clustered at the ITL3 level.
Our results show that adaptation strongly reduces the likelihood of floods, especially when investments accumulate over time. Specifically, a 1-percentage-point increase in adaptation expenditure as percentage of GDP (black bars) reduces the number of floods by 76.46 units after two years. Three caveats apply:
- The delayed effect of adaptation expenditure is in line with the concept of ‘time to build’ (Ramey 2020).
- Observed expenditure levels are far lower than 1 percentage point of GDP – the median expenditure is 0.002% of GDP.
- The rarely significant coefficients are consistent with the fact that expenditure itself does not necessarily reduce flooding.
What appears to matter more is adaptation capital, and capital is only built by consistently investing in flood defences. The red bars in Figure 4 highlight this finding: a 1-percentage-point increase in the stock of adaptation capital as a percentage of GDP is associated with 23.7 fewer floods in year t, 33.3 in year t+1, 20.6 in t+2, 21.9 in t+3, 44.9 in t+4 and 40.8 in t+5. A county increasing its stock of adaptation capital by the median amount in year t will be flooded 0.4 fewer times by year t+5, which corresponds to a 3.5% reduction in the number of floods compared to the average local authority. These results strongly support the case for sustained investments in adaptation capital.
We next examine the intensive margin: can flood defences reduce economic damage when floods occur? While flood defences can prevent flooding, it could mean that in well-protected areas only extremely severe conditions trigger a flood, potentially still causing significant damages. To answer this question, we estimate a state-dependent model that distinguishes between high and low adaptation-expenditure counties (Figure 5).
Figure 5 Adaptation investments: Intensive margin
Note: Dynamic impulse response functions of gross value added (GVA) to a one-standard-deviation increase in the numbers of floods: high (blue line) and low (red line) adaptation-expenditure state. The state is defined using a regime-switch dummy as in Ramey and Zubairy (2018). Estimates are based on state-dependent LP-IV. All specifications include ITL3 and year fixed effects. Controls include population size and one lag of GVA. Standard errors are clustered at the ITL3 level. Shaded areas denote 90% confidence bands.
With the exception of the construction of buildings and manufacturing of textiles, differences in point estimates are substantial, though overlapping confidence bands prevent definitive conclusions. Nevertheless, counties with lower levels of adaptation expenditure seem to experience a significantly larger impact of flooding on the accommodation and civil engineering services, suggesting floods are more destructive there. Similarly, gross-value-added declines in wholesale trade and in food and beverage services primarily occur in counties with low adaptation investments.
When a flood happens, these areas are less protected and sustain larger economic losses. Adaptation capital likely reduces the destructive power of floods by limiting the overflow of water or simply slowing it down. The presence of public adaptation investments is also likely correlated with private adaptation measures, such as flood gates, which we don’t capture through the data and that further decrease the damages from flooding.
Policy implications
The evidence described in this column has three key policy implications. First, the heterogeneous impact across sectors calls for tailored responses to flood events. Given budget constraints, governments should take great care in identifying the hardest-hit sectors to maximise the effectiveness of post-disaster interventions. Second, while our results do not provide conclusive evidence as to the dynamics of aggregate inflation, they highlight significant price effects in core inflation components, which have important implications for monetary policy. Lastly, building adaptation capital is an effective tool to reduce the probability of flooding and, albeit with weaker evidence, the extent of damages once a flood occurs.
The decision to invest in adaptation, however, needs to account also for the significant associated maintenance costs to ensure it maintains its functionality. Furthermore, the choice needs to be weighed against policy alternatives that potentially encompass supporting and managing the relocation of households and businesses away from areas particularly exposed to flooding.
Source: cepr.org
