d'arge r., kogiku k., (1973) economic growth and the environment.pdf
TRANSCRIPT
-
8/10/2019 D'Arge R., Kogiku K., (1973) Economic growth and the environment.pdf
1/17
.. .(1)
conomic rowth
and
the
nvironment 1,2,3
R. C. D ARGE and K. C. KOGIKU
University
of
California, Riverside
The premise that man resides withina bounded and closednatural environment, an environ
ment with relatively fixed dimensions, has been discussed in many forms for at least three
centuries. The idea of a spaceship
earth
has recently caught the imagination and
interest of both economists and environmentalists in particular, and the public in general
[3], [5], [6] and [17].
basic tenet of this viewpoint is that whatever has been produced,
consumed, not recycledand discarded within the sealed spaceship earth is still here and
willcontinue to be. Of course, energydispersionmay be offsetby nuclear fusion or breeder
reactors so that what was previously discarded may be completely reused without seriously
impinging on the earth s natural energy reservoirs. Economically and technically-feasible
extraterrestrial transportation systems, destruction of wastes by symmetrical anti-wastes,
or increased efficiencies in solar energy conversion also may become realities. However,
such energy-augmenting technological changes, at least at this time, appear to be in the very
distant future. What is of pressing current importance, if we accept the premise of a
nearly sealed capsule earth with relatively fixed dimensions in terms
of
mass-energy and
waste assimilative capability, is how rapidly economic growth should proceed.
Rather than attempting to dissect the spaceship
earth
premises regarding limited
energy reserves and assimilative capability, we take these as given in this paper. We then
construct several simple paradigms of an economy confronted with a bounded, closed
natural environment and then we analyze how such an economy should optimally produce
and consume over time. The next section of the paper deals with an extremely simple
model of waste generation, based on the conservation of matter-energy principle, and with
the consumption behaviour of the economy s inhabitants assumed to be predetermined.
In subsequent sections, the model is generalized to an optimal control problem where
consumption and waste generation are allowed to be regulated, and an attempt is made to
integrate the non-mutually exclusiveprocesses of resource extraction and waste generation.
The materials balance view of a closed resource system indicates that tonnages of raw
materials extraction utilized by an economy are approximately equal to tonnages of waste
products generated by the economy in the long-run [3]. A basic identity derivable from
the principle of conservation of matter-energy, given the assumption of materials balance
and no recycling is:
R ~ W f W
F=
W
1 First version received December 1971; final version received Apri 972 1s.).
2
A portion
of
the research reported here was financially supported by Resources for the Future,
Inc., with no responsibility for results or interpretation. A previous draft of this paper was presented at
the 1970Econometric Society Meetings in Detroit, Michigan.
3
The authors wish to acknowledge the very helpful comments and suggestions of K. Oddson, K.
Hamada, G. Heal, and two referees.
61
-
8/10/2019 D'Arge R., Kogiku K., (1973) Economic growth and the environment.pdf
2/17
62
REVIEW OF ECONOMIC STUDIES
where R, F, and W denote material extraction, production consumed by consumers, and
total waste flows, respectively. Wi and W
J
are the amounts of waste flow originating in
the production and consumption sectors.
Given these fundamental identities and assumptions weare able to write an additional
identity provided the flow F of material is assumed homogeneous. Basically,we assume
that there is only one type of consumer good. Then:
F
=
y
.. .
2
where y is per capita income in units of material flowper capita and N is total population.
Finally, if production and consumption are joint products of waste emissions, total waste
flows can be assumed to be proportional to final product [3], [5]:
W
=
gF. ... 3
Then, by assuming that each variable previously defined refers to time
r ,
waste flows are
linearly related to total income measured in material units:
Wet = g .y t .N t . . ..
4
Thus, we obtain a relation between waste flows and output per capita. Actually, the
assumption
of
proportionality is even more restrictive than one might at first presume,
since it completely specifies an implicit technology relating output to raw materials. By
definition,
Wet
=
R t ,
so
Wet
=
g
.y t .N t yields yet = R t . Thus, in this most
g N t
simple case, production results only from the magnitude of raw materials and there is no
substitution between labour and raw materials in the production process. In terms
of
crude empirical approximation the number
g
yet for theconsumption sectori.e.,
Wit /
N t
is estimated to be greater than one ton
per
capita per year for the United States at current
production levels.
Our next task is to
U model
the environment in some semi-realisticand yet manage
able way. Environmental pollution, at least in its quantitative dimensions, is usually
expressed in terms of concentrations, i.e., parts per million ppm) of dissolved solids or
DDT, ppm suitably indexed) of carbon monoxide or oxides of nitrogen concentrations in
the air, or tons per cubic acre of solid wastes. Thus, a natural single unit of waste concen
tration appears to be average density. Of course, concern with densities only may be
misleading, particularly with such pollutants as carbon monoxide or methane gas generated
from solid wastes. or now, however, wewill assume that density is a reasonable abstract
measure of waste concentration.
Let D t and v denote waste density at time
t
and total environmental waste holding
capacity, respectively. We, in effect, identify the closed resource system by a simple fixed
volumetric magnitude,
v.
Then using the definition that waste divided by volume equals
waste density and applying it to 4) yields:
D t
=
D O
+
ft
y t N t dt. ... 5
v 0
Given exogenously determined percentage rates of growth in population of
l n
and
income per capita of l P waste density can be easily related to initial population levels
and material flowper capita:
p+n t
D t =
D O + fl_- y O N O . ... 6
v f3+n
Increasing density of wastes within the closed resource systemis not necessarily bad per se.
However, if the natural environment is relatively small in comparison to current and
expected future sums of waste flows, people may suffer physiological, psychic, or other
damages. As an example, we could presume for each individual a monotonic, continuous
-
8/10/2019 D'Arge R., Kogiku K., (1973) Economic growth and the environment.pdf
3/17
-
8/10/2019 D'Arge R., Kogiku K., (1973) Economic growth and the environment.pdf
4/17
64
REVIEW OF ECONOMIC STUDIES
(8)
to the form of the utility function in addition to separability which sidesteps the problem
of interpreting cross partial derivatives of the utility function:
uy>O
UD>O uT>O
UyyO UTT O
In essence, these conditions stipulate that per capita utility of material flowis increasing at
a decreasing rate, while
per capita
disutility generated by increasing waste densities is
increasing at an increasing rate. As the finiteplanning interval r is increased, it is assumed
that per capita utility at any time t is increased, but at a decreasing rate. This is perhaps
one of our most tenuous assumptions since it implies that, regardless of how close
t
is to
T,
individuals will be precommitted to applying a non-shifting evaluation to an increment
in r.
Consequently,
the optimal finite planning interval is deduced with no adjustment
occurring over time to account for potential future disobedience in the Strotz
[ ]
sense,
as
t approaches T(l).
From equation (5), we are able to write a relation between waste flow and waste
densities:
D= W
v
g
D
= -
yN
v
(9.1)
...(9.2)
Equation (9.2) is, in effect, a description of the changing state of the natural environ
ment as a result of the amount of material consumption at time t. Equations (7) and (9.2)
yield the pollution process model except for specification
of
initial and terminal con
ditions and inclusion of a social rate of time preference factor. The terminal time
r
is
conceptualized to be that period of survival which is optimum for the maximization to be
stated below. Given a presumed decimation density level (for example, 2,000
4,000 ppm
of carbon monoxide),
then
D r
=
To complete this exposition, the initial density is specifiedas
D O
=
l ,
and population
at
t as:
N t
=
N O e
nt
and the utility function is discounted with a discount factor a.
2
(10)
It
should
be
noted that the assumptions of a finite planning interval coupled witha finiteenvironmental
capacity for wastesare extremelyrigid. Ofcourse, infinite planning intervals in conjunction with an unlimited
natural environment assimilative capacity reduces to a trivial non-existent scarcity case unless scarcity is
introduced in some other way, i.e., available technology, knowledge, or capital. Alternatively, an infinite
planning interval in conjunction with a finite environmental waste assimilative capability appears to be
philosophically inconsistent, provided population is constant or increasing through time, material flow
per capita has a positive lower bound, rates of environmental waste assimilation are very low or zero, waste
generating technology is completely static in character, and wastes at some level of concentration become
lethal to the human species. These conditions quite obviously ensure that at some moment in the finite
future, decimation willoccur. The alternative case of a finiteplanning interval and infinitewaste assimilative
capability also appears to be philosophically inconsistent. Specifying an infinite planning interval for
mankind, of course, implies the imposition of a constraint on the decision process, namely that regardless
of how degraded man s habitat becomes, he must continue to exist. A more general and encompassing
decision process would stipulate that the survival period be optimally chosen with infinity as one possible
choice. We have chosen the course of specifying a finite planning interval and making it subject to choice
by current generations. However, such a decisionprecludes consideration of an infinite time horizon. The
marginal utility of an additional unit of time may approach or be at infinity for those concerned only with
preservation of the human species. In such a case, the length of
T
would dominate all other decisions in a
trivial manner since minimal consumption at each instant of time must prevail with finite resources,
specificallywaste assimilative capability.
2
Although weshall apply a discount factor which reflectsa positive or zero rate of time preference for
material-consumption flows,several potentially explosiveethical questions are brushed aside.
the human
race faces the possibilityof extinction bypollution (or other means)over a finiteinterval, are inter-generational
utility comparisons defensible? Within a different context, Dasgupta recently suggested that a small
positive discount of the
future
may be accepted as
ethical
[12]. However, if we assume that each
-
8/10/2019 D'Arge R., Kogiku K., (1973) Economic growth and the environment.pdf
5/17
D ARGE KOGIKU
GROWTH
AND THE
ENVIRONMENT 65
... (11.1)
Given the above definitions, an optimal control problem can be formulated as follows;
maximize
f
fT{
d 1_e-
at
}
J =
[u(y)-u(D)
u T)]e-atdt
=
[u(y)-u(D)]e-
at+
-
u t) dt
o 0 dt a
where
y
is the control variable and
is a state variable, whose equation of motion is
iJ
=
fl
yN = bye b
gN O)jv).
v
...(11.2)
The problem is to select the admissable control y so that a maximum integral of discounted
individual utility is achieved over the interval 0 t T where T is free to vary.
The Hamiltonian function for the problem can be written:
d 1_e-
at
H =
[u(y)-u(D)]e-
at+
- u(t)+
pbenty
dt
a
d
e
at
= [u(y)-u(D) pbye n+a)t]e-
at
-
u t)
dt
a
...(12)
where p is the costate variable corresponding to
According to the theory of optimal control, optimality calls for the following con
ditions :
iJ = bye D O
= 2 D T)
= 5
. aH D) -at
p = - - = U e
aD
y*
maximizes
u(y)+
pbye n+a t
[u(y)-u(D)+ pbye ? n) ]e-
at
=
f
{a[
u(y(s-u(D(s]
d
1_e-
at
- npbye n+a)S}e-asds - -
u t)j
dt
a t
d e
at
[u Y T-u D T
pbye n+a)T]e-
aT
+
-
u t)/t = O
dt
a
If we define a new variable
z = -
bpe n+a t
... (13.1)
... (13.2)
... (13.3)
... (13.4)
...(13.5)
generation can be exactly separated from others, then current generations with higher rates of time preference
may actually eliminate the existence of some distant future generation. The faster we consume in a closed
resource system, the more rapidly extinction occurs. While the utility of distant generations may seem
valueless now, if we were that distant generation, we may very well value our continued existence at or
approaching infinity. Even very high rates of time preference over a finite interval would not make it
worthwhile to consume currently above some basic subsistence level, provided the last future generation s
utilities were considered. In any case, we shall assume that such inter-generational utility comparisons can
be contrived so that each generation would be equitably considered, including potential but excluded future
generations, and in the remainder of the discussion emphasize the results where all generations are treated as
equals, i.e.,
a = O
1 See, e.g., Michael Athans and Peter L. Falb, Optimal Control, New York, N.Y.: McGraw-Hill,
1966, Theorem 5-7, p. 293.
E---40/1
-
8/10/2019 D'Arge R., Kogiku K., (1973) Economic growth and the environment.pdf
6/17
66
REVIEW OF ECONOMIC STUDIES
... (12 )
... (13.1 )
... (13.2 )
(13.3 )
then (12) and (13.1)-(13.5) become
d
l e
at
H
= [u(y)-u(D)-zy]e-
at+
-
u(t)
dt
a
dD/dt
= benty
D O
=
f} D -r:
= i5
dzldt
= (n+a)z-bentuD
u(y)-zy u(w)-zw for all w such that 0 Vo w VI
where
V
o
is a lower bound and
VI
an upper bound for consumption;
u z
0
[u(y(t)- u D t -
z(t)y(t)]e-
= fe- {a[u(y(s-u(D(s]nz(s)y(s)}ds
d 1_e-
at
- -
u(t) t
(13.4 )
dt a
... (13.5 )
l e
at
[u(y(r-u(D(r)-z(t )y(r)]e-
at+
-
u(t)lt
=
dt a
For (13.3 ) the control should be:
y
= Vo
if z u,(vo)
= [Uy]-I(Z) ifu,(vt) z u,(vo) ...13.6
= if z u (Vt).
That is, Y = F(z), which is Lipschitz continuous for all z and non-increasing, should have
the graph shown in Figure 1. Using F z the system (13.1 )-(13.2 ) can be replaced with
dD/dt = bentp(z) D O
=
f}
.. (13.1 )
dzldt =
(n+a)z-bentuD
z O = Zo
(13.2 )
where
Zo
is free. A solution exists and is unique for all Several possible timepaths are
shown in Figure 2.
From
the preceding two equations:
: ~ = n : ~ ; : r U D ~ 0 ... (14.1)
depending on
y =
F(z)
-
8/10/2019 D'Arge R., Kogiku K., (1973) Economic growth and the environment.pdf
7/17
D ARGE KOGIKU
GROWTH AND THE ENVIRONMENT 67
z
z
b
n a
D
I V
o
)
I V
o
)
b
n
... (14.2)
V
b
n a
UD
D
0
D
0
FIG
2
dz 0
dD
-
8/10/2019 D'Arge R., Kogiku K., (1973) Economic growth and the environment.pdf
8/17
68 REVIEW OF ECONOMIC STUDIES
a substantial undiscounted shadow price should be applied to waste density. Following
the initial large-scale shift the shadow price decreases over time.
Re-writing (13.3 ) for an interior solution
Hence in this case
u, y = z.
(15)
(17)
Y
= Zfuyiy
...
16
and
y is increasing when z is decreasing. Equation (13.2 ) shows that z will be falling
when
n a z
0, uQQ