Ecological Economics 68 (2009) 2549–2559

Contents lists available at ScienceDirect

Ecological Economics

j ourna l homepage: www.e lsev ie r.com/ locate /eco lecon

Analysis

Ecosystem services and regional development: An application to Sweden

Ing-Marie Gren a,⁎, Lina Isacs b,1

a Department of Economics, Swedish University of Agricultural Sciences, Box 7013, 750 07 Uppsala, Swedenb Kemikalieinspektionen, Box 2, 172 13 Sundbyberg, Sweden

⁎ Corresponding author. Tel.: +46 18 671753; fax: +4E-mail addresses: ing-marie.gren@kon.slu.se (I.-M. G

(L. Isacs).1 Tel.: +46 8 519 41 100.

0921-8009/$ – see front matter © 2009 Elsevier B.V. Adoi:10.1016/j.ecolecon.2009.03.023

a b s t r a c t

a r t i c l e i n f o

Article history:Received 16 June 2008Received in revised form 10 March 2009Accepted 24 March 2009Available online 6 June 2009

Keywords:Ecosystem servicesRegional green accountingGrowthSustainable developmentSweden

JEL classification:Q56Q57R11

This study carries out interregional comparisons of development and sustainable use of natural capital withand without inclusion of non-marketed ecosystem services. A simple dynamic model of an open economyshows that appropriate adjustment of conventional income accounts includes flow and stock components,but excludes explicit entrance of pollution. Calculations are made for Swedish regions and for two types ofecosystem services — pollution sequestration and recreational services — provided by three types ofecosystems: forests, agricultural landscape and wetlands. When comparing the adjusted and non-adjustedregional income accounts it is shown that the two measurements generate significantly different pictures ofregional income and growth; regions that are traditionally considered as relatively less growth promoting areshown to hold important sources of wealth from natural capital, while counties that are rich in conventionalaccounts fall behind when adjusting for values of changes in natural capital.

© 2009 Elsevier B.V. All rights reserved.

1. Introduction

The question on economic performance of different countrieswhen accounting for non-marketed environmental changes has beenraised in many studies (for reviews, see Heal and Kriström, 2005;Harris and Fraser, 2002; Asheim, 2003; Atkinson and Hamilton, 2007).This question has often been accompanied by a related inquiry onwhether or not economies are on a sustainable path. Inspired byWeitzman (1976), there has been a rapid development in thetheoretical literature on role of national income as an indicator ofwelfare changes and sustainable development (e.g. Weitzman, 1976;Asheim 1994; Aronsson et al., 1997; Brekke, 1997a,b; Dasgupta andMäler, 2000, 2001; Cairns, 2002). Empirical applications on adjust-ments of conventional income accounts with respect to environ-mental changes and/or measurements of sustainable developmenthave been carried out since late 1980s (Repetto et al., 1989; Pearce andAtkinson, 1993; Hamilton, 1994; Hamilton, 1996; Vincent 1997;Hamilton and Clemens, 1999; Hamilton and Dixon, 2003).

However, in spite of the relatively many empirical studies of greenincome accounts and sustainable development, there are only a few

6 18 673502.ren), lina.isacs@kemi.se

ll rights reserved.

applications on regions within a single country and these aremade fordeveloping countries (Vincent, 1997; Gundimeda et al., 2007). TheVincent (1997) study estimates marketed values of changes inmineraland timber resources for assessment of sustainable development indifferent regions in Malaysia. Gundimeda et al. (2007) include bothmarketed and non-marketed services of forest resources whenadjusting state accounts in India. The main purpose of this study isto present calculations of green regional income accounts for anindustrialised country, Sweden, where several non-marketed outputsfrom different types of natural capital are included. Another purpose isto use these accounts for making interregional income and growthcomparisons of the green and conventional regional income accounts.In addition, evaluations are made with respect to sustainable use ofnatural capital assets.

The green adjustment to conventional income carried out in thispaper differs from the practice used in most other empirical studies.A large number of studies apply the approach used by Repetto et al.(1989), which is among the first attempts to adjust conventionalgross domestic product (GDP). Environmental changes are thenmeasured as value of changes in natural resources, such as soilerosion, deforestation, and depletion of subsoil resources, and associal costs from pollution (e.g. Pearce and Atkinson, 1993; Hamilton,1994, 1996; Vincent, 1997; Hamilton and Clemens, 1999; Ahlroth,2000; Hamilton and Dixon, 2003). This paper uses an alternativeempirical approach for green national accounting where the value of

2550 I.-M. Gren, L. Isacs / Ecological Economics 68 (2009) 2549–2559

changes in natural capital is derived from its production of outputsor ecosystem services. By using a simple dynamic model for an openeconomy it is shown that this approach implies that pollution doesnot have an explicit entrance into green accounting, which is incontrast with several empirical studies where the cost of pollutionhas been deducted from conventional accounts. The empiricalapplication calculates the money metric of two ecosystem services— pollution sequestration and recreational values — for three typesof natural capital, or ecosystems; forest, wetlands, and agriculturalland. Calculations and comparison of conventional and adjustedregional accounts are carried out for four different regions inSweden.

The paper is organized as follows. First, we give a presentation ofthe simple dynamicmodel of a region, which provides the basis for thecalculations of green regional accounts, growth and sustainabledevelopment. Next, data retrieval is described for calculations of thevalue of change in ecosystems — forest, wetland, arable land — as themonetary value of associated changes in the supply of ecosystemservices. Section 5 presents interregional comparisons of adjusted andnon-adjusted regional income accounts, and section 6 investigateswhether the use of natural capital resources is sustainable in thedifferent regions during time. The paper ends with a brief summaryand a concluding section.

2. The model

Themain purpose of this section is twofold: i) to derive correctionsof conventional income measurements when accounting for values ofnon-marketed ecosystem services, and ii) to analyse the entrance ofpollution in green accounting. The second purpose arises from thecommon practice of negative entrance of monetary assessment ofpollution in green accounting. The main challenge in this section is tofind appropriate measurements of the changes in value of the non-marketed goods of natural capital. Since market prices of marketedgoods are most likely not to be optimal due to, among others,imperfect functioning of markets and lack of prices of negativeexternal effects, we apply the non-optimising modelling frameworksuggested by Brekke (1997a,b); Dasgupta and Mäler (2000, 2001);Arrow et al., (2004); Vouvaki and Xepapadeas (2008), and Dasgupta(2009).

However, most of these studies derive green accounting measure-ments for closed economies, which are of less relevance for regionswithin a country, which can be characterised as small open economieswith given input and output prices. As demonstrated in Brekke(1997a,b), this means that a region's production and consumptiondecisions can be separated since borrowing and lending can be madeat a given price of capital. In this paper, we focus on the value ofproduction because of two main reasons. One is that productionpossibilities put limits to consumption in the long run. The otherreason is based on a pragmatic concern due to the lack of data onregional consumption of commodities and services but availability ofofficial statistics on net value of production. We also define wealth interms of production values per capita since this a common unit ofcomparison in economic development among regions and also withina region during time.

For simplicity, all marketed goods and services produced in aregion j, where j=1,..,n regions, are suppressed in the compoundedgood Q j, and the non-marketed ecosystem services are represented bythe single compounded service E j. Both types of goods use naturalcapital, S j, as a production factor. The marketed good and ecosystemservice also need man-made capital, K j. The market good causespollutants, N j, as by-products, which are treated as inputs intoproduction and spread by wind and water streams into other regions.For example, the production of most goods use energy as inputs,which generates emissions of carbon dioxide, sulphur dioxide,nitrogen oxide and so forth. The separable production functions for

marketed and non-marketed goods are then written as Q j=Q(K j,S j,N j) and E j=E j(S j) respectively, which are assumed to be increasingand continuous in all their respective arguments. In this setting, jointproduction of marketed and non-marketed services by natural capitalis assumed.

The change in S j over time is determined by its intrinsic growth,β j, and deposition of pollutants, D j=aijN i, where aij=Nij/N i denotesthe share of deposition in region j of total emissions from region i. Inpractice, pollutant transports among regions are described by amatrix, such as the EMEP model, which presents transports ofnitrogen oxides, ammonium, sulphur dioxides and particulatematters among European regions (Tarrasón et al., 2007). Thedevelopment of K j over time is determined by gross investment, I j,minus depreciation, ρ jK j, where ρ j is the depreciation rate.Population in each region, P j, is assumed to grow at the constantrate θ j. For given initial stocks of K j and S j, the changes during timein these two assets are assigned simple functional forms. Thedynamics of the regions are then described by three equations ofmotion according to

:Sj= β jS j − α jD j ð1Þ

:Kj= I j − ρ jK j ð2Þ

:P

j= θ jP j ð3Þ

Sj0 = S j

; Kj0 = Kj

; P j0 = P j

where α j is the impact of deposition of pollution on the ecosystemfunctioning, and S0

j and K0j are the initial levels of natural and

reproducible capital assets respectively. In general, the impact of D j onthe growth in S j is negative as written in Eq. (1), but for somesubstances, such as nutrient, there is a positive impact on thefunctioning of the recipient ecosystem to a certain threshold level.Beyond this pollution level, bifurcation may occur where thecharacteristics of the ecosystem is changed by a shift into anothertype of ecosystem. One example is provided by the Laholm Bay at thewest of Sweden, which was heavily polluted by nitrogen during 1980s.The vegetation of large sea area bottoms vanished and species likecrayfish became extinct. Such changes usually imply non-continuousproduction functions and associated difficulties in assessing account-ing prices of ecosystems (see e.g. Mäler, 2000), which is disregardedin this paper.

In each period of time, net value of production per capita isdetermined by

π jt = ptQ

jt + u j

tEjt = P

jt ð4Þ

where pt is the constant net value per unit of marketed output, andutj is the constant net value of the non-marketed good. Sustainability

is now defined as non-declining net value of production per capita,W, during time. In order to simplify notations, super index denotingregion j is omitted in the sequel. Total wealth in time t is then, foreach region, determined by current and future streams of net valuesfrom production of both marketed and non-marketed goodsaccording to

Wt =Z∞

t=0

πte− rtdt ð5Þ

where r is the constant and given interest rate. The assumption of aconstant interest discount rate is likely to be theoretically unsound,but is of practical convenience for the empirical application. This isalso true for another assumption in this paper; non-autonomous

2551I.-M. Gren, L. Isacs / Ecological Economics 68 (2009) 2549–2559

growth in wealth. Admittedly, the exclusion of exogenous growth inwealth during time is strong, in particular for small open economiesconsidered in this paper where terms of trade are likely to changeexogenously. Given these simplifying assumptions and given initialK, S, and P, (5) can also be written in terms of initial stockparameters (see e.g. Dasgupta and Mäler 2000, 2001; Arrow et al.,2004). The state valuation function is then written as

W Kt = Pt ; St = Ptð Þ =Z∞

t=0

πte− rtdt ð5′Þ

In much of the literature, sustainable development is defined asnon-declining utility of consumption or as non-declining consumptionduring time (e.g. Brekke, 1997a,b; Asheim 2003; Heal and Kriström,2005). A similar concept is applied in this study, where sustainability ismeasured in terms of net value of production per capita instead ofconsumption. Non-declining value of production then implies intactincome and consumption possibilities per capita. Assuming that thefunctional in (5′) is continuous, change in wealth in time t andonwards is found by differentiating (5′) with respect to time (see e.g.see e.g. Sydsaeter 1978 pp.116 for differentiation of integrals), which iswritten as

1Pt

πt 1− θð Þ + mKt:K + mst

:S

� �= rWt ð6Þ

where vtK=WKt

and vtS=WSt are the accounting prices of Kt and St

respectively. Assuming constant accounting prices, we have from (6)that sustainable net value of production per capita in time t isdetermined by

1− θð Þ Aπt

At− θπt

� �+ mKt

A:KAt

+ mStA:S

At− θ mKt

:K + mSt

:S

� �

= r mKt

:K + m

St

:S − θ m

kt Kt + m

St St

� �� �ð7Þ

which is positive only if net investment per capita during a period, i.e.the right hand side of (7) is positive. It demonstrates the well knownresult that sustainability per capita can be can be obtained whenpopulation growth rate is negative evenwhen S and K are negative. Onthe other hand, positive Sand K do not guarantee sustainability whenθN0. Then, the rate of change in value of capital stocks needs to behigher than population growth rate in order to ensure a sustainabledevelopment in the period under study, which can be seen from theright hand side of (7). However, as discussed by, amongothers, Howarth(2007), the sustainability condition in (7) is only partial sinceexogenoustechnological change is not included. Furthermore, it assumes thatinvestment in man-made capital can replace degradation of naturalcapital, which is questionable for irreversible ecosystem changes.Nevertheless, it can be useful for the purpose of this study, i.e. to assesswhether the use of natural capital per capita is sustainable during aspecific period of time.

The conventional accounts of value of production include only themarketed goods and services in (6) evaluated at their market prices(see e.g. Dasgupta and Mäler (2000) for a discussion of the difficultiesof finding correct market prices of goods). It can then be noticed from(6) that the flow component of non-marketed ecosystem services,utEt, is not included in conventional national accounts. In order toidentify themissing non-marketed items of the stock components, theaccounting prices of the two capital resources, vtS and vt

K, need to befurther developed.

Values of marginal changes in S and K in period t are derived from(1) to (3′) by solving for St and Kt in (1) and (2) (see e.g. Seijerstadand Sydsaeter 1987 pp. 417 for solutions of differential equations). Thesolutions are then inserted into (5′), which is differentiated withrespect to St and Kt in order to obtain the accounting prices (see e.g.

Sydsaeter 1978 pp.116 for differentiation of integrals). The results arewritten as

mSt =Z ∞

t1= Ptð Þ utESt + ptQ St

� �e− r + θ−βð Þ τ− tð Þdτ ð8Þ

mKt =

Z ∞

t1 = Ptð ÞptQKt

e− r + θ + ρð Þ τ− tð Þdτ ð9Þ

According to (8) and (9), the accounting prices of the two capitalresources correspond to the current and future streams of utility frommarketed and non-marketed goods from a marginal change in St andKt respectively. The future values of these services are discounted bythe interest rate, r, and the population growth rate, θ, for bothresources. The accounting price of S also includes discounting offuture ecosystem services by the intrinsic growth rate of S, β. Whenthe change in the growth rate is positive, the discounting of future netutility is decreased as compared to when β=0. The latter is valid for anon-renewable resource, which thus is discounted only by the interestrate and population growth rate. Similarly, the marginal value of theinput K is discounted by the interest rate plus the depreciation rateand population growth rate.

Although the difficulty of measuring r and ρ is well known this isdisregarded in this paper since the focus is on the quantification ofmissing items in national accounts in an ecosystem service frame-work, and not on empirical measurements of included items. Anotherrelated issue is that production and wealth accounts may differ withrespect to inclusion of capital resources where production accountsmost often include only investment in reproducible capital whilewealth accounts cover several types of assets (see Hecht 2007 for areview of green accounting in practice).

Since vtK includes only the marketed good Q , there is by assumptionnomissing component in the accounting price of this asset. However, vt

S

includes both non-marketed and marketed items, which implies thatthe total value of a change in S is not fully included in the conventionalnet regional product accounts at market prices, NRP. It corresponds tothe Hotelling rent of non-renewable resources when β=0 and ES=0.Since the secondvariablewithin the parentheses in (8) reflects thevalueof marketed goods, the accounting price needed for adjusting NRP withrespect to changes in S, v t

SE, corresponds to

mSEt =Z ∞

t1= Ptð ÞutESt e

− r + θ−βð Þ τ− tð Þdτ ð10Þ

When including the non-marketed items of the flow and the stockcomponent, the environmentally adjusted regional product in periodt, ERPt, is determined by

ERPt = NRPt + utEt + PtmSEt ΔSt ; ð11Þ

which is divided by Pt in order to obtain ERPt per capita.Turning to the second purpose of this section, i.e. to identify the

role of pollution in the adjustment of NRP, we can observe thatpollution in a region, Nj, is not explicitly included in (10) or (11). Thisresult runs against the common practice of deducting damage costsfrom NRP (see e.g. Heal and Kriström, 2005, for a review). However,this practice implies double counting since the effect of pollution isalready included in the change in natural capital stock during a periodof time as a cause of ΔSt. This can be seen by solving for St, Kt, and Pt in(1)–(3), inserting the expression into (5′), and differentiate withrespect to N j. Since a change in Nj affects the functioning ofecosystems in all regions i=1,..,n as described by the assumed matrixof transport coefficients a ji, the total wealth impact is written as

Pi AW

i

ANj= −

XiPimSit a

jiai ð12Þ

2552 I.-M. Gren, L. Isacs / Ecological Economics 68 (2009) 2549–2559

where changes in the functioning of natural capital in a region i from amarginal change in N j are described by a jiai where i=1,..,n.Expression (12) shows that the wealth effects of a marginal pollutionchange in region j corresponds to the sum of values of associatedchanges in the functioning of the natural capital stocks for all regionsfor which a jiN0. The regional accounting prices in (12) include effectsof both marketed and non-marketed goods, where the former iscovered by conventional accounts. This has been demonstrated byBergman et al., (1995), who showed that the value of decreases in thegrowth of timber due to sulphur deposition on forest land entersconventional accounts. From (10) and (11) it can also be seen that if allregions apply the adjustment suggested by (10), this includes theimpact on non-marketed goods of pollution changes as written in(12).

In practice, the suggested adjustment in (10) comes most close tothe Environmental and Natural Resources Accounting Project(ENRAP) system, which accounts for both flow and stock componentsof environmental services (e.g. Peskin and Angeles, 2001). This is incontrast with the UN System of Integrated Environmental andEconomic Accounting (SEEA) where environmental services do nothave a positive entry (UN, 2003). This means that the flow componentin (10), utEt is included in the ENRAP system but not in the SEEAaccounts. A unique feature of the suggested adjustment mechanism in(10) as compared to the SEEA and ENRAP systems is that wastedisposal services, such as forests' carbon sink service, enter theaccounts as positive terms. Without this service, the pollutionconcentration in air would be higher and associated impact onclimate change would be larger. Only if the stock componentdecreases, such as a decrease in forest biomass, there is a negativeentry in the system.

3. Estimation of non-marketed flow and stock components ofnatural capital

Necessary data retrieval for calculations of the adjustmentcomponents of conventional regional accounts as expressed by Eq.(11) is divided into three interrelated tasks: i) quantification of thestock of each ecosystem and its change during time, ii) establishmentof the relations between production of ecosystem services andecosystem stock levels, and iii) estimation of unit net values of theecosystem outputs (e.g. Boyd and Banzhaf, 2007). As will be reportedin this section, the main empirical difficulty is associated with thesecond task. Unless otherwise stated, all data are found in Isacs(2007).

Under the first task, a practical difficulty arises with regard to thechoice of relevant natural capital assets and regions. The exempli-fication in this paper follows the UN (2003) ecosystem classification,which distinguishes between two broad classes of ecosystems:aquatic and terrestrial. The choice here is also dependent on theavailability of data. Since there have been relatively many studiesestimating recreational and pollutant sink values in Sweden withapplications on forests, arable land, and wetlands, calculations aremade for these ecosystem services and ecosystems (Sundberg andSöderqvist, 2004). Carbon dioxide sequestration values are estimatedfor forests and nutrient, nitrogen and phosphorus, sink values forwetlands. Recreational value estimates are obtained from studiesusing different methods for estimating willingness to pay, contingentvaluation and travel cost methods, for preserving recreational areasin forests, arable landscapes, and wetlands in Sweden (see Sundbergand Söderqvist (2004) for a review of methods and results).

The choice of regional division is guided by access to data on theecosystem stock variables. Annual data on regional gross incomeaccounts, forest and arable land use are available at the county levelscale (SCB, 2006). However, official statistics on wetland coverage arerelatively recent, less frequent and presented at lower resolution levelwhere Sweden is divided into four main regions, see Fig. A1 in the

Appendix. Therefore, calculations are made for these four regions andfor the most recent year with sufficient data coverage, which is foryear 2000. In the following, a brief presentation is given of dataretrieval on flow and stock values of wetlands, forests and agriculturallandscape in Sweden.

3.1. Forests

A forest ecosystem supplies a number of different marketed andnon-marketed services, such as timber, berries, biodiversity, carbonuptake and hunting possibilities. In Sweden, which is largely coveredby both boreal and deciduous forest, its associated values are closelyrelated to wealth, not the least since forestry has been an importantexport sector for a long period of time. In addition, the typical Swedehas a more or less intimate relation to the forest, in part sinceeveryone in Sweden has access to forest land through the nationalland ethic rule (Jämttjärn, 1996).

In the 1980s, some of the non-marketed values provided by aforest ecosystem became a global welfare issue owing to the severedepletion of rain forests and associated consequences in the form ofclimate change and biodiversity loss. Since then, the carbon sinkcapacity of forests has become a question of elevated importancethrough the establishment of the Kyoto Treaty. Via the photosynth-esis, forests absorb carbon dioxide (CO2), which is the mostimportant green house gas, and store carbon in tree biomass andin the ground. Since 2001, countries are allowed to include part ofthe carbon sink of their national forests owing to deliberatemanagement, in order to achieve their CO2 reduction commitmentsunder the Kyoto Treaty. Having the second largest forest land areaper capita in Europe, Sweden has a significant potential to use thisoption for achieving its emission reduction goals (LUSTRA, 2002).The forest's net carbon sink capacity depends mainly on biomassgrowth, but also on a number of different factors, such as type ofsoil, ground water level, tree species and climate (Hultkrantz, 1992;Morén 2003).

The consumption component of carbon sequestration in forestsconsists of the value of CO2 sequestration. The unit value ofsequestration is approximated by equilibrium permit price of carbonemissions at the EU market, which varied between SEK 0.06/kg CO2

and 0.3 since the introduction of the market in beginning of 2005until end of 2007 (Alberola et al., 2008). Calculations of regionaluptake of CO2 in Swedish forests are based on the assumption thatcarbon sequestration in trees is linearly related to biomass growth,which gives a sequestration corresponding to approximately 20 mil-lion tonnes of CO2 in year 2000 (see Table A2 in the Appendix). Thestock component determining changes in carbon uptake duringyears is related to different soil and sub soil processes, and thoseoccurring in tree biomass are currently most known in Sweden.Growth of a stand can, in turn, be described as a logistic or quadraticfunction of the biomass stock. Although it is not reasonable toassume tree felling with unchanged biomass stock for a single age, itis possible to harvest trees in a multi-age forestry at a sustainablebiomass stock. However, forest biomass growth functions useful forassessing impacts on carbon sequestration from changes in the stockis not available. Therefore, a simple linear relation between carbongrowth and stock is assumed. According to Morén (2003), averagecarbon sequestration owing to biomass growth corresponds 0.17 tonC/ha and the average carbon pool in Swedish forests is 85 tonnes ofC/ha. The average annual growth rate in the subsoil forest carbonpool, which is a measurement of the intrinsic growth rate β inEq. (8), is then approximately 0.002 for relatively small changes inthe carbon pool.

Recreational values of forests, such as hunting, picking ofmushrooms and berries, sporting and walking, are obtained from asurvey of studies estimating willingness to pay for recreation inSwedish forests (Jämttjärn, 1996). The average value per person and

2553I.-M. Gren, L. Isacs / Ecological Economics 68 (2009) 2549–2559

year amounts to SEK 3000, which corresponds to approximately SEK800/ha forest. The variation is, however, large among studiesranging from approximately 0.4 to 2 times of the average value(Jämttjärn, 1996). Due to lack of data, it is not possible to assigndifferent recreational values depending on location of forests. Asreported in, among others, Termansen et al. (2008), there can be asignificant spatial distribution of recreational values of forests withina country.

Table 1Flow and stock values of non-market services in forests, wetlands, and agriculturallandscape in different regions in Sweden, billions of SEK in 2000.

NorthNorrland

SouthNorrland

Svealand Götaland Total

Consumption valuesForests 3–14 2–11 2–10 2–10 9–45Wetlands 7–43 1–21 3–13 3–13 14–90Agricultural landscape 0.2–0.4 0.2–1 1–1 1–2 2–3Total consumption values 10–56 3–33 6–24 6–25 25–138

Investment valuesForests 15–53 5–25 7–38 4–23 31–139Wetlands 5–34 1–6 −(4–22) 1–4 3–22Agricultural landscape −1 to −1 0–0 0.2–1 0.2–1 0–1Total investment values 20–86 6–31 3–17 5–28 34–162

Total consumption andinvestment values

30–142 9–64 9–41 11–53 59–300

Source: Tables A1 and A2 in the Appendix.

3.2. Wetlands

Wetlands serve a useful purpose for society in many differentrespects. Besides sustaining important hydrological functions likebuffering of groundwater and regulation of water cycles, wetlands canprevent flooding and reduce the effects of drought, as well as functionas cleaning plants by absorbing chemicals and pollutives. Sincewetlands host a diversified biological life and are vital environmentsfor many rare species, they also attract people for recreationalactivities. Moreover, peat land and wetland sediments serve ashistorical archives by storing both cultural and biological information,making it possible to track everything from civilizational spread toclimatological change (SEPA, 2006).

Sweden is one of the most wetland-rich countries in the worldwith its 10 million hectares, which corresponds to almost onequarter of the total country area (SEPA, 2006). The largest share ofSweden's wetland area is situated in the northern Sweden, seeTable A2 in the Appendix. In southern Sweden, almost 90% of theoriginal wetland area has been converted into arable land. The mainnutrient sequestration value wetlands arises from the need ofcombating eutrophication in lakes and seas, in particular the BalticSea (SEPA, 2004). In principle, the value of nutrient sink can then beestimated as the abatement of each nutrient times the marginalvalue of water quality improvement. When such values are notavailable, cost effectiveness analysis can be applied for assessingreplacement value of wetlands in a cost effective allocation ofmeasures for given nutrient abatement targets (see Gren 1999 for adiscussion and comparison of these methods). Since values ofnutrient reductions are not available, this study relies on thereplacement value of wetlands, which are derived from minimumcost solutions for achieving nutrient reduction targets to the BalticSea. The replacement value, i.e. cost savings made by includingwetlands as an abatement options for reaching a specific target, thenvaries depending on target formulation, abatement capacity bywetlands, the opportunity cost of land and construction costs ofwetlands. These parameter values vary for different parts of Sweden,and marginal replacement values have been estimated by Gren et al.(2008) for the achievement of the Baltic Sea targets suggested by theinternational administrative Helcom body (Helcom, 2007). Annualnutrient sink values of wetlands then range between SEK 400 and22,500/ha being highest in southern Sweden due to the relativelylarge nutrient abatement capacity per ha in this region, see Table A1in the Appendix. The steady-state nitrogen sequestration by createdwetlands can be reached within a relatively short period of time,3 years (Kusler and Kentula, 1990), and the associated growthcomponent, i.e. β in Eq. (10), is therefore assumed to be zero.

Swedish wetlands have also been subjected to a few recreationalvaluation studies (see Sundberg and Söderqvist (2004) for a survey).Among the things investigated are people's willingness to pay forbiodiversity provision, improved fishing possibilities and walkingfacilities. The studies are based on contingent valuation estimates ofwillingness to pay (WTP) for construction of new wetlands in thesouth of Sweden. The appropriateness of transferring these benefits toother regions, where the access to wetlands and other forms ofrecreation areas might differ, is questionable. The use of upper andlower bound values can at least partly attend to this problem. The

value per unit of land varies considerably and range between SEK2500 and 16,100/ha.

3.3. Agricultural landscape

In addition to food production, the agricultural landscapeprovides society with a variety of valuable by-products. In Sweden,agricultural land use like grazing and wooded pasture can be tracedto time periods long before the Viking era and are among the mostbiodiversity rich ecosystems in the country. The agricultural land-scape thus creates a number of important services, such asmaintenance of an open landscape, cultural heritage and biodiversity(Drake 1992). Recreational values of agricultural landscape inSweden have been estimated in three studies (Drake et al. 1991,Drake 1992 and Hasund, 1997). Drake (1992) is the only study withregion specific estimates, where Sweden has been divided into threeregions; a northern, central and southern part. The contingentvaluation method was used for eliciting WTP. According to theresults, people in the northern regions with relatively less agricul-tural landscape has a higher WTP for its preservation than people inthe south, where agriculture is more abundant, see Table A1 in theAppendix.

4. Total estimates of non-marketed flow and stock values

Consumption and investment values as shown by Eq. (11) inSection 2 are calculated for each of the three ecosystems. Changes invalues of future ecosystem outputs are discounted with assumptionsof continuous time, unlimited time perspective, and a constantinterest rate. A more realistic interest rate might be changing withrespect to distant future (e.g. Weitzman, 2001). However, thediscount rate is difficult to determine empirically, and a constantreal discount rate of 3% is therefore assumed. This rate is close to thehistorical rate of interest on risk free Swedish governmental bonds.Sensitivity analyses are carried out for both higher and lowerdiscount rates.

Sweden has experienced a positive population growth duringdecades, but is has been unevenly distributed among the regions(Amcoff, 2006). Population in urban areas, such as the Svealandregion in our study, has grown at the cost of rural areas, which havefaced a negative growth. This pattern in population growth isexpected to continue during the next decades mainly due to themigration of young people from rural to urban areas. According toforecasts of population growth during 2005 and 2025, the annualgrowth rates in the regions in our study are expected to varybetween −0.0085 and 0.0075 (see Table A4 in the Appendix). They

Fig. 2. Change in percent in GRP/cap and ERP/cap for high and low values of non-marketed services (ERPlowand ERPhigh) between 1995 and 2000 in real prices. Source:Calculations based on Tables A1, A2, A4, and A5.

2554 I.-M. Gren, L. Isacs / Ecological Economics 68 (2009) 2549–2559

are negative for the two northern regions and positive for the othertwo regions.

Due to lack of data it is also assumed that there are nomanagement costs associated with the provision of the ecosystemservices included in our study. For forests and arable land theservices are obtained as by-products of production of marketedoutputs, and this assumption is therefore valid. However, theexclusion of management costs implies an overestimate of thevalue of services of wetlands since there are in general costassociated with the management of wetlands as pollutant sink(e.g. Gren, 1995; Byström, 1998).

Based on all assumptions and data presented in this section and inSection 3, the estimated consumption and investment values for fourdifferent regions are as presented in Table 1.

The results presented in Table 1 show large variations inconsumption and investment values among regions and ecosystems.The total value ranges between approximately 65 and 300 billion ofSEK (8.45 SEK=1 Euro in 2000) which corresponds to 3–14% of GDPin 2000 (see Table A4). The single most important component is theinvestment value in forest, which accounts for approximately 45% ofthe total value. Since investment in forest recreation is negative, thevalue of carbon sequestration investment represents most of theforest investment value, see Table A3 in the Appendix. The secondlargest source of total values is the consumption value provided bywetlands, where recreational values account for the major part, seeTable A3 in the Appendix.

It can also be noticed from Table 1 that North Norrland providesthe highest values of non-marketed ecosystem services, whichcorrespond to approximately 45% of the total value. It is more thantwice as large as for any other region. The reason is the relatively largeconsumption values of wetlands and investment values of wetlandsand forests.

Table 2Value of changes in natural capital between 1995 and 2000 for assumed low and high

5. Regional income and growth

Interregional comparisons are made for adjusted and non-adjusted regional income per capita. However, it is not possible toobtain reliable estimates of net investment at the regional level,which would be the correct welfare measurement as described inSection 2. Gross regional product (GRP) is therefore compared withenvironmentally adjusted gross regional product (ERP) for low andhigh estimates of consumption and investment values of the non-marketed ecosystem services. When calculating and comparing GRP/capita and ERP/capita interesting shifts occur in ranking betweenthe four regions, see Fig. 1.

When comparing the regions with respect to conventional incomemeasurement, GRP, Svealand has the highest per capita income, whichis explained by the highest per capita income in the capital of Sweden,Stockholm, located in this region. The most northern region has the

Fig.1.Gross region product (GRP) and environmentally adjusted regional product (ERP)per capita for low and high values of non-marketed ecosystem services (ERPlow andERPhigh) for different regions and for Sweden as a whole in 2000. Source: Calculationsbased on Table 1 and Table A4 in the Appendix.

lowest income, which is approximately 30% below that in Svealand.However, when comparing ERP/capita, the ranking is changed so thatNorth Norrland now instead becomes the region with the largest percapita income for the high value of non-marketed ecosystem services.This is explained by the relatively large value of ecosystem services,which is of the same order of magnitude as GRP/capita at the uppervalue of ERP. There are relatively small differences in ERP and GRP forSvealand and Götaland.

The difference in ERP/capita among regions is measured by isincreased when the discount rate is reduced by one half, i.e. to 1.5%(see Fig. A1 in the Appendix). The relatively high ERP/capita in thenorthern regions then influences that of entire Sweden by increasingGRP/capita by approximately 25%. A doubling of the discount rate to6% has the opposite effect since the value of investment in naturalcapital then decreases as compared to the discount rate in thereference case.

Economic growth is estimated and compared for a five yearperiod, from 1995 to 2000 for GRP/capita and ERP/capita. Growth isthen measured as changes in GRP/capita and ERP/capita from the1995 levels. Differences in estimated growth between regions aresimilar to those obtained for the interregional comparisons ofincome, see Fig. 2.

Growth as measured by change in GRP/capita in the northernregions corresponds to approximately 1/5 of the growth in the highestgrowing region, Svealand. However, growth is increasing significantlyfor the North Norrland when considering values of ecosystemservices. For assumed high values of the services, this region has thehighest growth in Sweden. The reason is the relatively large increasein areas of forests and wetlands in the northern counties, see Table A2

monetary values of recreational and pollutant sink services.

NorthNorrland

SouthNorrland

Svealand Götaland Total

Low High Low High Low High Low High Low High

Per capita changes, thousand SEKForests 4 9 −0.1 −0.8 0.8 3.8 −0.4 −2 0.2 0.6Wetlands 20 117 5 27 −5 −8 1 2 1 7Agriculturallandscape

−0.2 −1.2 −0.2 −0.7 0.1 0.2 0.1 0.2 0 01

Total 24 125 4 25 −4 −4 0.2 0.7 1.2 8

Total levels with no population growth, billions SEKForests 1.9 2.3 1.3 −1.4 3 16 −2 −14 4 3Wetlands 10 45 3 17 −6 −30 1.4 10 9 41Agriculturallandscape

−0.2 −0.4 −0.2 −0.4 0.2 0.7 0.3 1 0.2 1.5

Total 12 47 4 15 −3 −15 −0.3 −3 13 45

Source: Tables 1 and A3 in the Appendix.

2555I.-M. Gren, L. Isacs / Ecological Economics 68 (2009) 2549–2559

in the Appendix. For both Svealand and Götaland, growth in ERP/capita is lower then that in GRP/capita, which is due to the decline inforest areas for both regions. Decrease in wetland areas is anadditional explanation to the lower growth in Svealand and a decreasein biomass growth and thereby a reduction in carbon sequestrationcauses decline in growth for Götaland, see Tables A2 and A3 in theAppendix.

The growth figures for Sweden as a whole show a modestdifference between the two measurements. Growth in ERP/capitafor entire Sweden is approximately 1.5% higher or lower than growthin GRP/capita. However, a decrease of the discount rate to 1.5% has asignificant effect on the growth rate for entire Sweden, whichincreases by approximately 60% for high values of ecosystem services(see Fig. A2 in the Appendix). This is due to the increase in growth ratein North Norrland by approximately 20 times compared with thegrowth rate in conventional income.

6. Sustainable use of natural capital

Recall from Section 2 Eq. (7) that the condition for sustainable useof natural capital is determined by the value of stock changes andpopulation growth. A decline in the stock of the resources implies lessconsumption opportunities per capita for future generations, and istherefore not sustainable. However, a declining population growthrate, which promotes sustainable development of natural capital, maynot be desirable in rural areas where depopulation is regarded as athreat to the development of the regions. Therefore, changes in bothnatural capital per capita and totally between the years 1995 and 2000are calculated, see Table 2.

The calculations in Table 2 point to a sustainable use of total andnatural resources per capita for Sweden as a whole, but there aresignificant differences between regions and ecosystems. Whereas theuse of natural capital can be unsustainable in Svealand and Götalandit is clearly sustainable in North and South Norrland. Wetlandmanagement contributes to sustainable use in North Norrland but isthe main cause of unsustainable natural capital management inSvealand. The switch from slightly sustainable use of natural capitalper capita to unsustainable use of total natural capital in Götaland isexplained by the decrease in population during the period of timeunder study. However, whether or not the regions are on asustainable path depend also on net investment in other capitalresources not included in this study, which can counter act orenforce the results presented in Table 2. Nevertheless, it can beconcluded that natural capital investment contributes to sustainabledevelopment to a larger extent in North and South Norrland than inthe other regions.

7. Summary and discussion

The main purpose of this paper has been twofold: to derive ameasurement of non-marketed ecosystem services to be includedin regional income accounts and to apply this for calculation ofregional green accounts in Sweden. A specific feature of this paperis the treatment of natural capital assets as inputs into productionof ecosystem services. Consumption and investment values ofnatural capital were then derived from its current and futurestreams of ecosystem services. It was found that pollution does notenter the accounts explicitly, but only indirectly through itsimpacts on the functioning of natural capital with respect toproduction of ecosystem services. Differences in adjustments toconventional regional accounts among regions are then attributableto variations among regions in changes in natural capital assets,production of ecosystem services, and accounting prices of naturalcapital.

Not surprisingly, however, the estimation of accounting prices ofnatural capital turned out to be associatedwith considerable empirical

challenges. The problem of finding values of non-marketed goods andservices has been known and investigated since the 1970s (see Turneret al., 2003 for a review). Estimation of accounting prices also requiresthe quantification of the relation between production of outputs andnatural capital stock levels, which turned out to be the most difficulttask for the calculation of Swedish accounting prices of forests,wetlands, and agricultural landscape. The estimates were based on theavailability of existing valuation studies of ecosystem services inSweden. Therefore, only two outputs — recreational values andpollutant sequestration — were included. Regional estimates weremade for four Swedish regions, which differ with respect topopulation density, economic activity and natural capital assets.

Given all caveats associated with finding data, the adjustment ofconventional regional income accounts showed significantly differenteffects in the four regions. An interregional comparison shows that theenvironmentally adjusted income increases considerably for thenatural capital rich northern Swedish regions, while the impact ofadjustment on the southern regions, which have relatively highconventional income/capita, is modest. The result indicated a changein the regional pattern of income/capita, the environmentallyadjusted income being highest in the northern regions while theconventional income is highest for the southern regions. Similarresults were obtained for comparisons of growth among regions.Whereas growth in environmentally adjusted income was highest forthe northern regions growth in conventional income was highest forthe southern regions. However, the difference in adjusted and non-adjusted national growth was insignificant. The calculations alsopointed to sustainable use of natural capital in most regions, but todifferent degrees. Another result was the important role of the value ofcarbon sequestration for income adjustments and sustainable devel-opment, which is a cause of the increase in biomass growth in Swedenduring the years under study. Gundimeda et al. (2007) did alsoobserve the significant role of carbon sequestration in some states inIndia for reducing conventional regional income due to decreases inbiomass growth.

Although the inclusion of two ecosystem services and threedifferent natural capital resources implies an extension as comparedto the two earlier studies of regional accounts and sustainabledevelopment, the associated correction of conventional regionalaccounts is still only partial due to the exclusion of: i) values ofchanges in marketed outputs due to natural resource depletion, ii)other ecosystem services, such as biodiversity, and iii) other naturalcapital resources like lakes, urban areas, and coastal waters. Accordingto Skånberg (2001), the cost of resource depletion from forests,subsoil resources, and fishery amounts to 4 billion of SEK and the costof acidified lakes amounts to 11 billion of SEK in 1997 (in 2000 prices).By focusing on damages from nitrogen and sulphur deposition on theSwedish territory, Ahlroth (2000) calculated a negative adjustment ofapproximately 15 billion of SEK. If the effects of these negativeenvironmental items were considered, total environmentally adjustednational account for Sweden would be reduced by at the most 1%, butthe division of impacts among regions is likely to differ. The negativeeffects of forest and mine resource depletion are likely to have arelatively larger impact on northern regions, while the damages fromnitrogen and sulphur deposition, which mainly include health effectsand corrosion, occur to a relatively large extent in the denselypopulated Svealand and Götaland.

Improved estimates of spatial distributions of values of environ-mental changes are required in order tomakemore accurate estimatesof the impact of non-marketed ecosystem services on regionaleconomic growth and sustainable development. As shown by,among others, Bateman and Lovett. (2000) geographical informationsystems can be very useful in this respect. Nevertheless, the resultsfrom this study indicate that conventional regional accounts mayunderestimate provisioning of wealth and sustainable development innatural resource rich regions.

2556 I.-M. Gren, L. Isacs / Ecological Economics 68 (2009) 2549–2559

Appendix 1. Map of Swedish counties and provinces

Table A1Consumption unit values and accounting prices on non-marketed ecosystem pollutant abatement and recreational values.

North Norrland South Norrland Svealand Götaland

ForestsValues of carbon seq.

Consumption unit value, SEK/ton CO2a 60–300 60–300 60–300 60–300

Accounting price, SEK/kbm tree biomass growthb 2016–10,082 1555–7778 1267–6338 1036–5186

Recreational valuesConsumption unit, mill SEK/103 haa 0.34–1.7 0.34–1.7 0.34–1.7 0.34–1.7Accounting price, mill SEK/103 haa 15–78 12–61 10–49 7–40

WetlandsValues of nutrient sequestration

Consumption unit, mill SEK/103 hac 0.4–1.1 0.8–2.4 1.0–3.1 5.0–22.5Accounting price, mill SEK/103 hac 18–52 28–86 29–91 120–540

Recreational valuesConsumption unit, mill SEK/103 haa 2.5–16.1 2.5–16.1 2.5–16.1 2.5–16.1Accounting price, mill SEK/103 haa 116–752 90–580 73–473 60–389

Agricultural landscapeRecreational values

Consumption unit value, mill SEK/103 had 1.1–3.2 1.1–3.2 0.9–2.8 0.6–2.6Accounting price, mill SEK/103 had 50–151 39–117 27–82 14–43

a Isacs (2007).b Calculated with 0.73 ton CO2/m3 tree biomass change, with β=0.002 (Isacs, 2007 pp. 13, footnote 21).c Calculations based on Gren et al. 2008.d Drake (1992).

Table A2Ecosystem resources and changes in 1995 and 2000 for different regions in Sweden.

North Norrland South Norrland Svealand Götaland Total

1995 2000 1995 2000 1995 2000 1995 2000 1995 2000

Forests1000 ha 6627 6772 5811 5730 5293 5214 4971 4953 22,621 22,749Net change, 1000 ha 19 −48 8 −63 −49 44 36 −68 15 −135Net biomass growth, mill m3 6.55 7.70 4.66 5.15 5.85 7.84 8.39 6.90 25.45 27.58Net sequestration, mill ton CO2

a 4.78 5.62 3.40 3.76 4.27 5.72 6.12 5.04 18.58 20.13

Wetlands1000 ha 2298 2490 1134 1128 731 694 302 326 4637 4637Net area change, 1000 ha −35 44 −19 8 9 −38 −4 4 −48 16

Agricultural landscape1000 ha 125 134 213 206 937 966 2194 2240 3467 3546Net area change, 1000 ha −4 −8 5 1 7 14 −10 14 −2 21

Source: Isacs (2007).a Calculated with 0.73 ton CO2/m3 tree biomass change (Isacs, 2007, pp. 13, footnote 21).

Table A3Flow and stock values of non-marketed ecosystem pollutant sink and recreational values in forests, wetlands, and agricultural landscape in different regions in Sweden, billion SEK in2000.

North Norrland South Norrland Svealand Götaland Total

Consumption valuesForestsCarbon sink 1–2 0.4–1 0.4–2 0.4–2 2–7Recreation 2–12 2–10 2–7 2–8 8–37

WetlandsNutrient sink 1–3 2–3 1–2 2–7 6–15Recreation 6–40 −1–18 2–11 1–7 8–75

Agricultural landscape 0.2–0.4 0.2–1 1–1 0.4–1 1.8–3.4

Investment valuesForestsCarbon sink 16–57 6–29 7–36 5–26 34–148Recreation −(1–4) −(1–4) 0−2 −(1–3) −(2–11)

WetlandsNutrient sink 0–1 0.2–1 −(1–4) 1–2 0–0Recreation 5–33 1–5 −(3–18) 0.2–2 3–22

Agricultural landscape −(0.4–1) 0–0 0.4–1 0.2–1 0.2–1

Source: Tables A1 and A2 in the Appendix.

2557I.-M. Gren, L. Isacs / Ecological Economics 68 (2009) 2549–2559

Table A4GRP and population in Swedish regions in 1995 and 2000, and forecasted population growth rates.

North Norrland South Norrland Svealand Götaland Total

GRP, billions SEK in 2000 prices1995 105 139 769 817 18292000 109 146 979 984 2217

Population, 10001995 527 684 3378 4420 88272000 513 658 3448 4253 8872Forecasted annual population growth rate 2005–2025a −0.0085 −0.0025 0.0035 0.0075 0.004

Source: Isacs (2007).a Calculations based on Amcoff, (2006), page 14.

Fig. A2. Sensitivity analysis of growth in GRP and ERP for high and low values of non-marketed services (ERPlow and ERPhigh) for halved and doubled discount rates, r=1.5and r=6% respectively (growth between the years 1995 and 2000 in real prices).

Table A5Flow and stock values of non-marketed ecosystem services in forests, wetlands, and agricultural landscape in different regions in Sweden, billion SEK in 1995 (in 2000 prices).

North Norrland South Norrland Svealand Götaland Total

Consumption valuesForests. 3–12 2–11 2–10 2–10 9–44Wetlands 7–40 5–21 3–14 2–12 17–89Agricultural landscape 0.2–0.4 0.2–1 1–1 1–2 3–4

Investment valuesForests 14–50 5–27 5–25 7–33 31–134Wetlands −(4–26) −(2–13) 1–5 −(1–4) −6 to −38Agricultural landscape −(0.2–1) 0–1 0.4–1 −(0.1–0.4) 0–0

Total 19–75 10–48 12–56 11–53 54–233

Source: Tables A1 and A2 in the Appendix.

Fig. A1. Sensitivity analysis of ERP at low and high values of ecosystem services (ERPlowand ERPhigh) for halved and doubled discount rates, r=1.5 and r=6 respectively, ascompared with the reference case.

2558 I.-M. Gren, L. Isacs / Ecological Economics 68 (2009) 2549–2559

References

Ahlroth, S., 2000. Correcting net domestic product for sulphur dioxide and nitrogenoxide emissions: implementation of a theoretical model in practice. Working PaperNo. 73. National Institute of Economic Research, Stockholm.

Alberola, E., Chevallier, J., Chèze, B., 2008. Price drivers and structural breaks inEuropean carbon prices 2005–2007. Energy Policy 36 (2008), 787–797.

Amcoff, J., 2006. How to break the trend? A regional perspective on populationdevelopment until 2025. (in Swedish). Framtider 3, 5–11.

Aronsson, T., Johansson, P.-O., Löfgren, K.-G., 1997. Welfare Measurement, Sustainabilityand Green National Accounting. Edward Elgar, Cheltenham, UK.

Arrow, K., Dasgupta, P., Mäler, K.-G., 2004. Evaluating projects and assessing sustainabledevelopment in imperfect economies. In: Dasgupta, P., Mäler, K.-G. (Eds.), TheEconomics of Non-Convex Ecosystems. Kluwer Academic Publishers, Netherlands,pp. 149–187.

Asheim, G., 1994. Net national product as an indicator of sustainability. ScandinavianJournal of Economics 96, 257–265.

Asheim, G., 2003. Green national accounting for welfare and sustainability. Ataxonomy of assumptions and results. Scottish Journal of Political Economy 50 (2),113–130.

Aktinson, G., Hamilton, K., 2007. Progress along the path: evolving issues in themeasurement of genuine saving. Environmental and Resource Economics 37,43–61.

Batemen, I.J., Lovett, A.A., 2000. Estimating and valuing the carbon sequestered insoftwood and hardwood trees, timber products and forest soils in Wales. Journal ofEnvironmental Management 60, 301–322.

Bergman, L., Ahlroth, S., Skånberg, K., 1995. A CGE Analysis of Sulphur Deposition andSweden's Green National Product. Manuscript, Stockholm School of Economics,Stockholm, Sweden.

Boyd, J., Banzhaf, S., 2007. What are ecosystem services? The need for a standardizedenvironmental accounting units. Ecological Economics 63, 616–626.

Brekke, K.A., 1997a. Hicksian income from resource extraction in an open economy.Land Economics 73 (4), 516–527.

Brekke, K.A., 1997b. Economic Growth and the Environment. On the Measurement ofIncome and Welfare. Edward Elgar, Cheltenham, UK.

Byström, O., 1998. The nitrogen abatement cost in wetlands. Ecological Economics 26,321–331.

Cairns, R., 2002. Green accounting using imperfect current prices. Environment andDevelopment Economics 7, 207–214.

Dasgupta, P., Mäler, K.-G., 2000. Net national product, wealth and social well-being.Environment and Development Economics 5, 69–93.

Dasgupta, P., Mäler, K.-G., 2001. Wealth as a criterion for sustainable development.World Economics 2 (3), 19–44.

Dasgupta, P., 2009. The welfare economic theory of green national accounts.Environmental and Resource Economics 42, 3–38.

Drake, L., 1992. The non-market value of the Swedish agricultural landscape. EuropeanReview of Agricultural Economics 19, 351–364.

Drake, L., Kumm, K.-I., Andersson, M., 1991. Har jordbruket i Rottnadalen någonframtid? En analys av landskapsvärden och företagsekonomiska kostnader.Småskriftserien no. 48, Department of Economics and Statistics, Swedish Universityof Agricultural Sciences, Uppsala.

Gren, I.-M., 1995. The value of investing in wetlands for nitrogen abatement. EuropeanReview of Agricultural Economics 22, 157–172.

Gren, I.-M., 1999. Value of land as pollutant sink. Ecological Economics 30, 419–431.Gren, I.-M., Jonzon, Y., Lindkvist, M., 2008. Costs of nutrient reductions to the Baltic Sea—

technical report. Working Paper 2008:1. Department of Economics, SLU, Uppsala,Sweden.

2559I.-M. Gren, L. Isacs / Ecological Economics 68 (2009) 2549–2559

Gundimeda, H., Sukhdev, P., Sinha, R., Sanyal, S., 2007. Natural resource accounting forIndian states — illustrating the case of forest resources. Ecological Economics 61,635–649.

Hamilton, K., 1994. Green adjustments of GDP. Resource Policy 20, 155–168.Hamilton, K., 1996. Pollution and pollution abatement in the national accounts. Review

of Income and Wealth 42, 13–33.Hamilton, K., Clemens, M., 1999. Genuine savings rates in developing countries. The

World Bank Economic Review 13 (2), 333–356.Hamilton, K., Dixon, J., 2003. Measuring the wealth of nations. Environmental

Monitoring and Assessment 86, 75–89.Harris, M., Fraser, I., 2002. Natural resource accounting in theory and practice: a critical

assessment. The Australian Journal of Agricultural and Resource Economics 46 (2),139–192.

Hasund, K., 1997. Documentation of Three Contingent Valuation Surveys on Preserva-tion of Landscape Elements of Agricultural Land. Småskriftserien no. 107.Department of Economics. Swedish University of Agricultural Sciences, Uppsala.

Heal, G., Kriström, B., 2005. National income and the environment. In: Mäler, K.-G.,Vincent, J.R. (Eds.), Handbook of Environmental Economics, vol. 3. Elsevier B.V,New York.

Hecht, J., 2007. National environmental accounting: a practical introduction. Interna-tional Review of Environmental and Resource Economics 1, 3–66.

Helcom (Helsinki Commission), 2007. An approach to set country-wise nutrientreduction allocations to reach good marine environment of the Baltic Sea. HelcomBSAP Eutro Expo/2007. Helsinki Commission, Helsinki. Finland.

Howarth, R.B., 2007. Towards an operation sustainability criterion. Ecological Economics63, 656–663.

Hultkrantz, L., 1992. National account of timber and forest resources in Sweden.Environmental and Resource Economics 2 (3), 284–305.

Isacs, L., 2007. Monetary green accounts at the regional scale in Sweden. Report,Department of Economics. Swedish University of Agricultural Sciences, Uppsala,Sweden.

Jämttjärn, J., 1996. Den svenska skogens rekreationsvärde. Småskriftsserien nr 109.Department of Economics. Swedish University of Agricultural Sciences, Uppsala,Sweden.

Kusler, J., Kentula, M., 1990. Wetland Creation and Restoration. The Status of Science.Island press.

LUSTRA, 2002. Land use strategies for reducing net greenhouse gas emissions. ProgressReport 1999–2002. December 2002. Swedish University of Agricultural Sciences,Uppsala, Sweden.

Mäler, K.-G., 2000. Development, ecological resources and their management: a studyof complex dynamic systems. European Economic Review 44 (4–6), 645–664.

Morén, A.-S., 2003. Kolsänkor – en av lösningarna. Miljötrender 3, 6–7.

Pearce, D., Atkinson, G., 1993. Capital theory and the measurement of sustainabledevelopment. An indicator of weak sustainability. Ecological Economics 8, 103–108.

Peskin, H., Angeles, M., 2001. Accounting for environmental services: contrasting theSEEA and the ENRAP approaches. Review of Income and Wealth 47 (2), 203–219.

Repetto, R., Maghrath, W., Wells, M., Beer, C., Rossini, F., 1989. Wasting Assets: NaturalResource in the National Income Accounts. World Resources Institute, Washington,DC.

SCB, 2006. Regionala räkenskaper 1993–2003. Statistiska meddelanden NR 13 SM 0601.Sveriges officiella statistik. Statistics Sweden, Stockholm.

Seijerstad, A., Sydsaeter, K., 1987. Optimal Control Theory with Economic Applications.Elsevier Science Publishers, Amsterdam, the Netherlands.

SEPA (Swedish Environmental Protection Agency), 2004. Rikare mångfald och mindrekväve - Utvärdering av våtmarker skapade med stöd av lokala investeringsprogramoch landsbygdsutvecklingsstöd. Report 5362. SEPA, Stockholm, Sweden.

SEPA (Swedish Environmental Protection Agency), 2006. Nationell strategi förmyllrande våtmarker – underlagsrapport. SEPA, Stockholm, Sweden.

Skånberg, K., 2001. Constructing a partially environmentally adjusted net domesticproduct for Sweden 1993 and 1997: a presentation of the methodological steps andempirical data. Working Paper no. 76. National Institute of Economic Research.

Sundberg, S., Söderqvist, T., 2004. The Economic Value of Environmental Change inSweden: A Survey of Studies. Beijer International Institute of Ecological Economics,Stockholm.

Sydsaeter, K., 1978. Matematisk Analyse. Universitetsförlaget, Oslo, Norway.Tarrasón, L., Fagerli, H., Jonson, J.E., Simpson, D., Benedictow, A., Klein, H., Vestreng, V.,

2007. Transboundary adicification, eutrophication and ground level ozone inEurope in 2005. EMEP Status Report: August 8, 2007. Norwegian MeteorologicalInstitute, Bergen, Norway.

Termansen, M., Zandersen, M., McClean, C., 2008. Spatial substitution patterns in forestrecreation value. Regional Science and Urban Economics 38, 81–97.

Turner, R.K., Paavola, J., Cooper, P., Farber, S., Jessamy, V., Stavros, G., 2003. Valuingnature: lessons learned and future research directions. Ecological Economics 46,493–510.

United Nations, 2003. Integrated Economic and Environmental Accounting. http://unstats.un.org/unsd/envaccounting/seea2003.pdf.

Vincent, J., 1997. Resource depletion and economic sustainability in Malaysia.Environment and Development Economics 2, 19–37.

Vouvaki, D., Xepapadeas, A., 2008. Changes in social welfare and sustainability: theissues and empirical evidence. Ecological Economics 67 (3), 473–484.

Weitzman, M., 1976. On the welfare significance of national product in a dynamiceconomy. Quarterly Journal of Economics 90 (1), 156–162.

Weitzman, M., 2001. Gamma discounting. American Economic Review 91 (1), 260–271.