Policy Research Working Paper 8745 B eyond the G ap How Countries Can Afford the Infrastructure They Need While Protecting the Planet Background Paper Global Investment Costs for Coastal Defense through the 21st Century Robert J. Nicholls Jochen Hinkel Daniel Lincke Thomas van der Pol Sustainable Development Practice Group Office of the Chief Economist February 2019 Policy Research Working Paper 8745 Abstract Sea-level rise threatens low-lying areas around the world’s empirical observations of protection standards are limited. coasts with increased coastal flooding during storms. One Hence, the estimates should be considered as indicative, and response to this challenge is to build or upgrade coastal flood this remains an important topic for future research. Further, defenses. This report examines the potential investment building defenses is not a one-off capital investment. Over costs of such an adaptation strategy applied globally over the 21st century, the cost of a comprehensive protection the 21st century for sea-level rise scenarios consistent with strategy is dominated by maintenance costs in all the cases three Representative Concentration Pathways and 3 Shared considered in this report. This indicates that in addition to Socioeconomic Pathways. For all the protection models capital investment, the development of appropriate institu- considered, much less than half of the world’s coast is pro- tions and governance mechanisms to deliver maintenance, tected. The total defense costs are significantly higher than as well as the necessary funding streams, are essential for earlier estimates, amounting to as much as US$18.3 trillion. such a protection-based adaptation strategy to be effective. With cost-benefit analysis, there are large uncertainties and This paper was commissioned by the World Bank Group’s Chief Economist office for Sustainable Development Practice Group and is a background paper for the World Bank Group’s report: “Beyond the Gap: How Countries Can Afford the Infrastructure They Need While Protecting the Planet.”  It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/research. The authors may be contacted at R.J.Nicholls@soton.ac.uk. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Global Investment Costs for Coastal Defense through the 21st Century Robert J. Nicholls (University of Southampton, Southampton, UK), Jochen Hinkel, Daniel Lincke, Thomas van der Pol (Global Climate Forum, Berlin, Germany) JEL Classification: Q54 Keywords: Sea-level rise; coastal flooding; adaptation costs; protection Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 2 1 Introduction This report evaluates plausible cost estimates for coastal defense infrastructure against sea and related flooding that reflect present and future risks due to changes in population, the economy and climate-induced sea-level rise and their uncertainties. This includes consideration of a range of different possible defense investment strategies. Hence, this analysis recognizes that there is not a single protection cost estimate and actual protection costs will depend on multiple factors, including the aims of the defense investment. The analysis is reported at the scale of World Bank regions and the globe. The analysis considers three distinct defense technologies (1) sea dikes, (2) river dikes, and (3) surge barriers. These are combined into two defense approaches which we termed (1) open protection (a combination of sea dikes and river dikes only) and (2) closed protection (sea dikes and the lowest cost option contrasting surge barriers and river dikes. Existing defenses are estimated based on assumptions applied in earlier global estimates of global flood risk. The defense approaches are applied using adaptation strategies (or scenarios) rather than economic optimization approaches and ask what the protection costs would be if we followed a pre-defined strategy at a global scale. Hence, we aim to develop a set of capital investment and maintenance needs for coastal defense infrastructure for coastal flooding that provides a set of protection services for a range of realistic demands/conditions. All costs are reported in 2014 undiscounted US dollars. The report is structured as follows. Section 2 briefly reviews previous global assessments of protection costs, including their assumptions and the cost estimates. Section 3 presents the methodology, including the analytical framework, the DIVA model and how it was applied and the cost estimates that are employed. Section 4 presents the results, including the length of defenses and their costs including capital and maintenance costs. Section 5 discusses the implications of the results and the potential next steps, and Section 6 concludes. There are four Appendixes. Appendix 1 contains a range of additional results to augment those in the main report. Appendix 2 presents the new analysis of sea dike unit cost, drawing on all the available experience. Appendix 3 provides the unit dike costs by country. Appendix 4 provides summary details on the DIVA model. 2 Previous Studies Compared to other issues relevant to adaptation to climate change, there is a long history of assessments of the protection costs against sea-level rise. These go back to the pioneering study of Dronkers et al. (1990) which supported the First Intergovernmental Panel on Climate Change (IPCC) Assessment. This situation probably reflects the long history of coastal defense in places such as Northwest Europe and East Asia against storm-induced coastal flooding. Hence, as the threat of climate-induced sea-level rise emerged there was an evidence base and practical experience to draw upon. Hence, coastal zones were some of the first areas to consider climate adaptation, particularly protection and its costs. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 3 Table 1. A summary of previous estimates of global protection costs against sea-level rise. Study Cost Estimate Comments (2014 US dollars) Dronkers et $815 billion For 1-m rise, capital costs mainly reflecting flood al., 1990 protection, but other aspects (e.g. port upgrade) also considered. Hoozemans $1,630 billion For 1-m rise, as Dronkers et al. (1990) with a more et al., 1993 realistic consideration of storm surge hazard and resulting protection needs. Fankhauser, $284 billion For 1-m rise, using Dronkers et al. (1990) data and 1995 (OECD only) capital costs of optimal protection. Not global. Bijlsma et At least $590 For 1-m rise. Aggregation of 17 national studies in al., 1996 billion (NOT their Table 9.3, so not globally comprehensive. All global) types of adaptation considered, but floods dominates protection costs. Capital costs only. Tol, 2002 $1,524 billion For 1-m rise applying cost-benefit analysis to the protection decision using the FUND model. Capital costs only. Protects 348,000 km of the world’s coastline. Hallegatte et $50 billion per year Considers the 136 largest coastal cities to 2050, and al. 2013 (NOT Global; to scales up from a few recent city examples, rather 2050) than using unit cost estimates as most other studies considered here. Hinkel et al. $32-$84 Based on a demand for safety analysis for protection 2014 billion/year (for need (Yohe and Tol, 2002) using the DIVA model. RCP8.5) (costs for Reports capital costs and maintenance costs of the year 2100) dikes built since 2000. (For RCP2.6 corresponding Accumulated costs are $14-$37 billion/year in the year 2100 and costs: $1.9-$4.2 accumulated costs US$1.3 trillion - US$ 2.8 trillion). trillion for 21st Protects about 500,000 km of the world’s estimated century 1,000,000 km coastline length. Capital and maintenance costs. The available protection cost estimates are summarized in Table 1. They nearly all depend on estimates of the length of coast which requires protection and unit costs for that protection. The studies are not independent, and most studies build on earlier studies in terms of adding incremental improvement on issues such as the length of protection, except Bijlsma et al (1996) and Hallegatte et al (2003) which consider different approaches. The unit costs of defense types have often been shared between studies with the original Dronkers et al (1990) costs being influential (see Appendix 1). Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 4 While the costs reported in Table 1 are large in absolute terms, in relative terms they are quite modest, especially when compared to the value of assets and the size of population found in the coastal zone (e.g., Hinkel et al., 2014; Diaz, 2016). This shows that coastal protection has a great potential to reduce the human costs of sea-level rise, as the studies in Table 1 all generally conclude. A caveat on that conclusion is the loss of coastal ecosystems which are generally degraded by hard defenses. Assessments such as Hoozemans et al (1993), Tol (2002) and Diaz (2016) attempt to address this issue in their analysis by considering changes to coastal wetlands in addition to protection and other costs. 3 Methodology Here we follow a similar strategy to the studies in Table 1, and determine lengths of coast that require protection and then estimate the costs of this using unit cost estimates, as appropriate (Appendices 2 and 3). The analysis is conducted within the framework of the Dynamic Interactive Vulnerability Assessment (DIVA) Model (described in Appendix 4) which has been applied in earlier global assessments of coastal flooding (e.g., Hinkel et al., 2014), as well as contributing to global assessments of water security, including multiple forms of flooding (Sadoff et al., 2015). In this section, we first review the DIVA flood module. We then consider the adaptation measures and costs that are considered in the analysis. Then the methods to estimate current protection levels are considered: these define the baseline against which future defense investments occur. Next, we consider the adaptation strategies that we use which guide and illustrate the implications of the different investment choices that we might make, including the population living in the coastal zone and residual damages. Lastly, we consider the climate and socio-economic scenarios that are utilized. The distinction between defense technologies, defense approaches and adaptation strategies is outlined in Table 2 and explained in detail in the following sections. Table 2. A summary of the defense technologies, approaches and adaptation strategies employed in this study. Further details are explained in the text. Defense Technologies (1) Sea dikes (2) River dikes (3) Surge barriers Defense Approaches (1) Dike only protection (2) Dike and barrier protection Adaptation Strategies (1) Constant protection levels -- maintain current protection levels (2) Constant absolute flood risk -- maintain average Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 5 annual losses (3) Constant relative flood risk -- maintain relative average annual losses (4) Risk intolerance -- keep relative average annual losses below 0.01% percent of local (5) Cost-benefit analysis -- compute the optimum protection level balancing flood damage to assets with protection costs to 2100, including capital and maintenance costs 3.1 The DIVA Flood Module The methods for assessing global coastal flood risk are taken from Hinkel et al (2014). The impact of coastal extreme events is calculated with the DIVA flooding module (Figure 1) and the key data sets that are defined in Table 3. Impacts are expressed in terms of the mathematical expectation of flood damages under a given protection level for the 12,148 coastline segments defined in the DINAS-COAST database (Vafeidis et al., 2008) which describe the world’s coasts with approximately 100 parameters per segment. Population exposure is obtained by overlaying the elevation data with population data (Table 3). Exposed population is translated into exposed assets by applying sub-national GDP per capita rates to the population data, followed by applying an assets-to-GDP ratio of 2.8. Present extreme water level distributions are assumed to be uniformly displaced upwards with relative sea-level rise, following 20th century observations (see Church et al., 2013). In addition to climate-induced sea-level rise, relative sea-level rise includes glacial isostatic adjustment and deltaic subsidence (Table 3). Current protection levels are taken from Sadoff et al. (2015) (see Table 4), who took protection levels for the biggest 136 coastal cities from Hallegatte et al. (2013) and complemented these through expert judgement for segments not associated to one of these cities. Protection level zero is assumed if the population density in the 1-in-100-years floodplain is lower than 30 people per km². For residual damage and population living in the 100-year flood plain, the methods of Hinkel et al (2014) are used. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 6 Figure 1. The DIVA Flood Module (following the approach of Hinkel et al., 2014). Table 3: Data sets used within the DIVA flooding module in this study. Data Dataset Elevation data Shuttle radar topography mission (SRTM) (Rabus et al., 2003) Population data Global Rural-Urban Mapping Project (GRUMP) (CIESIN 2011) Extreme water levels Global Tide and Surge Reanalysis (GTSR) (Muis et al., 2016) Glacial isostatic ICE-5G (VM2) model (Peltier, 2004) adjustment Delta subsidence The DIVA delta dataset, taken from Ericson et al. (2006) where available (with some corrections such as Brown and Nicholls, 2015) and 2 mm/yr for deltas where no value is reported 3.2 Adaptation measures and costs considered This report considers three main adaptation measures against coastal flooding due to current and future conditions, including sea-level rise: a. Sea dikes: these are rigid coastal barriers built along the open coast and around lagoons to stop flooding as widely applied in Northwest Europe (e.g., the Netherlands), East Asia (e.g., China) and parts of North America (e.g., New Orleans). Synonyms include terms such as levees. They have been considered as the primary adaptation against coastal flooding in many earlier global assessments, including those described in Table 1. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 7 b. River dikes: In addition to sea dikes we consider the protection that is required along rivers that are influenced by coastal extremes and sea-level rise, and might be flooded due to the backwater effect of the sea (Dronkers et al., 1990; Nicholls, 2010). Thus, rivers need to be protected in the area of their river mouth as illustrated by numerous dikes in coastal areas such as along the Rhine in the Netherlands or along the major rivers in coastal China. Only dikes that are required to address sea-level rise are considered here. c. Surge barriers: This is an alternative approach to flood defense along rivers, and involves closing off rivers from the sea during an extreme event (Gilbert and Horner, 1984; Jonkman et al., 2013; Mooyaart and Jonkman, 2017). Globally storm surge barriers are quite limited in extent at the present time, only being found in a few places such as London (Thames Barrier), Rotterdam (Maeslantkering), Venice (Project MOSES) and New Orleans. However, there are many other places where surge barriers have been considered such as New York City (Hill et al., 2012), with these discussions intensified post Superstorm Sandy in 2012. Hence for the first time in global analysis of costs we consider surge barriers. Protection on the open coast is always provided by sea dikes. River protection can be provided either by river dikes to the upstream limit of coastal effects (termed open protection), or by storm surge barriers (termed closed protection, Figure 2). In this report we analyze two different defense approaches, with the difference reflecting how river reaches are protected (where protection is applied): a. dike only protection - all protection along all river mouths uses river dikes only, combined with river dikes on the open coast; b. dike and barrier protection - for river mouths, a least cost selection is made between open and closed protection. For the least cost analysis, accumulated surge barrier costs (construction and maintenance cost) through the 21st century are considered versus the accumulated river dike costs (construction and maintenance cost) through 21st century. The costs of each measure are assessed in terms of capital costs and annual maintenance costs. These costs are not discounted and are explained below. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 8 Figure 2: A comparison of open protection (left) versus closed protection (right), illustrated in the case of the Netherlands. 3.3 River and river distributary data set For this report we consider, for the first time in such analyses, surge barriers as an alternative to river dikes. To support this, we developed an improved spatially explicit data set of river distributaries and potential barrier positions with a focus on the large coastal cities where a large proportion of the economic coastal risk (and protection demand) is located (Nicholls et al., 2008; Hallegatte et al., 2013). The data set was based on the original DIVA river data set (Vafeidis et al., 2008), which was extended to a 245-river data set. For each river we identified the main distributaries using Google Earth, defining a total of 434 distributaries (Figure 3). Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 9 Figure 3: The river mouths (red dots) considered in this study. Each river distributary is linked to one coastline segment, which was selected so that it captures as much as possible of the exposure connected with the river. The protection decision for the river distributary is linked directly to the protection decision for the coastline segment. Hence, if the coastline segment needs to be protected, we assume that the river distributary also needs protection. The same protection levels are assumed for the associated river distributary as used on the open coast. 3.4 Sea and river dike costs 3.4.1 Sea dikes Sea dikes are built along the open coast where required using the length of the coastal segments in DIVA (Appendix 4). We estimate capital costs and also maintenance costs (1.0% of the capital cost per year) (Nicholls et al., 2010). In this work we updated our estimates of sea dike costs using new information on dike costs, as explained in Appendix 2. Unit costs of dikes vary within and between countries. Dike costs in earlier studies such as Hinkel et al. (2014) are based on older studies (Dronkers et al. 1990; Hoozemans et al. 1993), who multiply a point estimate of Dutch unit dike costs with a country-specific factor that was based on expert judgement. There are two main problems with this approach: (i) recent case study results suggest that the Dutch unit costs were underestimated, and (ii) some of the country-specific factors are inaccurate. A pragmatic approach to improve the dike unit cost estimates is to update the previous Dutch unit cost point estimate with the interval estimate for rural areas given by Jonkman et al. (2013). This improves cost estimates and stresses the uncertainty of defense costs, but individual national costs can be substantially in error. Hence the cost estimates an improvement for regional and global cost estimates, but at the national level significant errors may occur. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 10 This results in a new range of costs represented by low and high estimates of dike unit costs as documented in Appendix 3. The low costs are similar to the earlier unit costs in DIVA and the high costs are about three times higher. As the stock of dikes grows, so the maintenance costs also grow and can become significant, as shown in Section 4. Hence it is important to track these maintenance costs and make sure they are considered, rather than forgotten or ignored. The capital costs of existing sea dikes up to 2015 (the base year) are not considered in the analysis. It is simply assumed that defenses corresponding to the standards in Table 4 have been provided. However, the cost of post-2015 maintenance of these pre-existing dikes is considered. 3.4.2 River dikes River dikes are built on both sides of the lower coastal-influenced reaches of selected rivers (see Figure 2). To calculate the length of river to protect with dikes, it is estimated how far the 1-in-1,000-year extreme sea level event could penetrate inland along the river. This is done by dividing the water level of the 1-in-1,000-year event by the river slope. Hence, the impact length of the 1-in-1,000-year event increases with sea-level rise. For simplicity, we use the same unit costs as sea dikes. Due to geometric considerations, river dikes are on average half the height of sea dikes. This also implies that river dikes are built at sea-level elevation in the river mouth and their base elevation increases with distance from the river mouth. Due to the absence of wave action, the river dike maintenance costs are assumed to be half the maintenance cost of sea dikes (0.5% of the capital cost per year). As with sea dikes, existing river dikes’ capital and maintenance costs in 2015 (the base year) are not considered in the analysis, but post-2015 maintenance is considered. 3.5 Surge barrier costs Surge barriers require a location. This was developed using expert judgement to define a single potential surge barrier position for each river distributary. The selected position was based on a trade-off between minimizing the barrier length and maximizing the length of avoided river dikes, while considering local conditions (Figure 4). The few existing barriers are not considered, as for most of them there are already plans for substantial upgrade of even the construction of new replacement barriers (Lavery and Donovan, 2005; Tarrant and Sayers, 2012). Hence, all barrier capital (construction) and maintenance costs are considered in the analysis. As with river dikes, the same protection levels are assumed as for the open coast requirement. If surge barriers are built, existing river dikes are kept (maintained) but no longer raised with sea level rise. Surge barriers stay constant in width, but are incrementally raised with rising sea level. The unit costs for surge barriers are taken from Table 3 of Jonkman et al., (2013). We use the minimal (US$97,000 per meter height and meter width) and maximal (US$374,000 per meter height and meter width) unit costs as the low and high cost scenarios in this study, respectively. Maintenance costs are assumed to be 1%/year following the maintenance costs for sea dikes. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 11 Figure 4: Proposed river distributaries and surge barrier positions as illustrated for the southern Netherlands. 3.6 Initial protection levels There are no global data sets on current levels of protection and initial protection levels (taken as 2010) have to be estimated in an expert manner. Hence, current protection levels are taken from Sadoff et al. (2015) who applied current protection levels for the biggest 136 coastal cities from Hallegatte et al. (2013) and complemented these through expert judgement for coastal areas not associated to one of these cities (Table 4). Protection level zero (i.e. no protection) is assumed if the population density in the 1-in-100-years floodplain is lower than 30 people per km² and in rural areas in countries with low and low middle incomes. As the socio-economic scenarios assume substantial economic growth, this moves people in rural areas from no protection to protection over the 21st century in these poorer countries, and the global length of protection increases with time. Table 4: Protection standards adopted in this analysis (following Sadoff et al., 2015). Wealth Class (annual income per capita) Urban Rural Uninhabited (2014 US$ GDP per capita (PPP)) (>1000 (30 to 1000 (<30 people/km2) people/km2) people/km2) Low income (< $1,035) 1:10 no protection no protection Lower middle income ($1,035 - $4,085) 1:25 no protection no protection Upper middle income ($4,086 - $12,615) 1:100 1:20 no protection Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 12 High income (> $12,615) 1:200 1:50 no protection Special case: Netherlands 1:10,000 Special case: 136 large coastal cities taken from Hallegatte et al. (2013) Subsequently, protection levels evolve at five-year time steps following the Adaptation Strategies outlined below. 3.7 Adaptation strategies In this report, we assess the costs for the following five adaptation strategies: 1. Constant Protection Levels -- maintain current protection levels as defined in Table 4. As population and GDP change with time in the socio-economic scenario, so the length and standard of protection will increase. Once an area is protected, defenses are maintained to 2100. 2. Constant absolute flood risk -- maintain average annual losses for protected areas as defined under Strategy 1 (similar to Hallegatte et al., 2013). This strategy raises the protection level with both rising sea levels and socio-economic development (population, GDP) in order to maintain the current (2015) flood risk level constant in monetary terms. 3. Constant relative flood risk -- maintain relative average annual losses for protected areas as defined under Strategy 1. This strategy raises protection levels with both rising sea levels as well as socio-economic development in order to maintain the current flood risk constant in terms of percentage of local GDP (considered to be a socially acceptable loss). By local GDP we refer to the fraction of GDP that is produced within the low elevation coastal zone (LECZ -- which is the area below 10 m elevation) associated to a coast-line segment. 4. Risk intolerance -- keep relative average annual losses below 0.01% percent of local for protected areas as defined under Strategy 1. The GDP threshold of 0.01% is based on the losses in the cities of Amsterdam and Rotterdam in 2005 as calculated by Hallegatte et al. (2013) and applies this Dutch standard as a risk intolerant world. 5. Cost-benefit analysis (CBA) -- compute the optimum protection level minimizing the sum of residual flood damage to assets and protection costs to 2100, including capital and maintenance costs. This follows the methods of Lincke and Hinkel (2018). Strategies 1 to 4 are scenarios of how protection might be applied globally and are based on expert judgement, while Strategy 5 is based on economic optimization and balancing the costs of protection to the avoided damages. Under Strategy 1, the defenses are raised with relative sea-level rise, while under the other strategies, the defenses are generally raised more than the rise in sea level. Strategies 1 to 3 take a positive approach, while Strategy 4 takes a normative approach which allows us to consider the adaptation deficit, relative to Strategies 1 to 3. The adaptation deficit can only be assessed with respect to a normative assumption as to what is desirable -- here we ask what is required to give all protected areas following Table 4 the same level of safety as Amsterdam and Rotterdam. Similar analysis can be conducted with Adaptation Strategy 5. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 13 Note that the protected length in 2015 is maintained or grows over time under Strategies 1 to 4. Under Strategy 5 (Cost-benefit analysis) an optimal defended length (and defense standard) is determined in 2015 and maintained to 2100. 3.8 Socio-economic and sea-level rise scenarios Adaptation costs are assessed for a consistent set of socio-economic and sea-level rise scenarios over the 21st century. These scenarios are summarized in Table 5. For socio-economic scenarios, we draw on the Shared Socio-economic Pathways (SSPs), version 9 provided by IIASA and use three scenarios: SSP2, SSP3 and SSP5 (IIASA, 2016; O’Neill et al., 2014). Three global mean sea-level rise scenarios are taken assuming relatively low, intermediate and relatively high emissions, respectively. They are use the HadGEM-ES2 model. These are the 5th percentile of RCP2.6, the 50th percentile of RCP4.5, and the 95th percentile of RCP8.5 (all taken from Hinkel et al., 2014). Henceforth these are just termed RCP2.6, RCP4.5 and RCP8.5, respectively. Table 5: The socio-economic and sea-level rise scenarios applied in this study. Base year for sea-level rise is the 1985 to 2005 average. Year 2015 2030 2050 2075 2100 Global SSP2 7.4 8.4 9.4 9.7 9.2 population (billions) SSP3 7.4 8.7 10.1 11.6 12.8 SSP5 7.4 8.2 8.8 8.6 7.7 GDP per capita SSP2 14,400 20,800 30,000 46,700 72,600 (US$, global average) SSP3 14,300 18,500 20,800 23,000 26,000 SSP5 14,400 26,200 51,700 97,500 170,000 Sea-level rise RCP2.6 5th percentile 0.02 0.06 0.11 0.17 0.23 (global coastal average, m) RCP4.5 50th percentile 0.03 0.08 0.16 0.29 0.43 RCP8.5 95th percentile 0.04 0.13 0.28 0.59 1.03 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 14 4 Results The results are reported globally and for different groups of countries by income as defined by the World Bank in 2015 (World Bank, 2017). They are as follows: (1) East Asia & Pacific, (2) Europe & Central Asia, (3) High Income, (4) Latin America & Caribbean, (5) Middle East & North Africa, (6) South Asia, and (7) Sub-Saharan Africa. In this main report results are provided for the SSP2 and high adaptation unit costs, while in Appendix 1 the other results are reported for SSP3 and SSP5 and low adaptation unit costs. In Section 4.1 we consider the length of open coast and number of rivers/distributaries that require protection. In Section 4.2 the protection costs are presented in terms of annual and accumulated costs, the relative contribution of capital and maintenance costs and the adaptation deficit. There is a focus on the dike and barrier defense approach. In Section 4.3, the population living in the 100-year flood plain and the residual damages are reported. 4.1 Required Length/Quantity of Protection Based on the assumptions described in Table 4, we estimate that 24.2% of the world’s coastline is protected by dikes in 2010 (Figure 5). These comprise (1) East Asia & Pacific: 25.8% of the coast, (2) Europe & Central Asia: 54.3% of the coast, (3) High Income: 23.1% of the coast, (4) Latin American & Caribbean: 27.9% of the coast, (5) Middle East & North Africa: 46.6% of the coast. (6) South Asia: 23.9% of the coast, and (7) Sub-Saharan Africa: 11.8% of the coast (Figure 5). Tables 6, 7 and 8 summarize the length of protection required over time. From 2015 to 2100, the length of protection increases under Strategies 1 to 4. The global increase from 2010 to 2100 is 23% in length, while in Sub-Saharan Africa experiences a 133% increase and in East Asia & Pacific it is a 79% increase in length. Under Strategy 5, the protected length varies with both RCP and SSP scenarios -- it is longest under RCP8.5 and SSP5 (see Tables 7 and 8). This length is often longer than the other Adaptation Strategies in 2015, but this difference diminishes with time. In a few cases it is less than the estimated length of defenses in 2010 (Table 7). These abandoned defenses are assumed to have no costs in the period 2015 to 2100. Note that in all cases the length of protection estimated in this analysis is substantially less than that reported by both Tol (2002) and Hinkel et al (2014) in Table 1. Similarly, protection is required for 145 of 345 (42.2%) river distributaries in 2015. Globally, a total length of 8,600 km river dikes is required. Under constant protection levels, dike only protection and the high sea-level rise scenario, the length of river dikes increases to 11,500 km by 2100. This reflects that the dikes need to extend further inland as sea levels rise. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 15 Figure 5: Percentage of protected open coast in 2010 in the seven World Bank regions used in the study for SSP2, following the assumptions in Table 4. The results for SSP3 and SSP5 are almost identical and hence are not shown. Absolute values are given in Tables 6 and 7. Figure 6: Percentage of protected coast over time for SSP2, high adaptation unit cost and RCP 8.5 sea-level rise. Lengths are almost identical in Strategies 1 to 4 (illustrated here with Risk Intolerant Protection). Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 16 Table 6. The length of protected coast (in km) globally and by region over time (for selected years) under Strategies 1 to 4. 2010 is included to illustrate the large increase in protected length in Upper middle income countries in 2015. RCP 8.5 SLR and Risk Intolerant Protection are considered. Year 2010 2015 2030 2050 2100 East Asia & Pacific 31,300 31,400 32,100 37,000 49,100 Europe & Central Asia 4,600 4,600 4,600 4,600 5,000 High Income 104,100 104,100 104,100 104,500 108,600 Latin America & Caribbean 17,000 17,100 16,700 16,800 24,200 Middle East & North Africa 6,500 6,500 6,800 6,900 7,800 South Asia 3,500 3,500 3,900 4,800 6,700 Sub-Saharan Africa 4,100 4,100 4,100 4,200 6,400 Total (global) 171,100 171,200 172,300 178,800 207,800 Table 7. The length of protected coast under CBA Protection, RCP8.5 and SSP2 scenarios. Year 2010 Length of Protected Coast (2015 to 2100) (km) and percentage change relative to 2010 East Asia & Pacific 31,300 46,600 (+48.9%) Europe & Central Asia 4,600 5,600 (+21.7%) High Income 104,100 97,000 (-6.8%) Latin America & Caribbean 17,000 16,900 (-0.6%) Middle East & North Africa 6,500 8,100 (+24.6%) South Asia 3,500 9,300 (+165.7%) Sub-Saharan Africa 4,100 11,300 (+175.6%) Total (global) 171,000 195,000 (+14.0%) Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 17 Table 8. The global length of protected coast under CBA Protection under all the climate and socio-economic scenarios from 2015 to 2030. Antarctica is not considered. Scenarios Length Protected Distance (km) Proportion of globe Climate Socio-economic (%) 1.95E+05 SSP2 28.20% 1.69E+05 SSP3 24.47% 2.19E+05 RCP8.5 SSP5 31.75% 1.79E+05 SSP2 25.88% 1.57E+05 SSP3 22.76% 2.00E+05 RCP4.5 SSP5 28.88% 1.69E+05 SSP2 24.48% 1.52E+05 SSP3 22.06% 1.92E+05 RCP2.6 SSP5 27.80% Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 18 Of the 434 river distributaries considered in the analysis, we protect 232 of them (53.3%) in 2015. For dike and barrier protection, we protect 145 river distributaries with river dikes and 87 river distributaries with surge barriers. Dike only protection leads to longer lengths of defenses (Figure 2). This is summarized in Table 9. Table 9. River protection. The length of protected river mouths (in km) and the number of surge barriers globally and by region over time (for selected years) assuming open (dike only) and closed (surge barriers, if appropriate). The first line in each cell shows the length of river dikes and number of surge barriers for open protection and the second line in each cell shows the length of river dikes and the number of surge barriers for closed protection. This assumes RCP 8.5 SLR, SSP2 and Risk Intolerant Protection. Year 2010 2015 2030 2050 2100 East Asia & Pacific 3,600/0 4,000/0 4,200/0 4,900/0 5,700/0 3,600/0 3,700/26 3,700/27 4,100/31 4,600/31 Europe & Central 97/0 100/0 120/0 140/0 260/0 Asia 97/0 100/1 110/1 130/1 240/1 High Income 6,000/0 6,100/0 6,300/0 6,700/0 8,100/0 6,000/0 6,000/51 6,100/51 6,300/52 7,000/52 Latin America & 1,200/0 1,200/0 1,200/0 1,300/0 1,700/0 Caribbean 1,200/0 1,200/12 1,200/11 1,200/12 1,300/14 Middle East & 170/0 180/0 210/0 250/0 680/0 North Africa 170/0 180/2 190/3 220/3 340/5 South Asia 1,700/0 1,700/0 1,700/0 1,800/0 2,200/0 1,700/0 1,700/4 1,700/4 1,800/4 2,000/5 Sub-Saharan Africa 160/0 160/0 170/0 180/0 290/0 160/0 160/2 170/2 170/2 170/3 Total (global) 13,000/0 13,400/0 14,000/0 15,300/0 19,100/0 13,000/0 13,100/98 13,300/100 14,000/105 15,800/111 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 19 4.2 Protection Costs Figures 7 and 8 show the annual protection costs including the sum of capital and maintenance costs for the five Adaptation Strategies, the sea-level rise scenarios, the SSP2 scenario and the high adaptation unit costs. These costs are the sum of the capital costs of sea dikes, river dikes and surge barriers, and all maintenance costs. All the Defense Approaches produce similar estimates of annual protection costs rising through the 21st century. In 2100, global costs are about US$40 billion to US$110 billion per year under RCP2.6 and about US$60 billion to more than US$170 billion per year under RCP8.5. A large spike in the costs occurs in 2015 in all cases, being the biggest of several spikes in simulated cost time series for Strategies 1 to 4. It reflects a large increase in the protection standards of upper-middle-income countries due to rising living standards crossing a threshold value (especially in Brazil and China). This raises protection standards along large lengths of coast in 2015, following Table 4. Under the Risk Intolerance strategy, this is reinforced by the response to the adaptation deficit, which is discussed in more detail below. In all cases, this spike is the highest investment requirement, but if this strategy was followed, these costs would probably be distributed over a longer period, spreading the additional investment demand over time (see aggregated costs in Figure 8). Under RCP2.6, after 2015, annual costs are approximately constant, reflecting a near constant and slow rate of sea-level rise. Hence capital costs are linear and maintenance costs only rise slowly. In contrast, under RCP8.5 costs approximately double from 2020 to 2100, reflecting accelerating sea-level rise and a large increase in maintenance costs. Under Strategy 5 (Cost-benefit analysis) the broad costs are similar although often slightly lower than the Risk Intolerance strategy. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 20 Figure 7. Annual protection costs (for dike and barrier protection) for the seven regions and the total global costs over time for SSP2, high adaptation unit cost, the five adaptation strategies and three sea-level rise scenarios. Includes maintenance and capital costs. The spike in 2015 is discussed in the text. Results for other scenarios are shown in Appendix A.1.1. Tables 10 and 11 show the range of undiscounted protection costs (capital and maintenance) across all the different scenarios. Table 10 shows the total costs from 2015 to 2100, while Table 11 focuses on the near-term costs from 2015 to 2030. Across the five strategies these total US$0.6 trillion to US$2.5 trillion, US$1.0 trillion to US$3.7 trillion, US$0.6 trillion to US$2.6 trillion, US$1.7 trillion to US$5.5 trillion, and US$1.4 trillion to US$6.7 trillion from 2015 to 2030 and US$2.9 trillion to US$14 trillion, US$3.5 trillion to US$16.6 trillion, US$2.8 trillion to US$14.6 trillion, US$4.3 trillion to US$18.3 trillion, and US$3.8 trillion to US$18.2 trillion from 2015 to 2100, respectively. The biggest uncertainty in these ranges is the unit cost. The highest costs are found with the Risk Intolerant and CBA strategies. With the CBA strategy, the residual risks are much lower than the other strategies (Section 4.3) and overall it has the lowest cost. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 21 Table 10. The range of total undiscounted protection costs from 2015 to 2100 across all uncertainties under the five protection strategies (US dollars 103 billion). The main uncertainty is the unit cost. Adaptation strategy cost (capital and maintenance) 1. CP 2. 3. 4. RI 5.CBA CA CRF Scenarios FR R 4.6-13.2 5.5-15.6 4.8-13.5 6.1-17.3 6.8-16.0 SSP2 4.4-12.5 5.2-14.8 4.5-12.7 5.7-16.3 6.2-13.6 SSP3 4.9-14.0 5.8-16.6 5.2-14.6 6.4-18.3 7.9-18.2 RCP8.5 SSP5 3.5-10.0 4.2-11.9 3.5-10.0 4.8-13.6 5.0-11.4 SSP2 3.3-9.4 4.0-11.2 3.2-9.2 4.5-12.8 4.4-10.0 SSP3 3.7-10.5 4.5-12.6 3.8-10.8 5.1-14.4 5.7-13.2 RCP4.5 SSP5 3.1-8.8 3.8-10.7 3.1-8.8 4.5-12.8 4.3-9.8 SSP2 2.9-8.3 3.5-10.1 2.8-8.0 4.3-12.1 3.8-8.8 SSP3 3.3-9.3 4.0-11.4 3.3-9.5 4.7-13.4 5.8-11.5 RCP2.6 SSP5 Table 11. The range of total undiscounted protection costs from 2015 to 2030 across all uncertainties under the five protection strategies (US dollars 103 billion). The main uncertainty is the unit cost. Adaptation strategy cost (capital and maintenance) 1. CP 2. 3. 4. RI 5.CBA CA CRF Scenarios FR R 0.7-2.4 1.1-3.6 0.7-2.4 1.7-5.3 2.3-5.7 SSP2 0.7-2.4 1.1-3.6 0.7-2.4 1.7-5.3 2.0-4.6 SSP3 0.8-2.5 1.1-3.7 0.8-2.6 1.7-5.5 2.8-6.7 RCP8.5 SSP5 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 22 0.6-2.2 1.0-3.2 0.6-2.1 1.7-5.0 1.9-4.3 SSP2 0.6-2.2 1.0-3.2 0.6-2.1 1.7-5.0 1.6-3.6 SSP3 0.7-2.2 1.0-3.3 0.7-2.3 1.7-5.2 2.3-5.2 RCP4.5 SSP5 0.6-2.1 1.0-3.1 0.6-2.1 1.7-5.0 1.7-3.8 SSP2 0.6-2.1 1.0-3.1 0.6-2.0 1.7-5.0 1.4-3.3 SSP3 0.7-2.2 1.0-3.4 0.7-2.2 1.7-5.1 2.0-4.7 RCP2.6 SSP5 Figure 8. Cumulative protection costs (for dike and barrier protection) for the all RCP sea-level rise scenarios across the adaptation strategies. SSP2 and high adaptation unit costs are assumed. Results for other scenarios are shown in Appendix A.1.2 Figures 8 shows the cumulative costs for the same assumptions as in Figure 7 for three time periods: (1) 2015 to 2030, (2) 2031 to 2050 and (3) 2051 to 2100. Over the period 2015 to Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 23 2100, the total costs are about US$2,800 billion to US$11,500 billion for RCP2.6 and US$4,400 billion to US$18,300 billion for RCP8.5. In regional terms, most investment is in the High Income Countries followed by East Pacific and Asia. The costs considered here are composed of both capital and maintenance costs. As the stock of defenses increases with time, the absolute maintenance costs grow substantially. It is important to consider these cost requirements and make sure that the flood management governance institutions have sufficient funding available to support them. Maintenance is an area which can easily be underfunded or ignored, and if this occurs this leads to an increased chance of defense failure. Figure 9 shows the global capital and maintenance costs in relative terms for the sea-level rise scenarios across the adaptation strategies. Under RCP2.6, maintenance costs are substantial and constitute about 75% of costs per year throughout the century and are even higher under Strategy 5 (Cost-benefit analysis). Under RCP8.5, the relative investment cost rises towards the end of the century as sea-level rise accelerates and there are larger investment costs to keep pace. However, maintenance remain more than half the annual costs. It should be noted that the spike in protection costs in 2015 that is apparent in Figures 7 and 8 is also apparent here as a spike in relative investment cost. Figure 9. The share of capital versus maintenance costs (for dike and barrier protection) for SSP2, all SLR and all adaptation strategies from 2015 to 2100. Results for other scenarios are shown in Appendix A1.1.3. Figure 10 (and Appendix A.1.4) shows the adaptation deficit in 2015. It compares the cost of Strategy 1 Constant Protection to Strategy 4 Risk Intolerant and Strategy 5 CBA Protection. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 24 Globally, it amounts to as much as US$4 trillion. Regionally, the largest adaptation deficits are in the High Income countries, following East Pacific and Asia. While these costs are large, the capital costs are one-off investments -- once made, the defense standards can be maintained with similar maintenance costs to the other adaptation strategies -- global annual maintenance costs are increased by roughly US$10 billion to US$25 billion/year. Figure 10. Estimate of the adaptation deficit in 2015 for the Risk Intolerance and CBA Protection Strategies (for dike and barrier protection). The cost is derived from the difference between Constant Protection and the two stated strategies. Results for other scenarios are shown in Appendix A1.1.4. 4.3 Flood plain exposure and residual damage Figure 11 shows the flood plain population over the 21st century under all SSPs and RCPs. Flood plain population does not depend on the adaptation scenario. Note that these calculations assume uniform population change across each nation. At the end of the century it is highest under SSP3 and RCP8.5 sea-level rise (300 million people). Und SSP2 and SSP5 global population is falling towards the end of the century and thus global floodplain population is also declining (except for SSP2 and RCP 8.5). Annual residual damage costs are shown in Figures 12 and 13, and also in Appendix A.1.5. Residual damages grow with time in all cases, mainly reflecting both economic growth so asset values grow and sea-level rise which influences the flood depth and hence damage during flood events. Strategy 5 gives the lowest damages by far, reflecting the goal of the approach. The largest damages are estimated at less than US$400 million per year in 2100 (Figure 13), while with the other strategies, losses can be up to US$3,000 billion per year in 2100 under RCP8.5. Hence the CBA Protection approach would seem to deliver the overall lowest costs. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 25 Figure 11. Population living below the 1-in-100-year flood event per region over time (for all scenarios). These results neither depend on the adaptation strategy nor on the adaptation unit cost. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 26 Figure 12. Annual residual damage cost during 21st century per region for all adaptation strategies, sea-level rise scenarios and the SSP2 scenario. Unit costs only influence the results for CBA Protection (see Figure 13). Results for other scenarios are shown in Appendix A1.1.5. Figure 13. The annual residual damage cost for Strategy 5 (Cost-benefit analysis protection) for the SSP2 scenario and the low and high unit defense costs. Results for other scenarios are shown in Appendix A1.1.5. 5 Discussion These results provide a comprehensive set of estimates of the protection costs and related parameters given sea-level rise from 2015 to 2100 for a range of defense strategies. The differences in costs between most of the defense strategies are smaller than might be expected and the choice of using surge barriers or not similarly has a small effect on estimates of protection costs. Hence, we focus on Strategy 1 (Constant Protection) and Strategy 5 (Cost-benefit analysis) which is more distinctive from Strategies 1 to 4. For Strategy 1, the total defense costs from 2015 to 2100 are US$2.9 trillion to US$9.3 trillion and US$4.4 trillion to US$14.0 trillion for RCP2.6 and RCP8.5, respectively. For Strategy 5, the same costs are up to US$11.4 trillion and US$18.2 trillion for RCP2.6 and RCP8.5, respectively. These costs are higher than any earlier estimates in Table 1, reflecting the higher unit costs, the inclusion of river protection due to sea-level rise and more realistic maintenance costs. Here we include all defenses that need to be maintained, including the Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 27 legacy defenses built before 2015 which have often been ignored in earlier analyses. While these costs are high, the cost-benefit analysis finds that such investment makes economic sense, at least over the next 80 years. The cost-benefit calculations suggest that protection is the optimum response along about 22% to 32% of the world’s coast, depending on the scenarios. Hence, it is uncertain for 10% of the world’s coast if protect or retreat is the best option. The remaining 68% of the coast might be expected to be allowed to evolve more naturally, and low-cost ecosystem-based or nature-based approaches to coastal defense might be appropriate (e.g., Temmerman et al., 2013). This analysis shows that following a protection strategy commits the adapters to a long-term maintenance strategy with maintenance costs exceeding the capital costs under all the scenarios considered here. The Netherlands and London provide good examples of major flood defense systems that have a long history based on historic floods and are now being actively maintained and upgraded including allowance for sea-level rise as needed (Lavery and Donovan, 2005; Tarrant and Sayers, 2012; Ranger et al., 2013; Stive et al., 2011; van Alphen 2015). These efforts are linked to strengthening of flood management institutions, long-term planning that looks many decades into the future and securing the commitment to funding. For protection to be successful elsewhere, similar arrangements will be required, including the committed funding streams required for maintenance and the governance institutions to deliver the maintenance. Based on historic precedent, complacency in this regard can lead to disaster, with New Orleans and Hurricane Katrina being a recent example. Recognition of residual risk is also critical as defenses will sometimes fail even if they are well maintained. For example, Hallegatte et al. (2013) identified that if we protect the world’s large coastal cities, we should expect to see fewer, but bigger and more damaging coastal floods. Hence, protection does not avoid the need for other pillars of risk management such as flood warnings and disaster preparedness. Any coastal society following a protection approach needs to recognize these long-term commitments which will go on for centuries (Nicholls et al., 2018). If this commitment cannot be delivered, then alternative approaches to coastal adaptation are recommended. The construction of defenses under rising sea levels also has secondary impacts which include the removal of accommodation space for the migration of coastal wetlands (Schuerch et al., 2018). This degrades coastal wetlands as sea levels rise (termed coastal squeeze) and this effect should be considered when selecting adaptation options. In this analysis, we have considered relative sea-level rise due to the sum of climate-induced sea-level rise, glacial-isostatic adjustment and deltaic subsidence. Human-induced subsidence in coastal cities can also be significant, with multiple meters of subsidence observed in some coastal cities over the last few decades or longer (e.g., Nicholls, 1995; Kaneko and Toyota, 2011; Nicholls, 2018). In Asia, the World Bank (2010) saw subsidence as an equal threat for coastal megacities as climate-induced sea-level rise. While not analyzed here, city subsidence will not change the fundamental results, but it will enhance the costs of protection, including the maintenance costs, and it will increase the impacts if protection fails (Hallegatte et al., 2013). Mitigation of human-induced subsidence should be considered wherever possible as an immediate preventative response available to cities. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 28 6 Conclusion This represents the most comprehensive analysis of building coastal protection to sea-level rise completed to date. For all the protection models considered, one-third or less of the world’s coast is protected. Under Cost-benefit analysis (Strategy 5), the coastal length protected over the 21st century is 22 to 32 percent of the globe, depending on the scenarios. For Strategies 1 to 4, the protected length increases over time to about 30 percent of the globe by 2100. The population in the coastal flood plain responds to demographic change -- despite a rise in sea level, the coastal population falls under the SSP2 and SSP5 demographic scenarios, reflecting stabilizing and then falling global population. The residual risk grows with time under all scenarios and is substantially lower under cost-benefit assumptions, as the investments are optimized to reduce the sum of the residual damage and investment costs. The total defense costs are much higher than the earlier estimates summarized in Table 1. Under the highest cost estimates and the highest sea-level rise (RCP8.5) they are as much as US$18.3 trillion (Risk Intolerant Protection, RCP8.5 and high unit defense costs) from 2015 to 2100. This is about 10 times higher than most earlier cost estimates and up to three times higher than the last protection analysis by Hinkel et al (2014). The higher costs here reflect a higher range of unit costs and the inclusion of realistic maintenance costs, including for the defenses that existed in 2015, and that must be maintained to provide protection to 2100. The consideration of river defenses due to sea-level rise also raise costs. Cost-benefit analysis–based protection has high end costs from about US$8.8 trillion up to US$18.2 trillion (depending on the sea-level rise scenario) from 2015 to 2100. This demonstrates that while these investments in coastal protection are large, they are economically rational as the benefits outweigh the costs. Given the lower residual damage, CBA Protection is the lowest cost option, as might be expected. The differences in costs in employing surge barriers versus river dikes are not large at the global and regional scale presented here. While cost-benefit analysis offers lower costs over the century, early investment is larger as the defenses are not of an optimum standard in 2015. This might be interpreted as one estimate of the adaptation deficit -- this additional early investment is up to about US$4 trillion. However, there are large uncertainties and empirical observations of protection standards are limited. Hence, these should be considered as an indicative estimate and the size of the adaptation deficit remains an important topic for future research. Importantly, building defenses is not a one-off capital investment. Over the 21st century, the cost of a comprehensive protection strategy is dominated by maintenance costs in all the cases considered in this report (Figure 9; Appendix A1.3). This indicates that the development of appropriate institutions and governance mechanisms to deliver maintenance, as well as the necessary funding streams, are essential for such a protection- based adaptation strategy to be effective. The Netherlands and the Thames Estuary (London) are good examples of major flood defense systems that have been built and are now being actively maintained and upgraded as needed. They are linked to strong flood Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 29 management institutions and long-term planning looking many decades into the future. For protection to be successful elsewhere, similar arrangements will be required, including guaranteeing the funding streams for maintenance (see Hinkel et al., 2018). Any coastal society following a protection approach needs to recognize this long-term commitment. The danger of focusing on defense without this support is that society is lulled into a false sense of security, leading to bigger coastal disasters. If this commitment cannot be delivered, then alternative coastal adaptation approaches are recommended, such as accommodation or retreat. Further, even if well-maintained, defenses are always associated with residual risk and appropriate measures need to be put in place for its management, especially in coastal cities (Hallegatte et al., 2013). Hence appropriate flood warnings and disaster preparedness mechanisms remain essential even if a good protection and maintenance regime is in place. Defenses also degrade coastal wetlands as sea levels rise (termed coastal squeeze). Lastly, the world does not end in 2100 and protection is forever. Sea levels can be expected to continue to rise for centuries even if we fully follow the Paris Agreement and stabilize global temperature (Nicholls et al., 2018). Hence, protection upgrade and maintenance will need to continue into the 22nd century and beyond. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 30 7 References Bijlsma, L., Ehler, C.N., Klein, R.J.T., Kulshrestha, S.M., Mclean, R.F., Mimura, N., Nicholls, R.J., Nurse, L.A., Perez Nieto, H., Stakhiv, E.Z. Turner, R.K. & Warrick, R.A. (1996) Coastal Zones and Small Islands. In: Watson, R.T., Zinyowera, M.C. & Moss, R.H. (eds.), Impacts, Adaptations and Mitigation of Climate Change: Scientific-Technical Analyses, (The Second Assessment Report of the Intergovernmental Panel on Climate Change, Working Group II), Cambridge University Press, Cambridge, pp.289-324. Brown, S. and Nicholls, R.J. (2015) Subsidence and human influences in mega deltas: the case of the Ganges-Brahmaputra-Meghna, Science of The Total Environment, 527-528, 362-374 doi:10.1016/j.scitotenv.2015.04.124 Church, J. A., P.U. Clark, A. Cazenave, J.M. Gregory, S. Jevrejeva, A. Levermann, M.A. Merrifield, G.A. Milne, R.S. Nerem, P.D. Nunn, A.J. Payne, W.T. Pfeffer, Stammer, D. & A.S. Unnikrishnan (2013) Sea Level Change. In: Stocker, T. F., Qin, D., Plattner, G.-k., Tignor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex, V. & Midgley, P. M. (eds.) Climate Change 2013. The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York, NY, USA: Cambridge University Press. CIESIN, IFPRI, the World Bank & CIAT (2011) Global Rural-Urban Mapping Project, Version 1 (GRUMP v1): Population Density Grid. Data and Applications Center (SEDAC), Columbia University, New York, USA. Diaz, D.B. (2016) Estimating global damages from sea level rise with the Coastal Impact and Adaptation Model (CIAM) Climatic Change, 137: 143-156. https://doi.org/10.1007/s10584- 016-1675-4 Dronkers, J., Gilbert,J.T.E., Butler, L.W., Carey, J.J., Campbell, J., James, E., McKenzie, C. , Misdorp, R., Quin, N., Ries, K.L., Schroder, P.C., Spradley, J.R. , Titus, J.G., Vallianos, L., von Dadelszen, J. (1990) Strategies for adaptation to sea level rise. Report of the IPCC Coastal Zone Management Subgroup: Intergovernmental Panel on Climate Change. (Intergovernmental Panel on Climate Change, Geneva). Ericson J.P., Vörösmarty C.J., Dingman S.L., Ward L.G., Meybeck M. (2006) Effective sea-level rise and deltas: Causes of change and human dimension implications. Global Planet. Change, 50: 63–82. Fankhauser, S. (1995) Protection versus Retreat: The Economic Costs of Sea-Level Rise. Environment and Planning A, 27, 299-319 https://doi.org/10.1068/a270299 Gilbert, S. and Horner R. (1984) The Thames Barrier, Telford, London ISBN 0-7277-0249-1 Hallegatte, S., Green, C., Nicholls, R.J., Corfee-Morlot, J. (2013) Future flood losses in major coastal cities. Nature Clim. Change 3, 802–806. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 31 Hill, D., Bowman, M.J. and Khunda, J.S. (2012) Storm surge barriers to protect New York City. Against the Deluge. Papers from a conference held in Brooklyn, New York, 30-31 March 2009. American Society of Civil Engineers. Hinkel, J., Lincke, D., Vafeidis, A.T., Perrette, M., Nicholls, R.J., Tol, R.S.J., Marzeion, B., Fettweis, X., Ionescu, C., Levermann, A. (2014) Coastal flood damage and adaptation cost under 21st century sea-level rise. Proceedings of the National Academy of Sciences 111, 3292–-3297. doi:10.1073/pnas.1222469111 Hinkel, J., Aerts, J.C.J.H., Brown, S., Jiménez, J.A., Lincke, D., Nicholls, R.J., Scussolini, P., Sanchez-Arcilla, A., Vafeidis, A., Addo, K.A. (2018). The ability of societies to adapt to 21st century sea-level rise. Nature Climate Change, accepted. Hoozemans, F.M.J., Marchand, M., Pennekamp, H.A. (1993) Sea level rise: A global vulnerability assessment. Vulnerability assessments for population, coastal wetlands and rice production on a global scale. (Delft Hydraulics / Rijkswaterstaat, The Hague). IIASA (2016) SSP Public Database, Version 1. Available at: https://tntcat.iiasa.ac.at/SspDb Jonkman, S.N., Hillen, M.M., Nicholls, R.J., Kanning, W., van Ledden, M. (2013) Costs of Adapting Coastal Defences to Sea-Level Rise— New Estimates and Their Implications. Journal of Coastal Research 1212–1226. doi:10.2112/JCOASTRES-D-12-00230.1 Kaneko, S., Toyota, T. (2011) Long-term urbanization and land subsidence in Asian megacities: an indicators system approach. In: M. Taniguchi (ed.), Groundwater and Subsurface Environments: Human Impacts in Asian Coastal Cities, DOI 10.1007/978-4-431-53904-9_13, pp. 249–270. Lavery, S. & Donovan, B. (2005) Flood risk management in the Thames Estuary looking ahead 100 years. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 363, 1455-1474. (DOI:10.1098/rsta.2005.1579). Lincke, D. and Hinkel, J. (2018) Economically robust protection against 21st century sea- level rise. Global Environmental Change, accepted. Lowe, J.A., Howard, T.P., Pardaens, A., Tinker, J., Holt, J., Wakelin, S., Milne, G., Leake, J., Wolf, J., Horsburgh, K., Reeder, T., Jenkins, G., Ridley, J., Dye, S., Bradley, S. (2009) Marine and coastal projections, in: UK Climate Projections Science Report. Met Office Hadley Centre and Exeter, UK. Menéndez, M., Woodworth, P.L. (2010) Changes in extreme high water levels based on a quasi-global tide-gauge data set. Journal of Geophysical Research 115. doi:10.1029/2009JC005997 Mooyaart, L.F. and Jonkman, S.N. (2017) Overview and Design Considerations of Storm Surge Barriers. Journal of Waterway, Port, Coastal, and Ocean Engineering, 143, doi:10.1061/(ASCE)WW.1943-5460.0000383 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 32 Muis, S., Verlaan, M., Winsemius, H. C., Aerts, J. C. J. H. and Ward, P. J. (2016) A global reanalysis of storm surges and extreme sea levels. Nature Communications 7, Article number: 11969, doi:10.1038/ncomms11969 Nicholls, R.J. (1995) Coastal megacities and climate change. GeoJournal, 37, 369–379. Nicholls, R. J. (2010). Impacts of and responses to sea-level rise. In J.A. Church, P.L. Woodworth, T. Aarup, and W.S. Wilson (Eds.), Understanding Sea-Level Rise and Variability. (pp. 17-51). Chichester, GB: Wiley-Blackwell. DOI: 10.1002/9781444323276.ch2 Nicholls, R.J. (2018) Adapting to Sea-Level Rise. Resilience: The Science of Adaptation to Climate Change, Elsevier, Alverson, K., Zommers, Z. (eds) Paperback ISBN: 9780128118917, in press. Nicholls, R.J., Hanson, S., Herweijer, C., Patmore, N., Hallegatte, S., Corfee-Morlot, J., Château, J. and Muir-Wood, R. (2008) Ranking port cities with high exposure and vulnerability to climate extremes: exposure estimates. Paris, France, Organisation for Economic and Co-operative development (OECD), 62pp. (Environment Directorate Working Papers, (ENV/WKP(2007)1) ). Nicholls, R. J., Brown, S., Hanson, S., & Hinkel, J. (2010). Economics of coastal zone adaptation to climate change. (World Bank Discussion Papers; No. 10). Washington, US: International Bank for Reconstruction and Development / World Bank. Nicholls, R.J., Brown, S., Goodwin, P., Wahl, T., Lowe, J., Solan, M., Godbold, J.A., Haigh, I.D., Lincke, D., Hinkel, J., Wolff, C., Merkens, J. (2018) Stabilisation of global temperature at 1.5°C and 2.0°C: Implications for coastal areas. Philosophical Transactions of the Royal Society, A 376: 20160448.http://dx.doi.org/10.1098/rsta.2016.0448 O’Neill, B.C., Kriegler, E., Riahi, K., Ebi, K.L., Hallegatte, S., Carter, T.R., Mathur, R., van Vuuren, D.P. (2014) A new scenario framework for climate change research: the concept of shared socioeconomic pathways. Climatic Change 122, 387–400. doi:10.1007/s10584-013- 0905-2 Peltier, W. R. (2004) Global glacial isostasy and the surface of the ice-age Earth: The ICE- 5G (VM2) Model and GRACE. Annual Review of Earth and Planetary Sciences, 32(1), 111- 149. doi:10.1146/annurev.earth.32.082503.144359 Ranger, N., Reeder, T. & Lowe, J. 2013 Addressing ‘deep’ uncertainty over long-term climate in major infrastructure projects: four innovations of the Thames Estuary 2100 Project. EURO Journal on Decision Processes 1, 233–262. (DOI:10.1007/s40070-013-0014-5). Rabus, B., Eineder, M., Roth, A., Bamler, R. (2003) The shuttle radar topography mission: a new class of digital elevation models acquired by spaceborne radar. ISPRS Journal of Photogrammetry and Remote Sensing 57, 241–262. Sadoff, C.W., Hall, J.W., Grey, J.C.J.H., D. an. Aerts, Ait-Kadi, M., Brown, C., Cox, A., Dadson, S., Garrick, D., Kelman, J., McCornick, P., Ringler, C., Rosegrant, M., Whittington, Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 33 D., Wiberg, D. (2015) Securing Water, Sustaining Growth: Report of the GWP/OECD Task Force on Water Security and Sustainable Growth. Schuerch, M., Spencer, T., Temmerman, S., Kirwan, M. L., Wolff, C., Lincke, D., McOwen C.J., Pickering M.D., Reef R., Vafeidis A.T., Hinkel J., Nicholls R.J., Brown S. (2018). Future response of global coastal wetlands to sea level rise. Nature, 561, 231-234. DOI 10.1038/s41586-018-0476-5 Stive, M.J.F., Fresco, L.O., Kabat, P., Parmet, B.W.A.H., Veerman, C.P. (2011) How the Dutch plan to stay dry over the next century. Proceedings of the ICE-Civil Engineering, 164 (3), 114–121. Tarrant, O. & Sayers, P. (2012) Managing flood risk in the Thames Estuary – the development of a long-term robust and flexible strategy. In Sayers, P. (ed.) Flood Risk: Planning, design and management of flood defence infrastructure London, ICE Publishing, pp. 202-326. Temmerman, S., Meire, P., Bouma, T.J., Herman, P.M.J., Ysebaert, T., De Vriend, H.J., 2013. Ecosystem-based coastal defence in the face of global change. Nature 504, 79–83. Tol, R.S.J. (2002) Estimates of the Damage Costs of Climate Change. Part 1: Benchmark Estimates. Environmental and Resource Economics, 21, 47–73. Vafeidis, A.T., Nicholls, R.J., McFadden, L., Tol, R.S.J., Hinkel, J., Spencer, T., Grashoff, P.S., Boot, G., Klein, R.J.T. (2008) A new global coastal database for impact and vulnerability analysis to sea-level rise. Journal of Coastal Research 24, 917–924 Van Alphen, J. (2015) The Delta Programme and updated flood risk management policies in the Netherlands. Journal of Flood Risk Management DOI: 10.1111/jfr3.12183 World Bank (2010) Climate Risks and Adaptation in Asian Coastal Megacities: A Synthesis Report. World Bank, Washington DC, 120pp. Available at: http://documents.worldbank.org/curated/en/866821468339644916/Climate-risks-and- adaptation-in-Asian-coastal-megacities-a-synthesis-report (Accessed 21 March 2018) World Bank (2017) World Bank Country and Lending Groups. Available at: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and- lending-groups (Accessed 23 March 2018) Yohe, G. and Tol, R.S.J. (2002) Indicators for social and economic coping capacity—moving toward a working definition of adaptive capacity. Global Environmental Change, 12, 25-40. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 34 Appendix 1. Additional Results Contents A.1.1 Protection cost over time A.1.2 Cumulative protection costs A.1.3 Share of protection and maintenance cost A.1.4 Adaptation deficit A.1.5 Residual damage cost Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 35 A.1.1 Protection cost over time Figure A1: Protection costs per region over time as in Figure 7 for SSP3 (upper figure) and SSP5 (lower figure) and high adaptation unit cost. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 36 Figure A2: Protection costs per region over time as in Figure 7 for SSP3 (upper figure) and SSP2 (lower figure) and low adaptation unit cost. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 37 Figure A3: Protection costs per region over time as in Figure 7 for SSP5 and low adaptation unit cost. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 38 A.1.2 Cumulative protection costs Figure A4: The cumulative protection (capital and maintenance) costs as in Figure 9 for SSP2 and low adaptation unit cost. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 39 Figure A5: The cumulative protection (capital and maintenance) costs as before for SSP3 (high Adaptation unit costs in the upper figure) (low Adaptation unit costs in the lower figure) Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 40 Figure A6: The cumulative protection (capital and maintenance) costs as before for SSP5 (high Adaptation unit costs in the upper figure) (low Adaptation unit costs in the lower figure) Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 41 A.1.3 Share of protection and maintenance cost Figure A7: The relative share of capital versus maintenance costs as in Figure 11 for the SSP3 (upper figure) and SSP5 (lower figure) socio-economic scenarios Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 42 A.1.4 Adaptation deficit Figure A8: The adaptation deficit for each RCP/SSP under the risk intolerance strategy assuming high (upper figure) and low adaptation unit cost (lower figure). Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 43 Figure A9: The adaptation deficit for each RCP/SSP under the CBA adaptation strategy assuming high (upper figure) and low adaptation unit cost (lower figure). Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 44 A.1.5 Residual damage cost Figure A10. Annual residual damage cost during 21st century per region for SSP3. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 45 Figure A11. Annual residual damage cost during 21st century per region for SSP5. Appendix 2. Improved estimates of unit costs of sea dikes A.2.1 Introduction This section introduces determinants of unit costs of sea dikes, and approaches and limitations to improve global unit cost estimates. Current DIVA estimates are based on Dronkers et al. (1990) and Hoozemans et al. (1993). In recent years, several new unit cost estimates have been reported for case studies. Jonkman et al. (2013) and Lenk et al. (2016) provide unit cost overviews for several countries; these and other (meta) sources are summarized in Table A1. Table A1: Sources reporting unit costs for coastal defense measures. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 46 Spatial scale or Source Analysis Country location Dronkers et al. Global adaptation analysis. Any Country level 1990 Unit costs are based on estimates measures as applied in The Netherlands and assumed standard dimensions. Country cost factors are derived from expert judgment and applied to other countries Hoozemans et Global vulnerability Any Country level al. 1993 assessment. As in Dronkers estimates et al. (1990), improved by continuous cost functions of dikes and dunes, and inclusion of extreme sea levels Environment Technical feasibility and cost United Kingdom Thames Estuary Agency 2008 analysis for upgrading protection standards of fixed defenses in the Thames Estuary Linham, Green, Adaptation cost analysis of Australia, Country or state level and Nicholls the world’s largest port cities France, estimates 2010 Germany, Italy, New Zealand, South Africa, Spain, United Kingdom, United States - California, Mozambique Jonkman et al. Case study comparison. The Netherlands Country level 2013 Collection of project-based estimates for urban information and conversion and rural areas in comparable units United States New Orleans Vietnam Hai Phong / Nam Dinh Lenk et al. 2016 Regression analysis. Netherlands Country level Comparison with results of estimates other meta studies Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 47 Canada Vancouver, estimates for urban and rural areas A.2.2 Determinants of unit costs of sea dikes Unit costs of coastal protection measures are defined as costs per kilometer length per meter height increment of a flood protection measure. Jonkman et al. (2013) identify material and labor costs, design choices related to the alignment of the system, and the types of measures in an urban or rural environment as the main determinants of unit costs of coastal protection on the basis of three case studies. Moreover, land acquisition costs for dike construction will tend to be higher in urban areas than in rural areas (Delcan Corporation 2012), or involve structural solutions without increasing dike width in urban areas (Kok et al. 2008). In Dronkers et al. (1990) country cost factors ( ) have been applied to derive unit cost ( ) estimates of coastal protection measures at the country level: ⋅ , Eq. 1 where are Dutch unit costs of a standard one-meter dike increment of a dike of variable height, which comprises standard dimensions, construction materials and construction methods. The country cost factor of the Netherlands is normalized to one. The country cost factor is a correction for local conditions in material and labor inputs, mobilization of equipment, economies of scale and land acquisition costs. Differences in costs of financial capital are not included (Dronkers et al., 1990). Country cost factors have been informed by expert judgment of the (then) Delft Hydraulics. Hoozemans et al. (1993) improved these unit cost estimates by the development of continuous cost functions for stone protected and clay covered sea dikes, and sand dunes. They also included an allowance for extreme sea levels which influences initial dike heights, and roughly doubled global costs compared to Dronkers et al. (1990) (see Table 1). To set the stage, we analyzed possible determinants of unit costs of sea dikes with a country-level cross-sectional data set from the DIVA database. The total sample contains 248 observations. We removed 47 observations without low-lying land below 10 meters A.D., as these do not have sea dikes. Furthermore, eight outliers with very high asset densities (e.g. Monaco, Gibraltar, Bahrain) were excluded from the econometric analysis. A least squares regression analysis was performed to investigate how experts appear to have evaluated the importance of local differences in determinants of unit costs, as revealed Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 48 by the differences in country cost factors. A linear specification with initial dike height and no intercept was considered, as suggested by the results of Lenk et al. (2016) and some robustness checks. Country characteristics were also included: GDP per capita, log- transformed country averages of population densities and asset densities in low-lying areas, corruption and democracy variables, and a population-asset density interaction. Table A2: Unit cost results (OLS, no intercept). Estimate Std. Error t value Pr(>|t|) Initial dike height 7.79E-01 1.74E-01 4.48 0.00 (***) GDP per capita 4.64E-05 1.72E-05 2.69 0.01 (***) Log Population density 6.12E-01 1.06E-01 5.78 0.00 (***) Log Asset density -3.17E-01 2.70E-01 -1.17 0.24 Corruption -5.94E-01 3.36E-01 -1.77 0.08 (*) Democracy 2.28E-01 6.44E-02 3.54 0.00 (***) Log Pop. dens.*Log Asset dens. -6.71E-02 4.41E-02 -1.52 0.13 Table A2 reports the results; the adjusted-R2 is 0.77 and the F-test is 92.57 on 7 and 186 degrees of freedom, P-value is 0.00. Higher initial dike heights, higher GDP per capita and higher population densities are associated with higher unit costs. The results suggest that more expensive local labor costs and more expensive land inputs may increase unit costs of sea dikes. Asset density, in contrast, is insignificant. Democracy and corruption are associated with higher unit costs. The latter is only significant at α=0.1. The corruption variable has been scaled inversely, i.e. a higher score means less corruption, which explains the negative sign. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 49 A.2.3 Pragmatic approaches for improving the global unit cost estimates of sea dikes This section discusses a pragmatic approach to improve global unit cost estimates of sea dikes, and the current data limitations to achieve this goal. Dronkers et al. (1990) and Hoozemans et al. (1993) multiplied unit cost estimates of the Netherlands by country unit cost factors to derive unit costs for all countries. More recent unit cost estimates of sea dikes are now available for the Netherlands. Jonkman et al. (2013) report that unit cost estimates of sea dikes range from 4.9 M to 13.5 M euros (2014) for rural areas in the Netherlands, and from 16.9 M to 24.4 M euros for urban areas.1  One pragmatic approach to derive new rural and urban unit cost estimates at the global level would be to multiply the recent rural and urban unit cost ranges of the Netherlands with the original country unit cost factors. However, we hypothesize that applying the Dutch urban unit cost ranges would lead to an upward bias of unit cost estimates at the global level due to a set of specific circumstances in the Netherlands that have not been reflected in the country cost factors. These include the large national budget available for defense works, the large share of land below sea level, high protection standards, high protected value in rural areas, a strong solidarity which favors protection of rural areas more than in other countries, and the polycentric urban structure. Unfortunately, we cannot statistically test this hypothesis due to the low number of case studies reporting urban unit costs at the country level. Alternatively, we apply the rural ranges of the Netherlands to the rest of the world by multiplying them with the DIVA country factors and compare the results with unit costs found in available case studies. Table A3 shows this comparison for the United States, Canada, the United Kingdom and Vietnam. Appendix 2 contains a complete overview of DIVA country cost factors, previous DIVA estimates of unit costs and the newly estimated unit cost ranges. The case studies suggest that country unit cost factors are accurate for some countries, and inaccurate for others. The unit cost estimates reported for New Orleans, USA from the US Army Corps of Engineers in Jonkman et al. (2013), range from 4.0 M to 12.8 M euros (reference year 2014). These estimates resemble the unit cost estimate range in Table A3. Similarly, Canada’s dike unit cost may be in the range of 2 M to 12 M euros based on Lenk et al. (2015). However, the rural estimates of dike unit cost for Hai Phong and Nam Dinh, Vietnam reported in Jonkman et al. (2013) are 10 times lower than those reported in Table A3 for rural areas. 1 Converted to 2014 price levels.  Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 50 Table A3: Comparison between sea-dike unit cost derived from recent Dutch rural estimates. multiplied by DIVA country unit cost factors (left) and taken from available case studies (right). Unit costs are in M Euro km-1 m-1, reference year 2014 DIVA Case study results Case Location Current New study or area Country estimate estimate estimate type Meta source Netherlands 4.2 [4.9-13.5] [4.9-13.5] Rural Jonkman et al. (2013) [16.9-24.4] Urban Jonkman et al. (2013) USA 3.8 [4.3-11.9] [4.0-12.8] New Jonkman et al. Orleans (2013) Canada 3.6 [4.2-11.4] [2-12] Any Lenk et al. (2016) UK 4.7 [5.5-15.1] [16-20] Thames Environment Agency (2008) Vietnam 4.6 [5.3-14.6] [0.8-1.3] Rural Jonkman et al. (2013) Overall, the replacement of DIVA point estimates by ranges based on recent Dutch estimates has some appealing features. It clarifies that unit cost estimates of countries are highly uncertain, and it may improve unit cost estimates if country cost factors are correctly specified. However, the external validity of the method remains limited due to inaccurate country cost factors. Replacing the Dutch unit costs with recent estimates without updating country cost factors does not resolve systematic biases. A.2.4 Suggestions for further research on dike costs The case study results suggest that some country cost factors may be inaccurate. These may be improved by more empirical research on unit costs. A.2.4.1 Production factors of dike construction Case studies on dike unit costs indicate that the production factors labor (wages), land and physical capital (materials, machinery) are among the key determinants of the unit costs of sea dikes. Unit cost estimates may be improved by estimation of a production function of sea dikes: , , Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 51 For this, data on local input and output levels and input prices are needed. However, international surveys on factor use or costs, such as material surveys, for dike construction are currently absent (Jonkman et al. 2013). In some cases, such information is seen as commercially sensitive and hence even if collected it is not in the public domain. Some case studies are nonetheless available that provide illustrations of factor use for construction of coastal protection measures (e.g. Dijkman 2007). However, case study information cannot be generalized across countries, as this is largely explaining the variation in unit costs, and input data are largely lacking for sea dikes. A.2.4.2 Meta-analysis of case studies A meta-analysis as in Lenk et al. (2016), may be improved by more case studies, as well as by inclusion of country characteristics that have been omitted. However, this is a large effort, as available case studies are based on grey literature. Currently, the number of case studies remains insufficient for out-of-sample predictions of unit costs, and the sample of case studies needs to be increased. A.2.5. Other cost issues for further research A.2.5.1 Total costs Dronkers et al. (1990) apply dike length multipliers to coastline length "as the crow flies", to estimate total costs. This method is obsolete, and the applied dike length multipliers are likely to be inaccurate. Total cost estimates may not only be improved by updating unit cost estimates, but also by replacing previous dike length approximations by country totals of segment lengths as is already done in DIVA. However, the base coastline could be reviewed to test coastal lengths, especially where there are large populations and hence potential demands for defenses. A2.5.2 Other measures for unit costs Investment flexibility under sea level rise uncertainty is not addressed in the concept of unit costs, as gradual implementation of the work is not considered (Dronkers et al., 1990). However, the fixed cost component of structural flood protection investments is essential for economic decision-making, and it motivates anticipatory adaptation for the case of flood prevention by dike construction (van Dantzig 1956; Eijgenraam et al. 2016). It may therefore Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 52 be useful to distinguish between the cost components of unit costs of sea dikes that are constant independent of the size of a dike increment, and the cost components that are not. Without this distinction, unit cost estimates may be of less use to inform coastal adaptation decisions, and they may reduce the accuracy of cost estimates at the global (or smaller) level. A2.6 References Delcan Corporation 2009. Cost of adaptation - Sea dikes & alternative strategies. Final report EB3762. (British Columbia Ministry of Forests, Lands and Natural Resource Operations, Vancouver). Dijkman, J., 2007. A Dutch perspective on coastal Louisiana: Flood risk reduction and landscape stabilization. (United States Army: International Research Office of the US Army, London). Dronkers, J., Gilbert,J.T.E., Butler, L.W., Carey, J.J., Campbell, J., James, E., McKenzie, C. , Misdorp, R., Quin, N., Ries, K.L., Schroder, P.C., Spradley, J.R. , Titus, J.G., Vallianos, L., von Dadelszen, J., 1990. Strategies for adaptation to sea level rise. Report of the IPCC Coastal Zone Management Subgroup: Intergovernmental Panel on Climate Change. (Intergovernmental Panel on Climate Change, Geneva). Eijgenraam, C., Brekelmans, R., den Hertog, D., Roos, K., 2016. Optimal strategies for flood prevention. Management Science. Environment Agency 2008. Feasibility of upgrading fixed defences. Report and Appendices. (Environment Agency, TE2100 Project Thames Barrier, London). Hoozemans, F.M.J., Marchand, M., Pennekamp, H.A., 1993. Sea level rise: A global vulnerability assessment. Vulnerability assessments for population, coastal wetlands and rice production on a global scale. (Delft Hydraulics / Rijkswaterstaat, The Hague). Jonkman, S.N., Hillen, M.M., Nicholls, R.J., Kanning, W., van Ledden, M., 2013. Costs of Adapting Coastal Defences to Sea-Level Rise— New Estimates and Their Implications. Journal of Coastal Research 1212–1226. doi:10.2112/JCOASTRES-D-12-00230.1 Kok, M., Jonkman, B., Kanning, W., Rijcken, T., Stijnen, J., 2008. Toekomst voor het Nederlandse polder concept (in Dutch). (TU-Delft / HKV Lijn In Water / Royal Haskoning). Lenk, S., Rybski, D., Heidrich, O., Dawson, R.J., Kropp, J.P., 2016. Costs of sea dikes – regressions and uncertainty estimates. Nat. Hazards Earth Syst. Sci. Discuss. 2016, 1-21. Linham, M.M., Green, C.H., Nicholls, R.J., 2010. Costs of adaptation to the effects of climate in the world’s largest port cities. (AVOID programme). Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 53 Van Dantzig, D., 1956. Economic decision problems for flood prevention. Econometrica 24, 276-287. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 54 Appendix 3. Unit costs of sea dikes in M euros km-1 m-1, reference year 2014 New unit New unit Country cost cost DIVA unit unit cost estimate - estimate - Country cost factor Low High Aruba 1.6 0.4 1.8 5.0 Afghanistan 0.0 0.0 0.0 0.0 Angola 2.6 0.6 3.0 8.3 Anguilla 4.7 1.1 5.5 15.1 Aaland 2.1 0.5 2.5 6.8 Albania 2.0 0.5 2.3 6.2 Andorra 0.0 0.0 0.0 0.0 United Arab Emirates 3.3 0.8 3.8 10.4 Argentina 8.3 2.0 9.6 26.5 Armenia 4.2 1.0 4.9 13.5 American Samoa 3.8 0.9 4.3 11.9 French Southern Territories 2.8 0.7 3.2 8.8 Antigua and Barbuda 2.6 0.6 3.0 8.3 Australia 3.8 0.9 4.3 11.9 Austria 0.0 0.0 0.0 0.0 Azerbaijan 4.2 1.0 4.9 13.5 Burundi 0.0 0.0 0.0 0.0 Belgium 4.2 1.0 4.9 13.5 Benin 2.8 0.7 3.2 8.8 Bonaire, Saba and Saint Eustatius 1.6 0.4 1.8 5.0 Burkina Faso 0.0 0.0 0.0 0.0 Bangladesh 1.3 0.3 1.5 4.1 Bulgaria 2.0 0.5 2.3 6.2 Bahrain 3.3 0.8 3.8 10.4 Bahamas 2.8 0.7 3.2 8.8 Bosnia and Herzegovina 1.4 0.3 1.6 4.4 Saint Barthelemy 2.8 0.7 3.2 8.8 Belarus 4.2 1.0 4.9 13.5 Belize 2.9 0.7 3.4 9.4 Bermuda 1.6 0.4 1.8 5.0 Bolivia 0.0 0.0 0.0 0.0 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 55 Brazil 8.7 2.0 10.0 27.6 Barbados 2.6 0.6 3.0 8.3 Brunei Darussalam 2.8 0.7 3.2 8.8 Bhutan 0.0 0.0 0.0 0.0 Bouvetisland 2.6 0.6 3.0 8.3 Botswana 0.0 0.0 0.0 0.0 Central African Republic 0.0 0.0 0.0 0.0 Canada 3.6 0.8 4.2 11.4 Cocos Islands 3.8 0.9 4.3 11.9 Switzerland 0.0 0.0 0.0 0.0 Chile 2.6 0.6 3.0 8.3 China 6.4 1.5 7.4 20.3 Côte d’Ivoire 2.8 0.7 3.2 8.8 Cameroon 3.4 0.8 4.0 10.9 Congo, Dem. Rep. 3.4 0.8 4.0 10.9 Congo, Rep. 7.0 1.7 8.1 22.4 Cook Islands 2.0 0.5 2.3 6.2 Colombia 3.9 0.9 4.5 12.5 Comoros 3.1 0.7 3.6 9.9 Cabo Verde 2.9 0.7 3.4 9.4 Costa Rica 2.8 0.7 3.2 8.8 Cuba 2.0 0.5 2.3 6.2 Curacao 1.6 0.4 1.8 5.0 Christmas Island 3.8 0.9 4.3 11.9 Cayman Islands 4.7 1.1 5.5 15.1 Cyprus 2.1 0.5 2.5 6.8 Czech Republic 0.0 0.0 0.0 0.0 Germany 4.4 1.0 5.1 14.0 Djibouti 2.3 0.5 2.6 7.3 Dominica 3.3 0.8 3.8 10.4 Denmark 4.1 1.0 4.7 13.0 Dominican Republic 2.6 0.6 3.0 8.3 Algeria 2.1 0.5 2.5 6.8 Ecuador 2.3 0.5 2.6 7.3 Egypt, Arab Rep. 1.5 0.3 1.7 4.7 Eritrea 1.7 0.4 1.9 5.3 Western Sahara 2.2 0.5 2.5 6.9 Spain 2.1 0.5 2.5 6.8 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 56 Estonia 4.2 1.0 4.9 13.5 Ethiopia 2.8 0.7 3.2 8.8 Finland 2.1 0.5 2.5 6.8 Fiji 3.1 0.7 3.6 9.9 Falkland Islands 4.7 1.1 5.5 15.1 France 2.8 0.7 3.2 8.8 Faroe Islands 4.1 1.0 4.7 13.0 Micronesia, Federal State of 1.6 0.4 1.8 5.0 Gabon 5.7 1.3 6.6 18.2 United Kingdom 4.7 1.1 5.5 15.1 Georgia 4.2 1.0 4.9 13.5 Guernsey 4.7 1.1 5.5 15.1 Ghana 4.2 1.0 4.9 13.5 Gibraltar 4.7 1.1 5.5 15.1 Guinea 5.6 1.3 6.4 17.7 Guadeloupe 1.6 0.4 1.8 5.0 Gambia 4.1 1.0 4.7 13.0 Guinea-Bissau 5.6 1.3 6.4 17.7 Equatorial Guinea 2.9 0.7 3.4 9.4 Greece 1.8 0.4 2.1 5.7 Grenada 3.3 0.8 3.8 10.4 Greenland 4.1 1.0 4.7 13.0 Guatemala 4.1 1.0 4.7 13.0 French Guiana 2.6 0.6 3.1 8.4 Guam 3.8 0.9 4.3 11.9 Guyana 5.6 1.3 6.4 17.7 Hong Kong SAR, China 2.8 0.7 3.3 9.0 Heard Island and McDonald Islands 3.8 0.9 4.3 11.9 Honduras 4.4 1.0 5.1 14.0 Croatia 1.4 0.3 1.6 4.4 Haiti 2.6 0.6 3.0 8.3 Hungary 0.0 0.0 0.0 0.0 Indonesia 1.6 0.4 1.9 5.2 Isle of Man 4.7 1.1 5.5 15.1 India 5.7 1.3 6.6 18.2 British Indian Ocean Territory 4.7 1.1 5.5 15.1 Ireland 4.7 1.1 5.5 15.1 Iran, Islamic Republic 2.9 0.7 3.4 9.4 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 57 Iraq 4.9 1.2 5.7 15.6 Iceland 3.9 0.9 4.5 12.5 Israel 2.0 0.5 2.3 6.2 Italy 2.8 0.7 3.2 8.8 Jamaica 3.1 0.7 3.6 9.9 Jersey 4.7 1.1 5.5 15.1 Jordan 2.1 0.5 2.5 6.8 Japan 6.0 1.4 7.0 19.2 Kazakhstan 4.2 1.0 4.9 13.5 Kenya 6.2 1.5 7.2 19.7 Kyrgyzstan 4.2 1.0 4.9 13.5 Cambodia 5.6 1.3 6.4 17.8 Kiribati 2.5 0.6 2.8 7.8 Saint Kitts and Nevis 1.9 0.4 2.1 5.9 Kosovo 2.3 0.5 2.6 7.3 Korea, Rep. 18.5 4.3 21.3 58.7 Kuwait 3.3 0.8 3.8 10.4 Lao PDR 0.0 0.0 0.0 0.0 Lebanon 1.4 0.3 1.6 4.4 Liberia 2.8 0.7 3.2 8.8 Libya 3.4 0.8 4.0 10.9 Saint Lucia 3.3 0.8 3.8 10.4 Liechtenstein 0.0 0.0 0.0 0.0 Sri Lanka 1.1 0.3 1.3 3.6 Lesotho 0.0 0.0 0.0 0.0 Lithuania 4.2 1.0 4.9 13.5 Luxembourg 0.0 0.0 0.0 0.0 Latvia 4.2 1.0 4.9 13.5 Macau 3.4 0.8 4.0 10.9 Saint Martin 2.8 0.7 3.2 8.8 Morocco 3.6 0.8 4.2 11.4 Monaco 0.8 0.2 0.9 2.5 Moldova 4.2 1.0 4.9 13.5 Madagascar 2.8 0.7 3.2 8.8 Maldives 2.8 0.7 3.2 8.8 Mexico 5.2 1.2 6.0 16.6 Marshall Islands 1.9 0.4 2.1 5.9 Macedonia, FYR 0.0 0.0 0.0 0.0 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 58 Mali 0.0 0.0 0.0 0.0 Malta 2.0 0.5 2.3 6.2 Myanmar 8.7 2.0 10.0 27.6 Montenegro 2.3 0.5 2.6 7.3 Mongolia 0.0 0.0 0.0 0.0 Northern Mariana Islands 3.8 0.9 4.3 11.9 Mozambique 5.2 1.2 6.0 16.6 Mauritania 6.7 1.6 7.7 21.3 Montserrat 4.7 1.1 5.5 15.1 Martinique 1.7 0.4 1.9 5.3 Mauritius 2.8 0.7 3.2 8.8 Malawi 0.0 0.0 0.0 0.0 Malaysia 3.8 0.9 4.3 11.9 Mayotte 2.8 0.7 3.2 8.8 Namibia 6.2 1.5 7.2 19.7 New Caledonia 1.8 0.4 2.0 5.6 Niger 0.0 0.0 0.0 0.0 Norfolk Island 3.8 0.9 4.3 11.9 Nigeria 4.9 1.2 5.7 15.6 Nicaragua 6.0 1.4 7.0 19.2 Niue 2.0 0.5 2.3 6.2 Netherlands 4.2 1.0 4.9 13.5 Norway 2.6 0.6 3.0 8.3 Nepal 0.0 0.0 0.0 0.0 Nauru 3.7 0.9 4.3 11.8 New Zealand 2.0 0.5 2.3 6.2 Oman 2.9 0.7 3.4 9.4 Pakistan 4.4 1.0 5.1 14.0 Panama 4.4 1.0 5.1 14.0 Pitcairn Islands 4.7 1.1 5.5 15.1 Peru 2.1 0.5 2.5 6.8 Philippines 1.5 0.3 1.7 4.7 Palau 1.6 0.4 1.8 5.0 Papua New Guinea 2.5 0.6 2.8 7.8 Poland 4.2 1.0 4.9 13.5 Puerto Rico 2.5 0.6 2.8 7.8 Korea, Dem. People’s Rep. 2.8 0.7 3.2 8.8 Portugal 1.6 0.4 1.9 5.2 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 59 Paraguay 0.0 0.0 0.0 0.0 West Bank and Gaza 1.2 0.3 1.4 3.7 French Polynesia 1.7 0.4 1.9 5.3 Qatar 2.0 0.5 2.3 6.2 Reunion 1.5 0.3 1.7 4.7 Romania 4.1 1.0 4.7 13.0 Russian Federation 4.2 1.0 4.9 13.5 Rwanda 0.0 0.0 0.0 0.0 Saudi Arabia 2.9 0.7 3.4 9.4 Sudan 2.8 0.7 3.2 8.8 Senegal 4.2 1.0 4.9 13.5 Singapore 4.1 1.0 4.7 13.0 South Georgia and the South Sandwich Islands 4.7 1.1 5.5 15.1 Saint Helena 4.7 1.1 5.5 15.1 Svalbard and Jan Mayen 2.6 0.6 3.0 8.3 Solomon Islands 2.6 0.6 3.0 8.3 Sierra Leone 4.1 1.0 4.7 13.0 El Salvador 3.8 0.9 4.3 11.9 San Marino 0.0 0.0 0.0 0.0 Somalia 6.4 1.5 7.4 20.3 Saint Pierre and Miquelon 2.8 0.7 3.2 8.8 Serbia 2.3 0.5 2.6 7.3 South Sudan 2.8 0.7 3.2 8.8 São Tomé and Príncipe 1.8 0.4 2.0 5.6 Suriname 4.4 1.0 5.1 14.0 Slovak Republic 0.0 0.0 0.0 0.0 Slovenia 1.4 0.3 1.6 4.4 Sweden 2.1 0.5 2.5 6.8 Swaziland 0.0 0.0 0.0 0.0 Seychelles 2.9 0.7 3.4 9.4 Syrian Arab Republic 2.1 0.5 2.5 6.8 Turks and Caicos Islands 4.7 1.1 5.5 15.1 Chad 0.0 0.0 0.0 0.0 Togo 2.8 0.7 3.2 8.8 Thailand 5.6 1.3 6.4 17.7 Tajikistan 4.2 1.0 4.9 13.5 Tokelau 2.0 0.5 2.3 6.2 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 60 Turkmenistan 4.2 1.0 4.9 13.5 East Timor 1.0 0.2 1.1 3.1 Tonga 3.1 0.7 3.6 9.9 Trinidad and Tobago 3.8 0.9 4.3 11.9 Tunisia 6.2 1.5 7.2 19.7 Turkey 2.1 0.5 2.5 6.8 Tuvalu 1.5 0.3 1.7 4.7 Taiwan, China 0.9 0.2 1.0 2.8 Tanzania 2.8 0.7 3.2 8.8 Uganda 0.0 0.0 0.0 0.0 Ukraine 4.2 1.0 4.9 13.5 United States Minor Outlying Islands 3.8 0.9 4.3 11.9 Uruguay 6.4 1.5 7.4 20.3 United States 3.8 0.9 4.3 11.9 Uzbekistan 4.2 1.0 4.9 13.5 Vatikan 2.8 0.7 3.2 8.8 Saint Vincent and Grenadine 1.8 0.4 2.0 5.6 Venezuela, RB 5.9 1.4 6.8 18.7 British Virgin Islands 4.7 1.1 5.5 15.1 Virgin Islands, U.S. 2.5 0.6 2.9 8.1 Vietnam 4.6 1.1 5.3 14.6 Vanuatu 2.9 0.7 3.4 9.4 Wallis and Futuna 2.8 0.7 3.2 8.8 Samoa 2.5 0.6 2.8 7.8 Yemen, Rep. 1.2 0.3 1.4 3.7 South Africa 7.2 1.7 8.3 22.9 Zambia 0.0 0.0 0.0 0.0 Zimbabwe 0.0 0.0 0.0 0.0 Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 61 Appendix 4. The Dynamic Interactive Vulnerability Assessment (DIVA) Model The Dynamic Interactive Vulnerability Assessment (DIVA) model (Hinkel et al., 2009) currently offers, to our knowledge, both the most detailed global scale representation of the coastal zone and the implications of sea-level rise and the most comprehensive and advanced representation of relevant processes at the global scale, including adaptation. It includes a global database which represents the world’s coasts (excluding Antarctica), using more than 12,000 linear segments with more than 100 parameters per segment. The DIVA model focuses on various aspects including coastal erosion impacts and adaptation (Hinkel et al., 2013), coastal wetland change (Spencer et al., 2016) and coastal flood impacts and adaptation (Hinkel et al., 2014). Here we will focus on flooding and use the same model as Hinkel et al (2014). The flood module can assess the impacts of increased coastal flooding on the population and coastal assets by comparing results obtained using various available data sources and adaptation strategies under a comprehensive sample of state-of-the art socioeconomic and sea-level rise scenarios. Flood risk is considered in terms of expected annual damage to assets, expected annual number of people flooded, and adaptation costs in terms of the dike and other defense investments and the additional maintenance costs to maintain the new defenses. The maintenance costs can be significant and are important to consider as it is important that such maintenance is included in long-term plans. For adaptation, Hinkel et al. (2014) follow earlier studies, such as Hoozemans et al. (1983) and Nicholls (2004), and consider a common protection approach using dikes contrasting various adaptation strategies. One initial DIVA approach to adaptation was based on a demand for safety function (Yohe and Tol, 2002) which attempts to model human behavior based on empirical observation. However, the DIVA framework is flexible, and a wide range of alternative adaptation strategies can be formulated and explored (e.g. Sadoff et al., 2015). In each case, as sea levels rise, the capital and maintenance costs, as well as residual damages can be estimated. DIVA has been used to investigate the implications of sea-level rise in earlier assessments by the World Bank and the Asian Development Bank, including the global analysis of Nicholls et al. (2010), the East Asia analysis of Nicholls et al. (2013) and the national analysis of Kebede et al. (2010). A4.1 References Hinkel J. and Klein R.J.T. (2009) Integrating knowledge to assess coastal vulnerability to sea-level rise: The development of the DIVA tool. Global Environmental Change 19(3), 384– 395. Global Investment Costs for Coastal Defense through the 21st Century Report to the World Bank 62 Hinkel, J., Nicholls, R.J., Tol, R.S.J., and 7 others (2013) A global analysis of erosion of sandy beaches and sea-level rise: an application of DIVA. Global and Planetary Change, 111, 150-158. Hinkel, J., Lincke, D., Vafeidis, A.T., Perrette, M., Nicholls, R.J., and 5 others (2014) Coastal flood damage and adaptation costs under 21st century sea-level rise. Proceedings of the National Academy of Sciences, 111, 3292–3297. Hoozemans, F.M.J., Marchand, M., and Pennekamp, H.A., 1993. A Global Vulnerability Analysis: Assessment for Population, Coastal Wetlands and Rice Production on a Global Scale, 2nd Edition. Delft Hydraulics, the Netherlands. Kebede, A.S., Nicholls, R.J., Brown, S. and Hanson, S.. (2010) Coastal zone. In Mozambique: Economics of Adaptation to Climate Change, World Bank, Washington DC, pp. 47-58. Nicholls R.J. (2004). Coastal flooding and wetland loss in the 21st Century: Changes under the SRES climate and socio-economic scenarios. Global Environmental Change, 14(1), 69- 86. Nicholls, R.J., Brown, S., Hanson, S. and Hinkel, J. (2010) Economics of coastal zone adaptation to climate change, Climate and Development, World Bank Discussion Papers, 10, World Bank, Washington DC, 48pp. Nicholls, R.J., Hanson, S. and Hinkel, J. (2013) The economics of adaptation in the coastal sector - In Economics of Climate Change in East Asia - Westphal, Michael L., Hughes, Gordon A. and Brömmelhörster, Jörn (eds.) Asian Development Bank, Manila, pp: 60-83. Sadoff, C.W., Hall, J.W., Grey, J.C.J.H., D. an. Aerts, Ait-Kadi, M., Brown, C., Cox, A., Dadson, S., Garrick, D., Kelman, J., McCornick, P., Ringler, C., Rosegrant, M., Whittington, D., Wiberg, D., 2015. Securing Water, Sustaining Growth: Report of the GWP/OECD Task Force on Water Security and Sustainable Growth. Spencer, T., Schuerch, M., Nicholls, R.J., Hinkel, J., Lincke, D., Vafeidis, A.T., Reef, R., McFadden, L. and Brown, S. (2016) Global coastal wetland change under sea-level rise and related stresses: the DIVA wetland change model. Global and Planetary Change, 139, 15- 30. Yohe, G. and Tol, R.S.J., 2002. Indicators for social and economic coping capacity—moving toward a working definition of adaptive capacity. Global Environmental Change, 12, 25-40.