Valuing the Multiple Benefits of Blue-Green Infrastructure for a Swedish Case Study: Contrasting the Economic Assessment Tools B£ST and TEEB

Kronan is located some 3 km northeast of the city center of Luleå in northern Sweden and will comprise several distinct residential areas, of which Kronandalen is one. Kronandalen is situated close to green areas (mainly forest) where the main recreational area of the city and two large lakes are situated. Direct access to these areas from Kronandalen is provided by bicycle and foot paths. Kronandalen was once a military area used by the Swedish armed forces until 1992. It was then taken over by the municipality and used for local activities (including the municipal recycling station, a film studio, a youth hostel, and several offices). These occupied the former army barracks and buildings. The development plans aim to make Kronandalen a dense urban area (2,000 apartments on ca. 25 ha) but with a number of parks and green spaces.

An assessment was made of the value of the current situation (baseline) as well as the value of three different options for the developed urban area (Table 1). The existing stormwater infrastructure comprises a rudimentary sewer network for drainage of parts of the impervious areas. No detailed design documentation exists, and only incomplete drawings were available. The first development option, Option 1, considered includes a separate sewer network based on current Swedish design guidelines, for design rainfall with a 20-year return interval. All of the proposed development options have floor levels of the buildings set above street level for flood protection. For Option 2, the municipality proposal, stormwater management relies on swales and a central pond, combined with new traditional sewer networks in the residential districts where no space for swales was considered to be available. In Option 3, in addition to Option 2, additional green roofs and bioretention facilities were considered for residential streets (Table 1). The USEPA’s Storm Water Management Model (SWMM) was used to assess the hydrologic and hydraulic performance of each option.

The former military area baseline represents a land use that is specific to Kronandalen. An alternative baseline (pristine forest) was considered (Table 1) as a benchmark. This alternative baseline represents a typical situation in Sweden when it comes to the development of new urban areas where forest regularly is present prior to construction.

Both baselines and the three options are summarized in Table 1.

The assessment of the two baselines over a period of 30 years showed considerable financial benefits (Table 2). Based on information from the municipal water department, maintenance and operational costs of the existing sewer system (present in the former military baseline) over this time were estimated to be around €80,000 (present value).

The predominant benefits under the baseline were assessed from carbon sequestration using B£ST and from capturing airborne particulate matter from TEEB. TEEB treats the latter as a health benefit due to improved air quality. The main benefits derive from the services provided by the large number of existing trees and green areas. In addition to carbon sequestration, B£ST also provided an evaluation of other benefits (Table 2), whereas TEEB valued only the health benefits in the case of the former military site baseline. With both tools, recreation was taken into account, but only in the forest baseline.

The overall results (Fig. 1) showed that, according to both tools, the forest baseline offered greater benefits and a higher NPV than the former military site baseline. This is due to the greater number of trees and the natural characteristics of the area (e.g., green pervious surfaces, existing habitats). Even though B£ST showed benefits from more benefit categories than TEEB, the TEEB results showed a higher NPV. This may be because TEEB does not use confidence scores or sensitivity testing, which in B£ST can lead to lower benefit values by explicitly allowing for uncertainties (Ashley et al. 2018c). Users of TEEB are expected to already have modified any input data to include uncertainties.

The two baselines were compared with the three options for the new development scheme at Kronandalen. Similar benefit categories were assessed for all options and baselines. However, recreation was taken into account only in comparison with the forest baseline (see subsequent discussion). In this section the results for the different options are presented in comparison with the former military site baseline. Similarities and differences resulting from comparison with the forest baseline are then presented and discussed.

B£ST results show a rather high NPV over the 30-year timescale for all options [Fig. 2(a)], with increasing benefits across the options. The construction, maintenance, and operational costs of the sewer system, as well as the BGI measures, were taken into account. As hypothesized, implementation of BGI results in a higher NPV for the benefits in comparison with the traditional (piped) approach. Although the extended BGI (Option 3) gave the greatest benefits and NPV, the proposed plan (Option 2) has the best benefit-to-cost ratio since the additional costs of the extended BGI implementation exceeded the benefits (Table 4).

TEEB valuation [Fig. 2(b)] shows a higher NPV than that determined using B£ST. For TEEB, there are only slight differences in value between the three options. In contrast with the B£ST results, in TEEB the proposed plan (option 2) has the highest NPV (Table 4).

B£ST shows that the main benefits relate to health and amenities, and the main disbenefit relates to carbon reduction and sequestration (Table 3 and Fig. 3). This is because the character of the site changes from being relatively green to becoming an urban area. The increase in the numbers of properties and inhabitants leads to major benefits (since the green areas are used by more people), while the reduction in the number of trees brings a disbenefit (less carbon sequestration). Across the options there is a considerable increase in the value to human health, resulting in an increase in overall NPV.

In TEEB the major benefit was found to be related to the home values (ca. 85 % of the total benefits) (Table 3 and Fig. 4). The BGI for Options 2 and 3 only affected the benefits marginally. Social cohesion (ca. 14 % of the total benefits) was found to be the next greatest benefit using TEEB. Both of these categories contributed to approximately the same NPV for both BGI Options 2 and 3. As with B£ST, these benefits occur due to development of the urban area. The categories in which the implementation of BGI gave the main relative benefit (i.e., where BGI generated the most relative additional value in comparison with the traditional drainage option) were health and energy. However, compared with the total NPV, this was relatively minor. The health category in TEEB showed disbenefits for the traditional drainage option, but the implementation of BGI (proposed plan and extended BGI) resulted in an increase in benefit value. This is due to the subcategory improved air quality, which, compared with the baselines, showed disbenefits. The implementation of BGI resulted in greater benefits in the health subcategory a greener environment and, thus, compensated for the disbenefits, so that the overall health benefits were positive. The energy category showed an increase in value across the options, but this was marginal ($<1\%$) to the total benefits. In summary, the higher costs of BGI implementation and maintenance resulted in a positive NPV for all three options, although the benefit values increased when increasing BGI implementation.

Comparing the three options with the forest baseline rather than the former military (existing) site baseline resulted in lower NPVs because the value of the forest baseline was greater [Figs. 5(a and b)]. Because an entirely new sewer system had to be installed, the associated higher costs led to lower benefit-to-cost ratios (Table 4). The estimation of the benefits and costs showed that, although the benefits could be maximized when choosing the extended BGI option, more economically beneficial results could be obtained with the other options [proposed municipal plan for B£ST (Option 2) and traditional drainage option (Option 1) for TEEB].

In comparison to the forest baseline, the three options showed an especially high negative impact (disbenefits) in reduced carbon sequestration when using B£ST and reductions to improved air quality (included in health) in TEEB (Supplementary Table S4) caused by the removal of a large number of trees. In addition, recreation (which was assessed as a category only in comparison with the forest baseline) also adds negative impacts, although being nearly negligible ($<1\%$) compared with the overall NPVs. Otherwise, the general distribution of the benefits is similar to the results for the former military baseline, and there are similar differences between the benefits found using the two tools (Supplementary Figs. S1 and S2).

Both B£ST and TEEB follow a similar approach in which first an impact created by using BGI is determined, and then this impact is monetized using standard values. B£ST offers more categories for assessment than TEEB, and the calculations and monetary values used, mostly based respectively on UK data [see see values library in the B£ST in CIRIA (2019)] or Dutch (Buck Consultants International 2016) studies, differ.

Some categories or subcategories assess similar benefit values in both B£ST and TEEB but do not lead to similar overall results (Table 5). This is because of the different monetary values and specific calculations and, thus, differences in the required data input in B£ST and TEEB (CIRIA 2016, 2019; TEEB 2010; Buck Consultants International 2016).

An important (advantageous) aspect of B£ST are the screening questions. These support the user in making decisions about relevant, and therefore subsequently monetized, impacts and benefits prior to expending resources inefficiently on obtaining data for categories that have only marginal overall benefits. The TEEB web tool does not include this, so there is a danger of blind input, so to speak, leading to a risk that the user simply fills in the requested data input fields without first carrying out a review of what are likely to be the most significant potential impacts.

Another important aspect are the confidence scores applied in B£ST, which can have a large impact on the monetized results. A high level of uncertainty was expected for the data input in the Kronandalen case study because this was for a new development scheme for a whole area, including housing and urban infrastructures as well as water and stormwater management measures. The precise details of the BGI and associated plans were not available at the time, so suppositions were required as to what the details of the options entailed. A further limitation was that the database of monetary values provided in B£ST and TEEB was used for the estimations (instead of specifically adapted Swedish data). The original intention was to use national or regional monetary values. However, although at least some Swedish data would have been available on benefits like those used in the tools, the format was found to differ substantially from the Dutch and UK data, which are used for the tools, and for some categories no suitable data were available at all. Thus, the goal of using local information was not fulfilled and would have necessitated redevelopment of (parts of) the tools. This is the main limitation of using the tools in the Swedish context.

The risks from the double counting of benefits are highlighted for B£ST users. The technical guidance for the tool (CIRIA 2016) offers a lot of information and decision support providing recommendations as to how best to avoid double counting. In TEEB, double counting is also a risk and is not highlighted in the way in which it is in B£ST. During the valuation process in this study, in TEEB a high risk of double counting in the value of homes subcategory was detected. It includes six subquestions that either value a more general change of the area (e.g., improvement in parks) or a more specific feature (e.g., construction of ponds, improvement to green spaces). If the overall improvement of the area already takes such features into account, there is no extra benefit expected for the same beneficiaries and the more specific subquestions are not assessed. Therefore, only two out of six subquestions were included in this study. There is always a subjective aspect to evaluation (Spangenberg and Settele 2010). In this case, for instance, the decision about the (most) relevant benefit or impact was initially made subjectively, and an important question arose when applying the tools: how many of the features available in B£ST (e.g., preliminary screening questions, double counting) should be included when applying the TEEB tool? For example, owing to the initial screening of the relevant categories, air quality was not assessed in B£ST because for that specific case no significant impact from the development, under any of the three options, was expected according to the environmental impact assessment of the area made by Luleå Municipality (2016). In contrast, TEEB always includes (since no initial screening is intended) the similar subcategory improved air quality in the health benefits. It is therefore important for the user to decide what kind of impact or benefit is actually achieved by the chosen BGI scheme. It could be argued that for this study, either the air quality category in B£ST should have been valued or the air quality subcategory in TEEB should not have been valued. According to the B£ST screening questions and impact pathway, the BGI to be implemented (e.g., trees, swales) would contribute to particulate filtering; however, no significant impact on air quality was expected from these measures in the context of the application. TEEB does not include such preassessments and requires only a quantitative input of the measure implemented; it calculates the benefits without including the initial conditions or context. For this study, both tools were applied using their respective approaches and implemented features (i.e., in TEEB there was no possibility of discarding assessed air quality, in contrast to B£ST). This emphasizes the value of having the screening questions in B£ST support initial considerations and deliberations as to what benefits need to be looked at in detail. In contrast, TEEB tends to support simple and unquestioning data input and so may yield unrealistic results.

Despite similarities in the corresponding categories (Table 5), the results obtained with B£ST and TEEB differ for the previously given reasons. Similar findings were presented by Gomez (2016), who assessed the multiple benefits of BGI in a case study in Elwood, Australia.

Other studies have shown that the results obtained with B£ST and TEEB indicate broad similarities in the distribution of benefits (e.g., Ashley et al. 2018a). In contrast, the results from the Kronandalen case study do not show this. B£ST showed a number of negative impacts, whereas with TEEB only one negative impact, in the health category, was apparent for the traditional drainage option. Furthermore, the main benefits differ (health in contrast to home values in B£ST and TEEB, respectively) and similar categories (e.g., property price increase–home values, health–a greener environment) show significantly different values (Table 5).

The Kronandalen case study in general was different from other studies: the main objective of previous studies evaluating the multiple benefits of BGI was often the assessment of flood risk reduction managed by the implementation of BGI (e.g., Hoang et al. 2018; Lawson et al. 2015; O’Donnell et al. 2017) or the improvement of existing urban areas in terms of recreational sites and a greener environment] e.g., urban regeneration (Ashley et al. 2018a)]. In these studies, the main benefits would be expected to lie in the corresponding categories (e.g., flooding, amenities, recreation). In Kronandalen, the creation of housing space is the main objective, and this results in large benefits in the related categories (e.g., value of homes, amenity, health, social cohesion), while the otherwise targeted stormwater-related benefits are less highly valued. The more environmentally related categories (e.g., carbon sequestration, biodiversity, and ecology) show negative impacts due to the comparison with baselines that reflect a near natural (forest) or relatively green (former military site) area. There is a contrasting outcome when BGI is retrofitted in existing urban areas that have mostly traditional piped drainage systems (e.g., Ashley et al. 2018a; Gomez 2016).

The water management–related benefits expected from the implementation of BGI (e.g., reduced flood risk, water quality) are less important in Kronandalen because of the predominant conditions in the area (e.g., sewer system of sufficient capacity) and therefore result in only small beneficial impacts. Furthermore, the height levels of the ground floors of buildings in the proposed development (all options) have been chosen such that any risk of flood damage to houses/properties is minimized, and this reduces the value of BGI with respect to flood protection, which is significantly different from evaluating retrofitting schemes. Moreover, the results indicate that TEEB focuses more on urban benefits (e.g., property price increase and benefits received by the inhabitants such as health), whereas B£ST also addresses the importance of environmental impacts (e.g., carbon reduction and sequestration, biodiversity, and ecology).

The low additional benefits due to BGI on reduced flood risk in this case show that, in general, there are differences between an assessment of a new development scheme (Kronandalen case study) in a predominantly green field site and assessments of retrofitting schemes in existing urban areas (previous studies). In an existing area with a sewer system, retrofitting BGI may target specifically the reduction of flood risk, which can lead to corresponding benefits. In the Kronandalen case, flood risks are already reduced in Option 1 given the higher ground floor levels of all buildings. Thus, the drivers and main objectives of implementing BGI differ, leading to different outcomes. Further, in contrast to a retrofitting scheme where one knows detailed parameters (e.g., number of inhabitants, number and type of houses), for a new development, there are greater uncertainties in the reliability of the data used because of, at least in the early planning stages, uncertainties in the proposed development. For instance, estimation of the number of beneficiaries is crucial for assessment with B£ST and TEEB. This number is based on the presumed numbers of people living and working in an area and therefore is subject to significant uncertainties because of the need to predict the dweller and visitor populations and their behaviors. In contrast, in a retrofitting case, the number of inhabitants in the area is usually known and much more certain, although, again, future numbers can only be estimated. Numbers and extents of existing and newly planted trees and green areas are also known more reliably for an existing area than one being designed as part of a development scheme.

While most earlier B£ST and TEEB studies evaluated retrofitting of BGI in existing areas (e.g., Ashley et al. 2018c; Gomez 2016), comparing the three options with the forest baseline involving assessing the new development of a natural environment. This assessment using both B£ST and TEEB was found to be especially difficult owing to the contrast in land use between the forest baseline and all three options. Numerous estimated values for input data were necessary, and as a result there were high levels of associated uncertainties. In general, none of the tools seems to be as well designed for assessing new developments in natural environments as for retrofit schemes. Both tools require estimates of the numbers of inhabitants or local users in the existing area before new development can be analyzed. However, it was found possible to utilize the tools for the forest baseline case. This was easier using B£ST versus TEEB.

For both tools, the data required posed a challenge when it came to estimating the beneficiaries and values of the benefits assessed. In particular, the importance of local green infrastructure for recreation, biodiversity, and ecology is difficult to assess in Kronandalen since there are many natural, green surroundings in the immediate vicinity of the proposed development. Also the apparent benefits of the effects on health and amenities (or home values) may also be questioned. For example, accessibility to a park or a permanent water body is usually already good or available in Swedish residential areas similar to Kronandalen (relatively green development, medium-sized city, relatively low population density, close proximity to natural areas), so that local schemes with new BGI do not necessarily add extra benefits, or these benefits may be lower compared with those in more densely populated areas. Nevertheless, this case study revealed considerable benefits in these categories because of the increased numbers of inhabitants and, thus, potential beneficiaries. This shows that, although the approach used in B£ST and TEEB is generally applicable when assessing the wider benefits of BGI, the input data have to be critically reviewed and adapted to the local situation, as demonstrated by Hoang et al. (2018), who stress the need to consider benefit proximity and intensity, which are parameters that relate to the physical adjacency of BGI and beneficiaries.

The effects of implementing BGI in specific contexts can differ from the original UK or Netherlands context in which the valuation tools were developed. A similar BGI scheme, but in a different location or developed in a different context, may not lead to the same benefits as in a UK, Dutch, or northern Swedish context, especially in categories such as recreation, health, and amenities, which are dependent on the nature of the society and the general availability of urban and countryside green spaces. Local impact assessments and studies should therefore be taken into consideration when valuing benefits. Hoang et al. (2018) use the concepts of benefit proximity and intensity in assessing the relative value of the benefits of BGI: proximity to populations and already existing BGI features and the diminishment of benefits based on the distance of the BGI from potential users. This is partially accounted for in B£ST and TEEB in certain categories of benefit (importance of local urban green spaces), as the distance to green spaces is included in assessing health benefits (e.g., Grahn and Stigsdotter 2003). Hence, in this case, the impact being evaluated is considered valid in a Swedish context, so the tools are applicable.

The main focus of this study was a general comparison of the use of the B£ST and TEEB tools in a Swedish context. The calculation of the monetized benefits was not expected to be directly applicable because local monetary values (which would be necessary for estimation in the Swedish context) were not used. Carbon prices, for example, differ in different countries, but it was not part of the study’s aims to define equivalent Swedish values. The quantitative output of financial value should therefore not be considered as absolute but, rather, only indicative of the relative potential value of the options.

Owing to the nature (new development) of the Kronandalen area and the challenges arising (described earlier), B£ST was found to be more suitable for the case study and would therefore be a preferred tool for similar future evaluations. However, assessment with B£ST is complex, and multiple aspects need to be considered (e.g., screening questions, double counting, confidence scores). This is important for adequate assessment and valuation of the benefits, but it is also dependent on the data available. Some important data may not be available with an appropriate level of confidence, so not all possibilities and potentials can be assessed. Depending on the objective of the assessment, TEEB could therefore also be appropriate as an assessment tool. For instance, when a first, more general, assessment of the possible monetized benefits is needed, TEEB offers a fairly straightforward and simple-to-use valuation method. In this case, it is important to keep in mind that, before monetizing benefits, an initial qualitative assessment (e.g., following the B£ST screening questions) should be carried out to determine possible benefits and disbenefits in order to reduce unwarranted analytical effort. Another advantage of B£ST is that the tool allows for user-defined inputs of monetary values, facilitating adaptation to non-UK cases (assuming the values are available). In contrast, the structure of the TEEB tool cannot be changed, meaning the categories and calculations need to be copied, e.g., into a spreadsheet, to come up with new monetary values.

Appropriate monetary values would be necessary to use either tool further in a Swedish context. Nearly all the same values are needed to calculate benefits in B£ST and TEEB (e.g., carbon prices, insurance, and health costs). However, B£ST monetizes more categories, so more values must be defined.

As a recommendation for further case studies, the possible benefits of a BGI scheme should first be assessed qualitatively (e.g., with the B£ST screening questions). Next, an initial quantitative evaluation could be carried out with TEEB to determine the most relevant solutions for a development or retrofitting scheme. The final evaluation of the favored schemes should then be carried out with B£ST. Such a procedure would make it possible to exploit the overall advantages of both tools.

The results of this study have implications for decision makers. The decision maker or funder, who in general and in this specific case of Kronandalen is the municipal water authority, is likely not to be the main beneficiary of a BGI scheme (Ashley et al. 2018b). The categories generating the largest NPV (amenities, health, home values, social cohesion) are not directly related to the core duty of the water authority. The question arises as to who pays and who benefits—the main beneficiaries are not the funder (the water authority). This raises the further question of whether this constitutes a significant obstacle to BGI implementation; however, such a question is beyond the scope of this paper. Furthermore, at least in Sweden, (storm)water management is financed by a water fee, which is to be used only in connection with issues directly related to water management. A strict interpretation of this rule would raise doubts about whether in the Kronandalen case the water authority would even be allowed to invest in BGI, given that most stormwater-related benefits can be obtained with a more traditional sewer solution. Tools such as B£ST or TEEB can be helpful to water authorities in highlighting additional benefits in discussion with other decision makers and funders (e.g., other departments in a municipal administration).