3. Size of the economy in flood-prone areas

The Netherlands is very vulnerable to flooding. Not only is a major part of the country below sea level, but there are also several large rivers that empty into the sea. That means the risk of flooding comes not just from the sea, but also from elevated river levels, as well as dyke failures¹¹⁾.  Climate change increases the risk of extreme amounts of precipitation and leads to rising sea levels, further increasing the risk of flooding. In this part of the study, we quantified the economy of the areas of the Netherlands that are most susceptible to flooding. Specifically, we quantified how much value added and production were generated, and how many employment opportunities , in flood-prone areas with various flood risks. 

3.1 How do we measure the size of the economy in flood-prone areas?

We made use of the LIWO maps of flood-prone areas to quantify the flood risks¹²⁾. Specifically, we used composites of the water depth maps, which show the maximum depth for certain flood chances, expressed in terms of predicted frequency. The depth was determined based on model simulations. We used three potential flood scenarios identified by the LIWO in our analysis: ‘extremely low risk’ (once every 100 thousand years or less), ‘low risk’ (once every thousand years), and ‘high risk’ (once every 10 years). These maps show potential floods that are unlikely to coincide in reality. They show the combined effect of four types of floods for protected and unprotected areas situated along primary and regional water systems: 

A. Flooding of areas close to water, unprotected by dykes
B. Failure of primary dams and dykes
C. Failure of non-primary dams and dykes
D. Flooding from regional water system

We marked all locations where the maximum depth of water is greater than 0 metres as flooded. Several flood scenarios are overlaid in Figure 3.1.1, which shows that, in the ‘high risk’ scenario, floods mainly occur in the areas along the rivers: Limburg, the centre of the Netherlands, and the area around the IJssel river. Flood type A, the flooding of areas unprotected by dykes, is the dominant type in that scenario. The dominant type in the ‘extremely low risk’ scenario is type B, the failure of primary dams and dykes; this is seen on the map as the flooding of Flevoland, Zeeland, Zuid-Holland, and larger areas around the rivers. 
 

Figure 3.1.1 Flood risk map for various scenarios, taken from the Environmental Health Atlas, flooding theme

In order to establish an accurate picture of the economy of flood-prone areas, we linked regional economic data to flood risk maps¹³⁾ in our analysis. The economic data that we used was sourced from CBS regional accounts, which includes value added, employment opportunities, and production, both per municipality and per economic activity¹⁴⁾.  For each of these variables, we show what amount is generated in flood-prone areas. However, further analysis is needed to translate that data into information on flood-related economic damage and loss. Such an analysis could also look for correlations between depth and damage, as well as the amount of time that would be needed to repair that damage. While interesting, that analysis is beyond the scope of this study. 

Our own analysis provides information on the economy in flood-prone areas for various flood risks. They are grouped according to economic sector and according to region, using the COROP-plus classification¹⁵⁾. 

3.2 National economy

We will first discuss the economic output generated by flood-prone areas, sorted by flood risk. Gross domestic product (GDP) is generally accepted as a good indicator of economic output, and this is why our analysis looked at the share of the GDP generated in the flood-prone areas shown on the different flood risk maps. As can be seen in Figure 3.2.1, that share was roughly 53 percent in areas with an extremely low risk of flooding. It was 34 percent for areas with a low risk, and a barely 1 percent for high-risk areas. Another point to note is that these percentages have been very stable over the past twenty years¹⁶⁾.

However, according to Figure 3.2.2, GDP generated in flood-prone areas is rising sharply in absolute terms. For instance, in the ‘extremely low-risk’ scenario, GDP rose by just under 200 billion euros: from 340 billion euros in 2010 to 525 billion euros in 2022. Given that the chances of flooding are constant, this rise can be explained entirely by the rise in total GDP (see Figure 3.2.1).

3.3 Differences between sectors

Our analysis of GDP can be supplemented by looking at the different economic sectors. We used economic data from 2022 for the following analysis. 

Table 3.3.1 shows the impact of the ‘extremely low risk’ scenario on various sectors. We calculated the full-time equivalents (FTEs), value added, and production for each flood-prone area. 

Value added generated in flood-prone area

Figure 3.3.2 shows the amount of value added generated in different sectors in areas likely to flood in the ‘extremely low risk’ scenario. As can be seen, in the ‘extremely low risk’ scenario the energy sector had the largest percentage of value added: almost 70 percent was generated in flood-prone areas. That percentage was around 50 percent for most other sectors, down to as low as 42 percent for the manufacturing sector. This is probably because much of the economic activity in that sector (15 percent of value added) is localised around Eindhoven¹⁷⁾, where little to no flooding occurs. This compensated for areas like the Rijnmond region (around Rotterdam), which contains 7 percent of the Dutch manufacturing sector and 85 percent of which would flood in the ‘extremely low risk’ scenario. 

Figure 3.3.3 shows the absolute GDP values generated in areas at risk of flooding. The values are grouped by sector and by risk scenario. The sector ‘Trade, transport, hotels, catering’ had the largest absolute GDP value: 102 billion, 69 billion, and 2 billion euros for the three different risk scenarios. The public sector and health care come in second place. In terms of value added, these sectors are also the largest at the national level.  

The largest share of value added in the energy sector comes from the Rijnmond and Groningen areas (including Eemshaven)¹⁸⁾. In each of those areas, more than 80 percent is likely to flood in the ‘extremely low risk’ scenario. The energy sector is also the only sector in which an increasingly large share of value added is generated in flood-prone areas, whereas that variable barely changes in the other sectors. For the energy sector, the share increased from 60 to almost 70 percent between 2010 and 2022 (in the ‘extremely low risk’ scenario). This is mainly due to growth in this sector in regions where flooding occurs over large areas, specifically the regions surrounding Amsterdam¹⁹⁾, Flevoland, Rijnmond, and Groningen (including Eemshaven).  

Employment in risk areas

We performed a similar analysis on employment figures, as expressed in FTEs. Like the earlier analysis, this one shows which sectors are relatively sensitive to flood risks and which sectors generate the most jobs in risk areas. 

As was the case for value added, the energy sector is the most sensitive sector in terms of employment, along with the information and communication sector. In the ‘extremely low risk’ scenario, almost 60 percent of the employment opportunities in those sectors were generated in flood-prone areas. Over 33 percent of employment in the information sector came from the Amsterdam and Utrecht areas, of which 64 and 72 percent would be flooded, respectively, in this scenario. This means that 40 percent of total employment in risk areas can be attributed to these two regions. However, the differences in percentages are much smaller for employment than they are for value added. In almost all sectors in this scenario, between 50 and 60 percent of employment was generated in flood-prone areas, the only exception being manufacturing at 43 percent. The two regions with the most employment in this sector (Eindhoven and Twente, 11 and 5 percent of the national total, respectively) hardly ever flood.  

For all three scenarios, public sector and health care had the largest absolute number of jobs in risk areas; it was also the largest sector in terms of jobs in general.  

3.4 Highest-risk areas in the Netherlands

The method that we chose to measure the economy of flood-prone areas made it possible to analyse various economic sectors, and also to look at the economy of each individual region. 

Figure 3.4.1 and Figure 3.4.2 show the variation in the value added in the different risk areas in the ‘extremely low risk’ scenario. The provinces of Flevoland and Zuid-Holland stand out in particular when looking at the at-risk percentage of value added per region. That area is largest in the Betuwe-²⁰⁾ and Delfzijl regions, where respectively 96 and 95 percent of value added is generated in areas with a significant flood risk. 

The Rotterdam (Rijnmond) and Amsterdam regions have the highest at-risk value added in absolute terms. These regions generate roughly 61 billion euros in value added in flood-prone areas in the ‘extremely low risk’ scenario. 

3.5 Major differences by sector and region

In addition to differences between regions and sectors, we can also look at the effect on sectors by region.²¹⁾

This report shows the results for the following three COROP-plus regions: Utrecht City, Southeast Utrecht, Southwest Gelderland (see Figure 3.5.1). These regions were selected because, despite bordering each other, they demonstrate major differences in the (relative and absolute) amounts of value added generated in at-risk areas. 

 Figure 3.5.2: Percentage of value added in flood-prone areas for three COROP-plus regions, by sector, ‘Extremely low risk’ scenario, economic data from 2022. Data on the energy and the water supply & waste management sectors cannot be shown for reasons of confidentiality.

Figure 3.5.3: Share of value added in flood-prone areas for three COROP-plus regions, by sector. ‘Extremely low risk’ scenario, economic data from 2022. Data on the energy and water supply & waste management sectors cannot be shown in this breakdown by region.

Figure 3.5.2 clearly shows major differences between these three adjoining COROP-plus regions. In Southwest Gelderland (the Betuwe region), 96 percent of the value added for all sectors is generated in flood-prone areas. The percentage varies from 60 to 80 percent in Utrecht City, depending on the sector. For most sectors, the smallest share of at-risk value added is generated in the Southeast Utrecht region: between 12 and 88 percent. 

The area of Utrecht City that is at risk of flooding is smaller than that of Southwest Gelderland, in relative terms. Nevertheless, Figure 3.5.5 clearly shows that Utrecht City has the highest absolute value added for flood-prone areas across all sectors. Public sector and health care and the trade, transport, and food & accommodation services sector have the largest absolute value added in flood-prone areas. The same is true at the national level (see Figure 3.3.3). In Utrecht City, the public sector and health care are the most at-risk, whereas at the national level this is the case for the trade sector.

3.6 Summary and further research

In this study, we have linked maps showing climate risks with economic data. More specifically, we have combined flood risk maps with several different economic variables. The approach adopted could also be used to investigate other climate threats, coupled with a wider range of economic variables. 

In this report, we measured the economy of flood-prone areas in the Netherlands. Among other things, the report shows to what extent value added is generated in areas at risk of flooding, as well as the absolute value added. Using our chosen methodology, we were able to investigate the variables stated, both by sector and by region. Our use of economic activity figures (i.e. value added, production, and employment) at a regional level complements earlier research on flood damage done by Deltares²²⁾ and De Nederlandsche Bank (DNB)²³⁾. DNB analysed the amount of damage incurred by companies to fixed assets due to flooding. Meanwhile, Deltares’ Damage and Fatality assessment model used the national accounts of Statistics Netherlands (CBS) to calculate the damages caused by business interruptions. For this study, we analysed several different possible flooding scenarios.

We found that, under the ‘extremely low risk’ scenario, roughly 53 percent of the Netherlands’ GDP came from flood-prone areas. This was only 1 percent under the ‘high risk’ scenario. Additionally, the percentages were found to have been stable over the past twenty years. However, GDP value in the flood zones did rise from 340 billion euros to 525 billion euros between 2010 and 2022. This can be explained by the period of GDP growth that started in 2010. 

There were also major differences between the economic sectors in terms of the share of value added in flood-prone areas. That share was nearly 70 percent for the energy sector under the ‘extremely low risk’ scenario, compared to slightly over 40 percent for the manufacturing sector. 

The different scenarios greatly affected regions’ vulnerabilities relative to each other. Under the ‘extremely low risk’ scenario, the Betuwe and Delfzijl regions were the most vulnerable of all regions: 95 percent of their value added came from areas at risk of flooding. The value added in at-risk areas in the Rijnmond and Amsterdam regions amounted to 61 billion euros per region, making them the highest in absolute terms. Limburg was especially vulnerable under the ‘high risk’ scenario, both in relative and absolute terms: 17 percent of North Limburg’s value added, roughly 2 billion euros, came from at-risk areas.   

Possibilities for further research

The notion of linking regional economic data to climate risk maps offers many opportunities. For instance, it can be used to show the Netherlands’ vulnerability to climate risks. To demonstrate this, we investigated which amounts of value added, employment, and production came from areas at risk of flooding. This analysis could be further expanded upon in several ways.

The use of flood data means that there are many maps available for analysis. While our study used the aggregated risk of four different flood scenarios, future research might focus on one specific scenario, such as ‘Flooding of areas next to water, unprotected by dykes’. Another possibility would involve looking at the (combinations of) local flood risk scenarios available from the LIWO. We published economic data at the COROP-plus level for the current study, but these regions are too large to show the economic effects of more localised flooding. Unfortunately, it will take considerable work to find out which municipalities have enough economic data available to be able to carry out research at that level. 

A final possibility involves using our analysis for other climate threats like drought, extreme precipitation, and heat. However, this would require investigating how to demarcate the risk areas for each of those threats. And, as with the potential study on local scenarios, it would first be necessary to ascertain whether there is enough economic data available for study (and publication) at the right level of aggregation.

3.7 Technical explanation

Flood maps

As explained in section 3.1, this research used flood maps from the National Information System Water and Floods (LIWO). These maps show the risk of various kinds of flooding in the Netherlands; the map resolution is 25 by 25 metres. More specifically, we used the map entitled Maximum depth of flooding in the Netherlands (‘Maximale overstromingsdiepte Nederland’). This map shows the maximum water depth across the Netherlands for various flood risks. There are five different versions of the map: 1) extremely low risk, 2) very low risk, 3) low risk, 4) medium risk, and 5) high risk. Our analyses focused on maps 1, 3, and 5. 

• 1) Extremely low risk: shows which areas should flood once every 100 thousand years (or less).
• 3) Low risk: shows which areas should flood once every thousand years.
• 5) High risk: shows which areas should flood once every 10 years.

All these maps show flooding that in reality would be unlikely to occur simultaneously. The maps are based on four types of flooding of protected and unprotected areas, situated along primary and regional water systems.

Before performing our analysis, we first discretised the flood map. Every location where depth of water was greater than 0 metres was marked as flooded, which provided us with the maps shown in Figure 3.7.1. These maps therefore no longer differentiate between various depths.

Determining the flooding percentage for each municipality

For each municipality, we determined how much would be subject to flooding. This was not done using the flooded surface area for a given municipality²⁴⁾, but rather using the number of businesses flooded in a municipality. The GIS analysis was done using data from the Basic Register of Addresses and Buildings (BAG). As part of the analysis, every building with the status ‘Object in use’ (Verblijfsobject in gebruik) and any function other than ‘residential’ (woonobject) was counted as a business. This is how we calculated the percentage of businesses flooded in each municipality; this was a much more suitable method for linking economic activities to flood maps than using the surface areas flooded.

Figure 3.7.1 Flood maps after discretisation. From left to right: 1) extremely low risk, 3) low risk, 5) high risk. The lines on the map represent municipal boundaries

Economic data

The economic data we used was sourced from the CBS regional accounts. Those accounts contain economic information broken down by year, municipality, and REGKOL level; it includes figures on value added, production, and employment, among others. Where necessary, the figures for specific companies were broken down by municipality, based on the number of jobs in each municipality. The REGKOL classification is based on the SBI, but on a lower level of aggregation than the national accounts classification, which is itself based on the SBI. Regional accounts are published at the COROP-plus level, aggregated to 21 economic sectors. We opted to use the same level of aggregation for this publication.

The economic data also contains data on extraterritorial regions, whose economic activities cannot be ascribed to a particular region or municipality within Dutch borders. Those activities include work in Dutch embassies abroad, military missions, or mining and quarrying on the Dutch part of the continental shelf. This data was not included in this study.

Linking flood maps to economic data

Linking the maps of flooded areas to economic data was an important step in analysing the data from those areas. Both the maps and the data are available at the municipal level, and were linked at that same level. The percentage of flooded business per municipality was applied to all economic variables. All the time series in this analysis used the same flood risk maps, combined with economic data from the relevant year.