Professional Documents
Culture Documents
of Public Economics
34 (1987) 1944.
North-Holland
PERFORMANCE
VERSUS DESIGN STANDARDS
IN THE REGULATION
OF POLLUTION
David
BESANKO*
received
May 1987
1. Introduction
A common
form of regulation
in the presence of externalities
or information
problems
is standard-setting.
Standard-setting
is widely used in
environmental
regulation,
regulation
of product and occupational
safety, and
regulation of product efficacy and durability.
An important
issue in devising a standard is whether the standard should
specify performance
goals (a performance
standard)
or should specify the
procedures
used to meet the performance
goal (a design standard).
In the
context of air quality regulation, a performance
standard would place a limit
on a factorys emission of sulfur dioxide per million parts of air, while a
design standard would specify the minimum efficiency of a stack gas scrubber
designed to reduce SO, emissions.
Although a performance
standard and a design standard may be directed
toward the same goal (cleaner air, fewer workplace accidents), their economic
consequences
may be quite different. Breyer (1982) notes that although
design standards are easier to enforce than performance
standards:
*This work was begun while 1 was visiting the Economics
Research
Group
at Bell
Communications
in 1985. I would like to thank David Baron, David D. Martin, Martin Perry,
and David Sappington
for helpful discussions
on this paper, and two referees for their useful
comments and suggestions.
OQ47-2727/87/$3.50
B.V. (North-Holland)
20
21
industry
ftxi,
(1)
eih
where f is a twice-continuously
differentiable
quasi-convex
function,
with
f, >O and f, ~0. The function f(.) is a reduced-form
pollution
production
function. To see how this reduced form is derived, suppose that each firm in
the industry
manufactures
a joint product (x,q) according
to production
functions:
x =A+, 4,
q = &, 4,
22
To develop
the specification
in
c(eJxi+ Neil,
where c(e,) is the marginal
cost of output, and k(ei) is the fixed cost of
installing e, units of the e.c.i. The marginal cost function c(.) is assumed to be
twice-continuously
differentiable,
nondecreasing
and (weakly) convex in e; i.e.
~20, ~20. The emissions control cost function k(.) is assumed to be twicecontinuously
differentiable,
increasing,
and (weakly) convex in e; i.e. k >O;
k 2 0. In addition, k(0) = 0.
Firms are assumed
to behave as Cournot
competitors
and thus set
outputs. The profit of a typical firm i is given by:
technology
discussed
above,
2;
s.t. x,=&,eJ,
where w is a vector of input prices. We are thus implicitly
assuming
that g is linearly
homogeneous
in a. The assumption
that ~20 would follow if g,sO,
i.e. if the e.c.i. had a
nonpositive marginal product of output.
23
5 qp,
where qp is the maximum level of pollution allowed per firm.4 Each firm then
chooses output xi and a quantity e; of e.c.i. to satisfy the constraint.
Under a design standard, the regulatory
constraint
for each firm is of the
form:
ei 2 e,.
a performance
standard
and a
31t is straightforward
to prove that the assumption
that P(X)X+P(X)
50 is a sufficient
condition
for a firms revenue function
P(x+X,)x
to be concave in x for X~E [O, X] and
x E [0, X]. The concavity of the tirms revenue function is used in the proof of the existence of an
industry equilibrium.
(See footnote 7 below.) The assumption
that P(X)X+P(X)<O
is also a
sufficient condition for a firms Cournot reaction function x(X,) to be decreasing in the output
X, of all other firms in the industry.
4The performance
standard
in this model can be thought of as arising in one of two ways.
First, we can think of the regulator as imposing a maximum level of pollution on each firm by
fiat. Alternatively,
the upper limit qp can be thought
of as arising because each firm in the
industry has purchased
q,, marketable rights to pollute. For this second interpretation
to be
convincing,
we need to include in our model the decisions by firms to trade these marketable
rights among themselves and the determination
of an equilibrium
price of rights. However, in
our setting a detailed theoretical
treatment
of these issues is unnecessary.
To see this, suppose
that before the product market opens each firm in the industry is awarded a certain number of
marketable
rights (which may or may not equal qP but which add up to nq,). In an equilibrium
in marketable
rights, firms will trade the rights among themselves until the marginal benefit of a
right is equated across all firms (where the calculation
of marginal
benefit takes into account
profits in the subsequent
product
market
equilibrium).
Since all firms are symmetric,
by
assumption,
marginal benefits will be equated when each firm acquires an equal number qp of
the rights.
24
design standard
is that under a performance
standard
the regulator would
monitor the actual pollution emitted by a firm while under a design standard
the regulator
would monitor
the efficiency of the system the firm puts in
place to control
emissions.
For example,
suppose
a firm manufactures
railroad
ties and generates
hazardous
creosote waste as a byproduct.
A
performance
standard
would involve monitoring
the quantity
of creosote
waste generated by the firm while a design standard would involve monitoring the procedures followed by the firm to dispose of or store the hazardous
creosote waste. A firm may do a poor job of disposing of the waste (low e)
but may not generate much pollution because its output x is low. Intuitively,
a performance
standard gives a firm more freedom because the standard can
be met either by increasing the efficiency of the emissions-control
system or
by cutting back on output, whereas a design standard can only be met by
increasing
the efficiency of the emissions-control
system. In practice, the
distinction
between a performance
standard
and a design standard may be
quite subtle. Consider the example of a manufacturer
of electrical capacitors
that uses PCBs as an insulating
fluid. A design standard
might be a
specification
of procedures for disposing of defective capacitors that contain
PCBs. To monitor
compliance,
the regulator
might send an official to
observe whether these disposal procedures
are being followed. However, if
this type of monitoring
is costly, the regulator
may end up monitoring
compliance
by measuring
the amount
of PCBs in the soil where the firm
dumps its wastes. If the latter enforcement
procedure is used, the effective
standard
is a performance
standard
because compliance
depends on the
amount of pollution created by the firm, even though the nominal standard
is a design standard.
2.3. Welfare analysis and regulatory objectives
In analyzing
optimal performance
and design standards,
two regulatory
objectives will be considered.
The first objective is social welfare maximization (SWM) subject to a minimum
profit constraint.
Social welfare is
defined as the sum of consumers
and producers surplus minus the social
costs B(cqj) of pollution damage. The social cost function is assumed to be
increasing
and (weakly) convex, i.e. B>O, and B 20. Under SWM the
regulators objective can be written as:
v(X) -C
c(ej)Xj-1
k(ej)
-B(C 4j)3
25
problem can
(2)
x 2 0.
Afx(xp,eP),
Af,(x, eP),
(3)
(4)
The precommitment
assumption
is important.
In the absence of precommitment
an alternative formulation
would be to have the tirms and the regulator
move simultaneously
and to
analyze the Nash equilibrium
in outputs and a standard.
To see the implications
of this setup,
notice that in our model the regulators objective function under either SWM or CM does not
depend directly on the standard (either qP or e,). Instead, it depends on the choices made by the
firms (x, e, and q). This implies that there will be many Nash equilibria in the game in which the
firms and regulator move simultaneously.
One Nash equilibrium
is where each firm conjectures
that the regulator will set a standard
which is sufficiently lax so that the unregulated
Cournot
output and a zero level of e.c.i. are feasible. Given such a conjecture each firm will produce its
unregulated
Cournot output. In turn, if the regulator conjectures that the firms will produce this
output and choose zero e.c.i., the regulator (being indifferent across standards,
given conjectures
about output, e.c.i., and pollution levels) will have no incentive not to choose the lax standard.
6Where there is no ambiguity, the subscript designating
a firm will be dropped.
A proof of the existence of an equilibrium can be found in Besanko (1987).
26
(5)
1.
Under
a performance
emissions
standard,
Cournot
industry equilibrium
output
output Xc.
on
p(XC)xC+P(XC)-c(O)=O,
(6)
where xC=XCnm.
P(XC).
The assumption
that P(X)X+P(X)sO
P(X)Xn-
+ P(X) is strictly decreasing
(7)
en the profit-maximization
problem
of a typical
is follows:
d[P(X)Xnm+P(X)]/dX=P(X)(l+nm)+P(X)Xn-
<P(X) + P(X)XC.
If P(X)sO,
the right-hand
side of (Fl)
which P(X) SO. If P(X) >O, then
F(X)+P(X)Xn-sP(X)+P(X)X
is negative,
which
(since XC
SO,
which proves the assertion
(Fl)
proves
5X),
the assertion
21
(8)
e2eD,
x20.
Again using Friedmans (1977) Theorem 7.8 an equilibrium
can be shown
exist. In equilibrium,
the regulatory constraint
(8) will be binding.
The condition for the symmetric industry equilibrium
output XD is:
P(XD)xD + P(XD) = c(e,),
where xD-XDnm.
This yields the following
to
(9)
characterization
of the industry
equilibrium.
for
the
unregulated
Cournot
equilibrium
is
dXD/deD=c(eD).
is negative
that
Q.E.D.
(see footnote
8) and c(e,)zO,
so
28
that perfor-
Proposition 4. For any target pollution level q, emissions costs are lower
under a performance standard than under a design standard.
Proof: To prove the result it is sufficient to demonstrate
that ep(q)<eD(q).
The proof is by contradiction.
Suppose, to the contrary, that eD(q) 5 eP(q).
Let XD(q) denote the equilibrium
industry output under a design standard
that induces a pollution level per firm of q, and xD(q) = XD(q)n- . From (9):
PWD(dbDM
+ JVD(q))- 4eDM)= 0.
By contrast,
(10)
from (3):
pWp(q))xpM
+ P(Xp(d)- 4eP(q))
>O.
By assumption,
(11)
(12)
(12) implies:
XPM< XD(q).
(13)
But,
4=
f (xp(djeP(qh
(144
29
4=
f(xD(qLeD(d).
(1W
consequence
of Proposition
(14a)
and
(14b)
cannot
simulta-
4 is:
Proposition 5. For any target pollution level q, the industry output X(q)
under a performance standard is less than the industry output XD(q) under a
design standard.
Proof:
Because
4=
fbP(q),eP(q))
= f(xD(q),
eD(q
Q.E.D.
Propositions
4 and 5 are depicted in fig. 1. The upward-sloping
line 00 is
the set of (x,e) pairs that generate
a constant
pollution
per firm. The
equilibrium
under a performance
standard q puts a firm at point R, where a
firms isoprofit contour
is tangent
to OO, while the equilibrium
under a
design standard
puts a firm at point S where a firms isoprofit contour
reaches its peak. The intuition
behind Propositions
4 and 5 is that under a
performance
standard
firms substitute
output for the e.c.i. to achieve the
pollution target q. Under a design standard, no such substitution
takes place.
Propositions
4 and 5 indicate the policy tradeoff the regulator faces under
SWM. If a performance
standard is used, the regulator sacrifices output for
low emissions
control costs while under a design standard,
the regulator
sacrifices emissions control costs for higher output. For a fixed pollution
target q the social welfare under a performance
standard
may or may not
exceed the social welfare under a design standard.
It has been argued by
some authors [e.g. Roberts and Farrell (1978)] that the objective functions of
environmental
regulators
typically include other factors besides emissions
control and pollution
costs. For example, state environmental
regulators
in
the United
States may be sensitive
to the employment
implications
of
pollution control regulations.
To the extent that social welfare maximization
is a reasonable
approximation
of regulatory
objectives in pollution
control,
our analysis provides a rationale unrelated to differences in enforcement
costs
why environmental
regulators
may prefer design standards
to performance
standards.
We next compare
a firms equilibrium
profit under
both types of
standards.
In the example
chosen optimally.
J.P.E.
in section
are
30
t(x,e):
f(x,e)
= 41
1)x) -
c(e)x
=
- k(e)
l?(q):
fx
Fig. 1. Comparison
of equilibrium
xD
under
performance
target.
(15)
31
P( [ 1 -n-
Combining
result.
Q.E.D.
[CM-P]
s.t. nD(q,) 2 n.
Proof:
the optimal
solution
to [CM-D].
Thus,
7c(4;) 2 n.
From
Proposition
(17)
6, it follows that
(18)
to [CM-P].
Moreover,
nk(eP(qi9)
+ B(n&) < n4eDMi))+ B(nqi%
from
Proposition
4,
(19)
32
6. An example
To further compare performance
presented.
The specification of the example
and design
standards,
an example
will be
is:
f(x, 4 = x/e,
P(X)=a-X,
c(e) = 0,
k(e) = ke,
B( nq) = bnq.
6.1. Industry equilibrium under a performance standard
The equilibrium
quantity
output under a performance
ePkp)=
CaCqpl~-kCqpl~21Cn+11~,qp2ka-,
q,<kaC;
0,
[a-k[qP]-]n[n+
XP(qp)
industry
o
3
l]-,
qpz ka-,
q,<ka-.
These functions
are depicted
in fig. 2. Fig. 2 indicates
that if the
performance
standard is sufficiently stringent (qp < ka- ) firms produce zero
33
o/,,
ka-l
Fig. 2. Equilibrium
under performance
standard
for example.
output and use no e.c.i. As the standard is relaxed (qp is increased), each firm
initially
uses a higher quantity
of the e.c.i., but beyond a certain point
(q,,= 2ka-), further relaxation
of the standard induces each firm to reduce
e. The equilibrium
industry
output is a strictly increasing
function
of the
allowed pollution
qp per firm over the range where qp> ka-. As the
standard
is relaxed industry
output approaches
na[n + l] -I - the unregulated Cournot industry output.
6.2, Industry equilibrium under a design standard
The equilibrium
firm is given by
under
a design
standard
that
results
in pollution
qD per
34
xD(q,)=na[n+l]-,
llq] -
eD(qD)=a[[n+
qD per firm is a
output is inde-
pollution
industry
- nkeP(qp)- bnq,
s.t.
xP(q,)
2 0,
where, for simplicity, the minimum allowed profit is assumed to equal zero.
Fig. 3 depicts social welfare as a function of qp. The function is defined at
qp=O and for qpE[ka-, CD), and the minimum
profit constraint
can be
shown to be satisfied for all values of qp in the domain of W(.).
The
function
W(.) can be shown to have no more than one local maximum.12
The global maximum
for W(.) will thus either be at this unique local
maximum
if Wp is positive at this point, or it will be at qp=O if Wp is
negative at the local maximum.
Assuming an interior optimum, the optimal performance
standard can be
shown to satisfy the first-order condition:13
(20)
Eq. (20) defines
the
optimal
standard
as a function
qp(n,u,b,k)
of the
Social welfare is not defined at qr~(O, ka-) because if the regulator sets a standard in this
interval, firms will produce
zero output and use no emissions control input. The regulator
constraint
will thus be slack over this interval and the actual level of pollution per firm will be
zero.
*However, for some parameter values it may not have a local maximum.
jThe second-order
condition can be shown to imply that q,z(3/2)ka-.
It can be shown that
the cubic equation in (20) has at most one root such that q,h(3/2)ka-.
If such a root exists
and if the solution to (20) is such that W(q,)>O, then (20) describes the global maximum.
If
(20) has no root that exceeds (3/2)ka- or if it does and that root is such that Wp(q,)<O, the
global maximum is qP=O.
35
t
Fig. 3. Social welfare under performance
standard.
Differentiation
{k(n+2)(n+
1)p2q;3[3kq;
-kqi3(aq,-k)(5n+
-2a])
l)(n+
dq, /dn
1)-4=0.
result
is driven
by two
effects.
by the second-order
Q.E.D.
First,
below,
a benefit
this proposition
from
condition
relaxing
of
the
that
36
pollution
standard
is that a relaxed standard
induces more output. As n
increases, however, the marginal
benefit of output (i.e. the industry
price),
decreases. Thus, for any qr, the marginal
benefit of relaxing the standard
decreases in n. Second, as n increases the marginal social cost nb of relaxing
the standard
increases. Together,
these effects work to decrease qp as n
increases.
To further characterize
the optimal performance
standard some numerical
examples
are presented
in table 1. The examples
indicate
the optimal
standard becomes less stringent as demand shifts outward (a increases) and
as the marginal
emissions control cost k increases. The standard
becomes
more stringent as the marginal social cost b of pollution
damage increases.
Table 1 also indicates that even though pollution
per tit-m declines with n,
total industry
pollution
QP=nqp rises with n: a tighter
standard
is not
sufftcient to offset the pollution of an additional firm.
The numerical examples also indicate that under an optimal performance
standard, social welfare does not unambiguously
increase with the number of
Table 1
Optimal
a=lOO,
II
1
2
5
10
20
100
a=lOO,
k=lO,
qP
8.61
6.62
4.36
3.09
2.18
0.95
k=20,
b=lO
QP
8.61
13.24
21.80
30.90
43.60
95.00
qP
Q'
1
2
5
10
20
100
12.14
9.32
6.13
4.35
3.05
1.30
12.14
18.64
30.75
43.50
61.00
130.00
k=lO,
standards.
a=lOO,
W%P)
3,511
4,179
4,423
4,194
4,105
3,052
b=lO
a=200,
performance
n
1
2
5
10
20
100
k=lO,
qP
1.02
5.39
3.55
2.52
1.71
0.16
a= 100, k=20,
WqP)
3,506
4,069
4,243
4,078
3,746
2,280
b=2
?I
1
2
5
10
20
100
qP
9.90
7.60
4.99
3.53
2.47
1.04
a=3OQ, k=lO,
b=lS
QP
7.02
10.78
11.75
25.20
35.40
76.00
W%lP)
3,539
4,119
4,325
4,195
3,910
2,630
b= 15
QP
9.90
15.20
24.95
35.30
49.40
104.00
WqP)
3,452
3,986
4,105
3,883
3,472
1,702
b=2
qP
QP
wphJ
qP
QP
W%lP)
1
2
5
10
20
100
21.36
21.05
13.92
9.93
7.04
3.14
27.36
42.10
69.60
99.30
140.80
314.00
14,890
17,609
19,166
19,436
19,390
18,738
1
2
5
10
20
100
33.52
25.80
17.06
12.18
8.63
3.85
33.52
51.60
85.30
121.80
172.60
385.00
33,616
39,193
43,409
44,141
44,207
43,450
37
problem
is
- nkeD(qD)- nbq,
St.
The constraint
indicates the pollution
targets under which the equilibrium
profit of each firm is non-negative.
The function WD(qD), sketched in fig. 4,
can be shown to be strictly concave over the range [k[a/[n+ l]]-, co). The
global maximum will thus either be the local maximum qz in the figure or at
qn=k[a/[n+
l]]-.
Assuming an interior solution, the optimal pollution target per firm can be
shown to equal
q,={akb-[n+
which implies
1]-1)1/2,
(21)
design standard
is equal to
(22)
the following
analogue
between
economics
of scale
and
imperfect
competition
has
been
38
D. Besanko,
Performance
Proof
Parts (a) and (b) follow immediately
from (21) and (22). Total
industry pollution QD under the optimal design standard is given by
QD(n)-nqD=n[n+l]-lZ[akb-]Z.
Differentiation
(23)
dQD/dn=(1/2)[n+2][n+1]-32[akb~]2>0.
Q.E.D.
of performance
between
performance
and design
standards
under
SWM
will
D. Besanko,
Performance
39
OS.design standards
now be considered.
As a first step, the expressions
for the equilibrium
industry output and e.c.i. can be shown to imply that for q~ [ka-, co), the
difference W(q) - WD(q) between total welfare under a performance
standard
and a design standard for any pollution target q is given by:
W(q)-
Wyq)=nk[[n+
l]q]-2[(1/2)[n+2]k-aq].
Proof
for a performance
IL in the example,
10.
(n+2)(n+
the regulator
condition
(24)
1)12~2a32kP12bK12,
standard
(25)
to a design standard.
Define
(26)
WD(q3
< WD(qi3
+ 44%
= W(qE)
Now, if qI:=k[a/[n+l]]-,
and only if (25) holds.
[from (24)]
d(q:)>O.
If q~={akb-[n+1]-}12,
d(qg)>O
if
Q.E.D.
40
standard
counteracts
the upward distortion
of output, but the downward
distortion
remains. The design standard deals inefficiently with the pollution
externality
because it does not induce firms to take into account
the
marginal
effects of output
on pollution.
However,
it leaves open the
possibility that the downward output distortion due to imperfect competition
will be offset by the upward
output
distortion.
When the downward
distortions
are less serious (high n) this possibility
is less likely, and a
performance
standard is more likely to be preferred to a design standard.
A comparison
between
optimal
performance
and design standards
is
presented
in table 2. As the preceding
analysis
suggests,
performance
standards
dominate
design standards
when the number of producers in the
industry
is sufficiently
large, while design standards
dominate
when the
number of producers is small.
7. Performance
Suppose
that instead
of setting a performance
standard
or a design
standard,
the regulator
levied an emissions tax t on each firm. Each firm
would then face the problem:
max P(X)x-c(e)x--k(e)
-tf(x,e)
s.t.
Table 2
Comparison
of optimal performance
standards
and optimal design standards.
a=lOO, k=20,
?I
Wp*
WD*
1
5
10
20
60
80
100
3,452
4,105
3,883
3,472
2,413
2,032
1,702
3,505
4,154
3,914
3,215
-6,143
- 14,563
-25,399
Wp* = maximum
standard.
WD = maximum
standard.
b=15
sign[ Wp* - WD*]
+
+
+
+
welfare,
design
41
f(xT,eT).
8. Discussion
and extensions
Because the formal model presented in the previous section is fairly simple,
it understates
the richness of the standard-setting
problem in the real world.
The objective of this section is to discuss a number of important
issues in the
standard-setting
problem
that our model ignores and to suggest possible
directions for future research into the standard-setting
problem.
By modelling regulation
as a simple constraint,
we are implicitly assuming
that the regulator is able to continuously
monitor the firms compliance
with
the standard and that the regulator can impose on the firm suficiently
stiff
penalties for noncompliance
with the standard. An extension of our analysis
would be to assume that pollution
cannot be monitored
perfectly and that
monitoring
is costly. The regulator would then be faced with the problem of
how frequently to monitor compliance.
In a model with positive monitoring
costs, it seems plausible to expect that continuous
monitoring
is not optimal.
This will weaken the effects of standards
whose compliance
is difficult to
monitor. A very stringent pollution
performance
standard whose compliance
is monitored
only infrequently
could well induce the firm to invest less in an
e.c.i. than it would under a very lax design standard which can be monitored
much more frequently.
Thus, an important
consideration
in choosing
a
performance
or a design standard would be the relative costs of monitoring.
In pollution
regulation
it may be relatively easy to monitor certain forms of
pollution
(e.g. air pollution
by an electric utility) but relatively difficult to
monitor
other forms of pollution
(e.g. contamination
of soil by PCBs). In
many cases it is conceivable
that differences
in monitoring
costs could
dominate the economic tradeoffs identified in our theoretical model.
Another important
aspect of the enforcement
problem in standard-setting
is the existence of limitations
on penalties for violating standards.
In most
cases the regulatory
agency bears the burden
of proof in establishing
a
record of noncompliance.
In addition, the magnitude
of the penalty is often
left to the discretion of the regulator or the courts. For example, under the
Clean Air Act Amendments
of 1977, Congress left unspecified the penalties
42
for noncompliance
but directed the Environmental
Protection
Agency to
consider a variety of factors (including severity of the pollution damage and
the impact of the penalty on the violating firm) in assessing penalties [Farber
and Martin
(1986)]. Typically,
the agency must go through
extensive
administrative
and legal proceedings
before a line can be assessed. Sometimes the outcome of these proceedings
is a revision of the standard
and
not the assessment of a penalty [Breyer (1982)]. For first-time violators of
standards,
penalties are usually not assessed [Roberts and Farrell (1978)].
When penalties are assessed they are often small in relation to the assets of
the violating
firm. Because penalties
are often small, firms will find it
advantageous
to violate the standard, and the regulator must take this into
account
in the formulation
of the standard.
With respect to the choice
between performance
and design standards, the size of the penalty may be an
important
consideration.
Courts
will generally
not impose penalties
for
violations
of standards
when those penalties
do not bear a reasonable
relationship
to the expected damage resulting from the violation. Thus, one
might expect a tendency for the courts to impose a higher penalty for a
violation of a pollution performance
standard than for a violation of a design
standard
because in the former case the perceived damages from overpolluting may seem far more substantial
than the perceived damages from
installing the wrong type of pollution control equipment.
Another
issue ignored
in our model is that firms may be able to
successfully
circumvent
the standards
without legally violating
them. For
example, under a design standard,
a firm might install pollution
control
equipment
and then let it sit idle if the use of the equipment
adds to the
marginal cost of the final product. What prevents a firm from doing this in
our model is our implicit assumption
that the regulator not only monitors
the installation
of the e.c.i. but verifies that it is used and maintained
to its
fullest level of efficiency. In practice, if inspection resources are limited and if
the pollution
control inputs are sufficiently complex, verification
efforts by
the regulator may turn out to be quite superficial. Indeed, the regulator may
end up making
inferences
about the efficiency of the emissions
control
procedures by observing actual emissions. But in this case, what is nominally
a design standard in fact becomes a performance
standard.
The firms may also be able to legally circumvent
a performance
standard.
If, for example, the performance
standard
is viewed as an endowment
of
marketable
rights to pollute (see footnote 4) then two or more firms in the
industry may have an incentive to merge in order to relax the performance
standard. l6
?Salant
Switzer and Reynolds (1983) have shown that in a symmetric Cournot model with
linear demkd
and linear marginal cost horizontal
mergers are generally unprofitable
unless all
firms in the industry merge together to form a monopoly.
An open question is whether the
presence of performance
standards alters this result, at least in some cases.
43
A final important
issue that we have ignored in our model is the effect of
pollution standards on entry. The imposition
of standards raises the costs of
entering the regulated industry.
Because performance
standards
give firms
more flexibility
than design standards,
it would seem that performance
standards would have less harmful effects on entry than design standards. Of
course, in the spirit of Salop and Scheffman (1983) and Rogerson (1984),
incumbent
firms may have a preference for design standards
for this very
reason. An important
and worthwhile extension of our analysis would be to
examine the implications
of alternative
regulatory
regimes for entry and to
study the preferences
of both the regulator
and the firms for regulatory
regimes
when the full impacts
of alternative
types of standards
are
considered.
44