Social net benefits from aquaculture production: A comparison of net cage cultivation and recirculating aquaculture systems

Abstract This paper applies cost-benefit analysis to assess social and private net benefits from rainbow trout and European whitefish aquaculture under recirculating aquaculture system (RAS) and marine net cage technologies. In addition to private investment and operational costs, we include eutrophication damage from nutrients and value the fish produced by its producer price. The assessment is made in terms of annualized present value of net benefits. We find that net cage production outperforms RAS by a wide margin for rainbow trout production. For the European whitefish, RAS narrows this gap considerably, but net cage production still receives higher net benefits. We extend our discussion to the specific challenge of the Weser ruling of the EU’s Water Framework Directive, which effectively hinders expansion of aquaculture production, and examine whether one-time offsets would provide aquaculture the possibility to expand production. We find that one-time offsetting provides higher net benefits compared to not offsetting with both technologies but is easier to execute for RAS.


Introduction
During the past 60 years, world consumption of fish has increased more rapidly than the consumption of meat from all the terrestrial animals combined.Given that one-third of the world's marine fish resources are overutilized beyond their biologically sustainable levels, aquaculture has enabled the increased consumption (FAO, 2018).The growing demand for fish and other seafood creates incentives to further increase aquaculture production (Asche & Smith, 2018). 1 Aquaculture provides technologically managed growing conditions for fish stocks on restricted water areas capable of ensuring a large supply of fish.
Unfortunately, large aquaculture production systems, typically built-in natural water areas, can also cause harmful environmental impacts.Constructing aquaculture farms may deteriorate the state of natural ecosystems; change the hydrological patterns of the water bodies; and cause salinization, acidification, and pollution, making the area or the water resource unsuitable for alternative purposes.Aquaculture releases nutrients to the waterbodies in the form of fish excretions and, to a smaller extent, as uneaten feed, inducing eutrophication (Nielsen, 2012;Talbot & Hole, 1994).Intense aquaculture has led to increasing problems with pathogens, parasites, and pests (Overton et al., 2019).The misuse of therapeutants used to prevent these may in turn cause harm for the ecosystem as well as the workers and customers (Naylor et al., 2021).The degree of harmful environmental impacts differs with the chosen technology, as well as the adopted practices.
Degradation of the water bodies is also harmful to aquaculture via weakened growing conditions for fish.Thus, improving environmental performance is also ultimately vital for fish farming as well (Nadarajah & Flaaten, 2017).Nevertheless, private incentives toward clean environmental practices in aquaculture are not strong enough from society's point of view.Therefore, societies impose regulations on aquaculture to ensure that it meets the growing demand for fish in an environmentally sustainable way (Osmundsen et al., 2017).Command and control measures, for instance in the form of specified production licenses, regulate both nutrients and the volume of production, thereby also affecting the growth potential of the sector (Nadarajah & Flaaten, 2017).
The need to simultaneously expand aquaculture production and to improve water quality creates an obvious tension requiring a positive solution.This tension is evident in the Baltic Sea.The Baltic Sea is a shallow water body with brackish water, and if the fish farm is located poorly, nutrients from fish farming contribute to eutrophication, which is caused predominantly by terrestrial sources, such as agriculture, forestry, municipal wastewater, and industry (HELCOM, 2018).Despite forming a minority of the loading, fish farming is affected by having its licenses restrained in renewal procedures.Farm-per-farm designed environmental permits impose an upper limit on fish feed use.Together with improved feed efficiency, this system has reduced loads from fish farms but has also prevented the expansion of the aquaculture industry, demonstrating that the current permitting process has not provided the best incentives for developing new technologies and abatement processes.Shifting to loadbased incentives is clearly needed. 2 The need for incentive-based policy is also emphasized by recent EU legislation.The Water Framework Directive (WFD) promotes water quality improvements in the EU.The so-called Weser ruling (2015) has given WFD a higher role, as it tightens the permitting process.Weser ruling refers to a ruling by the Court of Justice of the European Union given in 2015 regarding the project of deepening of the River Weser for navigation.According to the ruling the member states are to prohibit projects that deteriorate the status of a waterbody or hinder achieving its good status.Deteriorating is considered to start in the surface water as soon as one indicator shows signs of weakening, even if the overall status itself does not change (Soininen et al., 2019).Given that most of the Baltic Sea is affected by eutrophication (HELCOM, 2018), the possibilities for expanding aquaculture production have grown more limited, calling for more environmentally beneficial production technologies and abatement measures. 3 While traditional net cage production can still be improved, the technologically evolving recirculating aquaculture systems (RAS) present an interesting opportunity in production technologies with less nutrient discharges.RAS rears fish in a more enclosed system, thus minimizing water use and contacts with the environment, allowing for high control of nutrient discharge.Furthermore, RAS prevents the spread of parasites to wild fish and provides safe farming of non-endemic species (Ebeling & Timmons, 2012).In addition, RAS facilities can be located close to markets, and given that culturing takes place in indoors with full control over the water temperature, the weather conditions do not affect maximal feed conversion ratio (FCR) and faster growth (d 'Orbcastel et al., 2009;Tal et al., 2009).
Currently, RAS covers 1.5%-2% of the farmed fish production in Europe (EUFOMA, 2020), and its expansion is hindered by the high-cost structure and immature technological solutions.Despite the current state of affairs, many countries, such as Norway, the US, Denmark, Canada, China, France, Switzerland, South Korea, and South Africa, have shown great interest in land-based salmon farming, especially in RAS facilities (Bjørndal & Tusvik, 2019;Craze & Culeac, 2020).There are still many unresolved issues relating to this new technology.RAS facilities have significantly higher investment costs, and despite the efficiency provided by optimal culturing conditions, operational costs are likely to rise over those of the prevailing cage-based production (Bjørndal & Tusvik, 2019).Furthermore, RAS also uses a large amount of electricity and typically discharges to more shallow and sensitive water areas on or near land.In addition, off-flavors may occur in RAS production, giving the fish a musty and earthy taste and odor.Such undesired characteristics affect consumer demand.Depuration-that is, keeping the fish in clean water prior to harvesting to remove the unwanted flavor-helps but induces additional costs (Lindholm-Lehto et al., 2020).For production to be sustainable requires achieving environmental, economic, and social sustainability (Frankic & Hershner, 2003).
Unfortunately, economic studies focusing on aquaculture and the comparison of alternative technologies are scarce (Anderson et al., 2019).Answers are needed to the following questions, among others: How do the higher production costs of RAS relate to its lower nutrient loads when environmental impacts are valued in monetary terms?Furthermore, how do they compare with traditional net cage production?These questions constitute the research problems addressed in this paper.
We compare RAS technology and traditional cage-based marine aquaculture using social cost-benefit analysis (CBA).While net cage cultivation is a mature technology, RAS is hardly so.Individual RAS solutions differ greatly and the variation in costs is large (for Finland, see Vielma et al., 2022).Therefore, we develop a generic model of RAS and assess investment and production costs, drawing on advanced and well-working solutions.In addition, we assess how problems in achieving production goals affect costs and profits.We value the produced fish according to its producer price, representing the price received by the fish farmer as a difference to the market price paid by consumers.Investment and operation costs, excluding transfer payments such as taxes, constitute the private production costs.We add environmental damage caused by nutrient loads to private production costs and use the resulting annualized present value of social net benefits to compare RAS to net cage technology.Non-nutrient-related environmental impacts are not in the scope of this paper.
We extend our discussion to the challenges that the Water Framework Directive and the attached Weser ruling cause aquaculture in the Baltic Sea and examine whether one-time offsets (nutrient compensations) would provide a solution to promote aquaculture.One-time offsets refer to the possibility that a regulated company offsets some or all loads by buying the required reduction from other sources (see King & Kuch, 2003, Shabman & Stephenson, 2007, and Woodward & Kaiser, 2002 for the experience from the US and Canada, and Shortle et al., 2021 for a broader discussion).There are both land-based and onsite methods to compensate the nutrients released by aquaculture production (Filippelli et al.,2020;Racine et al., 2021;Soininen et al., 2019;Stephenson et al., 2010).As a cost-efficient method, gypsum amendment of arable fields was chosen as the offsetting method used in our analysis (Ollikainen et al., 2020).This possibility may open the path for expanding production without deteriorating the quality of water.

Key features of net cage and RAS cultivation
Fish can be farmed by using production systems differing by intensity, life stage of the fish, and interaction with the environment.The technologies in the scope of this paper, net cage and RAS are both intensive, meaning the stocking densities are quite high and require, for example, external feeding and disease control measures.However, whereas the stocking densities in net cage farming are typically less than 25 kg/m 3 , in an RAS facility rearing rainbow trout, they can be up to 100 kg/m 3 or even above.In the analysis, the fish are reared from fingerlings to market size under both technologies.The chosen production technologies differ in openness; net cages are open systems with direct interaction with the surrounding waterbody, while RAS facilities are more closed systems requiring water inlets and outlets (Lekang, 2007).

Net cage farming
Cage culturing has found its place as a major aquaculture technique around the world, especially in salmon farming (Asche & Bjorndal, 2011;Lekang, 2007).Net cages are typically located in the sea.Therefore, when constructing a net cage aquaculture system, one must consider multiple site-specific physio-chemical characteristics, and the system must be designed in such a way that it would withstand all possible marine forces simultaneously (Kumar & Karnatak, 2014).Proper design of the equipment is particularly important in the Baltic Sea, given its unique circumstances, the annual freezing of the sea, and the movements of the ice (Kankainen & Mikalsen, 2014).
Net cages consist of frames, nets, and mooring systems, which come in many shapes and sizes made from various materials.The frame of the cage is often made from flexible (polyethylene) plastic pipes, which can be filled with floating material and used as a working platform (Lekang, 2007).A circular shape is preferred as it requires less constructing material and increases durability (Kumar & Karnatak, 2014), while the space is best for letting the fish swim in a natural way.In open sea areas, cages can be larger, creating economies of scale (Asche, 2013).The net should hold its form in the water and be durable against weathering.The characteristics of both the nets and the fish also affect the stocking density.The mesh size is adjusted according to the size of the fish species (Piccolotti & Lovatelli, 2013).Synthetic materials are typically preferred due to their high breaking strength and uniformity (Lekang, 2007).The setup is held still with a mooring system including anchors, mooring grid, and ropes to connect them.
Feeding may vary from more laborious hand feeding with blowers or dispensers; to demand feeders, where fish are accustomed to collect the feed by themselves; to automatic feeders and optimal feeding programs utilizing an electrical supply; and ultimately to comprehensive feeding systems performing all the phases, from serving to distributing and monitoring (Lekang, 2007). 4 Recirculating aquaculture system (RAS) Recirculating aquaculture locates on land, typically inside facilities.The fish are reared in tanks with controlled conditions adjusted to be optimal for the species in question.This enables well-designed feeding and fast growth of the fish.The use of water is low; in general, over 90% of the water is purified and reused (Blidariu & Grozea, 2011).However, because pumps are needed to circulate the water and multiple pieces of equipment to purify it, the production requires considerable amounts of equipment and energy.
Fish generate carbon dioxide with their respiration, which requires special attention, since accumulating CO2 causes stress to the fish.Depending on the stocking density, aeration or oxygenation may be necessary because the first problems when intensifying the RAS are likely to emerge with oxygen.Air can be transferred to water by bubbling it through the water or in turn, directing water drops through the air (Ebeling & Timmons, 2012).Disinfecting the water with underwater ultraviolet lamps is an option for even, thorough purification.Alternatively, ozonating is typically used in hatcheries and with small vulnerable fish but is increasingly favored in all types of RAS farming (Bregnballe, 2015).
When it comes to the water treatment process, large particles of solid waste, such as uneaten feed, can be removed from the tanks by draining, while mechanical filters and settling tanks are used for removing the smaller particles (Bregnballe, 2015).Afterward, water is channeled through a biological filter as the finest particles and dissolved compounds, phosphates, and nitrogen pass the mechanical filter.Phosphates are not toxic, but nitrogen in its form of free ammonia (NH3) is toxic for the fish.The biological filter, i.e., nitrifying bacteria, converts nitrogen from ammonia to nitrate in a natural process called nitrification.Nitrification decreases water pH, thus requiring the use of, e.g., alkalinity-and pH-increasing chemicals such as lime (Ebeling & Timmons, 2012).Solids removal is a critical phase of the process because when it is insufficient, the solids can block the biofilter or decrease the share of favorable nitrifying bacteria.A blocked biofilter does not perform the nitrification as it should, and fatal compounds of ammonia and nitrite remain in the culturing water, endangering the fish (Badiola et al., 2012).
RAS farms typically have two primary discharge flows.A smaller volume consists of primary sludge originating from sedimentation systems (inside and outside fish tanks), particle filtration, and system backflush and has high contents of phosphorus and organic matter.A larger volume of discharge, often called system overflow, has a lower solids level but contains most of the nitrogen discharge as nitrates (van Rijn, 2013).These two streams, "primary sludge" and "overflow," are typically processed separately to reduce RAS nutrient discharge.
Primary sludge is further concentrated by sedimentation or filtration processes, typically with the aid of coagulants and flocculants also used at municipal wastewater treatment plants (WWTPs).Chemicals are mixed with the primary sludge, and then concentrated sludge is separated by belt filters, flotation, or screws.Concentrated sludge can be composted, used in a biogas process, spread to fields, or led to WWTPs.Depending on the regulation, overflow and the clarified water from the sludge thickening can be discharged without further treatment.However, nitrogen removal would remain poor, and further denitrification process is typically required.Recently, woodchip reactors have proven to be reliable and cost-efficient at nitrogen removal compared to technologically more complicated denitrification processes (von Ahnen et al., 2018).
Controlling and monitoring is also needed as maintaining the temperature and water quality parameters optimal can be challenging with intensive water recycling rates due to the interactions between the changing parameters (Murray et al., 2014).

Framework and data
We apply social cost-benefit analysis (CBA) to assess the private and social benefits of the two technologies, including the negative water externalities from fish farming.CBA is a useful tool for social decision-making when society compares alternative projects and plans (Boardman et al., 2017).Costs and benefits are determined over the life span of the project, and the discounted net value is typically used to assess investments.Formally, the net present value (NPV) is defined by the following formula for a project, which takes place n periods for a discount rate r.
In the NPV formula, investment costs are denoted by C 0 ; they occur immediately, while the annual production costs C t and revenues B t start running from year 1 or 2, depending on the production technology (Boardman et al., 2017, p. 141).Cost component C d denotes the social cost of eutrophication, expressed in monetary units.Note, though, that we will not include C d when determining the private profitability of aquaculture.We use a time span of 20 years for the facilities according to their life span and a discount rate of 3%.
We gathered information on the costs of fish farming from the Finnish aquaculture sector, and therefore assess the net cage and RAS technologies by using the most relevant species in the Finnish food fish markets, rainbow trout (Oncorhynchus mykiss) and European whitefish (Coregonus lavaretus, referred to here as whitefish).As the dominant species, rainbow trout covers approximately 95% of Finnish aquaculture production in volume, while the share of European whitefish is around 4% (OSFa, 2021).Rainbow trout is typically reared in intensive systems.Artificially reproduced eggs are incubated in nurseries.After hatching and the yolk-sac stage, fish are moved to grow-out production facilities.The first grow-out period may occur in inland freshwater or indoor facilities, after which fish are transported to sea for the remaining grow-out periods (Mustaj€ arvi, 1999).The market weight of 1.5 to 3 kg is reached after one or two grow-out periods in the sea (Janhunen et al., 2019).European whitefish are salmonids as well, and therefore, the final grow-out production is quite like rainbow trout culturing.However, the growth is slower than with rainbow trout.They are also less affected by stress, require a different type of feed, and weigh less.Like rainbow trout, they reach market weight in two to three years, but their size is approximately half of similar-age rainbow trout (Koskela et al., 2002).The producer price of farmed whitefish has been twice the price of rainbow trout in Finnish food fish markets, which provides incentives for the producer to focus on whitefish farming.However, Kankainen et al. (2014) report that the whitefish market price is very volatile because of the thin market; production has been annually about 1,000 tonnes in total in Finland.Therefore, we use 200 tonnes as the targeted annual production yield for whitefish.

Investments
Investment costs differ between technologies and production capacities.We chose 1 000 tonnes as the targeted annual production volume for the rainbow trout dominating the Finnish food fish markets and 200 tonnes for whitefish.We assume that investment costs occur at the beginning of the production period.In cage farming, feeding is performed from a boat.The investment costs include all necessary equipment such as net cages and their mooring, vessels for feeding and monitoring, construction and cargo, as well as a gutting facility on land (Kankainen & Mikalsen, 2014).Investment costs of gutting facilities are assessed based on discussions with experts in the field.The life span of equipment is expected to be 5 years for net cage equipment used in the sea, 10 years for the vessels and 20 years for the gutting/harbor area on land.Table 1 condenses the information on investment costs.
Establishing an RAS facility requires suitable property for fish farming.Prepping the land as well as construction and pipe work on the building determines a relatively high starting price for the production.The RAS facility also requires a considerable amount of technology, which increases the investment costs further compared to the net cages.The facility includes tanks for rearing, mechanical filters and biofilters, devices for oxygenation and ozone sterilization, pumps, feeders, and backup generators.RAS facilities do not represent mature technology.Thus, practical solutions and the equipment used may differ greatly.Consequently, investment costs reflecting these choices differ as well.The investment costs here illustrate a generic version of a RAS facility.We assume that the production consists of individual 100-tonne modules placed optimally inside one facility.For a large-scale rainbow trout facility, we altered the initial investment costs from Wright and Arianpoo (2010) by utilizing a funding application to the European Maritime and Fisheries Fund (EMFF) regarding construction costs of an existing facility in Finland, since the construction costs are higher than in several other countries.We also scaled the facility to be suitable to produce large rainbow trout.Private information from an entrepreneur from the field is in line with the level of the hypothetical investment costs.
For a smaller-scale European whitefish facility, there was a difference between the literature and recent information from the field.Therefore, we used an average.The equipment costs in Wright and Arianpoo (2010) are given in dollars; thus we converted them to euros using the conversion rate of 0.92 (16 April 2020).We also included a gutting facility to the investments.The estimated life span for the technology is 10 years and for the tanks and gutting facility, 20 years.To reach high nutrient reduction and to avoid considerable wastewater disposal costs due to high volumes of discharge water, we assume the RAS facilities to establish their own wastewater treatment units consisting of phosphorus removal in solids and nitrogen removal in the woodchip reactor.
A comparison of Tables 1 and 2 reveals a great difference in the investment costs between the two technologies.RAS investment costs for rainbow trout cultivation are around 3.5 times higher than those of net cage production.Whitefish farming in net cages is assumed to require an investment for a similar vessel, thus lacking the economies of scale received by larger-scale rainbow trout production.The investment costs are still around 2.65 times higher for RAS.

Production costs
Production costs are paid every year and can be divided into costs for livestock, i.e., fingerlings; feed; energy; labor; repair and maintenance; and other operating costs (STECF, 2018).
Fingerling costs depend on the initial weight, price, mortality, and growth rate of the fish.In the Finnish market, 20 g rainbow trout fingerlings can be bought approximately for e20/kg, whereas whitefish fingerlings of the same size cost e30/kg.The fingerlings reared in net cages are commonly vaccinated, which adds to the costs by approximately e0.18 per fish.
We assume that the fish reach the market size within 24 months in net cages (Kankainen et al., 2020), whereas fish reared under controlled rearing conditions in RAS facilities are ready within 12 months. 5The expected harvest weight is 2.3 kg for rainbow trout and 1 kg for European whitefish.We expect that the fish reared in RAS facilities are also kept in fresh water prior to gutting to remove the off flavors.For rainbow trout, this depuration takes 2 weeks, and for more sensitive whitefish, 4 weeks.A mortality of 10% is applied for fish reared under both technologies.
Feed costs depend on the price of the feed and the feed conversion ratio (FCR) of the fish.Feed prices are calculated based on the prices of commercial feeds available and the estimated use of certain feeds because the fish are fed with different-sized pellets based on their size.For net cage production, at the time of writing, the price of rainbow trout feeds varies from e1.28 to e1.50/kg and for whitefish from e1.29 to e1.43/kg.Fish reared in RAS facilities require specific types of feeds, the prices of which vary between e1.34 and e1.44/kg for rainbow trout and e1.41-e1.53/kgfor whitefish (Raisioaqua Oy, n.d.).We expect the FCR to be 1.2 for both species under both technologies, meaning all the fish will grow 1 kg after being provided with 1.2 kg of feed.
Energy costs are derived from the consumption and price of energy.Energy consumption is low in net cage facilities, where electricity is needed mostly for cooling the fish after slaughter.Fuel is needed for operating the boats, which adds to the energy costs (Lepp€ anen et al., 2017).In recirculating aquaculture systems, energy is needed for pumping, heating, oxygenating, and, for example, ozonating the water; the need may vary significantly between facilities (Badiola et al., 2018).Energy consumption per kg of produced fish also varies between different fish species.As European whitefish is more sensitive to increased concentration of carbon dioxide, slightly more energy is needed for the removal of the gas.The price is set according to the average price from 2018 and 2019 (Statistics Finland); for net cages, the average is based on only the winter months.
Labor costs are a product of unit labor costs and the amount of labor needed.For net cage aquaculture production, we assume that 6 people are needed to produce 1,000 tonnes of rainbow trout and 3 people to produce 200 tonnes of whitefish annually.Respectively, 10 and 6 people are assumed to be needed for RAS production.Salaries of the workers are calculated based on STECF (2018) data and appear to be slightly higher in RAS production.
Because repair and maintenance costs do not directly correlate with production volume, they can be calculated as a share of initial investment costs (O'Rourke, 1996).We employ 1.5% in line with a similar study (Bjørndal & Tusvik, 2019) considering land-based aquaculture of salmon in Norway.
Other operation costs include the rest of the operation costs.Of these, medicines, fuel, transportation, ice, gutting, and packing correlate with the production volume, while licenses, environmental monitoring, bookkeeping and financial administration, marketing, infrastructure and equipment insurance, office electricity and heat, traveling to seminars and meetings, road fees, and research and development expenses are fixed.To estimate these costs, we used the default values of a calculating tool by Kankainen et al. (2014) with the help of expert opinions.For RAS production, the operation costs also include oxygen used in oxygenation as well as the use of the woodchip reactor for nitrogen removal.In addition, chemicals are needed for pH adjustment and sludge water treatment.Furthermore, financiers may require business interruption insurance.Other costs are specified in Appendix A.
The environmental damage costs of nutrients released from aquaculture derive from the nitrogen (N) and phosphorus (P) concentration of the feeds.Some of the nutrients are absorbed in the growth of the fish, while the released nutrients may induce eutrophication.For net cage production, the nutrient load of grow-out is on average 40 g of N and 4.4 g of P per kg of fish produced (Ministry of the Environment, 2020).For RAS facilities, we derive nutrient loads directly from the use of the feeds using the reduction rates of 90% for phosphorus and 80% for nitrogen.
We employ damage valuation estimates for eutrophication in the Baltic Sea.The estimated monetary value varies between studies, chosen methods, and analyzed sectors (for a comprehensive discussion on abatement costs in all sectors, see Gren et al., 2008; and for marketable nitrogen permits in aquaculture, see Nielsen et al., 2014).The damage cost of N used in our calculations is e9/kg (L€ otj€ onen & Ollikainen, 2019).This information can be converted when calculating the environmental cost of P because according to the Redfield ratio, algae use in their growth 7.2 units of N for one unit of P (Lankoski et al., 2006).This yields the damage cost of P as e64.8/kg.
The benefits of cultivation are simply the selling revenues defined by the producer prices of the fish.The producer prices of fish vary between years; therefore, we used averages of the past five years.The producer price of reared whitefish has been approximately double relative to the producer price of reared rainbow trout (OSFa, 2021).In addition to fish, Finnish marine fish farmers also sell roe of rainbow trout, which is included in the net cage farming of rainbow trout.Roe production increases revenues by approximately 11%.Due to the fast-growing cycle under RAS, the fish do not meet the maturity prior harvesting.Thus, there are no roe to bring revenues to RAS.All prices are given in terms of e/kg for gutted fish, meaning the share of gutting waste is extracted from the final weight of the produced fish (Table 3).
Despite the high investment and production costs of RAS technology, it has an advantage when it comes to nutrient emissions.With each kilogram of fish reared in RAS instead of the sea, nutrients are prevented from eutrophicating the waters.For this reason, we also include in the analysis reduction in the social costs from avoided nutrient releases in the sea for RAS.
Table 4 shows the amount of avoided nutrients and their value in monetary terms, in the previously presented manner.As RAS bears additional costs in exchange for releasing fewer nutrients, costs can be compared to the reduced social costs.Deducting the value of the avoided nutrients from the operational costs, the costs of rearing rainbow trout in an RAS facility reduce by e0.54/kg and whitefish by e0.52/kg.Next, we will present the results using the actual production costs, the social net benefits, and the social net benefits in the case where the reduction in social costs due to avoided nutrients are taken into account.

Private and social net benefits of aquaculture
Calculation of the NPV of investment and production yields net present values to the technologies and fish types reported in Table 5.A dominant  outcome is that producing rainbow trout in RAS is not profitable, while whitefish brings positive, albeit much smaller profits relative to net cage cultivation.
We next disaggregate these results to individual cost and revenue items for private profits and social net benefits.These items are discounted to the first period and divided by respective produced amounts.This approach mimics the net present value calculation and yields the same aggregate net revenue, as presented above.Table 6 reports investment and production costs, environmental damage costs, and revenues in their net present values over per 1 kg of produced fish.Table 6 defines the net benefits assuming that the facilities produce the targeted amounts.If a facility does not achieve the targeted, ideal production, the average costs will be higher, and especially so for an RAS, which has high investment costs, as discussed later.The table facilitates a comparison of both production technologies for rainbow trout and whitefish.
Table 6 shows again that RAS facilities have higher investment and production costs than net cage technology for both rainbow trout and whitefish.For the chosen capacity, production costs of rainbow trout are 1.7 times higher, and when including the additional costs of renewing some of the equipment, investment costs are around 2.6 times higher in RAS production than in net cage production.For whitefish, with smaller capacity, production costs are 1.7 times higher, whereas investment costs are around 1.8 times higher than in net cage production.The difference between the technologies is considerable, despite the fact that for net cage facilities a larger share of the initial equipment investment must be renewed within the 20-year production cycle.Thus, the difference in unit production costs suggests higher profitability for net cage production.In contrast, RAS production outperforms net cage production in environmental damage estimates.They are 4-4.9times higher in net cage production; the amount depends on the species but does not change the difference in favor of RAS.
Table 7 illustrates the changes in social production costs, when the reduction in social costs from avoided nutrients are taken into account.
The social production cost of rainbow trout farming decreases from e3.43/kg to e3.01/kg.Accordingly, the social production costs of whitefish farming decrease from e5.03/kg to e4.63/kg.The NPV of selling revenue of rainbow trout for RAS is e2.94/kg, suggesting that the running annual net benefits without investment costs are slightly negative.To make NPV positive would require radically lower investment and variable costs than those reported in Table 2.For whitefish farming, taking the reduction in social costs due to the avoided nutrients into account, the NPV of social net benefits increase by five-fold, still leaving it far from the net cage arming of whitefish.
Both technologies receive the same producer price for all fish, but rainbow trout production under net cages also benefits from the roe production increasing the revenues.For private producers, net cage production is profitable with both species, while RAS reaches net profits only with the higher-value whitefish.Accounting for environmental damages for social net benefits considerably reduces this difference but does not change the ranking.In any case, a recirculating aquaculture system generates positive NPV with whitefish in our calculations, suggesting that focusing on highervalue fish is economically sensible when the production process goes as planned.
Table 6 is based on the assumption that both net cage farming and the generic RAS facility achieve their production goals.Under this assumption, average production cost is e2.68/kg for net cage farming and e4.51/kg for generic RAS farming (assuming 10% mortality of fingerlings for both technologies).For the RAS farming we need to account for the difference between theory and actual practice of RAS production in the Finnish data from RAS companies.It is typical that the current existing facilities have not reached full production volume.At present Finnish RAS-companies producing rainbow trout have at their best reached 65% of their planned production capacity.Along with prices, fish yields are the main factor behind economic performance (Ngoc, 2016).Lower fish yields mean the unit costs in Table 6 increase.Table 6 allows us to examine how different type of failures affect realized production costs.Consider first a fatal case in which a technological problem before harvesting decreases the final production yield by 40%.In this case the relative production costs would increase from the initial e4.51 to e7.52/kg of farmed fish.Problems during the initial stages of production would entail different costs.For instance, if a share of fish is lost as fingerlings, increase in production costs would be smaller, as variable costs such as feed costs would also decrease.
It is interesting to compare above costs to other studies.Using actual data from companies' balance sheets and financial statements Kankainen et al. (2020) assess average production costs in Finland using a formula (revenue-profit)/volume.It yields average production costs of e4/kg for net cage farming and e15/kg for RAS farming.Both figures reflect the realization of actual bottlenecks and harms in production and are higher than those in Table 5 because they already reflect the actual mortality and other possible negative shocks in production.Consider first net cage farming, Table 6 suggests roughly e2/kg lower average costs than Kankainen et al. (2020).For instance, assuming 33% mortality would make the figures identical.Unfortunately, the loss in yields in net cage farming is not known in Finland (see Bendriem et al., 2021, for technical and biological failures among net cage farms of Atlantic salmon in North America).
Turning to RAS, the difference with Kankainen et al. (2020) is large, roughly e8/kg.Interestingly though, Vielma et al. (2022) found that RAS companies suffer from high losses during growing stage and the best performance by an RAS company in Finland carry production costs of e8.9/kg when rearing rainbow trout.The company in question is producing 100 tonnes of rainbow trout annually and has been developing its production for 10-15 years.Costs reported in Table 6 under 40% loss of yields are quite close to the advanced facility reported in Vielma et al. (2022).Making them equal to Kankainen et al. (2020) would require 70% losses in fish yields.Thus, we conclude that some RAS facilities must have utterly failed in their production, making the average cost very high.See Appendix B for more realized costs and returns on Finnish RAS production.
Next, we determine the private profits and social benefits by accounting for the possible yield losses.The optimal production would reflect 90%-100% yields but roughly 60%-70% is the realized production volume currently achieved under Finnish RAS culturing.Table 8 shows how private and social net benefits change when the yield declines from the optimal 100%-50%.
Table 8 shows that receiving 90% of the planned production under RAS, the NPVs of social net benefits are negative with both species.The problem of not reaching full production capacity is a higher one for RAS facilities, but yearly variation in weather conditions also affects net cage farming.Net cage farming is more resilient against uncertainties.Interestingly, though, from the social perspective, even a 10% drop in fish yield decreases the NPV of rainbow trout net cage farming to e0.01/kg of farmed fish.In contrast, the private NPV is still e0.50/kg of farmed fish, and depending on the time of the losses, a decrease in production yield leads to a decrease in nutrient load as well.
In light of these results, we conclude that the prevailing marine net cage production is privately more profitable.Assessing environmental damage costs of eutrophication shows that social net benefits are still in favor of net cage production but with a narrower margin, thanks to the higher nutrient load of net cage production relative to RAS.Given the fact that RAS represents an evolving technology and environmental attitudes are strengthening, it is of interest to ask how changes in the key parameters impact the social net benefit comparison of the two technologies.We therefore provide sensitivity analysis concerning the expected key variables: discount rate, investment costs, damage valuation, and producer price (Table 9).
In the case of aquaculture, revenues come in the future, while significant investments occur immediately.Decreasing the discount rate to 2% gives more weight to the future stream of revenues.Thus, a lower discount rate increases the net present value of aquaculture, as here whitefish farming with both technologies and net cage farming of rainbow trout.However, since in our calculations rainbow trout farming under RAS is generating losses and a lower discount rate gives more value to these, the NPV actually decreases along the discount rate.Vice versa, increasing the discount rate has an opposite impact.
Changes in investment costs affect more RAS facilities with higher investment costs.However, as the operational costs of rainbow trout rearing under RAS are also so high, a decrease of 20% in investment costs does not make the production profitable.For the smaller-scale whitefish farming under RAS, an increase in costs of 20% also decreases the benefits to negative.
In case of increased valuation of waters, the monetized damage value caused by the nutrients would increase as well.An increase in the environmental damage values from the initial e9/kgN and e64.8/kgP decreases the social net benefits more from net cage farming due to the much larger nutrient releases.However, with the increase of 20%, the NPV of social net benefits remain higher from net cage production.
Producer price seems to be the most sensitive factor regarding the profitability of fish farming.A decrease of 20% decreases the NPV of whitefish farming under RAS to negative.The NPV of marine rainbow trout farming also decreases to negative despite the roe production.Still, an increase of 20% is not enough to make rainbow trout farming profitable under RAS.For a more comprehensive sensitivity analysis, we find the values for the given parameters, where the production yields zero profits.Table 10 presents the break-even values regarding investment costs and producer price with discount rates of 2%-4%.
Throughout the paper, we have pointed out that high production costs are a drawback of RAS.The net revenues are not enough to cover the operating expenses, not to mention the large investment of a rainbow trout RAS facility.The break-even values for the investment costs are negative, meaning that even if the RAS facility were free to construct, it should receive money to cover the running costs.A RAS facility growing the more valuable whitefish can in turn cover its operational costs and investments; however, when increasing the discount rate to 4%, the investment costs become too high.With a discount rate of 4%, the RAS facility rearing whitefish should have 2% smaller investment costs for it to break even.This suggests that the resilience against unexpected additional investment costs is much lower with RAS facilities than with net cage farming.
As in the sensitivity analysis, the break-even values show that rainbow trout farming is sensitive regarding the producer price.There has been variation in the annual producer prices, with an ascending trend.We used an average of 5 years, which is above the calculated break-even value; however, for example, in 2020 the average producer price for reared rainbow trout was e4.35/kg (OSFa, 2021).In our calculations, the given price would lead to the negative net present value of social net benefits with both technologies.Taking into consideration the roe production increasing the revenues of marine rainbow trout farming by approximately 11%, the NPV increases to positive.When producing rainbow trout in a RAS facility, the producer price should increase by 30%-32% depending on the discount rate, for the production to break even.Within the last 10 years, the highest average producer price has been e5.63/kg(2017) which is still 16%-18% lower than the break-even price.Another point of interest is how much higher a producer price should the producer yield from fish produced under RAS to receive net benefits equal to those from net cage production.In the base case situation, a significant price premium of 57% would lead to the same social net benefits of e0.33/kg of farmed fish.Despite whitefish being a valuable fish species in the Finnish aquaculture sector, its RAS production is also sensitive when it comes to producer price.With the given costs, the break-even values are quite close to the 5-year average used in our analysis with discount rates 2% and 3% and higher than the initial producer price with the discount rate of 4%.To reach net benefits of e2.17/kg equal to net cage farming, the RAS producer of whitefish should receive a price premium of 30%.
Our results are in line with previous studies (Bjørndal & Tusvik, 2019;Boulet et al., 2011), with production costs being higher under RAS.However, Liu et al. (2016) present RAS as financially viable from an operating perspective, because it generates a positive margin, and Boulet et al. (2011) concluded that despite RAS having higher production costs, it may still be profitable.Still, with the discount rate of 7%, both Liu et al. (2016) and Boulet et al. (2011) end up in negative NPVs.

Offsetting as a mechanism to cope with environmental impacts of aquaculture
The aquaculture industry is in a kind of stalemate in the Baltic Sea region.Given that most parts of the Baltic Sea are not in an ecologically good state, the Weser ruling causes challenges in the permitting process in those areas.This has made the expansion of traditional production difficult in some sea areas and requires comprehensive and expensive impact analysis.Thus, aquaculture lies in the confusing situation: societies wish to expand fish production, but the Weser ruling constrains growth opportunities.This situation raises a question.Does the Weser ruling change the relative profitability of the study's production technologies?
We next explore this issue by asking if aquaculture could pay and use nutrient reductions from other sources to compensate for its nutrient loads.This is a natural question, given the tight interpretation of the Weser ruling.The US and Canada provide experience from the use of offsets as a cost-efficient solution to cases in which abatement is very costly (Shabman & Stephenson, 2007).Looking at the technologies from this angle, we note that the two technologies differ in one important respect: RAS can be established on land where loads are released through a river to coastal waters of the sea, whereas net cage production takes place in the open sea.For RAS this means a broader set of possibilities to reduce nutrient loads to the sea; in addition to its own abatement, offsetting of loads also becomes possible.For net cage production offsetting is more difficult, because damage should be offset at the production site in the sea.Using filter-feeding organisms, such as mussels, in mitigating the negative impacts of net cage aquaculture near the production site may offer one solution but it must be ecologically and economically feasible (Maar et al., 2015;2020;Whitmarsh et al., 2006). 6Unfortunately, and unlike in the North Sea, the low salt content in the brackish Baltic Sea does not provide a good environment for mussel farming (Gren, 2019;NutriTrade, 2018).
RAS can offset the increase in loads by buying nutrient load reductions from farms and other nearby polluters so that loads to the same coastal water are reduced by the amount of increase caused by RAS production.This is not as easily possible for net cage operations, because they are located in outer seas, and reductions from external sources do not necessarily show up in the production site.This suggests that it may be easier for RAS farmers to use offset methods on land.
We illustrate the possibility of one-time offsets with some calculations for both technologies.We choose a gypsum amendment to arable fields as the offsetting measure since it has been proven to be cost-efficient in preventing phosphorus releases from agriculture.When applied to fields, gypsum increases the ionic strength and calcium concentration of the soil, condensing the soil and thus preventing the phosphorus runoff from the fields.Furthermore, gypsum is a relatively cheap, but temporary, way to abate the runoff (Ekholm et al., 2012;Ollikainen et al. 2020).
In our offsetting scenario, the aquaculture operators, both net cage and RAS, offset their nutrient loads by buying one-time offsets from agricultural farmers to match the estimated amount of nutrients derived from aquaculture production.We assume that they make the contract efficiently without any transaction costs.We present nutrient loads as phosphorus equivalents, and they are higher from net cage than from RAS production.Offsetting one kgP-eq.requires 0.78 hectares to be treated with gypsum.The accrued costs, e223/ha, are paid once, while the beneficial impact of gypsum is effective for five years.The time span is 20 years with the assumption that the first gypsum treatment is done in the first year, following another one every five years, the discount rate being 3%.To simplify the comparison, Table 11 also presents the cost variables in annual terms.
Offsetting nutrient loads decreases private profits in accordance with the initial nutrient load, thus for net cage production the impact is larger.When the nutrient load is offset, the difference between private profits and social benefits disappears.The results show that in the case of net cage farming, the environmental benefits of offsetting are higher than the costs of doing so.However, if the reduction cannot be verified in the production location, offsetting on land for net cage aquaculture is not feasible.For the RAS facilities, the benefits of on-land offsetting are positive, although less than a cent/kg of farmed fish.Nevertheless, if it enables aquaculture production in the first place, it surely is worthwhile to execute.

Conclusions and policy implications
We examined private and social net benefits from aquaculture production of rainbow trout and whitefish under recirculating aquaculture systems (RAS) and marine net cage technologies using cost-benefit analysis.Social net benefits were determined by including a eutrophication damage cost from nutrients to private investment and operational costs, and by valuing the fish produced by its price.We examined the possibility of using one-time offsetting of nutrient loads from production as a solution to growth obstacles stemming from the Weser ruling of the EU's Water Framework Directive.
Our main finding concerning net benefits is as follows.Private net benefits of net cage production outperform those of RAS by a wide margin and social net benefits by some margin for both cultivated fish species.An RAS provides positive net benefits only for whitefish production.If fish yield decreases to 80% from the optimal volume, cultivation of rainbow trout produces negative social net benefits for both technologies, and cultivation of whitefish yields positive social net benefits but only for net cage technology.According to the sensitivity analysis, changing damage value relating to eutrophication does not change the findings; however, even a slight increase of interest rate from 3% upwards makes the social net benefits from producing whitefish under RAS also negative.
Our findings concerning RAS differ from literature that found positive net benefits for this technology (e.g., Wright & Arianpoo, 2010).The study by Wright & Arianpoo reflects lower investment costs and fully optimized production processes, mostly in the North American context.Our results are based on cost information obtained from Finland and reflect a generic version of RAS.Our theoretical investment calculations for average costs in RAS are much lower than estimates obtained using data from financial statements.Given that our average costs are for full capacity, we concluded that current RAS facilities encounter high yield losses with the most advanced one experiencing a roughly 40% yield loss.The two technologies differ in a fundamental way.RAS is based on the idea that the amount of waste is minimized to interact with seawater, while net cage farming takes place in the seawater.comparison to net cage cultivation, RAS represents a kind of abatement technology.We approached this feature by taking into account the benefits from avoided nutrient releases and then added these benefits to RAS, but they did not change the overall picture.However, regulation is changing, as indicated by the Weser ruling.Interestingly, consumers preferences are changing as well.For instance, in Vancouver consumers are willing to pay a premium for RAS-cultivated fish (Vinci et al., 2015).More recent experiences on largescale salmon RAS production and related market prices have been gathered by Atlantic Sapphire production in Miami, Florida, but so far, such data are not publicly available.That type of tiered pricing may lie ahead in Europe as well.We determined the critical values of producer prices in terms of social benefits and found that RAS producers should receive a price premium of 57% for rainbow trout to make the social net benefits equal to the net cage.Such a price premium is unlikely to be received in Finland currently because rainbow trout competes with imported farmed Norwegian Atlantic salmon and the price of salmon determines the price of large rainbow trout (Guillotreau, 2004;Landazuri-Tveteraas et al., 2021).
We provided analysis on the possibilities of one-time offsets of nutrient loads from both technologies as a way of overcoming the obstacles of the Weser ruling.We found that social net benefits from aquaculture production increase by offsetting, and more so for the net cage, which has higher nutrient loads.From society's point of view, both technologies can afford one-time offsets for the expansion of production, but it is technically easier for RAS.Methods that reduce nutrients at the production site exists for both technologies, but for RAS located inland, they are easier to apply.In marine areas other alternatives, such as the role of nutrient-recycling fish feeds, could be evaluated (Nielsen et al., 2019).In any case, our analysis shows that authorities need to decide whether to allow offsetting as a solution to the tension between society's wish to expand production and to tighten environmental policy.
The current situation is experienced as unfair among the fish farmers, given the intensive efforts and success in reducing phosphorus loading from fish farms. 7This calls for a tighter and better-designed water policy toward agriculture.
All in all, aquaculture in the Baltic Sea (and other EU waters that are not in a good ecological state) faces challenges that go beyond traditional technological analysis.Water policies play an important role for this going forward and require new responses in aquaculture and policies promoting sustainable production.Stronger incentives for everyday environmental considerations can be created by shifting toward performance-based policies and sea area loading targets instead of the current plant-to-plant production regulation.The Danish freshwater aquaculture sector is an interesting example of such reform.When the novel farms using recirculating technology were introduced in 2005, they were allowed to increase their production in accordance with confirmed nutrient reductions.This led to a structural change into larger farms benefitting from economies of scale and vertical integration.Simultaneously, the environmental impact per kg of farmed fish as well as the production costs decreased.Using the Danish RAS technology in Finland would, however, require alterations (Nielsen et al., 2016).
Furthermore, a more comprehensive approach to water quality maintenance is needed, with an accounting of all sources of pollution.The reduction requirement for various sectors should be allocated by drawing on a cost-efficiency principle.These angles open up multiple issues for future research.

Notes
1. Asia has been providing 89% of farmed fish over the past two decades, China being the greatest producer.Simultaneously, many developed countries are currently importing significant amounts of fish (FAO, 2020).2. Finnish aquaculture has decreased its nutrient loading by 70%-80%.Its share of Finland's total nutrient loads in the Baltic Sea is 1%-2%.More than 60% of phosphorus loading in Finland comes from agriculture, showing no reduction over time.3. The unresolved tension between environmental protection and production may be one reason for the stagnation of growth in the European Union aquaculture sector despite efforts to boost it (Guillen et al., 2019).Between 2000 and 2014, the European Union invested e1.17 billion in its aquaculture sector.In 2016, the volume of production was 8% below that in 2000.EU is the largest importer of fish (FAO, 2020).4. The feed industry has developed feeding tables that produce information on optimal feeding as a function of temperature and fish size that leads to favorable growth and minimizes the feed conversion ratio.5.In net cage farming, the production is consisting simultaneously of first-and second-year fish, so that the annual growth equals 1 000 tonnes.In the last year, no fingerlings purchases are made, and feed purchases are halved only for the remaining second-year fish.6. Filippelli et al. (2020) have identified the economic and bio-physical conditions under which farming mussels would be a cost-efficient method in mitigating the impacts of excessive nitrogen in Danish agricultural catchments.
7. Considering the climate impacts, especially farmed bivalves (e.g., oysters and mussels) have low impacts in relation to their nutritional value.Compared to terrestrial animal source food, half of the seafood of which enough information was available, performed better than beef, pork, or chicken (Bianchi et al., 2022).By our calculations and accounting also for eutrophication damage, aquaculture provides proteins with social benefits that exceed that from beef by e40/kg for rainbow trout and e60 for whitefish in net cage farming, and by e20 and e40 in RAS (details available from the authors upon request).

Table 3 .
Data on production variables, costs, and prices.

Table 4 .
Reduction in social cost from avoided nutrients when using RAS instead of net cage technology.

Table 5 .
Net present values of private profits and social net benefits.
a The numbers may not match due to round ups.

Table 7 .
Social production costs of RAS including reduction in social costs from the avoided nutrients.

Table 8 .
Private and social net benefits with lower fish yields.

Table 9 .
Sensitivity analysis of social net benefits.

Table 11 .
Offsetting the nutrient load derived from aquaculture production.