The most remarkable finding is that the inventory yielded relatively few cases: only 37 decisions. It is difficult to quantify the total number of road and railway projects planned or implemented since 1999; however, it can be noted that the Administration allocates more than 4 billion euros each year to the building, operation and maintenance of infrastructure. The earliest case found was from 2004; since 2008, the average number of cases per year has been slightly below seven. Of all 37 cases identified, 12 related to railways, 22 to roads and 3 to both. Further, CABs accounted for 36 of the 37 decisions found while only a single one was made by a municipality.

By way of comparison, it can be noted that compensation requirements seem to be more common in Spain. Of all road and railway projects approved by Spanish national and regional authorities in 2006 and 2007, 40% (85 out of 214) included provisions regarding compensatory measures (Villarroya and Puig 2010). This percentage dropped to 22% (16 out of 72) for infrastructure projects approved between 2009 and 2011 (Villarroya 2012).

Table 1 shows the Swedish public authorities having made the decisions identified as well as the provisions of the Environmental Code referred to by them. Note that a decision (right-hand column) may refer to several provisions (the central, lighter columns). For some projects, a first decision was followed by a second, more specific one. In such cases, the second decision overrules the first one (meaning that this does not count as two decisions). By contrast, where a stretch of road was divided into several parts for which separate applications were submitted, this has been counted as one decision per application. The vast majority of decisions were associated with habitat protection (i.e. protection of small ponds and cairns), which the decision-making authority referred to in 28 out of 37 cases (76%), and with encroachment on nature reserves and Natura 2000 areas. Of all 37 decisions, 34 (92%) involved one or both of these categories. No decision related to compensation for damage to the everyday landscape (i.e. non-protected nature). It also turned out that most CABs had not taken a single decision ordering compensation measures. Of the 21 CABs, 13 had never ordered compensation and 6 had done so in a single case only, meaning that 90% of the CABs had adopted either no such decision or only one. The CAB of Västra Götaland accounted for 24 of the 37 decisions (65%) on its own.

Of the 37 decisions, 32 lay down specific requirements to be met by the compensation measures while 2 of them state that the measures are to be designed later and 3 of them provide for compensation in the form of an amount of money being allocated to a fund. Further, the inventory showed that the compensation measures tended to resemble the damage done and that the prevailing view is that the measures should be on a par with the intervention, i.e. that the compensation ratio should be 1:1. A typical example is a habitat-protection case in eastern Sweden (Lilla Edet on route E45) where an encroachment entailing the loss of 152 m of stone walls and 176 m of open ditches was compensated for through the addition of 150 m of new stone walls and 150 m² of new wetland (Länsstyrelsen Västra Götalands län 2011).

Two cases were studied in greater detail: (1) Röbäck/Röbäcksdalen, a project to build a new bypass on route E12 west of Umeå in relation to which the CAB ordered compensation measures; and (2) Järfälla, a project involving the expansion from two to four tracks of the railway stretch between Barkarby and Kallhäll in conjunction with the Mälarbanan line between Stockholm and Örebro, where the compensation order was issued by the municipality. The purpose of these two case studies was not only to illustrate the practicalities of environmental-compensation cases, but also to highlight a few concrete examples of implementation problems and opportunities for improvement from a planning perspective.

The Röbäck case is a rather traditional habitat-protection case, but a large and complex one. Its outcome was that an encroachment on a Natura 2000 area was compensated for through the establishment of a new nature area elsewhere. It was not possible to do this within the actual Natura 2000 area because of difficulties relating to land ownership, as revealed by the interviews – a large number of private land owners would have been involved, the situation regarding financial compensation and management issues was unclear, and the Swedish Transport Administration lacks the power to expropriate land for compensation measures. Therefore it was decided to use land owned by the municipality instead. The Järfälla case, by contrast, represents a broader problem complex because the expansion of the railway will entail both an encroachment on a nature reserve and impacts on the natural and cultural environment as well as on opportunities for outdoor and recreational activities. The Swedish Transport Administration was therefore ordered to compensate for the encroachment and impacts involved, for example by moving cultural remains and by building new bridges for better access to the recreation area by foot or bicycle as well as eco-passages. The compensation process in the Järfälla case has proceeded smoothly and the various interested parties have co-operated well. Measures have begun to be planned and implemented, and issues of maintenance are being discussed.

One general conclusion drawn from the case studies was that both projects were highly dependent on the skill of the parties involved and on their will to carry through the compensation measures in the best possible way. Rundcrantz (2007) draws similar conclusions from her case studies. However, an environmental expert from the Transport Administration noted that it was difficult even for those involved to grasp the whole process, and that the actual outcome was finally determined by a ‘tug-of-war’ between different actors as well as by legal and practical issues (Persson et al. 2014). Both projects seem likely to attain good final results, but it is very difficult for external reviewers to form a clear picture of the entire planning process based on the official documents.

A further conclusion specific to the Röbäck case concerns the fact that, in recent years, the region around Umeå – especially the many bird-protection areas to be found there – has seen the implementation of a large number of major road and railway projects giving rise to orders for environmental compensation. However, in the Röbäck case it was difficult to gain access to the various formal documents that circulated in relation to each individual part of the road project, not to mention experiences made by the various involved parties that had not actually been written down anywhere. One might think that by now some form of regional co-ordination of the environmental issues concerned would have evolved. This should have been able to bring about not only a more coherent approach both to the assessments and to the interaction between them, but also a package of environmental measures yielding maximum nature-protection benefit. That would have greatly enhanced both efficiency and transparency.

As regards the Järfälla case, it turned out that sub-projects that should in theory be capable of co-ordination, such as the EIA for the work schedule and the compensation case, were in fact carried out separately and in parallel, with no systematic cross-referencing. Indeed, sometimes the same individuals were involved in preparing parallel information for different documents. As a result, those documents may be very similar in part while there are also clear differences between them; in the absence of an explanation, this may be confusing to external reviewers (indeed, it was not clear to us until we had the opportunity to ask some follow-up questions in a second interview). In all likelihood, each individual sub-project has its own planning, budget, staff, etc., meaning that the work is governed by a project-based logic rather than being part of a coherent process where the site and the problems are in focus. From the perspectives of transparency and efficiency, it should be possible to achieve improvements through a clearer formal description of how compensation cases are to be incorporated in the overall road- or railway-planning process, as previously discussed by e.g., Rundcrantz (2007). It might also be a good investment to ensure that there is a regional project co-ordinator who knows the region well and is aware of all ongoing road and railway projects, in order to avoid duplication of work, further enhance transparency and increase efficiency.

Before discussing the policy implications of environmental compensation, it might be a good idea first to ask why anyone should undertake compensatory efforts in the first place. A list of arguments for and against compensation can be found in Persson (2011). These arguments are important to keep in mind when advocating the use of voluntary compensation. The principal arguments in favour are the following:

• There is an ethical responsibility to preserve common (i.e. public, in the sense that they are enjoyed by many people) goods such as nature. This is in line with the idea of ecosystem services: nature provides humans with benefits, and destroying the capacity of nature to do so will decrease the quality of life of humans (Cuperus et al. 2001, BBOP 2009).

• People should make amends for their actions. This argument rests on the polluter-pays principle (European Commission 2001, Beder 2006).

• The overall stock of environmental assets should not be reduced. This is also the rationale behind the principle of the mitigation hierarchy. This hierarchy consists of four (sometimes counted as three) consecutive steps: first, avoid causing impacts; second, minimise any impacts that cannot be avoided; third, restore any impacts that cannot be minimised; and fourth, compensate for or offset any remaining damage that could not be avoided, minimised or restored (see, e.g., BBOP 2009). Skipping any one of these steps may undermine environmental sustainability.

To this should be added that there is reason to believe that a smart use of environmental compensation will shorten planning processes and reduce conflicts (Persson 2006, Samferdseldepartementet 2013). Not unexpectedly, surveys on attitudes have shown there to be strong support for environmental compensation, even though they also identified some concern that the system may be abused by authorities and developers (Rundcrantz 2005, Persson 2013). In other words, there are both factual arguments and popular support in favour of the claim that the Swedish Transport Administration and corresponding authorities in other countries should pursue the issue of environmental compensation.

One way of increasing the use of environmental compensation is to improve the relevant legislation. However, considering the arguments above, it may well be that an even more important action to take in order to increase its use is to draw attention to all of its advantages so as to promote voluntary compensation.

The inventory showed that compensation had mainly been undertaken by reference to habitat and species protection, and also that no compensation order had been issued in relation to encroachment on the everyday landscape, i.e. unprotected areas, whose importance has been increasingly stressed by numerous authors (e.g., Rundcrantz 2007, ICF GHK 2013). This focus on biodiversity and on specific objects (such as wetlands and cairns) is a result of the approach presently taken in the Swedish Environmental Code, which focuses, as mentioned above, on habitat and species protection but ignores social, health and cultural aspects. This means that a number of environmental assets and ecosystem services are not taken into account. Examples of neglected issues include:

• the impact on cultural environments or characteristic features of the landscape;

• the risk of ill-health (caused, for example, by noise and air pollution);

• opportunities for recreation;

• encroachments that affect farming, forestry, fisheries or other land-dependent economic activities, such as by reducing the amount of arable land.

The legislators adding provisions on environmental compensation to the Environmental Code may not have included these issues, but they still underpin not only the concept of sustainable development but also initiatives such as the European Landscape Convention. It should also be noted that an emphasis on habitat and species protection similar to the one in Swedish legislation exists in countries such as Norway and Germany, whereas aesthetical values are given more weight in the corresponding British legislation (Thompson et al. 1997, Samferdseldepartementet 2013). Hence, even though there is a general focus on biodiversity, it is up to each country to decide what will be compensated for. When it comes to voluntary compensation, obviously, there is even greater freedom of choice.

The inventory showed that the level of compensation ordered was on a par with the extent of the intervention – in practice, the compensation ratio was thus 1:1. There is a Swedish legal precedent in the form of a decision by the Land and Environment Court of Appeal (Miljööverdomstolen 2002) from which it follows that authorities granting permits may not demand too much: the amount of compensation must be reasonable for both sides, i.e. both for the developer and for the public interest. What, then, is ‘reasonable’ in this context? Below is a list of good reasons for compensating by more than 1:1.

• Time. The economic way of looking at the allocation of benefits and costs is based to a large extent on discounting. This means that a compensation measure is worth more today than in ten or twenty years – not because, say, an actual wetland today could not have the same functions as an identical future wetland, but because the services provided by the wetland can be used immediately (Cole 2010).

• Uncertainty. Whenever a compensation measure is taken, there is some uncertainty as regards how the relevant functions will develop and work in the future (Maron et al. 2012). Even if the best available knowledge is used, there are many potential problems, ranging from the specification, implementation and maintenance of the work to the possible failure of nature to abide by the scientific models believed to be valid. Such shortcomings in models may relate, for example, to how a certain species goes about establishing itself or to how a water system reduces the amount of nutrients.

• Quality. Many US states have guidelines for how to weight different technical solutions when comparing them. For example, restoration is seen as more valuable than the creation of new environments, and the creation of new environments is seen as more valuable than the granting of protection for an area (National Research Council 2001).

• Actors involved. When the reasonableness of measures is discussed from the perspective of the different actors involved, issues specially focused on are whether the assessment of reasonableness differs between actors, how they perceive situations differently and how specific mitigation measures would work in each of their contexts (e.g. in out-of-kind and off-site compensation). This is especially important when it comes to compensating for social aspects (Persson 2006).

When ecological criteria are applied, high compensation ratios may be needed to guarantee a fair exchange (Maron et al. 2012). It should be noted that such ratios may be perceived as ‘over-compensation’ by certain actors, since other criteria (such as hydrological models) will typically yield lower ratios. Also, high compensation ratios may be needed to compensate for damage to recreational and other social areas. Further, it should be kept in mind that environmental disputes typically involve several perspectives and interests besides the matter of the correctness of scientific models (Söderbaum 2000, Persson 2006). Ideally, a ratio that is high enough to satisfy all environmental aspects that need to be compensated for should be chosen.

It is clear from the inventory that Swedish decision-making authorities have consistently proposed compensation measures of the same type as the intervention and usually also located close to the intervention (i.e. ‘in-kind’ and ‘on-site’ compensation). This was also the case in Järfälla (however, an exception could be seen in the Röbäck case, where the mitigation measures were out-of-kind and located off-site). This is in line with a principle which is advocated in the guidelines of the Business and Biodiversity Offset Programme (BBOP 2009) and which has been very influential not only within the EU (Tucker et al. 2013) but also in non-EU countries such as Norway (Samferdseldepartementet 2013). By contrast, others have argued that it is important to take a flexible view and adjust the compensation measures to suit each case, because each case is unique – or, as Cuperus et al. (1999:46) put it, ‘[g]eneral standards and guidelines for choosing between “on-site/off-site” and “in-kind/out-of-kind” compensation cannot be given, as these depend on the availability of suitable compensation sites and must therefore be determined on a case-by-case basis’. Paradoxically, alongside the BBOP’s focus on on-site compensation there is a worldwide trend for compensation to be provided off-site because of problems finding places for the environmental measures. This trend is likely to continue in Europe and elsewhere, given that mechanisms such as ‘habitat banking’ are also promoted by the European Commission (Conway et al. 2013) and national governments such as the ones in Germany (Wende et al. 2005) and the United States (Race and Fonseca 1996, Federal Register 2000, Stein et al. 2000, U.S. Army Corps of Engineers 2010).

The core of this issue can be summed up as follows: on-site and in-kind compensation is a way to make a contribution close to the site of the damage in order to safeguard local assets (such as ecosystem services), while off-site and out-of-kind compensation may be a necessary option for the promotion of assets of a more general type deemed to be more important or in cases where it is not possible to find an on-site and in-kind option, i.e. in the context of a strategy governed by objectives (Villarroya et al. 2014). The use of the kind of overall landscape perspective characteristic of the latter approach leads to greater flexibility and makes it easier to find locations for compensation measures. In addition, landscape-scale planning allows a more effective implementation of the mitigation hierarchy and prevents overlooking cumulative impacts (Kiesecker et al. 2010). It is true that objections have been made to off-site and out-of-kind compensation based on the hypothesis that it may be easier to gain the acceptance of the general public for on-site and in-kind compensation (ten Kate et al. 2004). However, no studies confirming this have been found and a study of preferences as regards environmental compensation actually showed the opposite hypothesis to be closer to the truth (Persson 2013).