Cotton Boll Weevil: An Evaluation of USDA Programs : a Report (1981)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

APPENDIX C* ADDITIONAL COMMENTS ON THE USDA ECONOMIC EVALUATION Arnold Paulsen This appendix will comment on l} the goals and approach; 2) the data base; 3) the method or model used to estimate beltwide benefits and costs; 4) a graphical interpretation; 5) the accuracy and the risk of the estimates; and 6} the policy inferences and public choice recommendations that can be based on the estimates of the Economic Evaluation Report (Economics and Statistics Service l98lb). Goals and Approach The overall goal of the economic evaluation was to facilitate the choice of a beltwide cotton insect management program and more specifically "to estimate and evaluate the current and future local- ized and beltwide economic impacts of alternative boll weevil/cotton insect management programs" (see page 2, Economics and Statistics Service l98lb). The beltwide focus of the economic evaluation was appropriate to "facilitate the [public] choice." This meant the economic analysis could not draw all data from the trial areas. The future focus of economic evaluation was also appropriate but implied that projections or forecasts of the future had to be made and that these would be unverifiable and speculative. The economic analysts recognized and *This appendix was prepared by Dr. Arnold A. Paulsen, a member of the NRC Committee, to provide additional evaluations of the USDA Economic team's report beyond those included in Chapter 5. The views ex- pressed in this appendix are those of Dr. Paulsen and are not neces- sarily those of the National Academy of Sciences or of the National Research Council. This appendix was not subjected to the customary review process of the National Research Council. ll0

lll reported a great deal of geographical variation in program impact among areas of the boll weevil belt. USDA did not report variation in cost of implementation due to weather, or variation of payoff or benefits due to weather or other factors. Only the normalized esti- mates are presented. The USDA Economic Evaluation correctly perceived that the imple- mentation of alternative public cotton insect strategies is similar to the adoption of technological advances. Each of the proposed cotton insect management programs is assumed to be fully adopted and successful, which they may not be. USDA apparently expects the benefit to continue forever which it may not. A summary of USDA economic feasibility findings is contained in a table reproduced in this report (Table 5.2). In this table the present value of estimated public costs are compared with the present value of estimated future national benefits. These market benefits of full implementation of each alternative are estimated to far exceed the estimated public costs of implementing BWE or OPM. The difficulty is that the probability of the benefits actually being so large as estimated may be small. The probability of the program costs being as small as estimated also appears to be very small. Thus the probability of the B-C ratio being accurate may be exceed- ingly small. For the public choice among BWE, OPM, or CIC the question of economic feasibility or rate of return to public expenditure is important. There are, however, several additional considerations in deciding among public eradication, public management, or continuation of current insect policy. First, public resources are limited, and only those projects expected to be feasible may be carried out. Second, non-market costs and returns of BWE and OPM may make the social benefit to cost ratio larger or smaller than the market or economic B-C ratios. For example, no cost or benefit is included for change in human health, aquatic life, or environmental quality in the economic feasibility calculation. Third, income redistributions are estimated by USDA and reported, but of course, not included in eco- nomic feasibility. The prospect that economic growth may be stimu- lated in the boll weevil belt relative to other cotton areas as a consequence of public boll weevil management may make public action more or less desirable. Fourth, other consequences may favor more public participation in cotton insect control such as "the joy of being free from boll weevil," or limit public participation such as the fear of urban taxpayers that public costs may overrun. The Data Base The task of the economic evaluation team was to provide esti- mates of the economic consequences of beltwide implementation of each of the proposed boll weevil eradication and management strategies. Data that could be used to make these estimates were accumulated through a Delphi technique. The USDA Delphi method was well designed to identify probable insecticide-use and lint-yield changes (Eco- nomics and Statistics Service 1981d) .

ll2 The data on future BWE and OPM program actions and costs were generated by an informal "Delphi method" not by hard science measure- ment. The required future insect management and eradication actions and costs are "speculation" or expert opinion rather than scientific data. The economic report does not describe completely the procedure and qualification of experts used to generate program costs. The experts formulating program and public cost data seem to have been less numerous, less diverse in opinion, and less representative of the full spectrum of expert opinion than the panel of experts in- volved in generating the data on cotton yield and control cost change. As a result the data on the program cost were obtained by a method less systematically designed and thus more uncertain. The full range and uncertainty of the volume of BWE and OPM activities required to obtain boll weevil eradication or management below the economic threshold remains unreported. The program cost data provided by USDA cannot be used to accept or reject hypotheses about the level of future public costs with known confidence. The USDA team can be criticized for not reporting in more detail the uncertainty and diversity of expert opinion about program cost. The program costs reported by USDA completely ignore the important issues of probabil- ity of program success and rate of adoption and opposition. The Method or Model No itemization is given by USDA of the sources of consumer benefit (Table 5.2). An approximation can be inferred by multiplying expected price changes by estimated l980-l98l quantities, see Table C.l. From this approximate itemization it seems that only one third of consumer benefits are directly from additional lint. The equa- tions of the USDA model seem to have transferred most of the conse- quences of reduced cost and increase yield on cotton into price depressing effects on cotton seeds, vegetable oil, and grains. This is a puzzling distribution since the gain in lint is 2.8 percent and the market for lint—at least domestically—usually is unwilling to absorb more lint even at sharply lower prices. The gain in cotton seeds is also 2.8 percent, but the market for oil seeds is very large and usually willing to absorb more material at a slightly lower price. The USDA estimates total U.S. cotton lint production would increase 2.8 percent, and the U.S. cotton lint price would be de- pressed 3.6 percent. These relative price-to-quantity changes indi- cate a slightly inelastic demand. Since the U.S. cotton market is relatively open, U.S. cotton prices are very similar to world cotton prices. Most additional cotton produced by better control of boll weevil probably would be exported or stored rather than sold on the U.S. market. If nearly all the additional cotton, i.e. 0.3 million bales, were exported, this additional U.S. cotton would raise U.S. exports 5 percent, but would add only about l.5 percent to world trade or add 0.5 percent to total world production. A price decline for U.S. cotton lint of 3.6 percent when 2.8 percent more is produced would be much too little if only the U.S. market were considered.

ll3 Table C.l Approximate Itemization of the Benefits to Consumers Commo dirty Expected Price Change3 Basic Simulation Quantity13 (x l06) Change in Value of Cropa (l06 dollars) Cotton - 2.73 «/lb. ll. 7c bales -l60 Corn - 0.8 C/bu. 6,l00 bu. - 49 Soybeans - 2.0 C/bu. l,560 bu. - 3l Small grain - 0.3 C/bu. 2,600 bu. - 8 Grain sorghum + l.4 C/bu. 780 bu. + l0.9 Cotton seeds -l9 $/ton 4.7 tons - 89 Cotton seed meal - 8.3 $/ton l.74 tons - l4.4 Cotton seed oil - l.62 C/lb. l,460 lbs - 23.7 Soybean meal - 3.8 $/ton 2l.3 tons - 80.9 Soybean oil - .02 C/lb. l0,l20 lbs. - 2.02 ANNUAL TOTAL $ 447 A PRESENT VALUE $6268 From Column OPM-NI-BWE in Table 6, p. 34 (Economics and Statistics Service l98lb) Dl980-l98l production from an unpublished description of the econo- metric model by Robert Taylor obtained at the April l, l98l meeting of the USDA Economic Evaluation Team in Arlington, VA "From Table 8, p. 39 (Economics and Statistics Service l98lb) If the annual total of $447 is capitalized at 7.l25 percent as a stream of perpetual benefits the present value is obtained.

ll4 But when the 2.8 percent more of production is translated to 0.5 percent more cotton produced in the world the 3.6 percent price decline seems reasonable. The increase in U.S. and foreign consumer surplus at $l60 million per year from expanded U.S. cotton production as a result of boll weevil eradication seems reasonably accurate. Generally since l974, the U.S. opportunity to export has expand- ed. Thus, the lint price depression estimated by USDA does not imply an absolute reduction in U.S. cotton lint price from l980 to l990 if OPM or BWE is implemented. The price change anticipated by USDA is relative to what would be without the improvement in U.S. cotton insect control. If long run world cotton prices rise, the contem- plated improvement in U.S. cotton insect control would merely prevent world cotton prices from rising as much as they otherwise would. The USDA seems to estimate an overly large benefit to consumers and loss to producers from 2.8 percent more cotton seeds and some small percent change in total grain. This would seem to arise from a faulty set of demand equations (or calculations) for cotton seed, vegetable oil, grains, and feeds. These few additional cotton seeds and grains would seem to add only slightly to the very large U.S. and world total supply of oil seeds, vegetable oils, and high protein feeds and grains. The aggregate of U.S. oil seeds is about 6l mil- lion tons. Thus, the anticipated l30 thousand tons more of cotton seeds would be only a 0.2 percent increase in oilseed supply. This small change would depress U.S. and world prices of oilseeds, via additional vegetable oil and high protein feeds, but probably by less than 0.5 percent. The cotton seed price reduction estimated by USDA was l5.5 percent. USDA estimates a reduction of about $89 million per year in the value of a crop of cotton seeds to cotton producers. This seems to be an overestimate. It seems to this reviewer more likely that cottonseed price would decline by a smaller percent than the percent by which cottonseed production expanded. The result would be an increase in cottonseed value of about 2 percent under OPM-BWE. The USDA demand equations estimate a loss in value of cottonseed oil of about $23 million per year or 2 percent. The volume of cot- tonseed oil is expected to rise 40 million pounds or 2.8 percent, but the price is expected to fall about 4.8 percent. This seems exces- sive. The vegetable oil market of the world is large—about 56 million metric tons—and rather open and competitive. The additional 40 million pounds or 0.2 million tons more U.S. cottonseed oil is almost negligible. Even the 20 percent smaller U.S. soybean crop in l980 did not raise the world price of vegetable oil. From better U.S. boll weevil control only about 0.l percent would be added to the world vegetable oil aggregate. This probably would depress world oilseed prices less than 0.5 percent and not 4.8 percent as USDA estimates. It seems probable that the total value of the oil compo- nent of U.S. cotton seeds would actually rise as a result of in- creased cottonseed production. The USDA estimates imply that much of the annual consumer bene- fit will be derived from depressing prices in the large U.S. feed- grain-livestock complex. The likelihood of significant price declines in the large feed-livestock aggregate by so small a technological

ll5 advance in cotton seems remote to this reviewer. Apparently USDA anticipates some more feed will be produced, but the quantity changes are not reported. The anticipated gain in efficiency in cotton is only a gain of about 2.5 percent and is applied to only 3 to 4 percent of total U.S. agriculture. The gain emerges only after l0 years of implementation. After full implementation, the Delphi panels anticipated about $50 million savings per year in cotton insect control cost on historical acreages. Additional lint yields would provide the same total lint from fewer acres, thus saving additional production costs of about $75 million per year. USDA expects more acres of land to be used for cotton because more cotton will be produced on lower-yielding, non- irrigated acres in Texas. The net release of land and other resources from cotton will be small and can have only a small impact on total U.S. agricultural capacity. A productivity gain of $l25 million in cotton—a big crop valued at $5 billion—is only 2.5 percent over l0 years or 0.25 percent per year. $l25 million in cotton productivity gain is ex- ceedingly small compared to the very large total of $l4l billion value of all U.S. agricultural production. It is only 0.09 percent after l0 years or 0.009 percent per year. Productivity gains in U.S. agriculture occur regularly from diverse technological, organiza- tional, and management improvements. Each year these raise the productivity of U.S. agriculture about l.5 percent or l5 times as much in one year as the impact of improved boll weevil and cotton insect management is expected to raise productivity over l0 years. Probably this relatively small gain in cotton resource productivity would not depress feed livestock prices and the total value of the products by as much as anticipated by USDA. Additional capacity in U.S. agriculture will be needed and can be absorbed without signifi- cant disruption of the feed-livestock markets or income loss to U.S. agriculture. The demand for U.S. agricultural products is expected to grow about 2 percent per year over the next decade or so. The real terms of trade for agricultural products may rise. Any agricultural re- source productivity improvement from cotton insect management, if accurate, might not reduce feed livestock prices absolutely but only relative to what they would be without this improvement. A Graphical Interpretation The market consequences from improvements in cotton insect management can be explained or interpreted graphically (see Figure C.l). The increased public involvement via OPM-BWE moves the supply curve down and to the right—more cotton will be supplied at the same price because the boll weevil is no longer a key pest. Before public eradication of the boll weevil USDA estimates that at $0.76 per pound ll.5 million bales of cotton lint will be offered (the intersection of the supply curve before eradication, SQ, and the Demand Curve, D, in Figure C.l). After eradication USDA estimates that 2.8 percent

ll6 I i.20r i.oo 080 LU E 06° a. § O 0.40 0.20 - t> 468 MILLION BALES OF COTTON J I I I I I I 10 12 14 16 FIGURE c.l Approximate cotton lint market in the United States before and after boll weevil eradication.

ll7 more cotton will be produced or ll.8 million bales will be offered. This larger quantity will be offered willingly by producers at a 3.6 percent lower price of $0.732 per pound (note the intersection of the Supply Curve, SA, and D, in Figure C.l). The benefit to the consumer is the change in the area under the demand and above the price line (labeled Area A in Figure C.l). This area is equal to the reduced expenditure by consumers on cotton plus the value of the additional cotton consumed. The value of this area is the consumer market benefit and can be approximated by $0.028 per pound or $l6l million per year for ll.5 million bales at 500 pounds per bale. If this benefit continues to accrue to consumers each year forever and the interest rate is 7.l25 percent, the stream has a present value of about $2.3 billion. This benefit to consumers in the form of a drop in the price is offset by a loss of the same size in net income to producers (see Area A in Figure C.l). Producer net income is the area above the supply curve and below the price line. If the price did not fall when production expanded 0.3 million bales, producers would receive $0.76 per pound times ll.8 million bales, or a gross revenue of $4.48 billion. However, at $0.732 they receive $4.32 billion or $l60 million less per year. If this loss to producers continued and the interest rate is 7.l25 percent, the stream of loss to producers has a present value of $2.3 billion. By expanding production the improve- ment in cotton insect management redistributes each year via the operation of the market $l60 million of income or welfare toward consumers and away from producers. The Area labeled A is a transfer and does not appear and should not appear in USDA calculations of economic feasibility. The costs of production for some acres are reduced by the im- proved management of cotton insects. In parts of Texas, lint yields are estimated to be improved 25 to 50 pounds per acre as a result of public involvement in cotton insect management. In parts of Georgia, Alabama, and Texas the cost of insect control is reduced $l0 to $40 per acre. The aggregate value of these yield increases and reduc- tions in cost of production over the whole belt can be indicated graphically. Note the area labeled B in Figure C.l. This is the area below the old supply curve SB and above the new more efficient supply curve SA. The dollar total value of this area each year can be approximated by multiplying the estimated saving in insect control cost by historical acres ($44 million for OPM-I and $57 million for OPM-NI-BWE) plus the lint increases of l23 million pounds for OPM-I and l68 million pounds for OPM-NI-BWE valued at $.50 per pound (the data to make these calculations are in Tables 3, 4, and l2 in Econom- ics and Statistics Service l98lb). The sum is about $l05 to $l25 million per year. This is an indication of the productivity gain to the U.S. economy of improved cotton insect management. Area B is the net market benefit of Table 5.2. Area B represents the gain in productivity or the opportunity to get more cotton for the same resources or the same cotton for less resources. USDA's presentation was not graphical but tabular. Logically the calculus done by USDA can be restated as follows in terms of

ll8 areas in the graphical presentation of Figure C.l. Consumer benefit = A Producer loss = Net before less Net after = (A + C) - (C + B) = A - B Net market benefit = Consumer benefit - producer loss = A - (A - B) = B Thus, graphically the net economic or market gain to the society from public cotton insect control is Area B, namely, the gain in yield and the saving in control cost. Area A is the gain to consum- ers but is offset by an equal size loss to producers and represents the redistribution effect of this productivity gain. The increase in cotton yield plus savings of insect control resources is the only net economic gain for the entire society and market. The size of A indicates the amount of income redistribution. Area B estimates economic welfare gain to the whole society or economy. USDA's model and tables quantify the concepts graphically presented in Figure C.l. The Accuracy and Uncertainty of the Estimates The consumer benefits and producer losses presented by USDA and illustrated graphically as Area A in Figure C.l seem inflated. Specifically the price depression in the animal feed and vegetable oil complex seems excessive. Thus, an overestimate may have been made of the redistribution consequences, namely toward consumers and away from producers. USDA estimates that if $240 million were appropriated and placed in a fund at an interest rate equal to the future inflation rate plus 7.l25 percent that the balance in that fund would pay for all the public costs of efforts needed to successfully eradicate the boll weevil forever. The probability of success from such an appropriation seems to this reviewer extremely small in light of the costs for the North Carolina trial and the Stanford Research Institute estimates. USDA's program cost estimates shown in Table 5.2 do not address the probability of success. There is uncertainty in weather, peoples' cooperation, and rate of program execution. Through 80 years of eradication history and 70 years of extension history efforts have not always been l00 percent successful. The uncertainty of the estimated ratios of costs to benefits are the product of the uncertainty of the costs and of the benefits. When both terms of the ratio are uncertain constructing a ratio multiplies the uncertainty. For example, if the benefits have a low probability (e.g., 0.l) of being so high and the costs have a low probability (e.g., 0.l) of being so low the probability of the combi- nation occurring is very low (e.g., 0.0l).

ll9 The Policy Implications and Public Choice For policy makers to allocate scarce public resources responsibly it is necessary they know not only the benefit-cost ratio but also the probability of success. Budget makers have to take informed risks, but they need to have analysts provide reasonable estimates of the risks involved with each proposal. USDA was unrealistic in providing only one benefit-cost ratio without associated probability of success. To present only one B-C ratio not clearly at the mode of the distribution and possibly with an extremely low probability of success, seems unbalanced. Several alternative program costs, each with an explicit probability of success, should have been provided. High risk operations like OPM-NI usually look more profitable than low risk operations like OPM-I until the relative risk is considered. The USDA did not report any critical review of the uncertainties in the OPM or BWE program proposals and program costs. Alternative benefit-cost ratios which had equal or larger probability of occurring should have been presented. The uncertainty of insect eradication costs, timing, and success are well known and widely understood. Because the l98l USDA evaluation did not present alternative costs and alternative benefit-cost ratios with a probability of success associated with each, an informed and intelligent public choice cannot be made at this time. The situation is too uncertain to make a rational choice among alternative proposals for public cotton insect control.