You are on page 1of 17

Estimating Pollution Abatement Costs of Salmon Aquaculture: A Joint Production

Approach
Author(s): Yajie Liu and U. Rashid Sumaila
Source: Land Economics, Vol. 86, No. 3 (AUGUST 2010), pp. 569-584
Published by: University of Wisconsin Press
Stable URL: https://www.jstor.org/stable/27821444
Accessed: 25-09-2019 10:37 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

University of Wisconsin Press is collaborating with JSTOR to digitize, preserve and extend
access to Land Economics

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
Estimating Pollution Abatement Costs of Salmon
Aquaculture: A Joint Production Approach
Yajie Liu and U. Rashid Sumaila

ABSTRACT. Salmon aquaculture generates good sediments and water columns, on benthic
output (i.e., salmon) and bad output (e.g., pollu communities, and on some fishery resources
tion). A joint production function approach is (e.g., Milewski 2001; Levings et al. 2002;
applied to model both outputs simultaneously. Two
Brooks and Mahnken 2003; Naylor, Eagle,
environmental production technologies are specified,
and Smith 2003). For instance, pollution in
namely, regulated and unregulated technologies.
Two production functions with different mapping the form of nitrogen and phosphorus may
rules are applied. Pollution abatement costs are increase the risk of eutrophication, and alter
estimated based on a series of data from the species composition and phytoplankton
Norwegian salmon aquaculture industry. Results density in the water column (Statistics
indicate that pollution abatement costs vary among Norway 2006). Pollution in the form of
observations and models. On average, pollution organic matter may change sediment chem
abatement cost is estimated to be about 3.5% in istry, resulting in changes in sediment flora
terms of total farmed salmon production, and 6.5% and fauna in affected areas (e.g., Mazzola et
in terms of total revenue of farmed salmon. (JEL al. 2000; McGhie et al. 2000; Pohle, Frost,
D24, Q22) and Findlay 2001). Some impacts are
measurable near-field changes in sediments
I. INTRODUCTION and water variables that are sensitive to
organic matter and nutrient additions,
Salmon aquaculture is known for pro while some are far-field effects, which are
viding highly valuable products to the difficult to observe and measure, such as
growing seafood market, creating employ eutrophication and effects on food webs
ment and income opportunities as well as (Hargrave 2003). There is an extensive
foreign earnings (Tidwell and Alan 2001;
Garcia and Grainger 2005). However,
The authors are, respectively, postdoctoral fellow,
salmon aquaculture is currently under Department of Economics, Norwegian University of
scrutiny and criticism because it appears Science and Technology; and associate professor and
to generate negative economic and environ director, Fisheries Centre, University of British Colum
mental impacts (e.g., Naylor et al. 2000; bia. This paper is part of the first author's Ph.D.
dissertation conducted at the Fisheries Economic Re
Pauly et al. 2002). Pollution is one of the search Unit, Fisheries Centre of the University of British
environmental concerns associated with Columbia. The authors would like to thank Dr. Carl
salmon aquaculture. Pollutants from salm Pasurka from the U.S. Environmental Protection Agency
for helping with GAMS, and extremely useful comments
on aquaculture consist of uneaten feed,
in the earlier version of this manuscript. Thanks also go to
feces, and other organic matter from Dr. Sumeet Gulati of the University of British Columbia
salmon farms. They are directly discharged for helpful comments and discussions. The authors are
into the marine environment because there also indebted to John Rune Selvik from the Norwegian
are no solid and effective barriers between Institute for Water Research and Mr. Svein Erik Stave
from Statistics Norway for providing the information on
open net-cage production systems and the pollution and sharing the nitrogen and phosphorus data
surrounding environment. Pollution may with us. We are also grateful to two anonymous reviewers
potentially create negative impacts on for their truly helpful comments and suggestions on an
earlier version of this manuscript. Canada's Research
Network in Aquaculture (AquaNet) is acknowledged for
Land Economics ? August 2010 ? 86 (3): 569-584 financial support to the first author. Rashid Sumaila
ISSN 0023-7639; E-ISSN 1543-8325 thanks the Sea Around Us Project for their support. Any
? 2010 by the Board of Regents of the errors, opinions, or conclusions expressed are those of the
University of Wisconsin System authors alone and should not reflect the funding agencies.

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
Land Economics August 2010

literature that documents these ecological -Aquaculture

and environmental impacts, especially in I 35 Agriculture


Europe, North America, and Chile, where I 30 Sewage
most farmed salmon are produced (e.g.,
125
Industry

Tlusty et al. 2000; Pohle, Frost, and Findlay


2001; Levings et al. 2002; Brooks et al. 2003; 1 20
Naylor, Eagle, and Smith 2003). These
negative impacts may be considered small 115io
at a large scale, but they can be very
significant locally, especially in areas where
salmon farms are concentrated. I I I I

In the case of the Norwegian salmon 1985 1992 1995 1998 2001 2004
aquaculture industry, pollution from fish Years
farms together with discharge from house
-Aquaculture
holds, industry, and agriculture have posed
Agriculture
a potentially serious risk in the coastal
waters and fjords (Statistics Norway 2006). ?Sewage
Figure 1 shows that nitrogen and phos - Industry
phorus from aquaculture have increased
rapidly over the last two decades, and their
contributions to the overall nitrogen and
phosphorus production have become larger
over time. Today, Norwegian aquaculture
is the largest source of phosphorus, and the o. o
second largest source of nitrogen in the 1985 1992 1995 1998 2001 2004
coastal areas of the country (Statistics Years
Norway 2006). Although pollution has not
increased at the same rate as the rapid FIGURE 1
Nitrogen and Phosphorus from the Norwegian
growth of aquaculture production, it is still
increasing. Salmon Aquaculture Industry and Other
Sources into the Marine Environment
While the impacts of pollution on the Data source: Statistics Norway 2006.
environment and natural resources have
been widely acknowledged, the economic
analysis of pollution is limited and needs to determined by estimating the cost of the
be studied further. Therefore, in this paper, direct damage to others that can be
we estimate the environmental costs associ attributed to pollution from a given eco
ated with pollution from the salmon aqua nomic activity. Second, it can be estimated
culture sector. According to the polluter by calculating the pollution abatement or
pays principle, salmon producers should prevention costs imposed on polluters by
bear the environmental costs of pollution. improving their production practice and/or
In other words, salmon producers should investing in abatement activities. In most
pay for the cost of removing the pollution, cases, the environmental costs estimated
or provide compensation to those who have from these two approaches differ. This is
been affected by the pollution. Hence, because they are estimated from two
estimating environmental costs is the first dimensions and administered by different
and essential step to establish public dam environmental policies. The first method,
age abatement and compensation programs however, is more difficult to execute than
or formulate economic-based environmen the second approach, due to complex
tal policies. ecological and economic consequences.
In general, environmental cost can be Therefore, in this paper, we estimate
measured in two ways. First, it can be pollution abatement cost by applying a

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
86(3) Liu and Sumada: Pollution Abatement Costs of Salmon Aquaculture 571

joint production approach. These pollution (e.g., Fare, Grosskopf, and Weber 2006),
abatement costs can be assumed to be productivities (e.g., Chung, F?re, and
proxies for environmental costs associated Grosskopf 1997), and pollution abatement
with salmon aquaculture. costs (e.g., Pasurka 2001; Verdanyan, and
A production process, such as salmon Noh 2006; F?re, Grosskopf, and Pasurka
aquaculture, produces desirable or "good" 2007). Also, joint production models can be
outputs (salmon), while simultaneously used to test the economic effects of different
generating undesirable or "bad" outputs environmental policies on different sectors
(e.g., pollution). Bad outputs are the by (e.g., Br?nnlund, F?re, and Grosskopf
products of good outputs (i.e., good out 1995; Martinez-Cordero and Leung 2004).
puts cannot be produced without producing For instance, if the government sets a
some bad outputs). Generally, good out pollution level as the pollution reduction
puts are marketable, while bad outputs are target, we can use joint production models
not traded in the market. In conventional to test how much production, revenue, or
production theory, the productivity and profit producers have to give up in order to
efficiency of a firm or an industry are meet the target.
generally measured based on good outputs. In this study, production technology that
However, joint production approaches have results in the joint production of good and
been recognized and developed to incorpo bad outputs is specified based on the
rate bad outputs along with good outputs assumptions of strong and weak disposabil
for measuring efficiency and productivity ity for bad outputs. If dumping bad outputs
(e.g., Fare et al. 1989, 2005; Chung, F?re, is not regulated, dumping is costless to the
and Grosskopf 1997; Chambers, Chung, firm, in other words, bad outputs can be
and Fare 1998; F?re, Grosskopf, and "freely" disposed of without any cost;
Pasurka 2007). The joint production ap hence, this can be modeled as strong
proaches have several advantages compared disposability of bad outputs. On the other
to conventional production approaches. hand, if dumping of bad outputs is regulat
First, the approaches do not require infor ed, dumping becomes costly to the polluter.
mation on pollution abatement technology It implies that environmental regulations
and its associated costs. The cost of pollu force producers to engage in pollution
tion abatement is captured by the reduction abatement activities (F?re et al. 1989).
in good output production. Second, they Thus, abating pollution becomes a costly
avoid the difficulties associated with directly activity, and producers have to internalize
measuring changes in the production process pollution abatement costs into their pro
through survey efforts. Also, the production duction process. This can be modeled as
technology automatically captures the syn weak disposability of bad outputs.
ergies among the abatement processes of We assume that salmon aquaculture
two or more pollutants (Pasurka 2001). operates under certain forms of regulatory
Joint production models, especially the constraints. This means that salmon pro
directional distance output function model, ducers, especially those in Norway, have
have gained growing interest and become engaged in pollution abatement activity.
the favorite kind of model for this kind of This assumption is appropriate because, in
analysis because of their flexibility of fact, there are some regulatory frameworks,
mapping rules and clear connection to both proposed and implemented, for the
traditional production functions (e.g., Norwegian salmon aquaculture industry.
Chung, F?re, and Grosskopf 1997; Cham For instance, Norway has implemented a
bers, Chung, and F?re 1998; Pasurka 2001; number of regulations such as limitations
F?re et al. 2005; Verdanyan and Noh 2006; on production level, farm size, and fish
F?re, Grosskopf, and Pasurka 2007). The density, and feed quota (Hjelt 2000; Maroni
directional distance output function model 2000). Feed is the key factor that contrib
has been used to estimate shadow prices utes to pollution. Feed formulation and

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
Land Economics August 2010

feeding technology have been greatly im 4. Weak disposability in good and bad
proved over the years. For example, the outputs: If (y,z)eP(x), and 0</l<l,
feed conversion ratio, a measure of the then ( , ) e P(x).
efficiency of converting feed into increased
body weight of farmed fish, has been greatly
The first and second properties are
reduced from around 4.0 in the 1980s to standard assumptions in production theory
about 1.2 at present (Asche, Guttormsen, (Shephard 1970). The first assumption
implies that inactivity results in no outputs
and Tveter?s 1999; Bjorndal, Tveter?s, and
Asche 2002; Tveret?s 2002). Moreover, (i.e., no free lunch), and finite inputs
produce finite outputs. The second assump
feeding technology has improved from
tion is strong disposability for good outputs
hand feeding to automatic feeders (Bjorndal
and inputs implying that y'can be produced
et al. 2002).
as long as y' < y given an input vector x. It is
II. THEORETICAL FRAMEWORK also called free disposability (F?re et al.
1989, 2005; F?re, Grosskopf, and Pasurka
2007). The third assumption is null-jointness
Let us assume a production process that
between good and bad outputs, implying
employs a vector of inputs e Rn+ to yield a that if no bad outputs are produced, then
set of good outputs denoted by a vector
good outputs will not be produced as well.
yeR , and bad outputs denoted by a The fourth assumption is weak disposability
vector g R+ . The technology (T) for the of outputs, implying that both good and
production process is represented by bad outputs can be reduced proportionally.
T = [(x, y, ) : can produce (y, )]. [1] The third and fourth assumptions are of
special interests to this study.
The technology illustrates all technically If dumping of bad outputs is unregulated,
feasible relationships between inputs and dumping is costless to the polluter, and firms
outputs. For a given input vector x, the can divert all inputs into good production
output set P(x) represents all feasible while jointly producing bad outputs for
given inputs. In other words, in the absence
output vectors (y,z), that is,
of environmental regulations, producers can
P(x) = [(y,z):(W)eT] [2] freely dispose of bad outputs without
bearing cost. This can be modeled as strong
The production possibility set P(x) illus disposability of good and bad outputs. On
trates the trade-offs between good and bad the other hand, if dumping of bad outputs is
outputs along the production possibility regulated, dumping becomes costly to the
frontier. Since both good and bad outputs firm, as it should, in other words, producers
are included, the production possibility set have to engage in abatement activities to
P(x) is also an environmental output set reduce bad outputs. For a given level of
(Fare et al. 2005; F?re, Grosskopf, and inputs, this can only be done by diverting
Pasurka 2007). The environmental produc some inputs away from producing good
tion possibility set P(x) has the following outputs and into reduction of bad outputs.
properties: If the targeted level of bad outputs is very
low, large amount of inputs will be required
1. P(x) is convex and compact, P(x) e Rn+ to be put into the abatement of bad outputs,
and satisfies the condition of no free and therefore, a small amount of inputs will
lunch. That is, P(0) = (0,0). be available for the production of good
2. Strong disposability of good outputs outputs. This can be modeled as jointly
and of inputs: If (y,z) e P(x), then for weak disposability of good and bad outputs.
y'<y, (y',z)eP(x), and for x'>x, The environmental output set is illustrat
(y',z)eP(x)c=p(x'). ed in Figure 2. The production possibility
3. Null-jointness: If (y,z) g P(x) and = frontier P(x) is constructed from observa
0, then y = 0. tions of inputs and outputs. The points A

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
86(3) Liu and Sumada: Pollution Abatement Costs of Salmon Aquaculture 573

assumed that producers attempt to maxi


S1: Regulated technology
mize good outputs and minimize bad
outputs if there is an environmental regu
iB lation implemented. We specify two pro
duction models with different mapping
rules for outputs, known as (1) environ
F(x?,z?) mental production function (EPF); and (2)
directional distance output function
(DDOF). These two models serve as the
functional representation of environmental
(bad output)
production technology. EPF is constructed
for solely maximizing good output and
keeping bad outputs constant in a direc
tional vector (e.g., F?re, Grosskopf, and
Pasurka 2007). DDOF is used for expand
ing good outputs and contracting bad
S2: Unregulated technology outputs in a directional vector (e.g., Chung,
F?re, and Grosskopf 1997; Picazo-Tadeo,
A ? Reig-Martinez, and Hernandez-Sancho
2005; F?re et al. 2005; F?re, Grosskopf,
and Pasurka 2007). Both models are addi
tive, and their expansion and/or contraction
take place in a directional vector. In fact,
EPF is a special case of DDOF.
Environmental Production Function
(bad output)

FIGURE 2 Like traditional production functions,


Environmental Output Sets given the vectors of inputs and bad outputs
Source: Modified from Fare et al. 2005, 2007. (x,z), an environmental production possi
bility frontier is defined as F(x,z) and
constructed based on observations. We
and represent the combinations of good
and bad outputs given a set of input (x). assume that F(x,z) is the bounded line
Since nonparametric linear programmingsegments OABCO in graph SI of Figure 2.
methods are used to measure productionThe maximum amount of good outputs can
efficiency, the production possibility fron be produced based on the production
tier P(x) is piecewise linear. The environ possibility frontier F(x,z). Given a set of
mental output set is bounded by the inputs and bad outputs (x?,z?), the maxi
piecewise linear segments OABCO. If badmum feasible production of good outputs is
output is assumed to be as stronglydefined as F(x?,z?). Because good outputs
disposable as good output, it means that can be freely disposed, y is feasible if
the maximum feasible amount of goody<F(x,z). Then, the environmental pro
outputs (e.g., OA or AB) at a given level duction set is defined as P(x) =
of inputs (x) can be produced with any level {(y,z) : y<F(x,z)}. Hence, an environmen
of production of bad outputs equal to or tal production function is "a complete
less than OC, including zero (thus violating characterization of the single output envi
the null-jointness assumption). ronmental technology" (F?re, Grosskopf,
Based on these assumptions about inputs and Pasurka 2007).
and outputs, a joint production approach is There are parametric and nonparametric
developed to model both good and badmethods of specifying production models.
outputs in different mapping rules. It isThe parametric method specifies a mathe

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
Land Economics August 2010

matical equation for a production model, maximize good output


for example, a quadratic function for a tity in the direction
directional distance function (e.g., Fare et inputs and bad outputs.
al. 2005; Vardanyan and Noh 2006). The are unregulated, the obj
nonparametric method uses the data envel observation k' is written
opment analysis (DEA) technique to mea
sure the efficiency performance of produc
ers. DEA uses linear programming
FcKxV'^g^m
techniques that first identify the theoreti subject to
cally best producers based on observed data
(e.g., inputs and outputs). Then, a produc
tion possibility frontier is constructed as a 5^a*xJ;<x?f,
k=l
/i=l,...,JV, [5.1a]
piecewise linear envelope of all observed
outputs and inputs. The producers on the
frontier are assumed to operate efficiently.
Producers who are not on the frontier are J2m?>0k'^ + ykm, m=l,...,M, [5.1b]
k=l
regarded as inefficient. The production level
due to inefficiency is calculated by compar
ing the performance of each producer to the
best producer. Nonparametric methods
have advantages over parametric methods
]T?^>zf, j = l,..,/, [5.1c]
=1

because they can incorporate several inputs


and outputs without generating different
estimates. They can be performed with
limited data sets, and they also avoid the
?)a* = 1, k=?9...9K9 [5.1d]
biases brought about by different paramet
ric models. Also, they require only quanti where ^, ^ /, ^,^ , and a? are the
tative data on inputs and outputs; no intensity variables. They are weights as
specific price data are needed (F?re et signed to each observation when construct
al. 1989; F?re, Grosskopf, and Pasurka ing the production frontier. Since a? is
2003). Therefore, in this study, a nonpara nonnegative, variable returns to scale are
metric method is adopted. imposed, and the summation of a^is as
Assuming there is a sample of k= ,.,., sumed to be 1.
producers employing a vector of inputs x?, When bad outputs are regulated, the
n= l,...,iV to obtain a vector of good output objective function for observation k' is
, m= ,.,., , and a vector of bad outputs written as
1-, j? 1,...,/, the environmental production
function on the production technology is F(x*y^;g,)=max0*\ [5.2]
then defined by
subject to
F(x,y,z; ) = max [ : (y + 0,z) e P(x)]. [3]
J2a^n<^n, [5.2a]
k=l
Let g denote a directional vector, g = (gy)
for good outputs, where g^ e , and
gm#0. EPF on the production technology
is defined by ^ykm >ek'g + ykm, m = ,.,., , [5.2b]
k=\

5(x,y,z; gv) = max [ : (y + 0g,,z) e P(x)] [4]


"

X^zf = zf, i = l,..,J, [5.2c]


The objective function
k=l of EPF is to

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
86(3) Liu and Sumada: Pollution Abatement Costs of Salmon Aquaculture 575

?]a*
k=l
= 1, k=l,...,K, [SM] ^(x^yV'^-g^ max ?*, [7.1]
subject to
where x?,ym,z/,a^,?>0 are defined as
above. The right-hand side of the con J2?kXkn<x?,
k=\
/i=l,...,tf, [7.1a]
straints of linear programming problems
represents the actual amounts of inputs or
outputs employed or produced, while the m=\,...,M, [7.1b]
k={
left-hand side of the constraints represents
the amount of inputs or outputs used or
produced by the most efficient or best
producers. It should be noted that the signs
?>z*>z/-/?Vz, j=h-J, [7.1c]
of the constraints for bad outputs in two /s:
equations are different. In the third con
straint equation above (equations [5.1c] and
^a* = 1, fc=l,...,?\ [7.1d]
?:=l
[5.2c]), the equality sign means that bad
outputs are weakly disposable under a The objective function under regulated
regulated technology, in other words, the technology is
observed amount of bad outputs equals the
amount of bad outputs produced by the 5(x^y*V';gr-gz) = max^', [7.2]
most efficient producers. On the other
hand, the inequality sign means that bad
subject to
outputs are strongly disposable under an
unregulated technology, in other words, the
observed amount of bad outputs equals or
] a* *< *', =1.JV, [7.2a]
k=\

is less than the amount of bad outputs


produced by the most efficient producers.
VJa^>y*' + ^'g m=l,...,M, [7.2b]
Directional Distance Output Function

A DDOF has the quality that it can allow ?a^ = zf-^Vz, 7=W, [7.2c]
the expansion of good outputs and con
traction of bad outputs at the same time.
Let g denote a directional vector, 5^a = 1, ?=1,...,*, [7.2d]
k=\
g = (gr-gz)> for g?od and bad outputs,
where g, = R+, gz = R+, and gm+y^0. where, x?,ym,z/,a^,^>0, the input and
DDOF on the production technology is output constraints are defined as in the case
then defined by of EPF.

ZV(x,y,z; gy9 - gz)= max[? : (y+?gy* - ?gz)e P(x)], Pollution Abatement Costs
1 [6]
When bad outputs are not regulated,
where ? is the maximum attainable expan their disposal is costless for producers but
sion of good outputs along the +g^ not to society, and all inputs are used for
direction, and largest feasible contraction producing good outputs. When environ
of bad outputs along the ? gz direction mental regulations are imposed, disposing
vector. Applying the same principles for of bad outputs (pollution) becomes a costly
strong and weak disposability for bad activity for producers because it takes away
outputs as in EPF, the objective function resources from producing good outputs to
of DDOF under unregulated technology is reduce/abate bad outputs. In other words,

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
576 Land Economics August 2010

FIGURE 3
Illustration of Environmental Production and Directional Distance
Output Functions

some of the inputs that are used to producemodels are usually used to measure the
good outputs have to be diverted for technical inefficiency of producers. Given
cleaning/abating bad outputs. The reduc input vectors, technical inefficiency mea
tion in the production of the bad outputs
comes at a cost in the form of a reduction in sures are determined by the ratio of actual
good outputs to maximum potential good
the production of the good outputs. Hence,
pollution abatement cost (PAC) is defined outputs. If observed data points lie on the
as the lost good outputs to producers frontier, producers are defined as efficient,
related to pollution abatement activity. otherwise they are inefficient. The magni
PAC can also be seen as the opportunity tude of technical inefficiency measures the
cost of the environmental regulations be distance between observed data points and
cause it is measured by the difference of the the production possibility frontier.
forgone good outputs under unregulated Figure 3 illustrates an environmental
and regulated technologies (Fare et al. 2005; production function and a directional
F?re, Grosskopf, and Pasurka 2007). distance output function. The dotted line
Therefore, PACs under two productionOABCO represents the regulated technolo
functions are expressed as follows: gy, while the solid line OY2BCO represents
Environmental production function: the unregulated technology. Points A and
are on the frontier P(x) and are efficient.
PA C = ?u(xk'; zk' ; gy) - F(x*'; xk' ; gy\ [8.1] Point a, for instance, operates inside the

Directional distance output function: frontier and it is therefore inefficient. Any


point on or inside P(x) can be expanded
PAC^Du^^'^-gy-g;) and/or contracted in both(y,z). For in
stance, in the case of the EPF, keeping
-5(xVV';g,,-g.). [8.2] bad output unchanged, the producer oper
ating at point a can expand its good output
It should be noted that these production from a to a\ with the regulated technology

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
86(3) Liu and Sumada: Pollution Abatement Costs of Salmon Aquaculture 577

or from a to a2 with the unregulated effects of using different directional vectors


technology. Under the DDOF, the produc for inputs and outputs on the pollution
er can increase its good output and decrease abatement costs in the sensitivity analysis.
bad output by moving from a to b\ with the Linear programming is used to solve the
regulated technology or from a to b2 with maximization problem for each functional
the unregulated technology. form. The computer software General Alge
Thus, PACs can be determined by the braic Modeling System (GAMS) is used to
difference between the maximum good execute the calculation. GAMS is a high
output production associated with unregu level modeling system for mathematical
lated and regulated production technolo programming and optimization.1 The opti
gies, or they can be determined on the loss mal solutions are achieved for unregulated
of good output production related to and regulated production technologies for
technical inefficiency (F?re, Grosskopf, two production models based on explicitly
and Pasurka 2007). As can be seen in differentiating the assumptions on the con
Figure 3, if the EPF is used, the lost output straints regarding inputs and outputs, in
due to inefficiency is the distance between a particular, the constraints on bad outputs.
and a under the regulated technology and
between a and a2 under the unregulated III. THE DATA
technology; the PAC is the distance be
tween ax and a2. If the DDOF is used, the Since Norway has widely available eco
lost output due to inefficiency is the logical and economic data related to salmon
distance between a and b\ under the aquaculture, we use the Norwegian salmon
regulated technology and between a and aquaculture industry as the case of an
b2 under the unregulated technology; the empirical application of the joint production
PAC is the vertical distance between b\ and models proposed herein. Norway is the
b2, that is, the distance between and Y2. pioneer in salmon aquaculture development
Since farmed salmon are sold in the market, and production. The country has been the
the potential revenue losses are calculated number one farmed salmon producer in the
using average yearly market prices and world since the beginning of salmon farming.
potential output production losses. There However, due to the lack of farm-level data,
fore, PACs also can be determined by the we consider salmon aquaculture as a whole
potential revenue losses. and use the data collected at an aggregated
industry level, on an annual basis.
Directional Vector In this analysis, salmon aquaculture
operation needs four inputs (feed, smolt,
Before conducting the programming, we labor, and capital) to produce one good
output (salmon production) and two bad
need to specify the directional vector g(g^)
outputs (nitrogen and phosphorus). The
and g(grgz) for the EPF and DDOF quantities of salmon production and inputs
models, respectively. We choose the direc
are extracted from Statistics Norway and
tional vector g = (g^, ?gz) = (l,0) for the
EPF model, and g = (gv,-gz)-(l,-1) for the Fisheries Directorate of Norway (www.
the DDOF model. The reason for choosing ssb.no),2 whereas the quantities of nitrogen
the unity directional vectors (g^l and and phosphorus were estimated by the
gz = l) is that they indicate the "shortest" Norwegian Institute for Water Research
distance when optimizing over the direction (NIVA) and compiled by Natural Resource
to reach the production frontier (F?re and and Environment, Norway (Statistics Nor
Grosskopf 2000). In other words, the way 2006). Several methods are used to
estimates from the two models give the quantify nitrogen and phosphorus from
maximum unit expansion in good output
production and simultaneous unit contrac 1 See www.gams.com for more information.
2 See www.ssb.no for more information.
tion in bad output production. We test the

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
578 Land Economics August 2010

TABLE 1
Summary Statistics for Norwegian Salmon Aquaculture, 1986-2005
Units Minimum Mean Maximum
Good output Production Tonnes in thousands 44.9 286.1 582.2
Bad output Phosphorus Tonnes in thousands 0.7 3.1 6.1
Nitrogen Tonnes in thousands 2.5 13.7 27.3
Input Feed Tonnes in thousands 87.3 357.4 727.6
Labor Man-hours in millions 3.0 3.9 4.9
Smolt Numbers in millions 27.3 96.0 160.2
_Capital_NOKs in millions 402.7 1,723.8
aquaculture. The production parameters,
models. Over the years, PACs in ter
both production
such as production, feed used, nitrogen and and revenue have s
phosphorus contents in feed and farmed trend for both models. E
decreasing
salmon, treatment yield, wastewater vollast several years, the PA
ly in the
ume, nitrogen and phosphorus trivial.
concentraIn this study, the technolo
tion of samples, and number ofproduction
sampling possibility frontier (P
constructed
periods are used for these calculations, and from all the data sets
the detailed information is presented
period of by20 years, (i.e., series prod
Borgvang and Selvik (2000). The data set Pasurka (2001) indicates tha
processes).
ranges from 1986 to 2005 and is summa
technology for any year is cumula
rized in Table 1. Since both the EPF andthe producers use the prod
meaning
DDOF are additive models, theprocess
measurein that year or any preceding
ment unit and magnitude of inputs and the technology based on
Therefore,
outputs may affect the results (Picazo
series data can be considered as imp
Tadeo, Reig-Martinez, and Hern?ndez-San
technology. The PPF is constructed
cho 2005; F?re, Grosskopf, andon Pasurka
cumulative technologies. In other w
2007). To avoid these problems, wethe
scale all
individual observations in recent
aredividing
inputs and outputs into fractions by more closed to the boundary po
these by their respective maximum PPFvalues
for in
both unregulated and regu
the samples of inputs and outputs.frontiers.
In other Also, it indicates that the
words, the values of all the inputs and
are measured based on improved p
outputs are normalized between 0 tion
and 1.
technology. That is why the PAC
The lost good outputs associated withzero or at no measurable c
virtually
technical inefficiency are estimated for the
recent years. Further, he points ou
EPF and DDOF models using pollution
the same abatement costs based on
data set. The PACs are simply the series data may be underestimated b
difference
between the lost good outputs resulting
technical change tends to reduce me
pollution abatement costs (Pasurka
from technical inefficiency for unregulated
Although
and regulated production technologies. By our models do not exp
multiplying by the prices of good output,
specify the technical changes of the i
namely, the annual farmgate try,
price, thetechnical changes have been
these
losses of good outputs in terms of revenue
place and been captured in the prod
are calculated. Hence, the pollution abate
process through input- and output-
ment costs are expressed in both lostsuch as the improvement o
variables
production and revenue of the good output.
formulation and feeding technology
husbandry management (Bjornd
IV. RESULTS AND DISCUSSIONS
leras, and Asche 2002).
However, the PACs show a wide
Figure 4 shows the pollution abatement
tion depending on years (producers) a
costs over time for the two production
production models applied. In some

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
86(3) Liu and Sumada: Pollution Abatement Costs of Salmon Aquaculture 579

wild salmon populations in 18 rivers affect


ed (Johnsen and Jensen 1994).
We averaged the 20-year PACs (Table 2).
This will help us average out the variations
in PACs observed. On average, pollution
abatement costs expressed in terms of lost
production and revenue are about 10.1
thousand tonnes and 472 million NOKs
(1 US$ = 5.5 NOKs), respectively, over the
20 years. These comprise 3.5% and 6.5% of
total farmed salmon production and reve
nue, respectively. The PACs estimated from
Years the DDOF and EPF models are about 12.1
and 8.2 thousand tonnes, which corre
sponds to about 4.2%, and 2.9% of total
salmon production, respectively. In terms
of revenue, the costs are around 544 and
400 million NOKs, which works out to
about 7.5% and 5.5% of total revenues,
respectively. Out of a total of 20 observa
tions, 8 in the DDOF model, and 10 in the
EPF model do not incur pollution abate
ment costs.
Considering that salmon producers have
achieved low or even negative profit mar
1966 1969 1992 1995 1998 2001 2004 gins recently due to low market prices (e.g.,
Years 2001-2004), these PAC estimates are prob
ably quite large in absolute value terms
FIGURE 4 (Sumaila, Volpe, and Liu 2007). Especially
Pollution Abatement Costs (PACs) for Twofor those producers who have not internal
Production Models: Environmental ized the PACs into their production pro
Production Function (EPF) and Directionalcesses, the profit margins may disappear
Distance Output Function (DDOF) (Scale ofwhen they do fully internalize them. These
the Axis Differs on the Two Graphs)
results, however, suggest that without
environmental regulations, farmed salmon
could be theoretically increased by 3.5% in
PACs increase unexpectedly. For instance, terms of tonnes of salmon production and
in 1989, 1994, and 2002, PACs were much 6.5% in terms of gross revenues. Given
higher than in their respective adjacentcurrent production technology and envi
years. These sudden increases in PACs ronmental regulations, these potential in
may be caused by different factors, such ascreases in production and revenue have
investment in production, market condi been diverted for reducing pollution. If the
tions, regulatory changes, or biophysical target pollution level is lower than the
shocks (e.g., disease outbreaks and acci current level, reduction in good output
dents). Tveret?s (1999) and Tveret?s, and production is possible. In particular, salm
Heshmati (2002) indicated that these factors
on farms that are too closely clustered may
might result in technical changes from yearhave to be closed down partially or fully.
This has already happened in the North
to year. For instance, in 1989, Furunculosis,
a bacterial disease, became endemic in Sea. A recent North Sea Agreement de
Norway, hitting the salmon aquaculture clares that the pollution level in the North
industry hard, with 189 salmon farms andSea has to be reduced to the level in 1985 for

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
580 Land Economics August 2010

TABLE 2
Average Pollution Abatement Costs Associated with the Two Production Models
Production Revenue
Thousand Million No. of Years
Model_Tonnes % of Total NOK % of Total with PAC = 0
Environmental production function 8.16 2.85 400 5.53 10
Directional distance output function 12.09 4.23 544 7.52 8
Average_ 10.13 3.54 472 6.53 _ 9
Note: PAC, pollution abatement cost.

DDOF model, somein


all production sectors and households inputs
all have to be
diverted
the countries bordering the to reduce bad
North outputs,
Sea. Inwhile all
order to meet the target,inputs
Norwegian
are used to increasefish
good outputs in
the EPF
farming facilities have been model. But, both models
prohibited in are
appropriate
the North Sea region since 1997in(Statistics
the case of salmon aquacul
Norway 2006). ture because some producers are compelled
to reduce
In reality, these PACs have bad output, such as
partially, ifin Norway,
while by
not fully, been internalized other the
producers,
Norwesuch as those in
Chile, are not so compelled
gian producers into the production process to a significant
as a result of stringentdegree.
environmental
regulations and technical innovations
(Asche, Guttormsen, and Tveter?sV. SENSITIVITY ANALYSIS
1999;
Bjorndal, Tveter?s, and Asche 2002; Tver
As the mapping
et?s 2002). The trend of reduced PACsrules
mayfor directional
vectors
also imply that the technical have impacts on the estimated
improvement
pollution
of the industry is confirmed abatementextent
to some costs (e.g., Vardanyan
by the Norwegian salmon
and Nohaquaculture
2006), a variety of mapping rules
industry. Moreover, it may
are be
applied
argued
for the twothat
directional models,
EPF and DDOF.
using averaged PACs measured from First, time
it is assumed that
the directional
series data is inappropriate because vector the
for bad outputs
remains constant
PACs in the last several years at 1, and the directional
are insignif
vector for
icant. However, we do believe good output
that not all is assumed to
individual farms operate gradually
withincrease
the on a scale of 1 to 10.
same
Results show that the estimated
technical efficiency (or inefficiency) and PACs fall
with increasing
are regulated under the same directional vector for good
environmental
outputbased
policies. Thus, the results (Figure 5). on
This isthe
to be expected
because more
Norwegian data can be used as a inputs are diverted
basis to to pro
ducingin
understand similar industries goodother
output. The magnitude of the
juris
dictions and/or used by policy makers
decline gets smaller in vector
as the directional
other countries to formulate environmental
for good output gets bigger. This is the case
in both not
regulation policies that have models (Figure
been 5). in
place. Next, in the base case, we assume that
It is expected that the DDOF estimates input factors are not regulated. Within a
higher pollution abatement cost than the production process, assuming inputs are all
EPF. This is because different mapping used to produce good output when bad
rules for bad outputs are applied in these outputs are unregulated, some inputs have
two models. DDOF increases good output to be allocated for pollution abatement
and decreases bad outputs in a directional activity when bad outputs are regulated.
vector; EPF only increases good output and However, in some cases, inputs do change
keeps bad outputs unchanged. In the depending on the resource, state of the

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
86(3) Liu and Sumada: Pollution Abatement Costs of Salmon Aquaculture 581

50

Years 1 4 7 10 13 16
Salmon faims (in ascending or

FIGURE 6
Pollution Abatement Costs (PACs) under
Different Mapping Rules for Inputs
and Outputs

directly reducing inputs may be more


resource efficient than directly reducing
bad outputs, because pollution is reduced
at source in the former case. This result may
provide producers and policy makers a
useful insight for formulating environmen
tal regulations and developing abatement
1986 1989 1992 1995 1998 2001 2004 programs. As in the case of salmon
Years aquaculture, it is difficult to mitigate
FIGURE 5 pollution once it is discharged into the
Estimated Pollution Abatement Cost for environment. This is because pollution is
quickly dispersed into the open ocean, given
Various Mapping Rules (Scale of the y Axis
Differs on the Two Graphs) current production technology?an open
net-cage system. Since feed is the key input
factor in contributing waste discharges,
economy, and time. For instance, feed use
has gradually reduced with the improvecontrolling feed input may be the most
ment of feed formulation and feeding efficient means to regulate pollution dis
charges for salmon aquaculture. For in
technology. Thus, we test the effect on
stance, a feed quota program has been
PAC under an assumption of reducing
established in Norway since 1995. The aim
input uses and expanding good output.
of the program is to control production
Here, we use the DDOF to illustrate
through controlling feed (Hjelt 2000).
these effects on PACs. Two scenarios are
presented: Scenario 1, increasing good
output while reducing inputs and bad VI. CONCLUSIONS
outputs simultaneously g(?gx9 + gy9 ?g2) =
(?1,1,-1); and Scenario 2, increasing good
We develop a joint production function
output while reducing inputs and keepingto model both good and bad outputs from
bad outputs constant g(-gx, + g,;,-gz)the
= salmon aquaculture industry. This
(?1,1,0). The results are compared with the allows us to calculate PACs from a produc
base scenario g( - gx, + j> - g2) = (0,1, -1 ).tion process and from a private producer's
Figure 6 shows that PACs decline when
perspective. Two production models, EPF
and DDOF, are applied. Good and bad
inputs are reduced. This indicates that

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
Land Economics August 2010

outputs are treated in an asymmetrical way years. This study explores these costs for
in the models. EPF only maximizes good different years due to data constraints at the
outputs and keeps bad outputs constant. farm levels. Ideally, cross-sectional panel
DDOF expands good outputs and con data of salmon farms are more appropriate
tracts bad outputs simultaneously. The than using time series of data aggregated
analyses are conducted based on the as from all the salmon farms. However, in the
sumptions of strong and weak disposability case of salmon aquaculture, such cross
of bad outputs. Consequently, unregulated sectional panel data were just simply not
and regulated production technologies are available when the study was conducted.
specified. The empirical analyses are carried Thus, potential future research will be to
out on the Norwegian salmon aquaculture measure annual PAC using panel data sets.
industry. One good output (salmon), two And it would be interesting to investigate
bad outputs (nitrogen and phosphorus), how the PAC changes over time by
and four inputs (feed, smolt, labor, and comparing the annual PACs. Further,
capital) are included in the analyses. The technical efficiency and pollution abate
data are aggregated at the industry level, ment costs greatly differ from year to year,
covering a period of 20 years from 1986 to with a generally declining trend in the
2005. We choose a nonparametric approach Norwegian case studied herein. This paper
to solve the maximization problem that this has provided a framework and an analysis
analysis entails. PACs are calculated by that promises to be of help to policy makers
determining the difference between the fore in determining appropriate environmental
gone good outputs using two technologies. policy regarding sustainable salmon aqua
Although PAC is estimated based on a culture.
production process from a producer's
perspective, it can be viewed as the pollu References
tion damage costs on the environment and
resource users. Further, the PACs can be Asche, Frank, Atle G. Guttormsen, and Ragnar
used as the economic bases to establish Tveter?s. 1999. "Environmental Problems, Pro
public damage abatement and compensa ductivity and Innovations in Norwegian Salm
tion programs or to formulate economic on Aquaculture." Aquaculture Economies and
based environmental policies. For example, Management 3 (1): 19-29.
the PACs can be used as reference points to Bjorndal, Trond, Ragnar Tveter?s, and Frank
set pollution taxes that can be imposed on Asche. 2002. "The Development of Salmon and
Trout Aquaculture." In Seafood Market Studies
producers. The producers who fail to
for the Introduction of New Aquaculture Prod
implement environmental regulations in
ucts, ed. Philippe, Paquotte, Catherine Mario
their production decision making should jouls, and James Young, 101-15. Vol. 59,
be penalized. The level of penalty can be set Cahiers Options M?diterran?ennes. Proceed
based on the estimates of PACs. To the best
ings of the seminar of the CIHEAM Network
of our knowledge, this study is the first SELAM, Zaragoza, Spain, June 21-22, 2001.
attempt to use a joint production function Paris: CIHEAM.
approach to estimate PACs for aquacul Borgvang, Stig A., and John Rune Selvik, eds. 2000.
ture, in general, and salmon aquaculture, in Development of HARP Guidelines: Harmonised
particular, although a joint production Quantification and Reporting Procedure for
model has been applied earlier to estimate Nutrients. SFT Report 1759/2000. Oslo: Nor
wegian Pollution Control Authority.
productivity and technical efficiency for
Br?nnlund, Ru?ar, Rolf F?re, and Shawna Gross
shrimp farming in Mexico (Mart?nez-Cor
kopf. 1995. "Environmental Regulation and
dero and Leung 2004). Profitability: An Application to Swedish Pulp
It should be noted that this joint produc and Paper Mills." Environmental and Resource
tion approach provides a framework to Economics 6 (1): 23-36.
measure pollution abatement costs through Brooks, Kenneth M., and Conrad V. W. Mahnken.
technical inefficiency across farms and 2003. "Interactions of Atlantic Salmon in the

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
86(3) Liu and Sumada: Pollution Abatement Costs of Salmon Aquaculture 583

Pacific Northwest Environment. II. Organic ture in Asia: Proceedings of the First Interna
Wastes." Fisheries Research 62 (3): 255-93. tional Symposium on Cage Aquaculture in Asia,
Brooks, Kenneth M., Annette R. Stierns, Conrad ed. Chi?, I. Liao and C. Kwei Lin, 1-17. Quezon
V. W. Mahnken, and Dale B. Blackburn. 2003. City, Philippines: Asian Fisheries Society.
"Chemical and Biological Remediation of the Johnsen, Bj?rn Ove, and Arne Johan Jensen. 1994.
Benthos Near Atlantic Salmon Farms." Aqua "The Spread of Furunculosis in Salmonids in
culture 219 (1-4): 355-77. Norwegian Rivers." Journal of Fish Biology 45
Chambers, Robert G., Yangho Chung, and Rolf (1): 47-55.
Fare. 1998. "Profit, Directional Distance Func Levings, Colin D., Jim M. Helfield, Dario J.
tions and Nerlovian Efficiency." Journal of Stucchi, and Terri F. Sutherland. 2002. A
Optimization Theory and Applications 98 (2): Perspective on the Use of Performance Based
351-64. Standards to Assist in Fish Habitat Management
Chung, Yangho, Rolf F?re, and Shawna Gross on the Seafloor near Salmon Net Pen Operations
kopf. 1997. "Productivity and Undesirable in British Columbia, Vancouver: Fisheries and
Outputs: A Directional Distance Function Oceans Canada.
Approach." Journal of Environmental Manage Maroni, Kjell. 2000. "Monitoring and Regulation
ment 51 (3): 229-40. of Marine Aquaculture in Norway." Journal of
Fare, Rolf, and Shawna Grosskopf. 2000. "Theory Applied Ichthyology 16 (4-5): 192-95.
and Application of Directional Distance Func Martinez-Cordero, Francisco J., and PingSun
tions." Journal of Productivity Analysis 13 (2): Leung. 2004. "Sustainable Aquaculture and
93-103. Producer Performance: Measurement of Envi
F?re, Rolf, Shawna Grosskopf, C. A. Knox L?vell, ronmentally Adjusted Productivity and Effi
and Carl Pasurka. 1989. "Multilateral Produc ciency of a Sample of Shrimp Farms in
tivity Comparisons When Some Outputs Are Mexico." Aquaculture 241 (1-4): 249-268.
Undesirable." Review of Economics and Statis Mazzola, Antonio, Simone Mirto, Tiziana La
tics l? (1): 90-98. Rosa, Mauro Fabiano, and Roberto Danovaro.
F?re, Rolf, Shawna Grosskopf, Dong-Woon Noh, 2000. "Fish-Farming Effects on Benthic Com
and William Weber. 2005. "Characteristics of a munity Structure in Coastal Sediments: Analy
Polluting Technology: Theory and Practice." sis of Meiofaunal Recovery." ICES Journal of
Journal of Econometrics 126 (2): 469-92. Marine Science 57 (5): 1454-61.
F?re, Rolf, Shawna Grosskopf, and Carl Pasurka. McGhie, Tony K., Christine M. Crawford, Iona
2003. "Estimating Pollution Abatement Costs: M. Mitchell, and Dominic O'Brien. 2000. "The
A Comparison of 'Stated' and 'Revealed' Degradation of Fish-Cage Waste in Sediments
Approaches." Working paper series, Social during Fallowing." Aquaculture 187 (3-4):
Sciences Research Network. Available at 351-66.
http://ssrn.com/abstract=358700. Milewski, Inka. 2001. "Impacts of Salmon Aqua
-. 2007. "Environmental Production Func culture on the Coastal Environment: A Review."
tions and Environmental Directional Distance In Marine Aquaculture and the Environment: A
Functions." Energy 32 (7): 1055-66. Meeting for Stakeholders in the Northeast, ed.
F?re, Rolf, Shawna Grosskopf, and William L. Michael, F. Tlusty, Dave A. Bengston, Harlyn
Weber. 2006. "Shadow Prices and Pollution O. Halvorson, Sarah D. Oktay, Jack B. Pearce,
Costs in U.S. Agriculture." Ecological Econom and Robert B. Rheault, Jr., 166-97. Falmouth,
ics 56 (1): 89-103. MA: Cape Cod Press.
Garcia, Serge M., and Richard J. R. Grainger. 2005. Naylor, Rosamond L., Josh Eagle, and Whitney L.
"Gloom and Doom? The Future of Marine Smith. 2003. "Salmon Aquaculture in the
Capture Fisheries." Philosophical Transactions Pacific Northwest: A Global Industry." Envi
of the Royal Society, Biology 360 (1453): 21-46. ronment 45 (8): 19-39.
Hargrave, Barry T. 2003. "Far-Field Environmen Naylor, Rosamond L., Rebecca J. Goldburg,
tal Effects of Marine Finfish Aquaculture." In Jurgenne H. Primavera, Nils Kautsky, Malcolm
A Scientific Review of the Potential Environ C. M. Beveridge, Jason Clay, Carl Folke, Jane
mental Effects of Aquaculture in Aquatic Eco Lubchenco, Harold Mooney, and Max Troell.
systems, 3-11. Canadian Technical Reports on 2000. "Effect of Aquaculture on World Fish
Fisheries and Aquatic Science, Vol. 1. Ottawa: Supplies." NatureAQS (6790): 1017-24.
Fisheries and Oceans Canada. Pasurka, Carl A. 2001. "Technical Change and
Hjelt, Knut A. 2000. "The Norwegian Regulation Measuring Pollution Abatement Costs: An
System and the History of the Norwegian Activity Analysis Framework." Environmental
Salmon Farming Industry." In Cage Aquacul Resource Economics 18 (1): 61-85.

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms
Land Economics August 2010

Pauly, Daniel, Villy Christensen, Sylvie Guenette, ical and Economic Impacts and Contributions
Tony J. Pitcher, U. Rashid Sumaila, Carl J. of Fish Farming and Capture Fisheries."
Walters, Reg Watson, and Dirk Zeller. 2002. EMBO Report 2 (11): 958-63.
"Towards Sustainability in World Fisheries." Tlusty, Michael F., . Snook, Vern A. Pepper, and
Nature m (6898): 689-95. M. Robin Anderson. 2000. "The Potential for
Picazo-Tadeo, Andres J., Ernest Reig-Martinez, Soluble and Transport Loss of Particulate
and Francese Hernandez-Sancho. 2005. "Di Aquaculture Wastes." Aquaculture Research
rectional Distance Functions and Environmen 31 (10): 745-55.
tal Regulation." Resource and Energy Econom Tveter?s, Ragnar. 1999. "Production Risk and
ics 27 (2): 131-42. Productivity Growth: Some Findings for Nor
Pohle, Gerhard, Bruce Frost, and Richard Findlay. wegian Salmon Aquaculture." Journal of Pro
2001. "Assessment of Regional Benthic Impact ductivity Analysis 12 (2): 161-79.
of Salmon Mariculture within the Letang Inlet, Tveter?s, Ragnar, and Almas Heshmati. 2002.
Bay of Fundy." ICES Journal of Marine Science "Patterns of Productivity Growth in the Norwe
58 (2): 417-26. gian Salmon Farming Industry." International
Shephard, Ronald W. 1970. Theory of Cost and Review of Economics and Business 50 (3): 367-94.
Production Functions. Princeton, NJ: Princeton Tveter?s, Sigbjorn. 2002. "Norwegian Salmon
University Press. Aquaculture and Sustainability: The Relation
Statistics Norway. 2006. Natural Resources and the ship between Environmental Quality and In
Environment. Available at www.ssb.no/english/ dustry Growth." Marine Resource Economics
subjects/01/ (accessed February 2007). 17(2): 117-28.
Sumaila, U. Rashid, John Volpe, and Yajie Liu. Vardanyan, Michael, and Dong-Woon Noh. 2006.
2007. Potential Economic Benefits from Sable "Approximating Pollution Abatement Costs
fish Farming in British Columbia. Marine via Alternative Specifications of a Multi-Output
Policy 31 (2): 81-84. Production Technology: A Case of the U.S.
Tidwell, James H., and Geoff L. Allan. 2001. "Fish Electric Utility Industry." Journal of Environ
as Food: Aquaculture's Contribution?Ecolog mental Management 80 (2): 177-190.

This content downloaded from 185.122.225.44 on Wed, 25 Sep 2019 10:37:52 UTC
All use subject to https://about.jstor.org/terms

You might also like