S. However, if the marginal productivity of capital is bounded this condition can be satisfied only if the levels of harvest and emissions are high enough. Thus, a low rate of decay of pollution and low absolute growth rate of the resource stock may imply that there do not exist any combinations of [h(e),e] that satisfy the modified golden rule of capital accumulation. Furthermore. note that some level of e which satisfies the above two conditions must also simultaneously satisfy the modified golden rule condition for the pollution stock. We do not study this condition here because the above arguments are enough to establish the claim that the traditional necessary-and-sufficient condition is only necessary here.2 This means that in the general equilibrium framework with many capital stocks in interaction the conditions for an optimal sustainable resource harvesting policy may be more restrictive than the ordinary partial analysis of renewable resource harvesting suggests. Next we consider under what conditions the socially optimal solution can be realized in a market economy. et solution with externalities
Assume an economy
which consists
of a representative
consumer,
a
‘We defme growth potential as the marginal growth rate of the resource stock when the stc;k size approaches zero. ‘The interesting question here is what are the properties of the optimal solution if no steady states exist. Because of the inada conditions in the utility function consumption can approach zero only t + x . Without them, the optimal planning horizon may be finite and both of the optimal terminal stc.rLksare zero.
0. Tahvonen and J. Kuuluvainen, Optimal growth with renewable resources
655
representative firm, the government and the common property environment [cf. Brock (1977)]. The utility of the consumer increases with consumption and decreases with pollution. He sells a fixed labor input (1) at a given wage (w) and rents capital at the market interest rate (r) to a firm which maximizes the instantaneous profits (II) (received by the consumer). The representative firm produces a composite commodity using four factors of production: labor, capital (composite commodity), a natural resource and emissions. Perfect information and perfect foresight and similar functional forms as in Section 2 are assumed. The problem of the representative consumer is, therefore, to
max W= 3 U(c,z)e-“dt :csoI 0
(17)
s.t.
C;=n+rk+wl+T+f
-c,
(18)
where T and f are the redistributed externality taxes on harvest flow and emissions. The problem of the representative firm is to, max
n=Q(k,k,e)-rk-qlt-be--1.
(20)
&.h.I.e2$0
The government collects the externality taxes, f=re
and
T=qh.
(21)
Finally, nature produces raw material which is harvested by the firm. The emission stream accumulates as a slowly decaying stock which affects the growth of the resource stock. i =
2=e-a:,
F( x, z) - h,
X(O)= so, and lim s(t) 2 0, l-+X
(22)
and _iim s(t) 2 0. I’X
(23)
z(0) =Zo,
Assuming interior solutions, the first order conditions ing firm require that
of the profit maximiz-
0. Tahrcwn
656
and J K uuluvuinen,
Q,,=q,
Q k =r,
I
.
Q e = T,
Optimal
growth
with renewable
resources
(24)
which are the implicit instantaneous demand functions for capital, the renewable resource and pollution. The consumer solves the problem ( 17)-( 18) in or&r to determine optimal consumption, additions to capital, and the stock supply of capital. The present value Hamiltonian for the consumer’s problem is ff = U(C,2) + p( II + rk + M’I+ T + f - c), and the necessary conditions are H,=U,--p=O,
(25)
and eqs. ( 18)--(19). It is obvious from (25) and ( 10) that i, = p, while (26) is identical to (13). The government has to find values for the externality taxes, q and r, which maximize s&al welfare. It can do no better than the social planner in the previous section and if it does worse, it has not found the social optimum . By (11) and (24) it must hold that qy +,G. In order to find the optimal time path for the tax on harvest flow the government has to actually find a solution to the resource owner’s problem: maximize the present value of the resource subject to (22), which yields a necessary condition identical to ( 14) determining optimal q/q. Using (12) and (24) the emissions are controled optimally by setting the externality tax f = -- e/i. Thus the price of emissions in the profitmaximizing firm is the social cost of pollution. The evolution of r is obtained using (15), only then the necessary conditions for the infinite time market solution are exactly the same as those in the central planner’s problem. Thus, by Proposition 1 we know that the steady sta?e, if it exists, is a saddle point. We have actually established the following proposition: Proposit ion 3. Assume a society with a representatice utility-maximizing consumer, representative profit-maximizing firm. the government and a common property erwironment producing a renewable resource and assimilating pollution. If the consumer and the gowrnment hare perfect knowGdge and perfect foresight and the karwst flow and emissions are taxed optimally, and the assumptions made on utility and production functions hold, the perfect market solutir with market clearing equals the socially optimal solution.
The
above
information
discussion indicates that, even in the market solution;, the requirements (from the empirical point of view) are strin
0. Tahvonen and .I. Kuuluoainen,
Optimal growth
with renewable resources
657
indeed, as stringent as in the social planner’s problem. In fact, the government has to find the solution to the social planner’s problem of the previous section in order to be able to tax emissions optimally. But what happens if the tax on emissions and common property harvesting are set nonoptimally? We conclude by examining the consequences of nonoptimal environmental policy. It is possible to solve c, h, and e as functions of prices and stocks [cf. Arrow and Kurz (1970)]. In particular, at any point in time, harvest and emission generation are both decreasing functions of 4 and T. If the government sets both 4 and r, they can be considered instruments and before analyzing the steady state we give the following proposition. Proposition 4. At given levels of capital, the renewable resource and pollution, if either q or T are set too low compared to the social planner’s optimal values, the rate of harvest, emissions and the interest rate are higher than on the social planner’s optimal path.
This result follows immediately by comparing the marginal productivities of capital when taxes are set optimally and when they are not (see proof in Appendix 3). It means that in an economy in which the price of the harvest or emissions are set too low, the environment, as a raw material source and as a sink for waste, is in danger of becoming over-exploited. Further, the use of the market interest rate in cost-benefit calculations, as is often suggested, may lead to biased results [cf. Brock ( 1977)]. Fcr the steady state we can give the following proposition. Proposition 5. (i) If the emission 1~1.xis lower than the social optimum, the steady state level of the pollution stock will be higher and stock of the resource will be lower than in the social planner’s optimal solution but the effect on the capital stock and consumption remains ambiguous. (ii) lf the tax on the harvest flow is lower than the social optimum, the stock of-the natural resource will be higher and the stock of pollution lower, but the erect on the stock of capital . and the steady state rate of consumption is again ambiguous.
The proof (Appendix 4) uses the fact that the equilibrium conditions for harvest and production must be met simtiltaneously. The possibility that pollution control may increase steady state consumption and capital levels in (5) should be compared to the result which states that when renewable resources are neglected, pc>llution control always reduces steady state capital and and consumption levels (but increases welfare) (see Tahvonen Kuuluvainen ( 1990) corollaries I and 1’). The result in (ii) showing that a nonoptimally low tax on resource harvesting increases the ievel of the renewable resource stock may seem counterintuitive. The explanation is that
658
0. Tahronen and J. Kuulurainen. Optimal growth with renewable resources
a nonoptimally low tax on harvest flow causes distortions in pollution control activity in the form of too low levels of emissions and the pollution stock [for further discussion see Tahvonen (1989)]. Note that these last consequences hold only if the distortions do not cause the nonexistence of the steady state. If this is the case the perfect foresight market solution path exhausts and po’llutes the raw material bases of the economy.
4. Conclusions We have studied the optimal growth of an economy with a resource base that is affected by endogenous pollution. The important question considered is whether an optimal and sustainable consumption and resource harvesting policy exists. It turns out that, opposite to the case of partial renewable resource models, an optimal sustainable path may fail to exist although the growth potential of the renewable resource stock exceeds the rate of discount. The absolute growth rate may be too low to satisfy the modified golden rule of capital accumulation. The core of the problem is, whether, with the given production technology, preferences, growth function of the resource stock and the rate of decay of pollution, the equilibrium conditions for three stocks can be met simultaneously. We further show that uncontrolled pollution may decrease steady state consumption and capital levels, opposite to the case where the effects of pollution on the resource stock are ignored. Accordingly, free access harvesting causes distortions in pollution control activity.
Appendix 1
Proof of Proposit ion I
We apply Sorger ( 1989), Corollary 2c. Let us first derive MHDS. Conditions (lO)-( 12) define the optimal levels of consumption, emissions and resource harvesting as functions of the shadow prices and stock levels: c = c(i., z), C~
k),e($,&j.,
k)] -c(j.,z),
0. Tahvonen and J. Kuuluvainen, Optimal growth with renewable resources
659
tb=-U,-Cc(i,,z),z]-_F=(x,z)+~(S+a). The equilibrium of this system possesses the saddle point property ‘curvature matrix’
if the
W) I, -H$
1
’
where H* is the maximized Hamiltonian, i = k, x, z and j = i., 4, $, is negative definite. Matrix Q is negative definite with a small rate of discount because matrices HE and Hz are negative definite with minimum eigenvalues below zero. By Sorger (1989), corollary 2c, this implies that the equilibrium, given it exists, is globally stable for bounded solutions. 0
Appendix 2 Proof 0s Proposition 2
-&=;=4=$=0. (14) and (4) now define .u=.Y(u), At the steady state _t= Z-x’(e)<0 when e<6. Using this with (3) gives !r(e)=F[_x(e),e/r], K(e)<0 when e ~6. By ( 13) we now get S - Qk[k, It(e),e] =O. This equation has solutions with some k>O when e
fk e)z~$ = U,(c,e/a)+ &(c, e/a) { QJk(e), + Q,C&), Me), e](6 +-a)} = 0, J+i;&&()
h(e), e]
F&(e), e/a]
H(c, e) = & = Q[k(e), h(e), e] -c = 0. J++&+o
By using the concavity assumptions on U, Q, and F, it can be computed that &(c, e) ~0, &,(c, e) CO and thus f(c, e) =0 defines c as a declining function of e. Because for al1 c>O it holds that Cm,,, f(w)>0 and lim,,,f(c,e)eO,
660
0. Tahronen
and J. Kuuluoainen,
Optimal
growth
with renewable
resources
there exists with all c >O some level of e between 0 and G which satisfies f(c,e)=O. When e-+0 T(c,e)=O requires that c--,00. When e-G from below f(c, e) = 0 requires that c +O. Consider next n(c, e) =O. Because n,(c, e) = - 1 and /7&c, e) $0, function f7(c, e) defines c implicitly as a function of e with an ambiguous sign for the derivative. However, by the necessity assumption e = 0 must imply c =0 for f7(c, e) =0 to hold. Note next that when e = $ it holds that h =O. Thus by the necessity assumption I?(c, e) =0 implies I7(0, i) =O. By (16) U(c, e) =0 requires strict&y positive c between 0 ce &. with at least one (c,e) combination This means that if &d, 0 f(c,e)= U(c,e)=O exists. Appendix 3 Proof of Proposition
4
(i) Let i” and i* denote the market and central planner’s solutions respectively. Assume rm< T*. Since e,, h, ~0, for all k, .Yand z it follows that em> e* and hm > h*, and P ,J k, e*, h*) c Pk( k, em, h”) = r. Analogously if 4” c q*, it follows that em> e* and hm > h* since e,, h,
5
(i) Assume T< P. By (24) ey > e*,, and therefore z: > z*,. By (9), S > F,( x*, , z”). For S = F,( xy, z: ) to hold, xz < x*, . Therefore F( x:, z:) = h”, < h*,,, and S = P,(k”, ,ez, h:) may require a lower or higher level of capital than the social optimum. Thus c’!JSC+, . (ii) Assume 4 c q*. By (4.28) h”, > hS, and S > F,Jx”, , z*, ). Because _Y: is given, the stock of pollution has to decrease, otherwise the equilibrium condition for harvest cannot hold, i.e., z: cz*, and ez
0. Tahvonen and J. Kuulucainen, Optimal growth with renewable resources
661
Seierstad, A. and K. Sydsaeter, 1987, Optimal control theory with economic applications (North-Holland, New York). Sorger, G., 1989, On the optimality and stabtlity of competitive paths in continuous time growth models, Journal of Economic Theory 48.526547. Tahvonen, 0.. 1989, On the dynamics of renewable resource harvesting and optimal pollution control, dissertation, Publications of the Helsinki School of Economics A, 67. Tahvonen, 0. and J. Kuuluvainen, 1990, The existence of steady states in growth models with renewable resources and pollution. Discussion papers no. 299, Department of Economics, University of Helsinki, Helsinki.