Revised Draft February 6 2002 Lake Victoria Fisheries: Policy Conflicts Created by Predator-Prey Relations Gardner Brown*,# Brett Berger** Moses Ikiara*** *University of Göteborg University of Washington #Resources for the Future, University Fellow **Federal Reserve Board, Washington ***Kenya Institute for Public Policy, Research and Analysis Please address all correspondence to: gbrown@u.washington.edu Lake Victoria Fisheries: Policy Conflicts Created by Predator-Prey Relations Abstract Economic efficient resource use is just one policy goal. Foreign exchange earnings, employment for women and healthy people are other goals promulgated by Tanzania, Kenya and Uganda in the management of Lake Victoria Fisheries. That these social goals are in conflict is vividly portrayed in the bioeconomic predator-prey model where favoring a particular species (goal) reduces the sustainable harvest of another species (goal). Species interdependence is increasingly important to understand. However, traditional concepts such as maximum sustained yield take on a new meaning. To be operational, fishery biologists and other researchers must choose an explicit value for each fish, a rate of exchange. Data from Kenya are used to estimate the population dynamics equations and illustrate the opportunity cost of trading one goal for another. JEL Codes: Q22 Fishery and Wildlife Q28 Government Policy 2 Introduction In the world of practical affairs, managers of renewable resources such as fish pursue goals other than economic efficiency. Indeed, economic efficiency often seems not to rank very high in the scale of possible objectives. We begin first with a predatorprey model driven by profit maximization. The model is then extended to demonstrate clearly how pursuing an additional goal, such as increasing the harvest of one species to improve foreign exchange balances, for example, can decrease the harvest of another species whose important public role contributes to the nutritional (protein) requirements of an indigenous population. The conflict in goal achievement arising from the biological interdependence between species is illustrated by a case in Lake Victoria, Africa. The two interdependent species accounting for more than 88 percent of the harvest in a recent year are Nile perch (lates niloticus) and dagaa (Rastrineabola argentia). Natural scientists have given some attention to the study of multiple interacting species [Larkin, 1966, May, 1974, Pielou 1969, Maynard-Smith 1974] but there has been less study of multiple species in an optimization framework, particularly from a bioeconomic perspective. Multiple species models are not generous in the results produced. Clark [1990] concludes that while optimal steady state values often can be determined, exact solutions for the optimal path to the steady state are not known and may not exist for some problems. The simpler the model, the more likely clean results will emerge. Solow’s analysis of a Volterra model is insightful but the model has no intraspecific competition; i.e. natural resource capital has no diminishing returns. Still, he 3 concludes that even in the two species models, there are no easily obtained qualitative results resembling those gleaned from single species model. Wilen and Brown (1986) make some progress in characterizing the solution to a predator-prey optimization problem. Modest success comes at the cost of assuming a unidirectional coupled system. The lower organism enhances the growth of the upperlevel organism but the predator has no effect on the prey. This assumption seems to be rather strong and is not empirically credible for the perch-dagaa population dynamics studied below. Ragozin and Brown (1985) established the existence of a steady state and described the unique approach to it for a predator-prey system in which only the predator species is harvested. It is this model which most closely resembles the model of perch and dagaa, with harvest of both species set forth below. In addition to the analysis of multiple objectives, we contribute to the predatorprey literature by deriving some unambiguous comparative static results discussed in Sec 3. Under restrictive but reasonable conditions – met in the Lake Victoria case, a price increase for the harvested predator (prey) increases (decreases) the stock of both predator and prey. Just as in the traditional single species model with a stock externality, harvest can respond either way to price increase. In the case studied, predator (prey) harvest responds positively to an own price increase if the relative optimal stock is less than or equal (greater than or equal) to one-half the respective carrying capacity. [For final draft, see about the case of P2 h1 , h2 ] 2. The Model Perch (predator) population dynamics is given by (1.1) 4 dR R f ( R) h1 RD , dt f R 0 for R Rmsy , f R R Rmsy 0 where R= the stock of perch at time t; time subscripts are suppressed, h1 = harvest of perch, D = the stock of dagaa, and msy = maximum sustained yield (single stock definition). The dagaa enter the perch population dynamics linearly. is the effect of a unit change in dagaa on the percent growth rate of perch. In particular, assume the underlying population dynamics for the perch is logistic, (1.2) R R r1 R 1 h1 RD , R where r1 the intrinsic rate of growth for the perch, R = the carrying capacity for the perch. The logistic is introduced for illustrative purposes, but both biological and economic reasoning require any candidate growth function to exhibit diminishing returns. Dagaa (prey) population dynamics is specified as (2.1) dD D g ( D) h2 DR , dt dg dg 0 for D Dmsy , 0 dD dD D Dmsy where h2 = harvest of dagaa. Adopting the logistic form again, 5 D D r2 D 1 h2 DR. D (2.2) In this primitive model, it is assumed that the marginal cost of harvest is constant (perhaps 0), independent of the population size. The goal in the introductory model is to maximize the present value of profit ( ) e t P1 h1 P2 h2 dt , 0 where the discount rate, Pi the unit profit of harvested fish, i 1(perch), 2(dagaa). It is understood that harvest cannot exceed a certain maximum 0 h1 h1 max 0 h2 h2 max, at any moment, an assumption made for mathematical convenience. Other objective functions are possible. For example, those concerned about managing perch and dagaa might value foreign exchange at a unit value of v1 . Since most perch are exported the annual amount of foreign exchange (F) is simply F P1h1 . Then the annual payoff of perch harvesting is P1 v1 P1 h1. Similarly, since dagaa is an artisanal fishery and dagaa form a substantial fraction of the total protein consumed by dwellers around lake Victoria, planners might reasonably place a social value on nutrition that exceeds the price of dagaa since a diminution of protein gives rise to opportunity costs such as public health costs and other social concerns such 6 as absenteeism. If the social health benefit of a unit of dagaa is v 2 , then the annual payoff from harvesting dagaa is P2 v2 h2 . In both of these important cases, the social goals of increased foreign exchange and improved health formally are the same as an increase in the price of perch and dagaa respectively. Thus we can have a richer measure of welfare, yet have some economy in model formulation by remembering that the prices for perch and dagaa are placeholders for more than the mere ex-vessel price of fish. We will give concrete examples in the empirical section of how introducing multiple objectives changes the harvesting policy. Forming the current value Hamiltonian for the simple maximization problem, (3) H P1h1 P2 h2 1 f ( R) h1 RD 2 g ( D) h2 DR , with 1 and 2 the adjoint variables for (1.1) and (2.1). For a maximum of H, either we use the following bang bang controls, (3.1) h1 0 if 1 P1 , h1 h1 max if 1 P1 , (3.2) h2 0 if 2 P2 , h2 h2 max if 2 P1 , or singular controls when (3.3) 1 P1 , (3.4) 2 P2 . The adjoint equations are (4.1) 7 1 1 1 f ( R) D 2 D (4.2) 2 2 2 g ( D) R 1R In a steady state interior equilibrium, from (4.1) (4.2), (3.3) and (3.4) (5.1) f ( R) D P2 D, P1 (5.2) g ( D) R P1 R. P2 In equilibrium, according to the right-hand side of (5.1) and (5.2), the real marginal rate of return on each species has to earn the market rate of return ( ). It will be useful in subsequent analysis to rewrite (5.1) and (5.2) as (5.1') f ( R) P1 P2 D , P1 (5.2') g ( D) (P1 P2 ) R , P2 The terms to the right of f ( R) and g ( D) arise from the predator-prey relation and can be thought of as a biological technical externality. These additional terms vanish when the species’ interaction coefficients, 0. Just as the traditional stock externality (when unit harvest cost falls as stock increases) has economic parameters so too does the technical externality with the common terms, P1 P2 , in each equation. When P1 P2 0 , the technical externality is positive meaning that both optimal steady state perch and dagaa stocks are larger than for a model with independent stocks 0. The technical externality is increasing in P1 and decreasing as P2 increases. Note for future reference that when 8 P1 P2 P2 0, an increase in P2 , the price of dagaa, makes P1 P2 smaller, and contributes to reducing the optimal dagaa stock as a result of an increase in the price of dagaa. Rewrite (4.1) for 1 0 as (6) 1 1 f ( R) D 2 D . If the figurative owner of predator capital sells a unit for harvest, the rental rate of 1 is earned. The owner bears two future changes in perpetuity so they are capitalized by . First is the marginal contribution of that fish to the fishery whose unit economic value is 1 . Second, drawing down perch by one unit increases the marginal productivity of dagaa by D , which is valued at its shadow value of 2 each period to the figurative owner of the prey. Thus the owner of perch or society in general would gain 2 D by the dimunition of perch by one unit. A similar economic interpretation is possible by rewriting (5.2) usefully. Using the specific functional forms for the population dynamics in (1.2) and (2.2) in combination with (4.1) and (4.2) in steady state, yields (7.1) R R R D P1 P2 1 , 2 r1 2r1 P1 (7.2) D D D R P1 P2 1 . 2 r2 2r2 P2 (7.1) and (7.2) illustrate that the perch steady state population (R) is a linear and positive function of dagaa (D) and the steady state population of dagaa is a linear and positive function of perch, for P1 P2 0. 9 Solving, yields (8.1) D* (8.2) R* D 2(r2 )r1 P2 R (P1 P2 )( r1 ) 4r1 r2 P2 D R P1 P2 P11 2 R 2(r1 )r2 P1 D (P1 P2 )( r2 ) 4r1r2 P1 D R (P1 P2 ) 2 P21 It will be useful to determine maximum sustained yield (msy) for perch and dagaa. This is not a trivial determination since Rmsy and Dmsy are not independent. In fact, one must choose weights for perch and dagaa harvest to obtain a solution as we show in appendix 1. Assuming that the harvest of each stock is weighted by its price, then (9.1) Rmsy (9.2) Dmsy R 2r1 r2 P1 D r2 P1 P2 4r1 r2 P1 D R P1 P2 P21 2 D 2r2 r1 P2 R r1 P1 P2 4r1 r2 P2 D R P1 P2 P11 2 Note that the solutions to the MSY problems are just the equilibrium levels for = 0. That is, Rmsy R * and Dmsy D * when the discount rate is zero. Moreover, D * Dmsy R * Rmsy whenever (i) P1 P2 0 (ii) r1 , r2 1 because It is reasonable to assume that r1 , r2 ; the intrinsic rate of growth for each species is greater than the discount rate for the general case. If this is not true, then in the traditional “lumped parameter” model, with no stock externalities, when the predator-prey interaction term explicitly has been omitted, it would 1 10 (10.1) Rmsy R* R 2r2 P1 D( P1 P2 ) 4r1r2 P1 DR ( P1 P2 ) 2 P21 (10.2) Dmsy D* D* >0 D 2r1 P2 R ( P1 P2 ) 4r1r2 P2 DR P1 P2 P11 2 > 0. When conditions (i) and (ii) hold, positive D* and R* require positive denominators in both (8.1) and (8.2), an implication used subsequently. 3. Policy Analysis through Comparative Statics Introductory remarks warned us that not many analytical results can be easily extracted from a predator-prey model. However, new insights characterizing changes in steady state predator-prey equilibrium due to parameter changes are presented below. We would like to know how the optimal harvest of dagaa or perch vary with changes in parameters such as a price change in perch or dagaa. Such an exercise in comparative statics is useful for policy analysis. Since foreign exchange is positively related to the price of perch, then improving foreign exchange earnings is connected to changes in the optimal harvest of dagaa and its impact on nutrition levels, for example. We reach such conclusions in steps. How do the optimal stocks, and harvest of perch and dagaa respond to changes in the price of perch or dagaa? The comparative statics results are worked out in Appendix 2 and are summarized below. There are ambiguous results when P1 P2 0. Therefore, assume P1 P2 0. For a change in P1, For a change in P2, pay to extinguish the species because they are relatively unproductive capital. At no population level is the 11 dR dD 0, 0, dP1 dP1 dD dR 0, 0. dP2 dP2 Earlier we identified the predator prey technical biological externality in the steady state equations (5.1') and (5.2') where the common factor is P1 P2 . We noted that when P1 P2 0 , this pushed the optimal solution for R and D to the right of f R g D. See Figure 1. Thus when P1 increases, it increases the positive technical biological externality which increases optimal R. The reasoning is the same for the positive effect of P1 on optimal D in (5.2'). The population of both perch and dagaa decrease when P2 increases because this exogenous change decreases the positive technical biological externality. Additionally, if R* 1 R, 2 dh2 dh 0, 2 0 ; dP2 dP1 and if D* 1 D, 2 dh2 dh 0, 2 0. dP2 dP1 For all other cases, the signs are indeterminate. We can provide no economic intuition for these results. We know, for example, that R* Rmsy from (10.1) and if R* increased due to an increase in P1 in a single species model, h1 would increase. Here the condition for dh1 1 1 0 is that R* R and in this model Rmsy R as is the case in the single 2 2 dP1 species model. See Appendix 1. real rate of return greater than the return on other forms of capital. 12 4. Illustrative Solutions We have developed estimates of the parameters for the predator-prey model using data that vary from highly credible to speculative. Nevertheless, quantitative solutions do have merit for several reasons. First, we demonstrate numerically the minimum parameters necessary to reach any conclusions about harvest levels for perch and dagaa. Second, we provide estimates of the desired harvest and stock levels for perch and dagaa for Kenya using the best available information. Rational management of these two fisheries has to be guided by some empirical understanding of the biological interactions between these two fisheries and the underlying economics governing them. Policy makers can differ about the management goals for the two fisheries. Fishery biologists can disagree with the biological structure and parameter estimates of the model as can economists disagree with the economic anatomy. However, this model demonstrates the need for a basic data set and it is intended to provide the basis for building a more credible model. Third, changing the parameters gives a sense of which parameters play the most sensitive role in changing the solution values for the harvest of perch and dagaa. Parameter estimates P1 Kshs 21.45 the ex-vessel price per kg for perch , P2 Kshs 5.9 the ex-vessel price kg for dagaa . The prices are based on recent unpublished data developed by Ikiara. We assume .03 the discount rate. r1 0.15 the intrinsic rate of growth for perch provided by Schindler. r2 0.8 the intrinsic rate of growth for dagaa provided by Manyala. The species interactions coefficients are 13 = .0000002, = -.0000007. While fishery biologists agree that there is a predator-prey relationship between these two most important fisheries in Lake Victoria, no research on the interaction coefficients is available. The order of magnitude of the interaction coefficients is governed by the fact that if the interaction term is larger than 1 1 for the perch or for the dagaa, the D R predator-prey component in the dynamic population equations induce changes in the stock larger than the size of the stock. The particular values selected are based on relative values provided by fishery biologists in lake Victoria but scaled down to meet required order of magnitude values just mentioned. There are no estimates for the carrying capacity parameters but they can be estimated indirectly. Fishery biologists have beliefs about maximum sustained yield and informal discussion yielded the following estimates: h1msy 55,000 tonnes for perch, h2 msy 80,000 tonnes for dagaa Given these values we derived R 620,336 tonnes, D 882,805 tonnes, using the technique described in Appendix 1 which also yields Rmsy 343,825 tonnes Dmsy 465,518 tonnes 14 Steady State Economic Optimum Using equations (8.1), (8.2), (1.2) and (2.2) when D R 0, h1* 47,963 tonnes, h2* 101,800 tonnes, R * 280,272 tonnes, D * 444,508 tonnes. Recent harvest levels were 98,280 tonnes of perch and 65,520 tonnes of dagaa in 1995, amounts substantially in excess of optimal values. If we could magically reach the economic optimum instantaneously, the total revenue equals Kshs 1.63 billion per year or a present value in perpetuity of about Kshs 54.3 billion. See Table 1. If we were at Rmsy and Dmsy , the total revenue is Kshs 20 million more per year. See Table 2. However, at Rmsy and Dmsy it makes economic good sense to harvest down the stocks to R * and D * , earning a lump sum of Kshs 1.49 billion and put it in the bank at .03 rate of interest to capture an annual stream in perpetuity of about Kshs 44.6 million. This strategy is a superior choice by about Kshs 24 million in additional revenue per year, a gain of about 1.3 percent (See the first three rows of Table 1). It may strike some one as a little surprising that the economic optimum and the msy solution are so close. That is because the discount rate of 3 percent is close to zero, which is the implied discount rate of the msy solution, together with the fact that harvest is unresponsive to changes in the stocks in the neighborhood of Rmsy and Dmsy . 15 Optimum Path to the Steady State Unfortunately the two fisheries are neither at the economic optimum nor at msy stock levels because this is a free entry fishery. The approximate location of the two fisheries is not known. Since recent harvest levels in Kenya exceed our estimated optimum and there is widespread concern about over fishing, there is no reason to believe that the fishery is in a free entry equilibrium at these levels. There are no estimates of stock, so, there is no way to know confidently where to begin the path to the steady state. Parenthetically, there is no current research agenda which would enable researchers to reliably estimate the population levels of perch and dagaa. Suppose we start at one-half or one-fourth of msy levels for perch and dagaa. How long does it take to reach the optimum steady state values and what is the present value of this policy? Starting at one-fourth msy, it takes about 6 years to reach the optimum which yields a present value of about Kshs 47 billion. See Table 1. Starting at one-half msy stock levels, it takes about 3 years to reach the optimum which has a present value of about Kshs 51 billion. A free entry steady state at one-fourth or one-half msy, yields annual revenues of about Kshs 720 million and Kshs 1.23 billion or about Kshs 24 billion and Kshs 41 billion respectively, in present value terms. Moving to the optimum from these two initial free entry equilibria would increase the present value of total revenue in these fisheries by Kshs 23 billion for the one-fourth msy starting point and Kshs 9 billion for the one-half msy starting point. Table 1. Present Value of Revenue Streams Given Initial and Terminal Points 16 Path NPV (Kshs) Start and stay at ( Rmsy , Dmsy ) From ( Rmsy , Dmsy ) to 55,058,346,564 ( R* , D* ) 55,789,989,139 1.33% %Change Start and stay at ( R* , D* ) 54,314,764,924 From (0.5* Rmsy , 0.5* Dmsy ) to * * 50,555,118,458 ( R* , D* ) 47,141,260,190 (R , D ) From (0.25* Rmsy , 0.25* Dmsy ) to 5. Multiple Objectives First, consider foreign exchange. More than one-half the tonnage of Kenya’s fish harvest is attributable to perch in recent years and in some years, more than 60 percent of the perch catch has been processed for export, mostly by foreign owned firms financed by the World Bank and non-governmental organizations. [Ikiara] As pointed out earlier, clearly foreign exchange earnings are so very directly related to the price of perch that a linear relation is a good approximation. Thus putting special emphasis on foreign exchange is like assuming a price increase for perch. We know from the earlier comparative statics analysis when the price of perch increases, that both perch and dagaa stocks increase, perch harvest increases and dagaa harvest decreases under the empirical conditions of the Kenya fisheries in question. Specifically, assume a 20 percent increase in the price of perch so that the perch piece of the objective function is 1.2 P1h1. The steady state supply response is inelastic for both species. The harvest of perch increases by 9 percent (about 4,000 tonnes), and this increases steady state revenues by about Kshs 94.4 million. Foreign exchange earnings increase by 9 percent as well. The harvest of dagaa decreases by 13 percent, or about 13,440 tonnes, which reduces steady state ex-vessel revenues by Ksh 79.3 million. 17 There is additionally the impact of this loss on employment for women and the loss of protein dagaa provide. Although actual steady state revenues increase annually, this could only be achieved by stopping fishing for about a year [Brett], so the loss of revenue in present value terms is about Kshs 1 billion. Dagaa is a relatively cheap source of animal protein. Declining harvest of dagaa constitutes a public health concern because nearly 50 percent of the rural population contiguous to Lake Victoria consume less than WHO’s recommended minimum of 2250 calories per adult equivalent day [Ikiara]. The greatest fraction of urban people suffering food deficiency in Kenya live in Kisumu, the center of the fishing industry. More generally, there is ample documentation of pronounced protein deficiency among the people living around the lake, especially children [Ikiara]. Shortfalls in calorie intake have public consequences such as increased morbidity and its associated public health costs as well as lowered mental capabilities when caloric loss is incident on children. Since much of the perch are exported or supplied to the tourist industry, one way to acknowledge the public benefit of increased caloric consumption is to put more weight on the harvest of dagaa, v2 h2 . Then dagaa’s contribution to the objective function is P v2 h2 and can be analyzed as if the price of dagaa has increased. We have established from the comparative statics analysis that an increase in P2 increases the dagaa harvest and decreases the perch harvest. These changes increase women’s employment, another national goal, because it has been estimated that 75 percent of artisanal (dagaa) fish traders are women [Ogutu, G.E.M, 1988], and perch processing activities are relatively more capital intensive. 18 To account for the contribution of dagaa to reducing public health costs and improving women’s employment, suppose v2=0.2 which is equivalent to a 20 percent increase in the price of dagaa or the welfare weight put on a kg of dagaa. Equilibrium perch harvest and revenues fall about 8 percent as a consequence of this change and dagaa harvest and actual revenue increases by about 11 percent. The difference in present value between this option and the original solution is kshs _______. Against this loss of income must be compared the marginal improvement in public health benefits and in women’s employment, not known. Selecting welfare weights for social goals is a notorious difficult prospect. To see if the welfare weight of 0.2 makes sense, policy makers could compare the national income cost of increasing dagaa protein (by about 11,000 tonnes per year)______ and compare it to the alternative cost of equivalent protein, assuming no extra value is placed on women’s employment for this calculation. Self Financing an Optimal Fishery Suppose we want to provide a subsidy, s, while the harvesters are not fishing and a tax while they are fishing. The tax is designed to recover the cost of the subsidy in present value terms. Suppose, we start at ½ MSY levels. It takes 3 years to get to steady state and I will assume, for illustration, that there is no harvest. We want to subsidize for the opportunity cost of forgone fishing but we only know revenues. Suppose that the opportunity cost is a fraction, , of revenues, = s. Revenue at ½ MSY is Kshs 1.23 X 109. So the opportunity cost, basis 3 years 3 from now, is 1.23 x10 9 e rt dt. 83.75 x10 9. 0 19 In steady state, revenues are Kshs 1.63 billion annually. So the tax revenues in present value three years from now are 1.63x10 9 e rt dt 50.33x10 9. 0 Solving for , = .07. Disregarding political feasibility and transaction costs, the estimated maximum tax necessary to compensate the losers in order to achieve an optimal fishery is 7 percent. If canoes, nets and labor in the fisheries have alternative employment opportunities, the tax would be less. 20 Table 2. Equilibrium Values Given Various Parameter Values Rmsy 101,802 MSY Rev/yr. (Kshs) 1,651,750,397 Optimum(*) Rev/yr. (Kshs) 1,629,442,948 52,391 88,362 1,887,700,465 1,835,204,801 96,480 42,658 114,740 1,386,663,103 1,420,375,760 50,865 93,749 43,654 112,523 1,754,410,304 1,727,368,854 478,606 60,660 56,866 53,447 84,734 1,533,560,044 1,467,493,692 259,088 437,504 55,000 80,000 45,302 108,536 1,651,750,397 1,612,092,709 465,518 237,904 430,501 55,000 80,000 42,483 115,003 1,651,750,397 1,589,785,260 465,518 195,536 416,494 55,000 80,000 36,373 127,135 1,651,750,397 1,530,298,729 h1 msy 80,000 47,963 59,538 61,863 422,399 49,946 249,858 425,668 506,397 315,403 343,825 465,518 343,825 343,825 280,272 55,000 497,712 309,018 471,319 304,464 438,327 242,637 P2 +20%=7.08 311,959 442,424 P2 -20%=4.72 .04 .05 .07 381,343 P1 +20%=25.74 374,461 P1 -20%=17.16 21 * h2 343,825 h2 msy h1 R* D* 444,508 Dmsy 465,518 Parameters\Levels Standard * Figure 1 Optimal Solution for Perch When P1 P2 0 f R (P1 P2 ) D P1 f (R ) 0 (5.1') 22 R R* f ( R ) P1 P2 D P1 References C.W. Clark (1976), Mathematical Bioeconomics, New York: Wiley. C.W. Clark (1990), Mathematical Bioeconomics, New York: Wiley. O. Flaaten (1991), “Bioeconomics of Sustainable Harvest of Competing Species,” Journal of Environmental Economics and Management, 20: pp. 163-80. P.A. Larkin (1966), “Exploitation in a Type of Predator-Prey Relationship,” J. Fish. Res. Canad, 23:pp. 349-56. R. May (1974), Model Ecosystems, Princeton NJ: Princeton University Press. J. Maynard-Smith (1974), Mathematical Ideas in Biology, Cambridge: Cambridge University Press. G.E.M. Ogutu (ed.) (1988), Artisanal Fisheries of lake Victoria, Kenya: Options for Management, Production and Marketing, Proceedings of a workshop held in Kisumau, Kenya. E.C. Pielou (1969), An Introduction to Mathematical Ecology, New York: Wiley. D.L. Ragozin and G. Brown. (1985), “Harvest Policies and Nonmarket Valuation in a Predator – Prey System,” Journal of Environmental Economics and Management, 12: pp. 155-68. R. Solow (1976), “Optimal Fishing with a Natural Predator,” in (R. Grieson, Ed.), Public and Urban Economics, Essays in Honor of W.S. Vickrey, Lexington, Mass: Lexington Books. J. Wilen and G. Brown (1986), “Optimal Recovery Paths for Perturbations of Trophic Level Bioeconomic Systems,” Journal of Environmental Economics and Management 13: pp. 225-34. 23 Appendix 1. Calculation of Maximum Sustainable Yield The maximum sustainable yield is defined as the maximum of a weighted sum, where the weights are P1 and P2 (in this context P1 and P2 are two weights and not necessarily the prices). If the weights are each set to 1, then the problem is one of maximizing sustainable harvest in tons. Alternatively, adopting an economic perspective and weighting the harvests by prices, the problem becomes one of maximizing sustainable instantaneous revenue flow. Ph 1 1 P2 h2 Maximize over h1 , h2 , R, D subject to: R0 D0 (A.1) R h1 r1 R 1 RD R D h2 r2 D 1 RD D (ie. R 0) (ie. D 0) For the above problem, r1 , r2 , , , R, and D are taken as given. Substituting the constraints into the objective function, (A.1) becomes a problem only in terms of the stock levels. The solutions to this problem are Dmsy Rmsy D 2r2 r1P2 Rr1 ( P1 P2 ) (A.2) R 2rr2 P1 Dr2 ( P1 P2 ) (A.3) 4r1r2 P2 DR ( P1 P2 ) 2 P11 4r1r2 P1 DR ( P1 P2 ) 2 P21 Substituting (A.2) and (A.3) for D and R respectively in the harvest functions gives h1 and h2 in terms of the parameters values. It is therefore possible, given the msy harvest levels, to implicitly solve for two parameter values. It so happens that the carrying 24 capacities are quadratic functions of the other variables and therefore two solutions are possible. It was found, however, that given our specific harvest levels, and the other parameter values, only 1 reasonable solution exists when the weights are equal to the prices. 25 Appendix 2. Derivation of Comparative Status Results How do the optimal stocks and harvest of perch and dagaa respond to changes in the price of each species? Differentiate (7.1) and (7.2) with respect to P1 : D D P1 P2 R DR P1 2r2 P2 P1 2r2 P2 (11) D R P1 P2 D RDP2 2 P1 2r1 P1 P1 2r1 P1 (12) Substituting (12) into (11), D D P1 P2 R P1 P2 D RD P2 DR P1 2r2 P2 P1 2r1 P12 2r2 P2 2r1 P1 and rearranging, RDD P2 P1 P2 DR D 4r1r2 P12 P2 2r2 P2 . 2 P1 RD P1 P2 1 4r1r2 PP 1 2 (13) Assume P1 P2 0 which makes the numerator 0 . Since 4r r P DR P P P 0 for R 1 2 1 2 1 1 2 * 2 0 , the denominator is positive. Therefore D 0 P1 Then from (12): R R P1 P2 D RD P2 0. P1 2r1 P1 2r1P12 P1 Recall h1 r1 R 26 r1 2 R RD , R (14.1) h2 r2 D r2 2 D RD . D (14.2) Differentiating (14.1) with respect to P1 and rearranging terms, h1 R r R R R D r1 2 1 D R , P1 P1 R P1 P1 P1 R h1 R D . r1 1 2 D R P1 R P1 P1 Thus, (15.1) R 1 h1 0 if , and undetermined otherwise. R 2 P1 Differentiating (14.2) with respect to P2 and rearranging terms, D h2 D R . r2 1 2 R D P1 D P1 P1 Thus, D 1 h2 , and undetermined otherwise. 0 if D 2 P1 In order to sign R D h1 h , , , and 2 , begin by differentiating (7.1) and (7.2) P2 P2 P2 P2 R R P1 P2 D RD P2 2r1 P1 P2 2r1 P1 (16) D D P1 P2 R D RP1 2 P2 2r2 P2 P2 2r2 P2 (17) Substituting (17) into (16): R R P1 P2 D P1 P2 R DR P1 RD 2 P2 2r1 P1 P2 2r2 P2 2r2 P2 2r1 P1 27 (15.2) 2 RDR P1 P1 P2 RD R RD P1 P2 1 2 P2 4r1r2 PP 4r1r2 PP 2r1P1 1 2 1 2 R P2 RDR P1 P1 P2 RD 2 4r1r2 PP 2r1 P1 1 2 2 RD P1 P2 1 4r1r2 PP 1 2 (18) Again assume P1 P2 0 which makes the numerator negative. Since 4r r P DR P P P 0 for D 1 2 1 2 1 1 2 * 2 0 , the denominator is positive. Therefore R 0. P2 Then from (17), D D P1 P2 R DR P1 0 2 P2 2r2 P2 P2 2r2 P2 Recall h1 r1 R r1 2 R RD , R (19.1) h2 r2 D r2 2 D RD . D (19.2) Differentiating (19.1) with respect to P2 and rearranging terms, h1 R r R R R D r1 2 1 D R , P2 P2 R P2 P2 P2 R h1 R D r1 1 2 D R . P2 R P2 P2 Thus, 28 R 1 h1 0 if , and undetermined otherwise. R 2 P2 (20.1) Differentiating (19.2) with respect to P2 and rearranging terms, D h2 D R . r2 1 2 R D P2 D P2 P2 Thus, 29 D 1 h2 , and is undetermined otherwise. 0 if D 2 P2 (20.2)