The Economic Impact of Sea-level Rise on Nonmarket Lands in Author(s): Wei-Shiuen Ng and Robert Mendelsohn Source: AMBIO: A Journal of the Human Environment, 35(6):289-296. 2006. Published By: Royal Swedish Academy of Sciences DOI: http://dx.doi.org/10.1579/05-A-076.1 URL: http://www.bioone.org/doi/full/10.1579/05-A-076.1

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research. Article Wei-Shiuen Ng and Robert Mendelsohn The Economic Impact of Sea-level Rise on Nonmarket Lands in Singapore

such as beaches, marshes, and mangroves. Because they provide Sea-level rise, as a result of climate change, will likely pleasure to many people, they are often not traded on markets inflict considerable economic consequences on coastal and have no monetary market value. It is therefore not obvious regions, particularly low-lying island states like Singapore. how much should be spent to protect these resources from Although the literature has addressed the vulnerability of inundation. This study analyzes whether abandoning natural developed coastal lands, this is the first economic study to areas or protecting them is more efficient. Protection is advised address nonmarket lands, such as beaches, marshes and only if the benefits of protection are higher than the costs (11). mangrove estuaries. This travel cost and contingent We assess the benefits of protecting each coastal resource using valuation study reveals that consumers in Singapore a willingness-to-pay survey and a travel cost study. These attach considerable value to beaches. The contingent valuation study also attached high values to marshes and benefits are then compared with the costs of protection. mangroves but this result was not supported by the travel The next section describes the physical benefits associated cost study. Although protecting nonmarket land uses from with preserving beaches, marshes, and mangroves. The section sea-level rise is expensive, the study shows that at least following that describes the methodology used to value the highly valued resources, such as Singapore’s popular benefits of protecting these public resources. Results are beaches, should be protected. subsequently presented from the travel cost analysis and contingent valuation survey, and the last section describes the cost of protection for each resource. The paper concludes with policy recommendations for Singapore to address desirable INTRODUCTION AND LITERATURE REVIEW adaptation in natural areas to sea-level rise. The consequences of sea-level rise due to global climate change PHYSICAL BENEFITS have significant social and ecological impacts on coastal regions throughout the world. Global sea-level rise is caused primarily The shore zone has natural features that provide considerable by thermal expansion of seawater from rising ocean tempera- coastal protection. Sand and gravel beaches contribute as wave tures and also by melting of terrestrial ice, glaciers, and ice energy sinks, and barrier beaches act as natural breakwaters sheets (1, 2, 3, 4). Intergovernmental Panel on Climate Change (12). Coastal vegetation absorbs wind or wave energy, retarding (IPCC) has projected that sea levels will rise on average by five shoreline erosion. Marshes act as a sea defense (13), and mm per year over the next 100 years and could possibly rise mangroves are sediment traps (14) that act as a buffer zone even faster (5). These changes are important because they have between land and sea and play a significant role in protecting the potential to alter ecosystems and habitability in coastal both the coastal areas and coral reefs at the same time (7). If regions, which are home to an increasing percentage of the these ecological functions of the natural coastal systems are lost, world’s population, habitat for much of the world’s fisheries, coastal resilience would decline. As sea level rises, beach and vacation spots across the world (6). erosion, wetland displacement, and mangrove species inunda- Since the impact of sea-level rise varies among coastal regions, tion will occur, unless adaptation measures are implemented in it is important to assess each nation’s vulnerability to sea-level time. Beach erosion will move the shoreline, shifting the beach change (7). The degree of influence depends on the area that may profile closer inland (15). One method of preventing beach be inundated, the cost of potential adaptation measures, and the erosion is continual beach nourishment, which preserves ability to adopt such measures. Small island states are among beaches in their current conditions and discourages further some of the most vulnerable regions in the world to sea-level rise. erosion. Another approach is to build undersea seawalls and Climate change and sea-level rise will no doubt pose a serious backfill sand behind these hard structures. In both cases, the threat to small island states like Singapore (8). beach could continue to be used for recreational purposes. Global climate change and global sea-level rise are beyond Singapore has already installed hard undersea structures for Singapore’s control (9), but Singapore does have adaptation land reclamation (16). Beach protection is expensive, but it is a options that can reduce damages. IPCC 2001 argued that the feasible option. most serious considerations for some small island states is As for marshes and mangrove estuaries, natural inland whether they will have adequate potential to adapt to sea-level migration could be a protection measure. The impact of sea- rise within their own national boundaries (5). Singapore has the level rise on marshes and mangrove estuaries depends on financial capacity to protect its developed lands, and it will be vertical accretion rates and space for horizontal migration, able to carry out effective adaptation measures. In a recent because increasing sea level will destroy current habitats, hence, study, Ng and Mendelsohn (10) demonstrated that Singapore marsh and mangrove species would have to shift to new tidal should protect all of its market lands from inundation. The cost areas (5). Coastal ecosystems are threatened when marshes of protecting Singapore’s developed coast is much less than the drown because they do not accrete vertically fast enough (17). value of the potentially inundated land. They develop ponds as they drown and eventually can This study focuses on the economic impact of sea-level rise disappear entirely (6). The vertical accretion of the marsh on nonmarket lands in Singapore. Although the economic surface must occur at a rate at least equal to the rate of relative literature on sea-level rise has addressed developed lands, this is sea-level rise, in order to maintain an elevation for marsh the first study to address land that is not developed. Nonmarket vegetation to survive (17). Sea-level rise could thus easily alter lands are jointly consumed by many people and include areas marsh hydrology and affect the rate of net vertical accretion.

Ambio Vol. 35, No. 6, September 2006 Ó Royal Swedish Academy of Sciences 2006 289 http://www.ambio.kva.se Z ‘ Mangroves go through similar processes when sea level rises. WTP ¼ VðtiÞdt: Eq: 3 When the rate of sea-level rise exceeds the rate of sedimentation, Ti substrate erosion, inundation stress, and increased salinity will where willingness to pay (WTP) is the annual value to the occur (18), mangrove species zones will migrate inland, and seaward margins will die back (7, 19). These events could individual of the site. The social value of a site is the sum of the happen to the mangrove estuaries on the western and northern consumer surplus values of all the individuals who visit that coasts of Singapore with increasing sea level (20). particular site (26). Coastal wetlands would naturally migrate inland in response to relative sea-level rise. However, in Singapore’s case, highly Contingent Valuation Method developed land behind the wetlands will prevent adequate sediment supply to the mangrove estuaries (20) and will inhibit The contingent valuation method (CVM) uses surveys of public the natural migration of mangroves as sea level rises. This opinion to estimate the nonmarket values associated with changes creates a complex problem, since protecting nonmarket lands in the environment. Numerous previous studies have applied will often imply sacrificing developed lands. Previous studies CVM in various valuations of natural resources (27, 28, 29, 30). have estimated that as more developed lands are protected, The contingent valuation method asks the sample population there will be a greater loss of wetlands (21, 22, 23, 24). The hypothetical questions about their willingness to pay for a site. extent of the impact of sea-level rise will depend on the decision Since this method could cover both use and nonuse values, it to either protect or modify the coastline, therefore allowing was chosen in this study to measure the nonmarket values of coastal wetlands to migrate inland (25). A tradeoff exists beaches, marshes, and mangroves in Singapore. The average between preserving nonmarket lands and protecting developed willingness to pay observed in the sample was then extrapolated lands, as human infrastructure will prevent coastal wetlands and to the entire adult population in Singapore to obtain an mangroves from migrating inland as sea level rises. In this aggregate valuation: study, we assume that the cost of marsh and mangrove VðY WTP; P; QÞ¼VðY; P; QÞ: Eq: 4 protection is the market value of the property lost due to inland migration of the marshes and mangroves. where V is the indirect utility function of the individual, Y is income, P is the price of a trip, and Q is the number of trips METHODS OF MEASURING THE BENEFITS taken. OF PROTECTION Survey Data Two methods were used to determine the value of the coastal resources: travel cost and contingent valuation. The travel cost This study employed both a quantitative and qualitative research method is a behavioral technique that deduces values from what design. A survey with a sample size of 338 was conducted over a people actually do. Contingent valuation is an attitudinal period of ten weeks in the summer of 2002 in Singapore. approach that measures values from what people say, not what Questionnaires were evenly distributed to residents above the age they do. Each method has its own strengths and weaknesses. of 19 across the country (the questionnaire designed for this The travel cost method only measures the use value of a site, study is available from the authors upon request). The survey whereas the contingent valuation method measures both use was conducted personally so as to assure maximum positive and nonuse. We use both methods in this study to measure a response rate and to obtain reliable results of high quality. range of possible values for these coastal nonmarket sites. The questionnaire used in this survey consisted of 20 questions and took approximately 10 to 15 minutes to complete. Travel Cost Method Questions 1 to 4 were asked to support the travel cost study. The first question required respondents to mark their residential The travel cost method (TCM) is a tool that deduces the values area on the map provided. It was then possible to calculate the of resources based on the decisions of visitors to travel to the distance traveled to each site and, subsequently, the travel cost site from different distances. The travel costs they incur in order of each respondent. Questions 2 to 4 listed 11 specific beach, to visit the site is the price of admission (11). The further an marsh, and mangrove sites and asked for the respondents’ individual lives from the site, the higher the travel cost. By frequency of visits within the past year. Questions 5 and 6 tested observing how visitation changes with distance, the analyst can respondents’ knowledge of sea-level rise. estimate the demand for visitation. The analyst estimates the Questions 8 to 10 asked attitudinal questions that helped demand to visit each site (beach, marsh, or mangrove estuary) identify protestors. They asked whether the respondent felt from these cross-sectional data (26). The demand for visits to a responsible for protecting natural sites, whether polluters were beach, marsh, or mangrove estuary is given as: responsible for damages from global warming, and whether the government would actually protect resources if paid. If people Vi ¼ f ðCi; X1i; X2i; ...; XniÞ; Eq: 1 were hostile to these attitudinal questions and offered zero where Vi is visits by the ith individual, Ci is the cost of a visit by willingness to pay, they were considered to be protestors. individual i, and the X’s are demand shift variables. In this The contingent valuation questions began with a general analysis, we assume that demand has the following form: interest question. Question 7 asked: Will you agree to an increase in your annual personal income tax to be used to V ¼ a þ bC þ e ¼ a þ bðT Þþe ; Eq: 2 i i i i i protect each of the following three different types of nonmarket where ei is the stochastic component, assumed to be normally land in Singapore? and independently distributed, with zero expectation. Almost all the sites analyzed in this study had no admission fee, so only Beaches...... Yes No Don’t Know travel costs were taken into account when estimating the Marshlands...... Yes No Don’t Know demand function. Mangroves...... Yes No Don’t Know The value of a site to an individual is equal to the area under the individual’s demand curve for that site but above their travel If a respondent chose ‘‘Yes,’’ they were asked three close- costs, which can be calculated as: ended willingness-to-pay questions, with answer categories

290 Ó Royal Swedish Academy of Sciences 2006 Ambio Vol. 35, No. 6, September 2006 http://www.ambio.kva.se country. The geographical locations of the eleven tested sites are marked on Figure 2. The four beaches are , East Coast, , and West Coast Beach Parks. The beaches are distributed in the south and east and range from average to very popular beaches. The four marshes are Reservoir, Poyan, Sungei Khatib Bongsu, and Marshes. The marshes are on the north and west coasts. The final three sites are mangroves. They are Sungei Buloh Nature Park, Sungei , and Sungei . The mangroves are located on the north coast.

Travel Cost The total travel cost for each individual to each of the 11 sites was calculated using the estimated travel distance and the transportation cost at 0.33 USD per km. The visitation regression model was applied to each site, with number of visits (V) as the dependent variable and travel cost (TC), age (AGE), gender (GEN), number of children (CHD), educational level (EDU), family size (FAM), and income level (INC) as the independent variables:

Figure 1. The geographical location of the 338 survey respondents. V ¼ b0 þ b1TC þ b2AGE þ b3GEN þ b4CHD þ b5EDU þ b6FAM þ b7INC; Eq: 5 ranging from $0 to more than $500. Each question included a where bn is a constant. short description of the solution to beach erosion, marsh, and The visitation regression results are displayed in Table 1. mangrove retreat. The willingness to pay was measured in terms Only variables that are significant for at least one site are listed. of an increase in income tax. Age, gender, number of children, and income were only The remaining questions were demographic (age, gender, significant for one or two sites. Age was significant in Changi number of children, education level, family size, income level). Beach Park (t ¼2.17) and Kranji Reservoir Marsh (t ¼ 2.33), These were placed last in the questionnaire, because they involve but with opposite signs for the coefficient. Gender was only private information and respondents are more likely to reveal significant in one site, (t ¼ 2.51). Men such information if other questions have been answered first (31). tended to visit this beach site more frequently. The number of ArcView GIS (geographic information system) was used to children was significant in Sungei Punggol, and income was calculate the distance between each respondent’s residential significant in . area and each nonmarket land site. LimDep was used to analyze The travel cost coefficient was negative (the expected sign) and the data collected from the survey. All estimates in this study significant in only five out of the eleven sites. These successful are expressed in terms of 2002 United States dollars (USD) regressions include all the beaches and one marsh site, Pasir Ris unless otherwise stated (Singapore dollar: 1 SGD is approxi- Park. The travel cost coefficient on the remaining marsh sites and mately 0.59 USD). all the mangroves were either the wrong sign or not significant. Because valuation is based on this travel cost coefficient, the RESULTS results imply that these other sites did not have observable value. Survey That is, the behavior of visitors was low, effectively random, and Figure 1 shows the location of every respondent. As can be seen did not reveal statistically significant value for these sites. from Figure 1, the survey was distributed evenly across the Using the travel cost demand functions in Table 1 that had negative travel cost coefficients, we calculated consumer surplus. The consumer surplus for Sungei Khatib Bongsu (marsh), Sungei Buloh Nature Park (mangrove), and Sungei Mandai (mangrove) could not be estimated. It is reasonable to assume that these sites with positive travel cost coefficients have low use value. Average individual consumer surplus and total consumer surplus (Table 2) were calculated using the results generated from the visitation regression equation. The consumer surplus measured the economic benefits attached to each site. Two of the beaches had relatively low values per person, whereas Pasir Ris and beaches were valued at 65 and 426 USD, respectively. The two low-valued beaches were worth an aggregate value of 140 000 and 1.5 million USD, but the two most popular beaches were valued at 167 million and 1090 million USD per year. The consumer surplus estimates for marshes were much less. The most popular marsh was worth only 240 000 USD in total. The estimates for the mangrove estuaries were the lowest. The only valued mangrove was worth just 815 USD per year. This approach suggests that the beaches Figure 2. The geographical location of the 11 natural resource sites are worth far more than the other natural sites and that the studied. most popular beaches are worth considerably more.

Ambio Vol. 35, No. 6, September 2006 Ó Royal Swedish Academy of Sciences 2006 291 http://www.ambio.kva.se Table 1. Visitation regression model. Beach Marsh Mangrove

1 2 345678 9 1011

Changi West Kranji Sungei Pasir Ris Beach East Coast Pasir Ris Coast Reservoir Ponyan Khatib Park Sungei Buloh Sungei Sungei Independent Park Park Park Park Marshes Marsh Bongsu Marshes Nature Park Mandai Punggol Variables: (CBP) (ECP) (PRP) (WCP) (KRM) (PM) (SKB) (PRPM) (SBNP) (SM) (SP)

Constant 2.28 25.03 9.87 0.33 –0.10 0.16 –0.10 0.85 –0.12 –0.15 0.34 (3.61) (2.12) (1.36) (0.21) (0.21) (1.05) (0.67) (1.10) (0.42) (1.15) (2.05) Travel cost –0.04 –1.48 –0.53 –0.12 –0.02 –0.007 0.01 –0.07 0.006 0.005 –0.006 (2.37) (3.27) (2.11) (2.03) (1.31) (1.66) (1.22) (2.18) (0.61) (0.79) (0.68) Age –0.03 –0.17 –0.05 –0.004 0.02 –0.0007 0.002 –0.008 0.004 0.002 –0.0019 (2.17) (0.77) (0.35) (0.15) (2.33) (0.27) (0.69) (0.57) (0.86) (0.81) (0.63) Gender 0.48 –0.45 –0.02 0.13 0.10 0.08 0.06 0.09 –0.09 0.04 0.008 (2.51) (0.13) (0.007) ( 0.29) (0.75) (1.93) (1.52) (0.40) (1.08) (1.17) (0.16) Children –0.05( –0.004 –1.39 –0.53 –0.003 –0.01 –0.03 0.18 0.03 0.05 0.10 –0.33) (0.001) (0.88) (1.63) (0.03) (0.48) (0.95) (1.04) (0.52) (1.80) (2.87) Income –0.37E–06 –0.6E–05 –0.2E–04 0.7E–05 0.2E–06 –0.2E–07 0.1E–06 0.2E–05 0.3E–06 –0.6E–07 –0.2E–06 (0.33) (0.31) (1.66) (2.55) (0.29) (0.08) (0.49) (1.29) (0.61) (0.30) (0.62) R 2 0.06 0.04 0.04 0.04 0.02 0.03 0.03 0.03 0.02 0.03 0.03

OLS regression model, which estimated the relation between the number of visits, travel cost, and other demographic factors. The t-statistic is in parentheses. There were 338 observations in these regressions.

Contingent Valuation the protection of beaches, mangrove estuaries, and marshes (Questions 11–13), were identified as protesters. They were Figures 3 to 5 reflect the responses to the questions about removed from the estimation of contingent valuation (32). responsibility for protecting natural resources, such as the We next estimated the actual WTP of each respondent using beaches, marshes, and mangroves. These questions were aimed two approaches. A tobit model was used to estimate the at determining general public attitudes and specifically deter- combined probability that someone would pay and how much mining whether people were likely to be hostile to the survey. he or she would pay. The second approach used a two-equation Only 24% of the respondents agreed that it was their model that first used a probit model to determine whether responsibility to pay for the protection of beaches, marshes, someone would pay, followed by a conditional Ordinary Least and mangrove estuaries (Fig. 3). A significant number, 46%,of Squares (OLS) regression to estimate the amount that they the respondents did not know if it was their responsibility or would pay (Tables 3–5). This probit model created a dummy not. The remaining respondents, 30%, believed that they were variable, 0, for respondents who gave a zero willingness to pay not responsible for the cost of protecting natural resources. estimate, and 1 for respondents who gave a positive value. However, this does not imply that these respondents would not Hence, only respondents who agreed to pay a positive amount be willing to pay for the protection of natural resources. were included in the OLS regression. Both approaches The majority (90%) of the respondents agreed that polluters generated similar results across questions asking about protect- should bear the cost of damages resulting from global warming, ing (both 50% and 100% of) the beaches, marshes, and and hence, should protect natural resources from inundation mangroves. We consequently present only the results from the (Fig. 4). Residents of Singapore believe in the ‘‘polluter pays’’ 100% protection case in Tables 3–5. principle. They also believe that resources would not be In the regression for beaches, Table 3, family size was only protected, even if the public paid, as reported by 44% of the significant in the probit model, where larger families were more respondents (Fig. 5). About 25% of the respondents felt that likely to be willing to pay for protection. Gender and income their payment would not aid in protecting nonmarket lands. level were significant in both the conditional OLS and tobit The respondent’s attitudes suggested that many of the models. Men and higher-income families were willing to pay respondents might protest against a survey that asked them to more for protection of beaches. Age, number of children, and pay to protect Singapore against the damages of sea-level rise. education were not significant in any regression. Twenty respondents, who answered negatively to all of the The probit results in Table 4 suggest that families with above three questions and gave zero as the willingness to pay for children are less likely to be willing to pay to protect marshes,

Figure 3–5. The public should bear the responsibility for the protection of natural resources. Polluters should pay to fix the damages from global warming. The natural resources would not be protected even if the respondents paid.

292 Ó Royal Swedish Academy of Sciences 2006 Ambio Vol. 35, No. 6, September 2006 http://www.ambio.kva.se but larger families are more likely to be willing. According to Table 2. Consumer surplus (CS) values derived from visitation the conditional OLS and tobit models, males and higher-income regression model. families are willing to pay more for marsh protection, as with Total Value beach protection. Identical results apply to the mangrove CS/ (2005 models in Table 5. Name Characteristic person USD/year) The CVM responses were then used to derive the willingness Changi Beach Park Beach *0.053 *140 441 to pay of the public for the protection of nonmarket lands East Coast Park Beach *426 *1 090 341 680 (Table 6). Beach protection had the highest willingness to pay Pasir Ris Park Beach *65 *167 106 714 value. On average, people were willing to pay 23 USD per year West Coast Park Beach *0.58 *1 487 368 Kranji Reservoir Marshes Marsh 0.0059 20 562 to protect 50% of the beaches and 33 USD for 100% protection. Poyan Marshes Marsh 0.0003 686 Mangrove and marsh protection were valued only slightly less. Sungei Khatib Bongsu Marsh N/A N/A People were willing to pay about 18 USD per year to protect Pasir Ris Park Marshes Marsh *0.095 *239 875 Sungei Buloh Nature Park Mangrove N/A N/A 50% of the mangroves and marshes and 25 USD for 100% Sungei Mandai Mangrove N/A N/A protection. Summing these results across the entire adult Sungei Punggol Mangrove *0.0003 815 Singapore population gives the resulting aggregate estimates Statistically significant values are marked with an asterisk (*). All values are in 2005 USD. of WTP. The CVM results suggest much higher values for the mangroves and marshes than the TCM results. However, the CVM generated values for beaches were consistent with the range of values measured by the TCM.

COST OF PROTECTION Table 3. Probit, OLS, and tobit estimation of WTP for beach Beaches protection. The sea-level rise literature focuses on protecting coastlines by Dependent Variable: Willingness to Pay 100% Protection building sea walls (12, 33). Whereas that is a feasible strategy for Observations 318 278 318 developed land, it is not suitable for beaches, as it would not provide a continuous transition between the land and sea. Independent Variables: Probit Linear(OLS) Tobit Singapore is actually a frontier leader in reclaiming beach from Constant 0.86 48.35 36.02 the sea. Singapore has long explored building hard structures (7.46) (1.41) (1.13) under the sea in order to reclaim land for the island. After Age –0.04 –0.01 –0.12 (1.69) (0.02) (0.21) constructing the underwater hard structure, engineers raise the Gender –0.05 27.00 21.73 sand levels behind the structure. The result is a beach that (1.39) (2.45) (2.20) appears natural from the land. This approach requires two Children –0.01 5.73 1.79 (0.48) (0.75) (0.27) major costs: the hard structure and the sand. Education 0.01 –5.41 –4.17 We follow a careful adaptation strategy that builds sea walls (0.22) (0.33) (0.28) over time as they are needed (34). This approach takes into Family size 0.03 –2.43 0.17 (2.51) (0.63) (0.05) account the sea-level rise that has occurred up to each moment Income 0.24E–06 0.26E–03 0.26E–03 of time and designs the least-cost strategy for the expected (1.21) (4.32) (4.63) future. The faster sea levels rise, the quicker adaptations must R 2 0.06 0.09 0.12 be taken. However, if the sea rises slowly, decision makers have Figures in parentheses are the t-statistic values. more time to act. We examine applying this strategy to three sea-level rise scenarios over the next hundred years. The timelines involve a 0.2, 0.49, and 0.86 m increase in sea level by 2100 (1, 35). These three scenarios span the range of effects predicted by the IPCC. For each scenario, a hard structure is constructed when the sea level rises to a certain height. In the 0.2 m sea-level rise scenario, only a single structure will be built and it will be done in 2080. Table 4. Probit, OLS, and tobit estimation of WTP for marsh protection. With the 0.49 m scenario, the first structure will be built in 2040, and a taller replacement will be built in 2100. In the 0.86 m Dependent Variable: Willingness to Pay 100% Protection scenario, three structures will be built in 2020, 2060, and 2100. Observations 318 242 318 The more rapid the sea-level rise, the earlier the first structure Independent Variables: Probit Linear(OLS) Tobit must be built and the more frequently it must be replaced by a higher wall. Constant 0.45 18.8 –8.86 (2.87) (0.51) (0.29) Although the use of such structures to protect against sea- Age 0.001 0.31 0.42 level rise has not been suggested before, Singapore has had (0.39) (0.45) (0.75) extensive experience with the approach. Almost all the existing Gender 0.03 37.02 29.47 (0.69) (3.16) (3.15) beaches in Singapore are artificial and have been constructed Children –0.07 6.49 –1.88 with granite stones, which generally have a design life between (2.16) (0.79) (0.30) 50 and 100 years (17). Beach sand is not a scarce resource; Education 0.04 –2.96 1.89 (0.52) (0.17) (0.13) hence, this approach may be a practical and economical option Family size 5.6E–02 –1.76 2.06 to protect key beaches from sea-level rise. Beach nourishment (3.23) (0.42) (0.61) also requires maintenance in the form of constant sand Income 3.1E–07 2.8E–04 2.4E–04 (1.15) (4.39) (4.64) replenishment, every 5–10 years (36). The current cost of sand 2 R 0.06 0.13 0.10 is estimated to be 30 USD per cubic meter, while the present value of the stream of maintenance costs (MC) is assumed to be Figures in parentheses are the t-statistic values.

Ambio Vol. 35, No. 6, September 2006 Ó Royal Swedish Academy of Sciences 2006 293 http://www.ambio.kva.se Table 5. Probit, OLS, and tobit estimation of WTP for mangrove Table 7. Total impact of sea-level rise on beaches in Singapore protection. until 2100. Dependent Variable: Willingness to Pay 100% Protection Sea-level Assumed Benefit of Benefit of Observations 318 242 318 rise years of protection protection Cost of Independent Variables: Probit Linear(OLS) Tobit scenario Year protection (WTP) (travel cost) protection

Constant 0.48 26.69 –6.96 0.2 m 2080 80 5 352 81 236 7.37 (3.11) (0.73) (0.23) 0.49 m 2040 60 4 002 60 750 8.99 Age 0.002 0.12 0.46 2100 60 13 132 199 324 32.88 (0.75) (0.18) (0.83) 0.86 m 2020 40 2 841 43 130 6.41 Gender 0.02 38.32 29.19 2060 40 5 220 79 234 31.16 (0.32) (3.34) (3.11) 2100 40 11 526 174 952 55.62 Children –0.07 6.07 –2.18 Both cost and benefit of protection are shown in millions of USD, discounted to present (2.19) (0.74) (0.34) values, using a 4% discount rate. The frequency of protection measures depends on the Education 0.04 –5.19 2.44 sea-level rise scenario. (0.6) (0.3) (0.17) Family size 0.05 –2.53 1.15 (2.78) (0.62) (0.34) Income 0.12 0.34 2.6E–04 significantly higher, due to the high frequency of visits made to (0.45) (5.17) (4.94) R 2 0.04 0.15 0.11 the beach sites. In Table 7, we examine beach adaptation choices. With each sea-level rise scenario, there is a moment Figures in parentheses are the t-statistic values. when building a hard structure is required to protect the beach. For each construction choice, we look at the present value of benefits for that project evaluated at the moment the project is 4% of the construction cost. The volume of sand (VS) required to be executed. We assume that without the protection, the varies from beach to beach and is replenished at different times beach would lose its annual recreation value. For example, the according to the different sea-level rise scenarios. We estimate first sea wall built to protect against the 0.49 m sea-level rise the cost of protecting each of the four beaches used in the travel increase must be built in 2040. The structure is supposed to last cost study. The cost of beach protection includes the until 2100. We take the construction costs in 2040 and compare construction cost of the underwater hard structures (CS), the them to the stream of benefits from 2040 to 2100 evaluated in cost of sand, and the maintenance cost (MC): 2040 Singapore dollars. The present value in 2040 of the stream 3 : : of benefits is: TC ¼ CS þ VS 50 þ MC Eq 6 Z 2100 hi The total cost (TC) ranges from approximately 336 000 USD to rðt2040Þ Bt rð21002040Þ PVðBtÞ¼ Bte dt ¼ 3 1 e : 4.37 million USD for the four beaches in the 0.2 m sea-level rise 2040 r scenario. The cost of protection increases dramatically together Eq: 7 with the height of the sea wall. In the 0.49 m scenario, the The results in Table 7 reveal that the benefits of beach second wall required in 2100 costs from 1.50 million to 19.51 protection far outweigh the costs. Every single beach million USD, depending on the length of the beach. In the 0.86 adaptation project is justified whether one uses travel cost or m scenario, the cost of the third wall ranges from 2.54 million to contingent valuation. Singapore should protect its more 32.99 million USD. (Costs of beach, marsh, and mangrove valuable beaches against sea-level rise, regardless of the protection in 0.2, 0.49, and 0.86 m sea-level rise scenarios are scenario. Since the time line applied in this study is from the available from the authors upon request.) present to 2100, protection cost and benefit values are only estimated until 2100. The discount rate used in the benefit Marshes and Mangroves estimation for all three resources is 4%. A lower rate of 2% will In many countries, the most efficient method to preserve result in higher benefit values, but these would not be marshes and mangroves in their natural state is to allow them to substantial enough to alter the predicted year or frequency of migrate inland as sea level rises. This requires that inland- sea-level rise protection, particularly in the cases of marshes and developed land be sacrificed to advancing mangroves and mangroves. marshes. The cost of this approach is the lost value of the inland For marshes and mangroves, we take a decadal approach to area. In Singapore, the inland spaces are mostly developed, analyze protection decisions. Every decade, we examine what it which makes this approach too expensive to be feasible. would cost to move the marsh or mangrove back into land that is now developed. If the resource is allowed to move back, we assume that it will provide another decade of benefits. The cost BENEFITS AND COSTS of protection is the value of lost developed land. Both the costs In this section, we compare the costs and benefits of different and benefits are evaluated as though the decision must be made adaptation strategies to deal with sea-level rise for each at the beginning of each decade. resource. We compare the two different estimates of benefits The decadal costs and benefits of marsh and mangrove from the travel cost and contingent valuation study respectively. protection are shown in Table 8. The consumer surplus values The benefits of protection derived from the travel cost study are derived from the travel cost method are minimal for marshes and almost negligible for mangrove protection. The costs of protecting both resources exceed the estimated benefits using Table 6. Willingness to pay (WTP) for 50% and 100% protection of travel cost analysis, due to the high value of land in Singapore. natural resource. The travel cost analysis suggests that Singapore should not Beaches Marshes Mangroves allow either resource to migrate inland. The contingent valuation values in Table 8 for marsh and mangrove protection Willingness to Pay 50% 100% 50% 100% 50% 100% are much higher. These analyses suggest that it is reasonable to Average WTP/person 23 33 18 25 18 26 pay for inland migration at least initially, if sea level increases Total WTP 58 83 45 64 47 65 (millions of 2005 USD/year) follow the lower scenarios. However, with the 0.86 m sea-level rise, inland migration eventually becomes very expensive, and

294 Ó Royal Swedish Academy of Sciences 2006 Ambio Vol. 35, No. 6, September 2006 http://www.ambio.kva.se Table 8. Total impact of sea-level rise on marshes and mangroves substantial nonuse value. People may have enjoyed simply in Singapore. knowing that they were there. The study also examined the costs of protection. All these Cost of protection Benefit of protection Sea-level rise scenario resources can be protected with underwater hard structures and additional sand. In the long run, mangroves and marshes can Year WTP Travel cost 0.2 m 0.49 m 0.86 m also survive by migrating inland as the sea rises. Although in Marsh most tropical countries, migration may be the least-cost 2010 1823 7.44 40.66 99.53 174.79 alternative, the high level of development in Singapore makes 2020 2192 8.94 99.53 143.94 252.78 hard structures more attractive. Examining three sea-level rise 2030 2616 10.67 143.94 184.71 324.37 2040 3087 12.60 184.71 222.06 389.96 scenarios, we calculated what the costs will be to save the 11 2050 3579 14.60 222.06 256.19 449.91 sites examined in the travel cost study. 2060 4027 16.43 256.19 287.31 504.57 We compared the present value of the benefits of protection 2070 4298 17.53 287.31 315.59 554.24 2080 4128 16.85 315.59 341.22 599.24 from the travel cost and contingent valuation studies against the 2090 3013 12.30 341.22 364.35 639.86 costs. Both the travel cost and contingent valuation methods 2100 3673 14.99 364.35 385.15 676.37 suggested that it would be worthwhile to protect the beaches in Mangrove 2010 1 841 0.02 41.84 102.81 185.29 every sea-level rise scenario. The travel cost analysis, however, 2020 2 212 0.03 60.50 150.48 264.23 revealed that it would not be worthwhile to protect the marshes 2030 2 641 0.03 77.63 194.33 340.20 and mangroves. The cost of protection was orders of magnitude 2040 3 116 0.04 93.34 234.51 409.87 more expensive than the benefits of the lost resources. In 2050 3 612 0.05 110.98 267.93 473.56 2060 4 064 0.05 124.00 301.49 531.62 contrast, the contingent valuation analysis suggested that 2070 4 338 0.05 135.76 332.06 584.41 protection was worthwhile as long as sea-level rise was limited. 2080 4 167 0.05 146.41 359.77 629.30 Once sea-level rise exceeded approximately 0.6 m, however, the 2090 3 041 0.04 156.01 381.98 672.54 2100 3 707 0.05 164.64 404.55 711.47 cost of protection became too high. However, for milder sea- level rise outcomes, the contingent valuation analysis suggested All estimates are presented in millions of USD adjusted to present values using a discount rate of 4%. The benefit of protection remains the same regardless of sea-level scenario, that it would be worthwhile for Singapore to protect its marshes while the cost of protection changes according to different scenarios. WTP is willingness and mangroves. The results of the travel cost and contingent to pay. valuation analyses for marshes and mangroves were thus in conflict with each other. This is an important methodological issue to resolve in future studies. further migration beyond 2070 is not economically justifiable. Vulnerable coastal nations need to begin analyzing their With the remaining milder scenarios, the contingent valuation adaptation options against sea-level rise. They must weigh the analysis suggests that Singapore should fully protect both benefits of their nonmarket coastal resources against the cost of marshes and mangroves. protection. Although it may be too early for countries to institute adaptation measures, it is not too soon to begin DISCUSSION AND CONCLUSION planning. Adaptive strategies require time to implement. In general, knowledge regarding sea-level rise among the public Countries with important coastal resources need to start planning the strategies that they will adopt to reduce the in Singapore was found to be low. Further, there was a strong impacts of sea-level rise. belief that producers and the government should bear the responsibility of paying for any form of environmental References and Notes protection. Almost 90% of the respondents believed that 1. Church, J.A., Gregory, J.M., Huybrechts, P., Kuhn, M., Lambeck, K., Nhuan, M.T., polluters should pay, and only 24% agreed that they, as Qin, D., Woodworth, P.L., et al. 2001. Changes in sea level. In: Climate Change 2001 consumers, should pay for protection measures for natural The Scientific Basis. Hughton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J. and Xiaosu, D. (eds). Third Assessment Report of the Intergovernmental Panel on resources. About 44% of the respondents were skeptical Climate Change. Cambridge University Press, Cambridge, UK, pp. 639–694. regarding the ultimate use of their payment, because they felt 2. Jones, R.N., Pittock, A.B. and Whetton, P.H. 2000. The potential impacts of climate change. In: Climate Change in the South Pacific: Impacts and Responses in Australia, New that natural resources would not be protected even if they paid. Zealand, and Small Island States. Gillespie, A. and Burns, W.C. (eds). Kluwer Academic Publishers, Dordrecht, p. 385. Although these negative attitudinal responses led us to drop 20 3. Ahmad, Q.K., Anisimov, O., Arnell, N., Brown, S., Burton, I., Campos, M., Canziani, protestors, most of the population was willing to pay for O., Carter, T., et al. 2001. Summary for policymakers. In: Climate Change 2001 Impacts, Adaptation, and Vulnerability. McCarthy, J.J., Canziani, O.F., Leary, N.A., Dokken, protecting beaches, marshes, and mangroves. D.J. and White, K.S. (eds). Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, pp. 1–18. This study explored two measures of the benefits of 4. Neumann, J.E. and Livesay, N.D. 2001. Coastal structures: dynamic economic protecting natural sites from damage caused by sea-level rise. modeling. In: Global Warming and the American Economy—New Horizons in Environmental. Mendelsohn, R. (ed). Edward Elgar Publishing Limited, Chelton, UK, The travel cost measure relied on people’s behavior to infer pp. 132–148. values by examining where they choose to visit. The contingent 5. White, K.S., Ahmad, Q.K., Anisimov, O., Arnell, N., Brown, S., Campos, M., Carter, T., Liu, C., et al. 2001. Technical summary. In: Climate Change 2001 Impacts, valuation method directly asked respondents for their values. Adaptation, and Vulnerability. McCarthy, J.J., Canziani, O.F., Leary, N.A., Dokken, D.J. and White, K.S. (eds). Third Assessment Report of the Intergovernmental Panel on The two methods provided somewhat consistent measures of Climate Change. Cambridge University Press, Cambridge, UK, pp. 19–74. values for beaches. The travel cost method differentiated 6. Douglas, B.C. 2001. An introduction to sea level. In: Sea Level Rise: History and Consequences. Douglas, B.C., Kearney, M.S. and Leatherman, S.P. (eds). Academic between some popular beaches that were highly valued and Press, San Diego, pp. 1–10. some more remote beaches that were not. The contingent 7. Kaluwin, C. 2001. Adaptation policies: addressing climate change impacts in the Pacific region. In: Sea Level Changes and Their Effects. Noye, B.J. and Grzechnik, M.P. (eds). valuation questions led to only one value of beaches. Overall, World Scientific Publishing Co. Pte. Ltd, pp. 273–291. 8. Warrick, R.A. and Rahman, A.A. 1992. Future sea level rise: environmental and socio- the average value from the CVM was not that different from the political considerations. In: Confronting Climate Change Risks, Implications and TCM. However, the contingent valuation method suggested Response. Mintzer, I.M. (ed.). Stockholm Environment Institute, Cambridge University Press, Cambridge, pp. 97–112. that mangroves and marshes were far more valuable than the 9. Ministry of the Environment. 2000. Singapore’s Initial National Communication: Under the United Nations Framework Convention on Climate Change. Ministry of the travel cost method implied. The CVM placed almost the same Environment, Singapore, p. 34. value on all three coastal resources. The travel cost method 10. Ng, W. and Mendelsohn, R. 2005. The impact of sea level rise on Singapore. Environ. Dev. Econ. 10, 201–215. suggested that the marshes and mangroves had very little use 11. Tietenberg, T. 2000. Environmental and Natural Resource Economics (6th ed). Addison- value. People did not enjoy visiting the sites. However, the Wesley, Tietenberg. Boston, p. 646. 12. McLean, R.F., Tsyban, A., Burkett, V., Codignotto, J.O., Forbes, D.L., Mimura, N., CVM revealed that the sites nonetheless may have had Beamish, R.J. and Ittekkot, V. 2001. Coastal zones and marine ecosystems. In: Climate

Ambio Vol. 35, No. 6, September 2006 Ó Royal Swedish Academy of Sciences 2006 295 http://www.ambio.kva.se Change 2001 Impacts, Adaptation, and Vulnerability. McCarthy, J.J., Canziani, O.F., 28. Quah, E. and Tan, K.C. 1999. Pricing a scenic view: the case of Singapore’s East Coast Leary, N.A., Dokken, D.J. and White, K.S. (eds). Third Assessment Report of the Park. Impact Assess. Proj. Appr. 17, 295–303. Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, 29. Spash, C.L. 2000. Multiple value expression in contingent valuation: Economic and UK, pp. 343–379. ethics. Environ. Sci. Technol. 34, 1433–1438. 13. King, S.E. and Lester, J.M. 1995. The value of salt-marsh as a sea defense. Mar. Pollut. 30. Swallow, S.K., Opaluch, J.J. and Weaver, T.F. 2001. Strength of preference indicators Bull. 30, 180–189. and an ordered-response model for ordinarily dichotomous, discrete choice data. J. 14. Solomon, S.M. and Forbes, D.L. 1999. Coastal hazards and associated management Environ. Econ. Manage. 41, 70–93. issues on South Pacific islands. Ocean Coast. Manage. 42, 523–554. 31. Dillman, D. 1999. Mail and Internet Surveys: The Tailored Design Method. Wiley and 15. Sorensen, R.M., Weisman, R.N. and Lennon, G.P. 1984. Control of erosion, Sons, New York, p. 79. inundation, and salinity intrusion caused by sea level rise. In: Greenhouse Effect and 32. Dziegielewska, D.A. and Mendelsohn, R. 2003. Does ‘‘no’’ mean ‘‘no’’? In: Essays on Sea Level Rise A Challenge for This Generation. Barth, M.C. and Titus, J.G. (eds). Van Contingent Valuation and Air Quality Improvements in Poland. PhD Dissertation, Yale Nostrand Reinhold Company, pp. 179–214. University, New Haven, CT, p. 83. 16. Wong, P.P. 1985. Artificial coastlines: the example of Singapore. Z. Geomorphol. 57, 33. Watson, R.T., Zinyowera, M.C. and Moss, R.H. 1996. Climate Change 1995. Impacts, 175–219. Adaptations and Mitigation of Climate Change: Scientific-Technical Analyses. Cambridge 17. Reed, D.J. 1995. The response of coastal marshes to sea level rise: survival or University Press, Cambridge, p. 879. submergence? Earth Surf. Proc. Land. 20, 39–48. 34. Yohe, G., Neumann, J. and Marshall, P. 1999. The economic damage induced by sea 18. Ellison, J.C. 2000. How South Pacific mangroves may respond to predicted climate level rise in the United States. In: The Impact of Climate Change on the United States change and sea-level rise. In: Climate Change in the South Pacific Impacts and Responses Economy. Mendelsohn, R. and Neumann, J. (eds). Cambridge University Press, in Australia, New Zealand, and Small Island States. Gillespie, A. and Burns, W.C. (eds). Cambridge, UK, pp. 178–208. Kluwer Academic Publishers, Dordrecht, pp. 289–301. 35. Carter, T.R., Rovere, E.L., Jones, R.N., Leemans, R., Mearns, L.O., Nakicenovic, N., 19. Nurse, L.A., Suarez, A.G., Wong, P.P., Briguglio, L., Ragoonaden, S., Githeko, A., Pittock, A.B., Semenov, S.M., et al. 2001. Developing and applying scenarios. In: Gregory, J., Ittekkot, V., et al. 2001. Small island states. In: Climate Change 2001 Climate Change 2001: Impacts, Adaptation, and Vulnerability. McCarthy, J.J., Canziani, Impacts, Adaptation, and Vulnerability. McCarthy, J.J., Canziani, O.F., Leary, N.A., O.F., Leary, N.A., Dokken, D.J. and White, K.S. (eds). Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, Dokken, D.J. and White, K.S. (eds). Third Assessment Report of the Intergovernmental UK, pp. 145–190. Panel on Climate Change. Cambridge University Press, Cambridge, UK, pp. 843–876. 36. Amadore, L., Bolhofer, W.C. and Cruz, R.V. 1996. Climate change vulnerability and 20. Wong, P.P. 1992. Impact of a sea level rise on the coasts of Singapore: preliminary adaptation in Asia and the Pacific: workshop summary. Water, Air, Soil Poll. 92, 1–12. observations. J. Southe. Asian Earth 7, 65–70. 37. First submitted 1 August 2005. Accepted for publication 25 January 2006. 21. Park, R.A., Armentano, T.V. and Cloonan, C.L. 1986. Predicting the effect of sea level rise on coastal wetlands. In: Effects of Changes in Stratospheric Ozone and Global Climate. Titus, J.G. (ed). United Nations Environment Programme and U.S. Environmental Protection Agency, Washington, DC, pp. 129–152. 22. Armentano, T.V., Park, R.A. and Cloonan, C.L. 1988. Impacts on coastal wetlands throughout the United States. In: Greenhouse Effect, Sea Level Rise, and Coastal Wetlands. Titus, J.G. (ed). Environmental Protection Agency, Washington, DC, pp. 87– Wei-Shiuen Ng is a research analyst at the World Resources 126. 23. Smith, J.B. and Tirpak, D.A. 1988. Sea level rise. In: The Potential Effects of Global Institute. Her address: 10 G Street NE Suite 8000, Washing- Climate Change on the United States Draft Report to Congress 2. Smith, J.B. and Tirpak, ton, DC 20008, USA. Fax: 1-202-729-7775; Telephone D.A. (eds). National Studies U.S. Environmental Protection Agency Office of Policy, Number: 1-202-729-7722. Planning and Evaluation, Office of Research and Development, Washington, DC, (9) pp. 1–47. [email protected]. 24. Titus, J.G. 1986. Greenhouse effect, sea level rise, and coastal zone management. Coast. Manage. 14, 147–171. Robert Mendelsohn is a professor of economics at Yale 25. Rijsberman, F. 1991. Potential costs of adapting to sea level rise in OECD countries. In: Responding to Climate Change Selected Economic Issues. OECD Environment University School of Forestry and Environmental Studies. His Committee, Organization for Economic Co-operation and Development (OECD), address: 230 Prospect Street, New Haven, CT 06511, USA. Paris, France. Fax: 1-203-387-0766; Telephone Number: 1-203-432-5128. 26. Freeman, A.M. 1993. The Measurement of Environmental and Resource Values. Resources for the Future, Washington, DC, p. 516. [email protected]. 27. Ajzan, I., et al. 2000. Effects of perceived fairness on willingness to pay. J. Appl. Soc. Psychol. 30, 2439–2450.

296 Ó Royal Swedish Academy of Sciences 2006 Ambio Vol. 35, No. 6, September 2006 http://www.ambio.kva.se