CO ysx 330 / . Giannini X V Foundation y ECONOMIC EVALUATION OF MOSQUITO CONTROL AND NARROW SPECTRUM MOSQUITOCIDE DEVELOPMENT IN CALIFORNIA M.E. Sarhan, R.E. Howitt, C.V. Moore, C.J. Mitchell '-'^ Vis RE C. UBRARY Giannini Foundation Research Report No. 330 August 1981 Division of Agricultural Sciences UNIVERSITY OF CALIFORNIA ECONOMIC EVALUATION OF MOSQUITO CONTROL AND NARROW SPECTRUM MOSQUITOCIDE DEVELOPMENT IN CALIFORNIA M.E. Sarhan Assistant Professor of Agricultural Economics University of Illinois at Urbana R.E. Howitt Associate Professor of Agricultural Economics University of California at Davis C.V. Moore Agricultural Economist ESS, USDA, stationed at Davis C.J. Mitchell Research Entomologist, Center for Disease Control U.S. Department H.H.S. Public Health Service Ft. Collins, Colorado ABSTRACT A simultaneous equation model of the behavior of a mosquito abatement district based on biological and economic data is presented. Results indicate high long term costs if heavy reliance on chemical pesticide control methods continues, due to a pesticide resistance buildup in the mosquito populations. Physical source reduction methods were shown to be more efficient both in the short and long run. A linear programming model is presented which optimizes the mix of chemical and physical control methods. Results indicate increasing costs of mosquito abatement as pesticide effectiveness declines. Simulation results of narrow spectrum pesticide manufac- turing firms indicate negative returns to research, development and marketing for most firms even with significant subsidies. ACKNOWLEDGMENTS We would like to thank the many mosquito control district managers in California who took the time and effort to assist us in understanding the complexities of mosquito control, especially Dr. Don Murray of Delta MAD, Harman L. Clement of Kern MAD, and William E. Hazeltine of Butte MAD, who provided the empirical data for this study. Also, the contri- bution of Dr. G.P. Georghiou of U.C. Riverside and Don Womeldorf of the California Bureau of Vector Control is recognized. Finally, the funding provided by the Mosquito Research Coordi- nating Committee of the University of California was greatly appreciated. r TABLE OF CONTENTS Page SUMMARY AND CONCLUSIONS vi INTRODUCTION AND OBJECTIVES 1 METHODS AND PROBLEMS OF MOSQUITO CONTROL IN CALIFORNIA .... 4 Economic and Social Significance of Mosquitoes 4 Ecology of Mosquito Species Under Consideration 9 Methods of Mosquito Control . . 9 The Problem of Mosquito Resistance to Chemical Pesticides 11 Mosquito Control Districts 17 MOSQUITO ABATEMENT RELATIONSHIPS: Analytical Framework and Development of the Models 21 Introduction 21 Mosquito Control Districts Studied 22 Types of Models for the Mosquito Control Agencies .... 23 Kern MAD: Annual-Data Model 24 Estimation Procedures and Data Sources 26 Results 28 Comparison of Relative Efficiency of Control Methods 37 Linear Programming Model 44 THE CHEMICAL INDUSTRY AND PESTICIDE PRODUCTION 48 Background 48 Environmental and Health Regulations on Pesticide Use . . 52 Industrial Research and Development Costs 54 Investment Decisions for the Pesticide Industry: A Simulation Model 56 RESULTS OF THE SIMULATION MODEL 61 Discussion and Implications of the Simulation Results ... 65 Conclusions ...... 66 REFERENCES 68 APPENDIX A 73 APPENDIX B 81 APPENDIX C 87 APPENDIX D 88 ii LIST OF TABLES Table No. Page 1 Organophosphorus Resistance in California Mosquitoes by Species and Chemical Larvicides, 1971 17 2 Nominal and Real Value of Local Budget and State Aid for All Reporting from MAD, California - 1954-55 to 1974-75 19 3 Estimated Results of Annual Abatement Model, Data Base Kern, County MAD • • 29 4 Results of Independent 10 Percent Increases in Acres Sprayed, Locations Treated and Source Reduction Activities, Kern MAD 38 5 Rank in Physical and Economic Efficiency of Control Activities, Kern MAD 39 6 Comparison of Direct Controls ^3 7 Results of Programming Model at Original, One-Half Original and Double Original Maximum Acceptable Mosquito Numbers Per Light-Trap Night 47 8 Results of Programming Model Under Five Levels of Pesticide Effectiveness and Original Standards of Mosquito Numbers ^9 9 History of Last Two New Pesticide Chemical Products Cleared for the U.S. Market by Each Participating Chemical Company, by Date of Approval 55 LIST OF APPENDIX TABLES A-1 Synthetic Organic Pesticide Production and Sales in the U.S. , 1954-1972 73 A-2 The Chemical Industry's Time and Cost Distributions for Three Firm Sizes 74 A-3 Expected Net Present Values of Investment in a Narrow-Spectrum Pesticide, with and without Subsidy, for Three Chemical Firm Sizes Under 17- and 20-Year Patent Rights and a 6 Percent Discount Rate .... 75 A-4 Expected Net Present Values of Investment in a Narrow- Spectrum Pesticide, with and without Subsidy, for Three Chemical Firm Sizes Under 17- and 20-Year Patent Rights and an 8 Percent Discount Rate .... 76 iii r LIST OF APPENDIX TABLES Table No. Page A-5 NPV Frequency Distributions for Small Firms and Low Time Requirements at a 6 Percent Discount Rate Rate 77 A-6 NPV Frequency Distributions for Small Firms and Low Time Requirements at a 6 Percent Discount Rate 78 A-7 NPV Frequency Distributions for Small Firms and Higher Time Requirements at a 6 Percent Discount Rate 79 A-8 NPV Frequency Distributions for Small Firms and ' Higher Time Requirements at a 6 Percent Discount Rate 80 B-1 Calculation of Effectiveness Index 84 D-1 Kern MAD Monthly Data Model: Empirical Estimation Results 88 iv LIST OF FIGURES Figure No. Page 1 Documented Organophosphorus Resistance in Aedes nigromaculis, California, 1960-75 14 Documented Organophosphorus Resistance in Culex tarsalis, California, 1960-75 15 Use of Parathion, Malathion, Methyl Parathion and Fenthion in Mosquito Control, California 1955-71 . . 16 Insecticide Use by Control Districts, California 1955-71 16 Effect of Declining Pesticide Effectiveness on District Control Cost at Three Mosquito Control Levels 5° Flow Chart of Calculations Required to Simulate Net Present Value for Investment Proposal 59 LIST OF APPENDIX FIGURES B-1 Summary of Calculations of Effectiveness Index for Each Species C-1 Expected Coefficient Signs - Kern MAD 87 V SUMMARY AND CONCLUSIONS The purpose of this study is twofold: (1) To examine the responsiveness of the population levels of four mosquito species, Aedes nigromaculis, Culex tarsalis. Culex pipiens guinguef asciatus . and Anopheles freeborni . to various control methods and environmental factors and (2) to examine the chemical industry's past investment in narrow-spectrum pesti- cides (e.g., minor-use pesticides for mosquito control) and to explore the impact of public regulations on the industry's investment decisions regarding these pesticides. In this analysis, the primary focus is on the direct costs of pesticides and other methods used in abatement activities. Although the externalities involved are acknowledged to be important to the problems encountered in this study, they are clearly beyond the scope of this research. A. Summary of Principal Findings !• The mosquito abatement relationships; The data base includes abatement relationships for the period from 1955 to 1975 in the San Joaquin Valley in Delta Vector Control District and Kern Mosquito Abatement District and in the Sacramento Valley in the Butte Mos- quito Abatement District. Only Kern district is discussed in detail here. Two types of simultaneous equation models, a monthly-data model and an annual-data model, were developed and estimated. The empirical estimates thus obtained outline the effects of environ- mental and control factors on mosquito populations and provide a description of the control district's past behavior and decision- making processes. If the district's behavior is assumed to be consistent over time, the models can also be used to predict the VI effect of future actions. Simultaneous equation models provide a positive economic analysis. A linear programming (LP) model also developed for the Kern district involved using coefficients from the simultaneous equation model and additional data from the district's reports. The LP model provided a normative approach to the problem, i.e., what ought to be considered an optimal cost minimization economic solution for mosquito control. The LP model also provided informa- tion on the value of having an effective chemical pesticide for mosquito control - that is, simulated market signals to the chemical industry. a. The regression models. The important findings are as follows: (1) The monthly-data models are more reliable than the annual- data models in estimating the effect of environmental factors and previous mosquito population levels on the average current population as measured by the number of mosquitoes captured per light-trap night. (2) Both the annual and monthly models showed that the mosquito control districts in California are not environmentally homogeneous and that the effects of the variables differ from one mosquito species to another within a district and within the same species between districts. Therefore, it is not reasonable to make general statements for a vast geographical area regarding control of mosquitoes or their responsiveness to the environment. vii (3) Pesticide treatments of spot locations were generally more effective in reducing the average number of mosquitoes than was broad spraying of large areas. (4) Source reduction activities generally reduced mosquito population levels. (5) High temperature, rain, and high river levels were Important factors in increasing the number of mosquitoes. (6) The pesticide effectiveness index (or resistance of mosquitoes to pesticides) is an important factor in the long- run indirect effect of pesticide control measures. (7) Past exposure to pesticides was directly correlated with higher resistance to pesticides in mosquitoes (lower pesticide effectiveness). (8) Source reduction activities were associated with lowered resistance of mosquitoes to pesticides. (9) In Kern control methods for nigromaculis ranked in order of efficiency are: (1) ditch construction, (2) construc- tion of fills, levees, etc., and (3) locations treated with pesticides. The cost of a 1 percent reduction in nigromaculis numbers is 4.85 times higher for spot spraying than for construc- ting ditches, and 1.04 times higher than for constructing fills, levees, etc. In 1975-76 cost conditions the rank remains the same and the cost of a one percent reduction in average numbers A. nigromaculis would be 6.05 times higher for spot spraying than for constructing ditches, and 1.46 times higher than for construc- ting fills, levees, etc. For C. tarsalis efficiency rankings are: (1) ditch construction, (2) locations treated with pesticides, (3) construction of sumps, ponds, etc., and (4) construction of fills, levees, etc. The cost of a 1 percent reduction in the average number of Cj^ tarsalis mosquitoes by construction of fills, levees, etc., is 18.02 times higher than that for constructing ditches, 7.72 times higher than for treating lo- cations with pesticides and 1.53 times higher than for con- struction of sumps, ponds, etc. b. The linear programming model. The principal findings are: (1) The optimal mosquito control plans for three standards dictating the maximum acceptable number of nigromaculis and tarsalis mosquitoes indicate that source reduction activities can be substituted for pesticide control measures. (2) The total number of hours of labor involved in control activities, and hence the costs, increase in inverse propor- tion to the number of nigromaculis and Cj_ tarsalis mosquitoes defined as being acceptable. (3) As the effectiveness of pesticides decreases, more source reduction activities must be operated in order to keep A. nigromaculis and tarsalis mosquito population levels at or below the specified acceptable standard. (4) When high river flow (i.e., 2.5 times higher than the average) is present, more source reduction is allowed, and pesticide effectiveness declines from 100 to between 75 percent and 50 percent, labor hours and costs must be increased drastic- ally in order to keep the number of mosquitoes at or below the Ix level defined as acceptable. When pesticide effectiveness is 25 percent or below, no feasible control plan has been developed that can operate within specified restraints and resources. Also, as river flow increased, the preferable method of pesticide usage is spraying large areas rather than treating spot locations. 2. The chemical industry's investment in pesticides: A computer simulation model was developed to examine the effect of various regulations, patent-right periods, subsidies and interest rates on the chemical firm's investment decisions under risk and uncertainty. The important findings are: (1) Complying with government regulations has lengthened the time and thus the production costs for all sizes of firms from discovery of a pesticide to marketing date. (2) The market for narrow-spectrum mosquito pesticides is so limited that only the smallest firms are likely to be willing to produce them without substantial subsidies. (3) Even with three hypothetical subsidy levels, invest- ment in narrow-spectrum pesticides is generally unprofit- able for firms of any size under present regulations, patent- right periods, and 6 or 8 percent discount rates. The few exceptions to this conclusion are in the cases of small firms under less stringent government regulation (i.e., condi- tions in or prior to 1967) which have a 6 percent discount rate, the highest subsidy level (i.e., $4,013,518, or $2,829,129 when discounted) and both 17-year and 20-year patent rights; and for large firms under the same conditions as above but with only 20-year patent rights. Inferences and Recommendations 1. The findings of the abatement models lead us to infer that even though pesticides can generally reduce mosquito population levels, their use has been overemphasized. Although source reduction activities are generally more economically efficient in controlling mosquitoes, the effect of these activities has been underestimated and they have not been efficiently sub- stituted for chemical control. We therefore recommend that pesticides be de-emphasized by mosquito control districts and various source reduction activities be substituted for them as appropriate. However, in such emergency situations as epidemics pesticides must still be used in order to reduce mosquito pop- ulation levels immediately. 2. Study results also show that the continued use of pesti- cides has led to heightened resistance to chemicals among the mosquito species treated. Moreover in the past, replacement pesticides were more readily available than they are at the present time or will be in the future. We therefore again stress the recommendation that the use of pesticides be de- emphasized in favor of source reduction activities whenever possible. 3. Among other benefits, such a measure should help to pre- serve the effectiveness of pesticides by reducing the amount of selection pressure on mosquitoes. xi The regression models did not consider the value of an effec- tive and immediate control method in case of crisis, but the LP model showed that spraying is necessary in the case of high river flows. Therefore, although there is no single solution to this question, it appears that an effective pesti- cide should be available at all times. 4. Data from the National Agricultural Chemical Association and the results of the simulation model for three sizes of chemical firms showed Chat the cost and time required for dis- covery, R&D and registration of pesticides have increased with the addition and enforcement of government regulations. Because costs associated with producing and marketing broad- and narrow- spectrum pesticides are the same the potential market is much smaller for the latter, conditions do not favor their unsubsi- dized production. However, the model did not indicate economic grounds for the State of California alone to subsidize the in- dustry's production of pesticides to be used only for mosquito control. Limited data and time constraints did not allow us to investigate the feasibility of a subsidy by out-of-state users of pesticides which were developed for California use. xii INTRODUCTION AND OBJECTIVES The Problem The world, in general, and the United States, in particular, have become increasingly concerned about pollution of the environment. One area of major concern focuses on pesticides, primarily those persistent pesticides that circulate through and accumulate in certain parts of the environment. Pesticides, and the benefits and costs associated with their usage, have therefore been a controversial topic for more than a decade. In addition, since the early 1950's mosquito control agencies in California have noted that mosquitoes were becoming increasingly resistant to traditional pesticides [Gillies et al. 1973]. This biological situa- tion has forced control districts to use new and often more expensive chem- icals or to shift to nonchemical control methods, which are aimed at creating conditions unfavorable to mosquito production by altering their environment, thereby reducing their potential abundance and spread of the insects. However, nonchemical methods are geared for long term rather than immediate effects, they do not provide sufficiently prompt results for use in the event of emergency situations. Many mosquito control agencies therefore argue that development of new, effective chemicals is a necessity for adequate control, especially in the case of malaria or encephalitis epidemics. 1 Although nonchemical control methods were not widely used by mosquito abatement agencies in the past, integrated controli./ programs have been suggested and used by many districts in recent years. Many abatement district managers and entomologists argue that pesticides used in such programs should be toxic to only mosquitoes and that they must not cause important damage to the environment. If such pesticides were used only for mosquito control, they might remain effective for long periods. In the past, however, many broad-spectrum pesticides were widely used for both mosquito control and agricultural pest control, with the result that mosquitoes developed resistance to pesticides. Thus, the broad-spectrum pesticides were not effective as long as might be desired. Under today's regulatory conditions, chemical firms must invest large sums of money to develop new compounds before the first dollar is received from sales. In addition to ever-rising costs of research and development for new, effective pesticides, registration^/ costs have rapidly increased [Ernst and Ernst 1971, 1973; Little 1975]. In the past, registration of a new pesticide hinged on its efficacy and safety, with contamination of food stuffs and danger to workers having contact Ij Integrated control is used here to mean that biological, chemical and physical techniques for controlling pests are coordinated or "integrated," and are used simultaneously in such a way that each method is compatible with the other [Mulhern 1973]. 2j The registration of pesticides begins when the manufacturer submits a request to the regulatory agencies (e.g., EPA) to produce a certain pesticide. Manufacturers are required to submit detailed biological, toxicological and ecological data in order to obtain permission for final development and sale, and this registration stage of production takes several years to complete [Lever and Strong 1973; Little 1975]. with the product being major concerns. Recent legislation requires man- ufacturers to supply technical data on the environmental impact of all pesticides [Djerassi, Shih-Coleman and Diekman 1974; Hunter 1973]. Because of the large investment required, a firm must be assured of a substantial market potential for a new chemical before attempting to produce it [Fitzsimmons 1972]. While costs to the pesticide producer escalate, public health, abate- ment district and other officials insist that new and effective materials to replace those of waning usefulness should be forthcoming and available to the mosquito control agencies. These officials argue that without immediate effective control, major events such as floods or earthquakes could cause mosquito densities to greatly increase. Such situations may very well result in epidemics, substantial losses in livestock production, and continuous public complaints of annoyance caused by mosquitoes. However, economic incentives for chemical companies to produce pesticides in general, and narrow-spectrum^/ insecticides in particular, have been reduced to the point where those products, which might be urgently needed to meet short -run emergency situations, may no longer be forthcoming. For this reason, the U.S. Congress, the chemical firms and their customers, and particularly, the California abatement agencies, have shown an interest in the possibility of public subsidy of the research development and registration costs for minor- use insecticides [Brady 1972; Djerassi, Shih-Coleman and Diekman 1974; Fitzsimmons 1972]. 3/ "Narrow-spectrum" insecticides, by definition in contrast to ~ "broad-spectrum," are less harmful to nontarget species and organisms. The latter are more profitable to a firm, however, since they have a larger potential market. The former may be- come unavailable because of their small potential markets, even though they may provide benefits to society which are not reflected in their market value. Objectives The overall objectives were to: (1) Examine the responsiveness of the population levels of the predominant California mosquito species to alternative abatement methods and environmental factors and (2) examine the profit potential of narrow-spectrum pesticides for a chemical company and analyze the impact of public regulations on chemical industry invest- ment decisions. The following procedures were involved in the study: 1. To collect and summarize information regarding mosquitoes and abatement operations in California. 2. To identify and estimate the physical and biological mosquito abatement relationships, including factors which influence the effectiveness of pesticides. 3. To use an economic efficiency criterion to determine whether the control agencies were making optimal resource allocation decisions among alternative control methods. 4. To examine the chemical industry's investment in pesti- cides during the past two decades and the impact of public regulations on the industry's investment decisions. 5. To develop a model for the chemical industry's investment in pesticides. To provide information on the industry's future investment situation and to determine the potential profit for firms investing in pesticides - particularly in narrow-spectrum pesticides. METHODS AND PROBLEMS OF MOSQUITO CONTROL IN CALIFORNIA Economic and Social Significance of Mosquitoes Although there are many species of mosquitoes in California, Culex tarsalis, Culex pipiens quinquef asciatus . Anopheles f reeborni and 4 Aedes nlgromaculls are of particular concern to this research. Culex tarsalis is the most important vector (carrier) of encephalitis in California. Aedes nigromaculis , "the irrigated pasture mosquito," may adversely affect livestock production and is a nuisance to humans and other animals. Culex pipiens quinquef asciatus , "the southern house mos- quito," attacks man, invades homes, and is considered a great nuisance. It also is a vector of St. Louis encephalitis virus in some parts of its range. Anopheles freeborni, the "western malaria mosquito," an efficient malaria vector and nuisance, poses a continual threat of malaria epidemics in parts of California. (1) Economic losses in the agricultural sector. Three decades ago agricultural losses in California were attributed to mosquitoes as a result of the reduced weight gain of meat animals and reduced milk pro- duction of dairy cows. In spite of early recognition of the problem, there has been no economic evaluation of the full impact of mosquitoes on beef cattle or dairy and poultry production in California. The primary reason for this is the difficulty involved in obtaining basic data showing quantitative cause-and-ef f ect relationships. A few studies have been conducted elsewhere. Hoffman and McDuffie, [1963], estimated that cattle producers lost $231,250 due to mosquitoes during the midpoint of the 1962 mosquito season in Cameron Parish, Louisiana. Sanders, Rieme and McNeil [1968] reported that Texas Gulf Coast cattlemen who attempted summer grazing observed a reduced feed intake by their cattle as a result of the continuous irritation and blood losses from mosquito attacks. Deaths caused by suffocation from inhalation of mosquitoes were also reported in both young and weak cattle. Steelman, White and Schilling [1972, 1973] determined that mosquitoes had a substantial effect on the average daily weight gain of steers fed various energy rations in Southern Louisiana. However, this study was carried out under controlled indoor conditions [MacClelland 1975]. Husbands [1973], suggested that the agricultural losses can vary and include: "(1) Weight losses by beef cattle (2) Reduced survival in calves (abortion) (3) Reduced milk, production (4) Reduced poultry production (survival, weight, eggs) (5) Reduced efficiency of farm employees (6) Economic losses through: a. Taxes needed for mosquito control b. Reduced land values c. Indirect losses in tax monies resulting from community losses due to sick leave, etc. d. Losses due to fees charged for veterinarian services (7) Recreational and aesthetic values lost in rural areas." Another item can be added to this list: the difficulty which farmers may en- counter in hiring workers in areas with high mosquito infestation. Crop losses can occur due to the inability to handle perishable crops at the proper time. This was a difficulty faced by peach growers in Sutter and Yuba counties be- fore the formation of a mosquito abatement district in 1946. (2) Mosquitoborne diseases in California. Although the effect of mos- quitoes on public health is generally thought of in terms of transmission of disease agents, such as viruses, protozoa, and helminths, there are also in- direct health effects caused by annoyance as well as impacts on economic and food production losses. The causative agents of malaria, yellow fever, dengue fever, encephalitis and filariasis are transmitted by mosquito bites. Western equine (WEE) and St. Louis encephalitis (SLE) and malaria threaten the human population in California, and WEE may significantly affect the state's horse population. 6 In California, excessive rainfall and snowpack in the Sierra Nevada mountains may result in extensive flooding in the Central Valley of California [Sudia et al. 1969], thereby creating conditions favorable to a rapid increase in C. tarsalis populations. Reeves [1968] reported a positive correlation between excess river flow, high Cj_ tarsalis populations, and increased risk, of encephalitis virus transmission to people and horses in Kern County. WEE and SLE have been endemic in California at least since 1933 and are occasionally epidemic. Approximately 800 human cases were reported in California during the 1952 epidemic. Human cases range in severity from those with inapparent infections to serious illness resulting in stupor or coma, and severe fulminating illness and death in 24 to 48 hours. Malaria is a major cause of death worldwide. In California, freeborni and A. punctipennis can transmit the disease [Mulhern 1973], which was introduced into the state in the early 1800 's. A major epidemic occurred in the Central Valley in 1883 and minor local epidemics occurred near Lodi in 1934-1935 and near Winters in 1938 [Brady 1972]. Bailey [1972] reported a great increase in the number of recorded cases of malaria during the 1950' s and 1960's, due largely to imported cases among veterans from Korea and Vietnam. These cases were scattered throughout the state, but the majority occurred in metropolitan areas or locations lacking mosquito species which could carry malaria. Bailey therefore concluded that the potential for a major malaria epidemic in California is not great. Although human malaria is no longer endemic in California, imported cases occasionally serve as a source for localized outbreaks, as evidenced by the recent ones in Butte, Yuba and Sutter counties [Enterprise, 1974]. In 1974 and 197 5 malaria outbreaks were documented in these counties (11 and 19 cases, respectively). It is evident that a demand for relief from annoyance caused by mosquitoes does exist. Murray's records [Murray 1972], kept since 1948, show a correlation between the number of complaints regarding A. nigromaculis , "the pasture mosquito," and the number of female mosqui- toes (only females bite). Reeves [1965] submitted that the health of the population studied was adversely affected by the presence of mosquitoes. The above discussion of the social significance of mosquitoes clearly illustrates the negative effects of mosquitoes on the health and welfare of human and domestic animals. Due in large part to the activities of mosquito abatement districts, mosquito population levels have been kept low and there has not been a large outbreak of mosquitoborne disease in California for the last several years. However, Hardy and Reves [1973], Reeves [1965, 1970] and Sudia et al. [1971], among others, believe that there is a continuing threat of an encephalitis epidemic in California. This is particularly true in the Central Valley, which provides habitats favoring the development of large populations of mosquitoes and avian hosts for the viruses. It should also be emphasized that the small probability of a malaria epidemic in California is a valid assumption only in the context of the present situation. An epidemic could occur in the wake of a disaster, such as an earthquake, flood, or any other major disturbance which would expose the human population to large numbers of Anopheles mosquito bites, and reduced medical surveillance. 8 Ecology of Mosquito Species Under Consideration Several environmental variables have a direct effect on the growth, life cycle and abundance of mosquitoes. The most important are temperature, humidity, waterA^, topography!./ , food supply and shelter. Each variable may have a different effect on different species. Excess water increases the likelihood of survival of immature mosqui- toes being associated with a decreased density of predators, an increased availability of food, and less effective action of mosquito abatement districts [Moon 1975]. The food supply of larvae affects their mortality rate and also the size of the adults. Topography, irrigation water and shelter greatly affect both the growth of mosquitoes and the effectiveness of mosquito abatement. All mosquitoes develop in four stages: egg, larva, pupa and adult. However, the biology of different mosquito species varies and individual species must be considered when discussing their control. Methods of Mosquito Control Mosquito abatement programs are aimed at reducing known breeding sources and killing mosquitoes. Control methods can be classified into two main categories: (1) Short-term control methods (primarily chemical) and (2) long-term control methods (primarily nonchemical ). For the past quarter century, mosquito control agencies of California emphasized 4/ Meaning the amount of water; particularly excess flooding and standing water. 5/ Topography, i.e., the smoothness or unevenness of the fields or areas, can range from deep, to numerous undulations, to no undulations. When these undulations are deeper and more numerous, the mosquito potential is greater [Davis 1961]. 9 chemical control of mosquitoes far more heavily than nonchemical methods. However, nonchemical methods have been used more frequently in recent years. Short-term control: Short-term control of mosquito populations primarily involves the use of chemicals in liquid, dust, or granular form applied from the air or ground depending on topography and other conditions. Chemicals usually are used as larvicides or adulticides.^/ Chlorinated hydrocarbons, organophosphorus , and carbamate compounds have been used extensively in California both in agriculture and for public health purposes. In recent years a new class of chemicals acting as insect growth regulators has been developed. These are chemical analogues of hormones, and other chemicals which mimic hormonal action. When a growth regulator which mimics the effect of the juvenile horraoneZ/ is applied during an insect's tissue maturation when endogenous juvenile hormones are low, lethal deformities result [Bradleigh and Plapp 1974]. Although chemicals provide only temporary control, they are useful in both rural and urban areas when immediate results are necessary. Short-term control may be warranted in the suppression of mosquitoes during epidemics, the treatment of flooded areas, or in answering frequent bj Larvicides are chemicals applied primarily to kill larvae and/or pupae and can be synthetic insecticides or oil larvicides. Adulticides are chemicals applied to kill the adults. _7/ The juvenile hormone is one of three primary hormones necessary for an insect's growth and development. The other two are the brain hormone and the molting hormone [Bradleigh and Plapp 1974]. 10 complaints of annoyance. Chemical control is also used to treat mosquito breeding sites in agricultural or urban areas where occurrence is so in- frequent that other methods of control are not justified [Mulhern, 1973]. Long-term control: Nonchemical methods of control generally alter the environment and reduce mosquito breeding sources. These methods are effective in the long run, but may not have as immediate an effect on mosquito population levels as do chemicals. Nonchemical methods of control include: (1) Biological control. (2) Physical control (source reduction). (3) Mechanical barriers, including bed nets and screening of buildings. Recent studies concerning nonchemical methods of control in California are by Hoy and Reed [1970], Hoy, Kauffman and O'Berg [1971, 1972] and Murray [1972]. Although nonchemical control methods are essential for effective comprehensive mosquito control, they sometimes are not sufficiently effective to completely substitute for chemical control. Georghiou [1965] summarizes the situation as follows: "However plausible the new methods (biological control) may be, and they undoubtedly are, they do not obviate the need for insecticides." Thus, even though pesticides may play a diminishing role in future pest control strategies, chemicals are likely to remain the primary tools and an important element of integrated control for some time [Brady 1972; Spiller 1968]. The Problem of Mosquito Resistance to Chemical Pesticides The Central Valley and other parts of the state have become ideal mosquito breeding places because of complex agricultural enterprises and the continuously expanding acreage of irrigated land. In these areas there has been extensive use of insecticides for control of mosquito 11 larvae and, to a lesser degree, adults. Because each generation may require treatment, 15 to 20 treatments a year may be necessary [Spiller 1968], Such repeated exposure to insecticides applied for mosquito con- trol and chemicals used to control crop pests has resulted in the biological selection of resistant strains [Kauffman, 1975]. Thus, several economically important mosquito species, including two of the most important species, tarsalis and nigromaculis . have become resistant.^/ Georghiou [1965, 1966] emphasized that the dynamics of resistance are very complex since they are influenced by many factors including pop- ulation movement, history of selective pressure on the population, degree of dominance of each resistance factor, background environment, stages in life cycle of the insect exposed to the insecticide, and interaction among resistance mechanisms. Womeldorf et al. [1972] summarized the history of the problem of mosquito resistance to organophosphorus pesticides as follows: "In California, DDT resistance in Aedes nigromaculis , the irrigated pasture mosquito, has been known since 1949. By the early 1950's resistance against DDT and other organ- ochlorine compounds had become widespread in Aj_ nigromaculis and in the state's primary vector of St. Louis and western equine encephalitis, Culex tarsalis. The organophosphorus compounds were then substituted for the organochlorine materials. Entomologists usually use the LD50 (the lethal dosage for 50 percent of the population) as an indicator for pesticide effectiveness. The LD50 is a measure of the degree of toxicity of a pesticide, measured in ppm (parts per million) in larvlcide tests and percent concentration in adulticide tests, and is the amount of technical-grade material concen- tration required to kill 50 percent of the target pest. The higher the LD50 coefficient, the less toxic the chemical is to the organism. 12 "Parathion resistance in Aj_ nigromaculis was first documented in Kings County in 1958. Within the next few years, parathion and malathion resistance was found in many areas of the Central Valley and methyl parathion resistance also appeared. By 1970, parathion resistance had become commonplace, methyl parathion resistance was not far behind and resistance to fenthion and other organ- ophosphorus compounds had been recorded in several areas of the state in adults as well as larvae. Additionally, problems in obtaining adult control with the carbamate propoxur had begun to develop. "Malathion resistance in Culex tarsalis was dis- covered in 1956 in Fresno County. Malathion resistance progressed through the Central Valley and is now common in other parts of the state as well. Resistance against all available organophosphorus larvicides became appar- ent in the San Joaquin Valley in 1969. Resistance against malathion or against all organophosporus larvicides is now widespread. " Figures 1 and 2 show the extent of the spread of organophosphorus resistance in A. nigromaculis and C. tarsalis in California, through 1973 [Gillies et al. 1973]. (Inclusion of an agency does not necessarily mean that every mosquito population in it is resistant, but rather that some populations are.) Figure 3 shows the usage patterns of four organ- ophosphorus insecticides for mosquito control in California from 1955-1971. Instances of organophosphorus resistance in important mosquito species in California are summarized in Table 1. Figure 4 shows the usage and re- placement of one class of chemical compounds by another. 13 Figure 1: Documented organophosphorus resistance in Aedes nirgromaculis California, 1960-1975. Figure 2: Documented organophosphorus resistance in Culex tarsal is , California, 1960-1975. Figure 3: Use of parathion, malathion, methyl parathion and fenthion in mosquito control in California, 1955-1971. Figure 4: Insecticide use by control districts for mosquito control In California, 1955-1971. a/ 525 500 350 300 250 200 150 100 50 <;> Organophosphoruses Chlorinated hydrocarbons 1955 1960 1965 Carbamates 1970 a/ No data are available for 1957. Source: (Womeldorf, et al., 1972). 16 Table 1 Organophosphorus Resistance in California Mosquitoes and Chemical Larvicides, 1971. Resistance of Species to Given Chemical Larvicide; Mala- Para- Methyl Fen- Species thio n EPN thion Parathion thion ABATE Dursban Aedes nigromaculis X Aedes melanimon Culex tarsalis X Culex pipiens X Subspecies Culex peus X Source: Womeldorf, Gillies and White 1972. Because of the increasing insecticide resistance problem and inadequate substitution of other potentially effective alternative methods (i.e., long- term nonchemical controls) by the vector control districts, California is faced with the possibility of uncontrolled mosquito populations. Mosquito Control Districts Organization and funding. Attempts were made as early as 1904 to establish agencies for public control of mosquitoes in California; however, the first mosquito abatement district, in Marin County, was not organized until 1915. In 197 5 there were 77 mosquito abatement districts and municipal and county control agencies in California which served an area in excess of 40,000 square miles with a total budget of more than $12 million [California 1975; Mulhern 1973]. The districts vary in size and budget. All are publicly organized and administered with a manager and board of trustees. The trustees are responsible for setting the district's fiscal and operational policies, 17 XXX XXX X XXX XXX XX X XXX and carry the power to levy taxes and prior to 1978 had limited authority to increase taxes on properties in the district in order to finance abate- ment operations. Since state subvention was discontinued in 1967, districts are financed entirely through local funds. Local funds are raised by a levy on taxable properties within a district. The control district's budget trends and the proportion of local and state funds to the total budgets have changed over the years [California 1975]. Table 2 illustrates the changes in and sources of funding from 1954 to 1974. Monitoring. The major objective of mosquito abatement districts is to control mosquitoes and, to a lesser extent, other insects (e.g., flies and gnats). For vector control districts to decide which control method is the most appropriate they must have a population monitoring system for each mosquito species in the district. However, estimates of the total adult mosquito population of any species within a specific area are not routinely made due to the expense and effort involved. Reasonably ac- curate methods, based on mark-release-recapture techniques, are available for estimating absolute population densities, but these are impractical except for use in experimental studies and are not essential for effective operation of a control district. A widely used index of relative abundance is based on light-trap col- lections, i.e., the number of mosquitoes per light-trap night. Control districts usually operate light-traps in several urban and rural areas. The number of traps operated, the frequency of operation, and their place- ment may differ from district to district, thus limiting the usefulness of these data for making comparisons between districts. Another drawback to 18 Table 2 Nominal and Real Value of Local Budget and State Aid for All Reporting California Mosquito Control Districts from 1954-1955 to 1974-1975 Nominal Budget Deflated Budget Fiscal Year Local State Subven- tion Total De- flator^ Local State Subven- tion Total Index 1954-55 $ 2,790,553 $342,183 $ 3,132,736 .8770 $3,181,930 $390,174 $3,571,104 1955-56 3,446,851 360,555 3,807,406 .8925 3,862,017 403,983 4,266,001 1956-57 3,445,887 347,480 3,793,367 .9200 3,745,529 377,695 4,123,225 1957-58^ .9395 1958-59 4,114,879 390,368 4,505,247 .9470 4,345,173 412,215 4,757,388 1959-60 4,391,876 104,695 4,496,571 .9485 4,630,338 110,379 4,740,717 1960-61 5,309,808 115,376 5,425,184 .9470 5,606,977 121,833 5,728,810 1961-62 5,132,426 110,612 5,243,038 .9465 5,422,531 116,864 5,539,395 1962-63 5,904,423 143,202 6,050,625 .9465 6,238,164 154,465 6,392,630 1963-64 6,384,199 144,600 6,528,799 .9460 6,748,624 152,854 6,901,478 1964-65 6,673,971 79,660 6,753,631 .9565 6,977,491 83,282 7,060,774 1965-66 7,076,598 50,000 7,126,598 .9820 7,206,311 50,916 7,257,228 1966-67 7,428,742 50,000 7,478,742 .9990 7,436,178 50,050 7,486,228 1967-68 7,818,601 7,181,601 1.0125 7,722,075 7,722,075 1968-69 8,267,290 8,267,290 1.0450 7,911,282 7,911,282 1969-70 8,914,503 8,914,503 1.0845 8,219,919 8,219,919 1970-71 9,713,372 9,713,372 1.1215 8,661,053 8,661,053 1971-72 9,879,474 9,879,474 1.1650 8,480,235 8,480,235 1972-73 10,225,437 10,225,437 1.2690 8,057,869 8,057,869 1973-74 11,089,851 11,089,851 1.4735 7,526,196 7,526,196 1974-75 12,925,425 12,925,425 1.6000 8,078,390 8,078,390 a. Data for fiscal 1957-58 were not available. b. Consumer Price Index, U.S. Department of Commerce. Source: Calculated from information obtained from the California Mosquito Control Association, Yearbooks , Annual Issues. 19 using the light-trap index is that not all mosquito species are equally attracted to light. Other sampling techniques must be used to measure the relative abundance of certain species. For example, it would be unreasonable and ineffective to develop a control strategy for A. nigromaculis based solely on light-trap indices. Landing countSL^/ are probably the most accurate index for this species. Complaints by res- idents reflect the degree of mosquito annoyance in an area and hence are an indirect measure of mosquito population levels. Both landing counts (for A. nigromaculis ) and complaints (for all species) are often used to supplement the light-trap index in deciding on which abatement activity to implement and at what level. Pesticides and source reduction control efforts. Until recently vector control districts emphasized chemical control far more than biological control and source reduction methods. From 1962 to 1974, on the average, vector control districts increased their source reduction budgets from 21.2 to 25.6 percent of the total budget. The pesticide resistance problem, environmental and safety requirements, and circumstances such as weather and mosquito intensity may have contributed to the increase. The number of districts reporting source reduction activities increased from 27 in 1962 to 39 in 1974. The abatement district's emphasis on chemical pesticides has helped them to reduce mosquito populations, but it has also contributed to the problem of mosquito resistance to chemicals. Resistance, in turn, has - ^/ In a "landing count," a person stands in a selected area and allows mosquitoes to land on him while he counts the number landing per unit of time. 20 required a continuous substitution of one pesticide for another or of one class of chemicals for another. In any case, whether these chemicals are used in a manner similar to or different from that in the past, they will be produced under more difficult conditions of shrinking potential markets and expanding public safety regulations. MOSQUITO ABATEMENT RELATIONSHIPS: ANALYTICAL FRAMEWORK AND DEVELOPMENT OF THE MODELS Introduction Essentially, the problem of mosquito control is similar to that of pest management in the agricultural sector. The aim in both cases is to minimize a pest population subject to a variety of conditions or con- straints. With crop pests, the problem is to find an optimal control strategy which minimizes the pest population at minimum cost so that an optimum crop yield is produced. With noncrop pests (e.g., mosquitoes) the common objective is to find a minimum cost strategy which reduces the pest population to a level at which the incidence of diseases they transmit and the annoyance they cause to humans are tolerable. Studies on the economic impact of mosquitoes and other pests of public health importance have been relatively few. Economic studies of pest management have been dominated by interest in agricultural pests, with only minor studies concerning nonagr icultural pests (e.g., mosquitoes and gnats). Most of these studies are based on theoretical approaches to the problem; few are empirically applied. An empirical study in the United States on mosquito abatement was done in 1974 by D.V. DeBord [1974] in which he investigated the demand for and cost of salt marsh mosquito abatement for 30 East Coast mosquito 21 abatement agencies. DeBord used a four-equation simultaneous pest management model to examine the responsiveness of mosquito density to abatement activities by the agencies and to examine the incentives to collect taxes for mosquito control purposes. The four components or sub- models were mosquito abundance, temporary control (chemical), permanent control (source reduction), and abatement demand. Another model was util- ized to check for possible economies of scale in the control operations. Analysis of 30 abatement agencies from 1959 through 1971 revealed that mosquito populations were reduced significantly with both chemical and nonchemical control measures. There were economies of scale with respect to the construction of source reduction measures but not with pesticide spraying activities. Study results indicated that the use of pesticides was three to four times more effective in reducing mosquito density than permanent control measures. Finally, results showed that demand for abatement (as measured by local per capita expenditure on control measures) is affected by income and population of the district, state grants for abatement, tourism and mosquito population levels. However, DeBord 's study did not take into account the important vari- ations in environmental conditions and control activities within the season by taking the time unit for observation to be one year. Also, the study failed to take into account the buildup of chemical resistance in mosquitoes and made the assumption that the entire area studied, which in- cluded regions in five states, is environmentally homogeneous. Mosquito Control Districts Studied In evaluating mosquito abatement relationships in California we should recognize that mosquito life cycles and breeding habitats and the effects 22 of mosquitoes on humans and domestic animals vary according to the mos- quito species involved. In addition, the environmental characteristics encountered by different abatement agencies may differ in climate, pop- ulation, important mosquito species and agricultural background. The choice of districts was limited by the quality and availability of organized, reliable data needed for the empirical research and the number of districts chosen was limited by time. The three districts selected are the Delta Vector Control District (VCD) of Tulare County, the Kern Mosquito Abatement District (MAD) located in Kern County in the San Joaquin Valley and the Butte Mosquito Abatement District of Butte County in the Sacramento Valley. These districts operate as public agencies for both rural and urban areas. However, no distinction was made between rural and urban activities because there was no clear separa- tion in the districts' reports covering the period of this study. Types of Models for the Mosquito Control Agencies A theoretical model of the underlying biological, economic and phys- ical relationships should be defined which relates abatement strategies with policy decisions. Empirical estimation of these relationships could be used either directly to describe the control district's behavior and decision-making processes, which is a positive economic approach (as in DeBord's study), and/or the coefficients can be used in constructing a pro- gramming decision model, which is a normative economic approach. The two types of models developed and estimated for Delta VCD and Kern MAD [Sarhan 1976] were a monthly-data model and an annual-data model. These two types differ in their structure and number of equations and in the definition and nature of effect of their variables on the abatement 23 r relationships and activities. For example, the annual models include variables which are hypothesized to have a long-run influence on mos- quitoes (e.g., source reduction). For Butte MAD, the limited data avail- able and the relatively few years for which these data are available led to developing only the monthly-data type of model. Only the Kern annual model is discussed in detail here but we include a brief summary of the important results and implications from other models. It should be em- phasized at the outset that the model specification was selected after several others were tested and eliminated due to statistical or theo- retical problems and limitations. Kern MAD: Annual-Data Model Kern annual-data model consists of the following 10 equations: (1) K^ = hi(K]^j-_]^ ,K2,K3,K4,K5,K5,K7,K3,K9) ( Aedes nigromaculis population) (2) Kio = h2(Kiot-l.K:2,K3,K4,K5,K6,K7,K8,Kii) ( Culex tarsalis population) (3) ¥^12 = h3(K22t-l»'^2>''^5»'^13»'^14) ( Culex p. quinquef asciatus population) (4) K4 = h4(K4t._i,K2,K3,K]^o»'^15>'^16) (Acres treated with pesticides) (5) K5 = h5(K5{._2,K]^,K]^o>Kl2»^16) (Locations treated with pesticides) (6) K7 = h7(K4,K2^5,K]^7) (Sumps, ponds, etc., constructed) (7) Ks - h8(Ki,Kio>K4,Ki6,Ki8) (Ditches constructed) (8) Kg = h9(K]^ ,K]^7 ,K]^8>'^19 1^20^ (Effectiveness index- A. nigromaculis ).!£/ 10 / See Appendix B for a detailed explanation of method used to estimate the effectiveness index. 24 (9) Ki4 = hii(Ki2,K2o) (10) Ki4 = hii(Ki2,K20) (Effectiveness index- C. qulnquef asciatus ) (Effectiveness index- C. p. quinquef asciatus ) where K]^,Kiq,Ki2>K4,K5,K7,K8,K9,Kii and K14 are endogenous variables and all other variables are assumed to be predetermined. The definition of the variables can be summarized as follows: K]^ = average number of female nigromaculis mosquitoes per light-trap night in the year K^t-l = average number of female nigromaculis mosquitoes per light-trap night in the previous year K2 = total number of days in the year when temperatures equaled or exceeded 100° Fahrenheit K3 = total amount of river flow during the year (the sum of flows of Kern and Tule rivers) K4 = total number of acres treated with pesticides (larvicides and adulticides applied by air and ground spray) in the year K5 = total number of locations spot-treated with pesticides (larvicides and adulticides) in the year K5 = number of cubic yards of dams and levees constructed in the year K7 = number of cubic yards of sumps, ponds, etc., constructed in the year Kq = number of miles of ditch construction in the year Kg = average effectiveness index of pesticides used during the year per average application with a standard dosage (a proxy for the A. nigromaculis species' resistance to pesticides ) K^Q = average number of female tarsalis mosquitoes per light-trap night in the previous year 25 K]^0t-1 ~ average number of female tarsalis mosquitoes per light-trap night in the previous year ^11 ~ average effectiveness index of pesticides used during the year per average application with a standard dosage (a proxy for the C» tarsalis species' resistance to pesticides ) ^12 ~ average number of female Cj;_ p^ quinquef asciatus mosquitoes per light-trap night in the year Ki2t-1 average number of female Cj^ p^ quinquef asciatus mosquitoes per light-trap night in the previous year K23 = total number of inches of rainfall from January to October in the year ^14 ~ average effectiveness index of pesticides used during the year per average application with a standard dosage (a proxy for the C. p. quinquef asciatus species' resistance to pesticides) K25 = total number of irrigated crop acres considered important to mosquito production during the year Kx6 = total deflated budget for the year K]^7 = the accumulated sum of cubic yards of sumps, ponds, etc., constructed in the past 10 years Kj^S ~ the sum of the number of miles of ditch construction for the past 10 years (stock) ^19 the sum of acres treated with pesticides for all mosquito species in the past K20 = the sum of the number of locations spot-treated with pesticides in the past. Estimation Procedures and Data Sources In this section we describe the sources of data and the procedures used in the empirical estimation of the regression model's parameters, which are the basis for Kern MAD's linear programming model. The periods selected for this study were 1955 through 1974. The length of the time period is important because it allows a greater range in variation for both mosquito population levels and the important 26 environmental factors which affect mosquitoes. A longer period also permits the effectiveness of long-term abatement activities (source reduc- tion) to enter the analysis. Estimates for the parameters of the model developed were derived from time-series data for the district studies. The relationships pre- sented, because of their general interdependent nature, required the specification of simultaneous equation models. Each equation contains one or several endogenous variables which also occur in other equations. The exogenous variables are assumed to be stochastically independent of the disturbances of the system. Due to the simultaneous nature of the model specification, the two- stage least-square method is used to estimate the parameters .2.-^/ Application of the omitted variable Identification test to each of the equations shows that the order condition for identification is satisfied and each equation is overidentif led. In this study we relied on a number of sources to obtain data needed for the empirical application of the model already described. Environ- mental and climatological data were obtained from the relevant publica- tions of the U.S. Department of Commerce [1971, 1973]. River-flow data were obtained from the files of the State of California Department of Water Resources and from official bulletins of that Department [DWR, 1974]. 11/ Given the model specified, it is quite probable that some degree of serial correlation exists which would lead to inconsistency. However, reliable detection methods for serial correlation in the presence of lagged endogenous variables are only just being developed and were not available to the authors. 27 r Mosquito population index data were obtained from monthly reports of the mosquito abatement district or from the data bank of the School of Public Health, University of California, Berkeley. Data for the abate- ment operations' input amounts and costs were obtained from the control districts' monthly reports, interviews with the manager and the California Mosquito Control Association, Inc., Yearbooks. Results With 54 nonzero coefficients in the model, discussion of the expected signs on individual coefficients is precluded. The anticipated coefficient signs are presented concisely in Appendix C. For detailed discussion of the expected signs see Sarhan [19761. The results of the Kern MAD annual-data theoretical model are presented in Table 3. A brief analysis is presented below. (1) Aedee nigvomaculie population; The estimated equation indicated that the average number of A^ nigromaculis per light-trap night in any year was influenced by river flow and number of days when temperatures were of at least 100°F. Both had positive effects, as expected. The number of acres sprayed and such source reduction activities as the construction of sumps were not significant influences and had incorrect positive signs. The number of locations treated and ditches constructed had the expected negative effects, but not at significant levels. Average pesticide effec- tiveness and construction of fills, levees, etc., were not satistically significant factors but carried the expected negative signs. The average number of A^ nigromaculis in the previous year was not significant and had a negative sign. These results indicate that in Kern MAD the quality of pesticides and the number of fills, etc., constructed were the most 28 Table 3 Estimated Results of the Annual Abatement Model Data Base - Kern County Mosquito Abatement District Endogenous Variables Equation number Normalized Endogenous Varlablea 6 u C o -o « ■H 01 (A ^ B U o c o. o III - £ B o 3 4-" C U3 (U -H U . 1-t (0 a (A 3 CO IB 3 O o •rt & •H > > V e ^ CD 41 4J 41 U •o (A c ?^ 3 o o O •H CJ 0) (A •H 4-> . XI 3 41 4-> > CO O »4 CO 41 o •H U 4J 1^ o > < a P. CO cn 4) 4) >, •a •H to U 3 ■H O 4J -H m p. 4) 4J (i ^. ^ 4J M-( O O .^ CO A 1 CO CO •a 4) K4 l*H 3 O u CO u u 41 4; ^ a. B e 3 4) Z 4J 3 o ■-I u 4J > ■H a:. I g o C •o to 4-1 41 CJ 41 3 M > ^1 CO 4> W OJ ►J CO >. 12t-l At-1 5t-l CO o u o 01 41 U < 4J ec •o 3 CQ ^ CO o c 4-> O eg 4* o ■a « u u • "•S •> ti « U U 4-1 CO CO 4-1 4> CO <« e. CO o o. S 4-t 4> 3 -H ^ C/3 3 4-1 O (X I (0 -rt 4-1 4-> (0 CO CO C 41 CO O O. O. •H 4-> J3 41 a tJ £ U -H *J O » c •C3 ■H t4H 4J O 4J CO CO thus show the immediate indirect effect through the effectiveness equations of source reduction and pesticide use control methods. Of greater impact are the longer-term indirect effects of control methods. The recursive form of equation (2) clearly recognizes that pesticide resistance is genetically transmitted to future generations; and likewise a source reducing pond or sump will show positive results for a number of years. This intertemporal effect has been noted in the context of a user cost earlier in the report. The effect of a unit change in an exogenous variable sustained for a period of time on the expected value of an endogenous variable is 14/ For a review of impact and t period dynamic multipliers see Goldberger [1964], pp. 374-375 or Dhrymes [1970], pp. 521-525. termed a T period dynamic multiplier and is obtained by solving the reduced form stochastic difference equation (2) for a given time horizon. If a unit addition to a source eradication has an effective life of T years, the resulting matrix of multipliers is shown as: (3) = (I + + • • • • n^^)n2. Mindful of the effective life of sumps and ponds, T was set at eight years for an empirical comparison. In Table 6, the direct effect and immediate and long-terra indirect effects are compared for a pesticide using control method K5. All three mosquito species are tabulated despite the change in sign on the eight-year indirect effect coefficient for K5 on A. nigromaculis . For C_j_ tarsalis and 2j_ quinquef asciatus , the direct ■effect for the pesticide control method is negative as would be ex- pected. The indirect effects, however, show dramatic differences. After one year the indirect effects of pesticide use (K5) increase the change in light-trap index for tarsalis, C. p. quinquef asciatus , and A. nigromaculis . The net result of direct and indirect effects of the con- trol method, while being beneficial in the short run, diverge drastic- ally in the long run with the long-term costly indirect effects of pes- ticide use on tarsalis greatly outweighing the short-term beneficial effects. It is clear from Table 5 that the decision-makers represented by the abatement model are aware of the inequality of the marginal cost per unit population reduction factors associated with different control alternatives. For example, for C. tarsalis, the marginal cost of mosquito abatement using physical control methods is approximately 1/10 that of using pesti- cides. Physical efficiency shows similar advantages for source reduction 42 Table 6 Comparison of Direct Controls (and Short-Run and Long-Run Indirect Effects) Change in light- trap index of mosquito Location treatment with pesticides (K5) (thousands) Direct effect 1 year indirect 8 years indirect A . ni gvormou tis C, tarsatis C. p. quinquefasaiatus -0.00369 -0.0245 -0.00063 0.00077 0.00609 0.00009 -0.002661 0.53202 0.00067 methods but at a lower magnitude. Consideration of the indirect effects in Table 6 exacerbates the problem. Given these suboptimal decisions from an economic viewpoint, a normative constrained optimization model of mosquito control may be of value in setting policies. Linear Programming Model A linear programming (LP) model was constructed for a cost-minimizing mosquito control district by making use of the reduced form regression results adjusted for the exclusion of the lagged endogenous and intercept terms and constraints based on the experience of the control district. The normative approach is concerned with what ought to be, rather than a description of phenomena as they exist (i.e., a positive analysis). The purpose, then, is to develop a linear programming model for cost minimization. In the LP form, the model simultaneously selects the minimum cost combination values for the control factors and assures that the number of mosquitoes will not exceed a specified population level. Constraint Equations; The coefficients of the constraint equations (a^j's) and the constraining right hand side values (b^'s) are obtained from the reduced form (2) by expressing the mosquito growth relations as constraints. In matrix notation the set of constraints is expressed as : Ax j< b where x is a vector of activity levels of mosquito populations, control actions by the abatement district and exogenous variables such as rain- fall and river flow. The model has eight main structural features to be explained: (1) Mosquito populations, (2) specifications for maximum ac- ceptable number of mosquitoes per light-trap night, (3) upper and lower 44 bounds on the total number of acres sprayed with pesticides, (4) the upper bound on the number of locations treated with pesticides, (5) upper limits on source reduction activities, (6) specification of the annual river flow, (7) specification of the objective function (operating costs) to be mini- mized, and (8) total labor availability to the district. The mosquito population relationships with respect to environmental and man-made control activities must be satisfied. Preliminary solutions suggested that the equation for £^ quinquef asciatus be omitted because its inclusion led to an infeasible solution. However, Kern Mosquito Abatement District annual reports indicate that control activities which succeed in keeping A^ nigromaculis and tarsalis mosquito populations below certain levels simultaneously keep £^ quinquef asciatus at an acceptable level of control. Therefore, since the specification of the mosquito populations allowed for control activities to apply to all species, the omission of C. p. quinquef asciatus will not affect the optimal solution. Objective Function; A theoretical economic model would suggest that decision-makers optimize over a demand function for the output, in this case mosquito abatement. However, interviews with M.A.D. managers in California [Sarhan, 1976] revealed that the district officers do not operate as if they had hypothetical demand functions for mosquito abate- ment in their minds. Rather, they have a very inelastic standard of mos- quito nuisance they were prepared to tolerate from the principal species, under a given control technology. Under these conditions, the socially efficient objective function for the LP model is one that minimizes the sum of variable costs of control facing the M.A.D. Since the levels of 45 V the alternative control actions are a subset of the activities vector x , the objective function can be expressed in vector notation as: Minimize C'x .. : where the vector C has nonzero elements equal to the variable cost per unit for each control activity. Clearly the maximum permissible mosquito population standards cru- cially affect the outcome of the LP. The district managers were twice con- fronted with the standards used, which were based on the mean of the lowest ten years observed in a 20-year period and concurred with their levels. Subsequently, these standards will be varied to assess their sensitivity on total district control costs. The other constraints, except river flow which was set at its 20-year mean, were set at the maximum observed level in the district. To assess the sensitivity of the LP model to changes in the maximum allowable mosquito densities, the standards used in the basic model were first cut in half and subsequently doubled. Table 7 displays the results of this sensitivity analysis. Under Plan 2, even though the population density was reduced by 50 percent, the total direct costs increased only slightly. The major change occurring was a more than threefold increase in the miles of drainage ditches constructed. By doubling the allowable mosquito density standard. Plan 3, direct costs were reduced slightly below the basic least-cost solution. To meet these more relaxed standards, the annual level of ditching was re- duced to zero and fewer locations were sprayed which required less labor input. 46 Table 7 Results of the Programming Model at Original, One-Half Original and Double Original Maximum Acceptable Mosquito Numbers per Light-Trap Nighti' Ann ual level of activities/items under; 100% Activities/items in 50% )timal solution Plan 1 (original)2/ Plan 2 (below)£/ Plan 3 (above)£/ Plan 1 Plan 1 the opt Acres sprayed with pesticides 5,000 Locations treated with pesticides 500,000 Fills, levees, etc., constr. (1000 c.y. ) 49.45 Sumps, ponds, etc., constr. (1000 c.y. ) 0 Ditches constructed (miles) 5.50 A,n. mosquitoes per light-trap night .589 C,t, mosquitoes per light-trap night 2.67 Hours of labor 2,198 Total direct costs ($) 24,832 5,000 500,000 48.31 0 18.72 .29 1.33 2,277 25,295 5,000 418,067 48.76 1.17 5.34 2,083 23,973 a/ All other parameters are held at their original levels, b/ A. nigromaculis light-trap numbers £ .589; C. tarsalis numbers £ 2.67. c/ A. nigromaculis light-trap numbers _< .29; C. tarsalis numbers £ 1.33. d/ A. nigromaculis light-trap numbers _< 1.17; C. tarsalis numbers _< 5.34. 47 The loss of pesticide effectiveness through a resistance buildup and and a lack of new pesticides appearing on the market are a serious concern to control agencies. To estimate the impact on the district of increasing pesticide resistance, the pesticide effectiveness coefficient was varied in discrete steps from 100 percent down to 5 percent. The results of these runs are presented in Table 8 for the basic mosquito population control standard. These results are also shown graphically in Figure 5, along with traces depicting the effect of increasing the control standard one-half the original mosquito density standard (double the original allowable population). It should be recognized that since the LP model did not allow for an extended effect of source reduction activities the actual total benefits of those activities are underestimated in the results. However, it was shown that even with the one-year effect of these source-reduction activi- ties it is possible to achieve the desired control level (i.e., to keep mosquito poplation levels at or below the specified standard) by substi- tuting such activities for use of pesticides as the effectiveness of pesticides declines or by combining them with pesticide use in an inte- grated program. THE CHEMICAL INDUSTRY AND PESTICIDE PRODUCTION Background Until 1945, pesticide production was limited and the chemicals were applied almost exclusively on high-value crops. Since 1945 when DDT was introduced and since the development of the new synthetic organic pesti- cides, spraying and dusting operations spread to most agricultural crops and many public health activities. 48 Table 8 Results of the Programming Model under Five Levels of Pesticide Ef f ectivenessi./ and the Original Standards of Mosquito Numbers—' Annual level of activities/items under; Activities/items in the optimal solution Pesticide 100% effectiveness on acres and locations 75% 50% 25% 5% Acres sprayed with pesticides 5,000 5,000 5,000 5,000 5,000 Locations treated with pesticides 500,000 500,000 381,501 500,000 500,000 Fills, levees, etc., constr. (1000 c.y. ) 49.45 41.83 47.40 48.73 50.24 Sumps, ponds, etc., constr. (1000 c.y. ) 0 0 2.32 29.65 55.99 Ditches constructed (miles) 5.50 60 60 60 60 A. nigromaaulis mosquitoes per light-trap night .589 .589 0 0 0 C. tavsalis mosquitoes per light-trap night 2.67 2.67 2.67 2.67 2.67 Hours of labor 2,198 2,406 2,663 4,303 5,384 Total direct cost ($) 24,832 25,594 28,251 41,570 51,441 a/ In this table different effectiveness levels were generated from sensi- ~ tivity tests on the coefficients of acres sprayed and locations treated in the mosquito population relations. The changes in the coefficients were assumed to be linear, e.g., the coefficients of acres sprayed and locations treated at 75 percent pesticide effectiveness were in each case equal to .75 x the original coefficient. All parameters were held at their levels of the original model except those for acres sprayed and locations treated with pesticides. W A, nigvomaaulis < .589 per light-trap night and C. tavsalis < 2.67 per light-trap night. 49 FIGURE 5 EFFECT OF DECLINING PESTICIDE EFFECTIVENESS ON DISTRICT CONTROL COST AT THREE MOSQUITO CONTROL LEVELS 80 70 60 50 40 30 20 10 Pesticide Effectiveness Since the late 1940' s there has been a dramatic increase in the pro- duction and sales value of pesticides sold to domestic and international markets. From 1954 to 1972, pesticide production in the United States almost tripled, from 419,274,000 pounds in 1954 to 1,157,698,000 pounds in 1972. These figures are shown in Appendix Table 1. The total sales value for the domestic market and exports increased dramatically by more than eightfold, from $124.5 million in 1954 to over $1 billion in 1972 [USDA 1973]. Of the many U.S. companies involved in the formulating, manufacturing, distributing and selling of pesticides, about 35 are considered to be major innovators which conduct extensive research and development programs [Little 1975]. Most of these companies are multiproduct firms, with less than 20 percent of total sales attributed to pesticides. Two-thirds of the companies with sizable research and development (R&D) efforts are large chemical-or petroleum-based firms; several are multiproduct pharmaceu- tical companies. Of the smaller firms also involved in pesticide produc- tion, pesticides account for as little as 20 percent to as much as 100 percent of their total sales. The USDA and other public agencies also have a role in developing new chemical pesticides [Klassen and Schwartz 1973; Kramer 1969]. The USDA, particularly the Entomology Research Division of the Agricultural Research Service (ARS), assists in assuring the continuing availability of pesticides for major and minor uses and markets [Klassen and Schwartz 1973]. Two main areas in which the USDA could provide more assistance are: (1) Generating toxicological , residue and efficacy data needed for registration of candidate pesticides. 51 (2) Conducting research which complements R&D by the industry, thereby greatly reducing the industry's R&D costs. Historically, the chemical industry has played the primary role in R&D and pesticide production. All but a few pesticides have been synthe- sized first in the laboratories of chemical companies [Djerassi, Shih- Coleman and Diekman 1974]. It is possible to assume that private industry will continue to be the primary source, developer and manufacturer of new chemical pesticides because it has the necessary experimental and production facilities and the experienced personnel. However, future pest control efforts will require greater research and development input from public agencies than in the past. Environmental and Health Regulations on Pesticide Use In California, laws stipulating enforcement procedures and controls over the sale and use of pesticides can be traced as far back as 1901 [Post 1972]. Various old regulations attempted to protect the agricul- tural industry from the harmful effects of substandard insecticides. Since 1919, the California Department of Food and Agriculture has been in charge of setting such regulations. The principal national legislative actions that have affected pesti- cide research and development in recent years have been the Federal Insec- ticide, Fungicide and Rodenticide Act (FIFRA), in 1947; the Miller (1954) and Delaney (1962) Amendments to the Federal Food, Drug and Cosmetic Act; the policy changes in pesticide residue requirements enacted during 1966- 67; PR Notice 70-15 in 1970; the Federal Environmental Pesticide Control Act (FEPCA) in 1972; and finally the regulations proposed and promulgated 52 under FEPCA since its enactment [Djerassi et al. 1974; Hunter 1978; Little 1975]. The first important act, FIFRA, passed in 1947, called for USDA reg- istration of economic poisons prior to their interstate transport and sale. FIFRA required that all product labels contain instructions for use and warnings about safety hazard to humans, animals and plants. The Miller Amendment (1954) required the Food and Drug Administration (FDA) to establish tolerance limits for pesticide residues. The Delaney Amend- ment (1962) prohibited the presence of any known carcinogen in food pro- ducts and increased data requirements for approval. Data requirements for pesticide registration under FIFRA increased slowly but at a steady pace from 1947 to the 1970' s. An Arthur D. Little, Inc. , study [1975] summarized the provisions that have had the greatest impact on pesticide R&D as follows. 1. The data requirements for pesticide registration and labeling, for example toxicity tests and data in- cluding safety, physical/chemical properties, efficacy and labeling information. 2. Data required for the establishment of tolerances on agricultural commodities, e.g., chemical, toxicological, biochemical, reproduction studies, etc. 3. The experimental use permit program, which required additional data permits for field testing of potential pesticide products. In 1970, the Environmental Protection Agency (EPA) was created and was granted full regulatory powers over economic poisons. Under the EPA, new regulations shifted the emphasis in USDA and FDA registration and data requirements from efficacy and safety to safety, health and environmental aspects. The industry believes that this shift has had a great impact on the R&D of pesticides [Farm Chemicals 1970; Hunter 1973; Little 1975]. 53 Industrial Research and Development Costs The costs of R&D associated with the discovery and development of a commercially viable pesticide have increased dramatically in the last several years. In order to gain some insight into the industry's cost we should consider the stages involved in R&D and registration [Lever and Strong 1973; Little 1975]. The four stages of R&D and registration are (1) synthesis and screening, (2) advanced tests, (3) field evaluation and (4) registration. In the synthesis and screening stage, compounds are screened for useful biological pesticidal activity. In stage two, some selected compounds which passed stage one are subjected to advanced screening (laboratory and greenhouse tests are included). In the field evaluation stage, ecological and biological evaluation of products selected from stage two is carried out under various conditions in order to uncover any likely problems and/or limitations. This stage requires the acquisition of use permits from federal and state agencies. The registration stage, which may take three or more years to complete, detailed biological toxicological and ecological studies for final development and obtaining a use and safety label from the EPA. Time required for the development of new agents varies and largely depends on the regulations and data requirements. For example, the average time expended from synthesis and screening through approval has increased as registration requirements have increased. This can be seen in Table 9. Several studies and reports have been concerned with the cost of devel- oping a pesticide. In 1969, Wellman [1969] estimated the total cost at $4.1 million, whereas the National Agricultural Chemical Association (NACA) 54 Table 9 History of Last Two New Pesticide Chemical Products Cleared for the U.S. Market by Each Participating Chemical Company, by Date of Approval^.' Number of products reported: Stage: First screening to decision to commercialize Decision to commercialize to first registration submission First registration submission to approval Average total from screening to approval Prior to 1963 10 1963- 1967 22 1967- 1971 25 Average elapsed time (months ) 39 16 33 21 61 61 32 19 18 69 Average R&D man-years expended from screening to approval: Average Time (Man-years) 49 34 65 a/ Composite analysis of all reporting companies [Little 1975]. Source: 1970 Industry Profile Study [Ernst and Ernst 1971]. 55 estimated the same costs to range from $2.5 to $6 million.il/ Johnson and Blair [1972] reported in 1972 that costs of R&D increased from $1.2 million in 1956 to $4.1 million in 1969. Lever and Strong [1973] reported that these costs increased from $1.2 million in 1956 to over $10 million in 1972. In 1975, Arthur D. Little, Inc. [1975] reported the estimated cost for discovery and development of pesticide was about $7.5 million. A new industry profile report was published in 1975 by NACA [Ernst and Ernst 1975]. A comparison with previous studies indicates that there has been a consistent increase in cost of discovery and commercialization, from $3.4 million in 1967 to $5.5 million in 1970 and to $6.1 million in 1973. Investment Decisions for the Pesticide Industry; A Simulation Model A model for investment under risk and uncertainty should permit the firm to calculate the expected net value of a project from a simulated joint probability density. The structure of the proposed model, which is concerned with the in- dustry's investment in narrow-spectrum pesticides for mosquito control in California, can be divided into four elements. The first is the criterion function. In this study the discounted net present value of the project will be used. The second element is the variables specified and their interrelationships. The third is the parameters (e.g., the mean and standard deviation of probability distributions). The fourth element is the development of the computational techniques to simulate the net present 15 / This figure is calculated excluding the opportunity cost of development capital. With an 8 percent interest rate, the development cost would be $11 million [Djerassi, Shih-Coleman and Diekman 1974]. 56 value of distribution. These four elements of the model's structure are described below. a. The criterion function - The net present value (NPV) formula consists of two segments: the discounted present value of the sequence of returns (DPVR) and the discounted present value of costs (DPVC), where NPV = DPVR - DPVC. In general, the discount- ing formula (assuming a constant discount rate over time) can be written as: NPV = T. — E 1 t=o TTTTy^ where Rj- and C^- are returns and expenditures over time and r is the discount rate. b. Specification and relationships of variables - The key variables that management is assumed to consider in this model are: (1) the total annual revenue from the sale of a pesticide to all California mosquito abatement agencies, Jl^/ (2) the time spent by the typical pesticide-producing firm (or a firm in the small, medium, or large firm in this category) from first discovery of the product to marketing and (3) the total cost of developing a new pesticide ( including R&D, cost of unsuccessful 16 / For estimation of this segment in equation form, the base market size is initially taken to be the state of California; however, it can increase if we include the possibility of ex- ports to other states. Data for the revenues were calculated from the quantity of malathion used by all mosquito control agencies in California. Malathion, an organophosphorus pesti- cide, was chosen for the estimation because it was used throughout the state and seemed to be a representative pesti- cide and because there was sufficient information available regarding its use [Sarhan 1976]. 57 products, etc.). These factors must be determined and combined to obtain a measure of the attractiveness of the proposed inves tment . The parameters. These are the constants of the system. There are three types: (1) the coefficients of the revenue equation, (2) the parameters of the different probability distributions, i.e., under normal conditions the mean and variance, and (3) the assignment of fixed values. The third category includes assign- ing a constant value to the length of the useful life of the product or to the period of the potential rights from the firm's viewpoint. It also includes the constant value assigned to the revenue from the first year's sales, which is needed to calculate the second year's sales. Finally, it includes assigning a con- stant value to the interest rate used in the discounting procedure Development of computational techniques. The calculations nec- essary to simulate the various values in the model are outlined in Figure 6. The program begins by reading in values of constants and rules. The computer generates random variables from a time- f rom-discovery-to-sale distribution and then generates a random total cost figure from a cost distribution. The choice of a cost distribution depends on the value of the time variable. It is assumed that the lower the number of years spent from first discovery to marketing, the lower the cost. Therefore, it is assumed that management develops a number of probability distributions of the costs, each corresponding to a different 58 Figure 6: Flow chart of calculations required to simulate net present value for investment proposal (1) (2) (3) (4) (5) START Read in parameters and rules Number run i = 0, 1, 2, i+1 ., K Generate a random number of years from discovery to market- ing of a product using a specified normal distribution. Round & call this X* E X* Xi? ,No X* X2? Yes Generate a random (12) cost figure from (H) cost distribution Generate a / random cost figure from (13) (M) cost distribution Generate a random cost figure from (L) cost distribution Compute annual cost and Z dis- counted costs stream (P VTC) Compute annual revenues and the discounted sum revenue stream (PVTR) Compute the discounted net present value (NDPV) 2(PVTR)-2(PVTC) (14) Advance Store one run value Yes , K 1 (10) STOP (20) I Compute the expected net present value of the investment (19) Construct frequency distribution for the net present values stored. (18) (6) (7) (8) (9) Stop advancing runs No K (17) (16) r range of values for the time required. In this model it is assumed that there are three cost distributions, with high, medium and low means corresponding to high, medium and low ranges in number of years, respectively. Next, the computer divides the total cost figure by the number (rounded) of years chosen and derives the annual cost.iZ./ The cost stream is then discounted to compute and store a present value of total cost (PVTC). Calculation of the annual revenue streams then begins. The revenues are computed according to the relationships specified for each of the fixed market horizons. The re- venues then are discounted and summed to give a present value of total revenue (PVTR), which is stored. Next, a net discounted present value of the investment (NDPV) is computed by simply subtracting the sum of the present value of total costs (PVTC) from the sum of the present value of the total revenues: NDPV = PVTR - PVTC. The NDPV is then stored and the above procedure is repeated the desired number of times. The stored NDPV's are then used to derive a frequency distribution, mean and variance of NDPV. In the last step, the values and their relative frequencies are used to com- pute the expected net present value of the investment. The assumption of even annual costs is important to simplify the discounted cost stream calculations. In addition, no other production costs are assumed after the marketing begins. 60 the other divided over four years, 11/ were calculated from Kern MAD's simulation model. The aggregate subsidies potentially available from the State of California were calculated by assuming that the value obtained by Kern MAD was equal to 5.5 percent of the total potential value of control cost savings to the state if effective pesticides were available. (This proportion was used because 5.5 percent was the average proportion of Kern MAD's budget to the total budgets of all control districts [CMCA, 1975]). Therefore, it was assumed that the total potential state subsidy is equal to 18.18 times the level for Kern MAD. Hypothesizing that only one pesticide is available to a mosquito con- trol district, with no replacement and that the pesticide is 80 or 60 or 40 percent effective, the amount of money the district would save if a 100 percent effective pesticide were available is equal to the sum of saving in their direct control costs over an extended period equal to the number of years needed for the 100 percent effective pesticide to reach the effective- ness level of the district's current pesticide. The number of years required for the effectiveness to drop from 100 to 40 percent is assumed to be proportional to the historical time interval used in calculating the effectiveness indices (see Appendix B) for several pesticides. With six years as a conservative estimate of the period required from discovery to marketing of a pesticide, the subsidies must be paid to the 19/ Preliminary runs showed that a lump-sum subsidy given at the time of discovery always added more to the total revenue, and thus to the NPV, than did the total discounted payments divided over four years. The lump-sum subsidy is therefore superior and only the results of the calculations under the policy are reported. 63 » chemical industry six years before the time that a 100 percent effective pesti- cide will be available. The subsidies then enter the firm's calculations as re- venues received in the beginning of year seven and therefore must be discounted. The NACA reports also provided material upon which to base three different probability distributions for the cost of discovery, development and production of a pesticide. These three distributions, which correspond to the cost conditions in 1967, 1970 and 1973, are designated as low, medium and high cost distributions. After a time distribution is specified (for any firm size), a random number of years can be drawn and its value determines which of the three cost distributions is to be used by the pro- gram to draw a cost figure. All time and cost probability distributions were assumed to be normal with known means and standard deviation. Appendix Table A-2 summarizes the different time and cost distributions. The results will be discussed in the following section. For each firm size, 100 simulated net present values were obtained from computer runs under two discount rates for two patent-right periods. These data were then used to construct a frequency distribution for the NPV's and to calculate the relative frequency distribution for each of the outcomes in an interval. (The interval used was $250,000.) The expected net present values were calculated from the relative frequency distribution and the values of the mid-points of the intervals. The results of the calculations are presented in Appendix A tables. The 96 computed combinations of the NPV frequency distributions which resulted in the expected net present values will not be reproduced here. However, representative illustrations of the computer output are given in Appendix Tables A-5 through A-8. 64 Fitzsimmons, K.R. , "Role of Industry in Advancing New Pest Control Strategies." In. Pest Control: Strategies for the Future, National Academy of Sciences, Washington, D.C. , 1972, pp. 352-361. Georghiou, G.P., "Insecticide Resistance with Special Refer- ence to Mosquitoes." Proceedings and papers, 33rd Annual Conference of the California Mosquito Control Association Inc., 1965, pp. 34-40. , "Research on Mosquito Resistance to Insecticides at the University of California, Riverside." Proceedings and papers, 34th Annual Conference of the California Mosquito Control Association, Inc. , 1966. Gillies, P.A. , D.J. Womeldorf, CP. Zboray and K.E. White, "Insecticide Susceptibility of Mosquitoes in California: States of Organophosphorus Resistance in Larval Aedes nigvomaeulis and Cutex tavsalis through 1973." Proceed- ings and papers of the 42nd Annual Conference of the California Mosquito Control Association, 1976. Goldberger, A.S., "Econometric Theory." John Wiley, 1964. Hardy, J.L. and W.C. Reeves, "Emergin Concepts of Factors that Limit the Competence of Cutex tavsalie to Vector Encephalitis Viruses." Proceedings and papers, 41st Annual Conference of the California Mosquito Control Association, Inc., 1973, pp. 7-10. Hoffman, R.A. and W.C. McDuffie, "The 1962 Gulf Coast Mosquito Problem and the Associated Losses in Livestock." Proceedings of the 50th Annual Meeting of the New Jersey Mosquito Extermination Association, 1963, pp. 421-424. Hoy, James B. and David E. Reed, "Biological Control of Cutex tavsalis in a California Rice Field." Mosquito • News, Vol. 30, No. 2, June, 1970, pp. 222-230. Hoy, J.B., E.E. Kauffman and Allen G. O'Berg, "The Mosquito- fish as a Biological Control Agent Against Cutex tarsatis and Anophetes fveebormi in Sacramento Valley Rice Fields. Mosquito News, Vol. 31, No. 2, June, 1971, pp. 146-152. , "A Large-Scale Field Test of Gambosia Af finis and Chlorpyrifos for Mosquito Control." Mosquito News, Vol. 32, No. 2, June, 1972, pp. 161-171. Hunter, Robert C. , "Federal Environmental Pesticide Control Act of 1972 - What Does It Mean to the Chemical Industry? Proceedings of the Annual Meeting of the Entomological Society of America, 1973. 69 Husbands, R.C. , "Economic Studies: Areas of Research." Memorandum to Donald J. Womeldorf, Vector Control Service, Sacramento: State of California Department of Health, November 15, 1973. Johnson, J.E. and E.H. Blair, "Cost, Time and Pesticides Safety." Chemical Technology, 21, 1972, pp. 666-669. Kauffman, Eugene E. , Manager-Entomologist, Sutter-Yuba Mosquito Abatement District, personal communication, 1975. Klassen, W. and P.H. Schwartz, Jr., "The Role of the USDA in Developing New Chemical Pesticides." Entomological Society of America, Bull. 19 (2), June, 1973, pp. 98-99. Kramer, Joel, "Pesticide Research: Industry, USDA Pursue Different Paths." Science , Vol. 166:12, December, 1969, pp. 1383-1386. Lever, B.C. and W.M. Strong, "Evaluation of a Pesticide by Chemical Industry." Proceedings of the Conference on Plant Protection Economy sponsored by the European and Mediterranean Plant Protection Organization, Brussels, Belgium, May 16, 1973. Little, Arthur D. , Inc. , "Evaluation of the Possible Impact of the Pesticide Legislation of Research and Development Activities of Pesticide Manufacturers." Final Report, Vol. 1, Contract No. 68-01-2219, Prepared for the EPA Office of Pesticides, Washington, D.C. , February, 1975. McClelland, G.A.H. , Department of Entomology, University of California, Davis, personal discussion. Moon, T.E. , "A Statistical Model of the Basic Infection Cycle of Western Equine Encephalitis Virus." Unpublished Ph.D., Dissertation, University of California, Berkeley, 1975. Mulhern, Thomas D. , (ed.), A Training Manual for Personnel of Official Mosquito Control Agencies. California Mosquito Control Association and Vector Control Section, California State Department of Health, 1973. Murray, W, Donald, "Pasture Mosquito Control Without Spray." Proceedings and papers, 25th Annual Meeting of Utah Mosquito Abatement Association, Salt Lake City, Utah, 1972, pp. 14-17. Post, A. Alan, "Pesticide Regulation in California Registration and Use Control. " Report to Chairman and Members of the Joint Legislative Budget Committee, State Capitol, Sacramento, California, December, 1972. 70 Reeves, William C. , "Developing Balanced Programs in the University of California for Mosquito Control — Medical Aspects." Proceedings and Papers, 33rd Annual Conference of the California Mosquito Control Association, Inc. , 1965, pp. 46-49. Reeves, William C, "A Review of Developments Associated with the Control of Western Equine and St. Louis Encephalitis in California During 1967." Proceedings and Papers, 36th Annual Conference of the California Mosquito Control Asso- ciation, Inc. , 1968, pp. 65-70. Reeves, William C. , "Current Observations on Mosquito-Borne Viruses of Concern to Mosquito Abatement Districts in California." Proceedings and papers, 38th Annual Conference of the California Mosquito Control Association, Inc., 1970, pp. 26-28. Sanders, D.P., M.E. Rieme and J.C. McNeill, "Salt Marsh Mosquito Control in Relation to Beef Cattle Production: A Preliminary Report." Mosquito News, Vol. 28, No. 3, 1968, pp. 311-314. Sarhan, M.E.S., "An Economic analysis of Mosquito Abatement in California and the Chemical Industry's Investment in Narrow Spectrum Pesticides." Unpublished Ph.D. Dissertation, University of California, Davis, 1976, pp. 342. Smith, Roy F. , "Pesticides: Their Use and Limitations in Pest Management." In Concepts of Pest Management, edited by R.L. Rabb and F.E. Guthrie, Proceedings of Conference at North Carolina State University, Raleigh, North Carolina, 1970. Spiller, Donald, Mosquito Problems in California's Central Valley. University of California, Division of Agricultural Sciences, April, 1968. Steelman, CD., T.W. White and P.E. Schilling, "Effects of Mosquitoes on the Average Daily Gain of Feedlot Steers in Southern Louisiana." Journal of Economic Entomology, Vol. 65, No. 2, April, 1972, pp. 462-466. , "Effects of Mosquitoes on the Average Daily Gain of Hereford and Brahman Breed Steers in Southern Louisiana." Journal of Economic Entomology, Vol. 66, No. 5, October, 1973, pp. 1081-1083. Sudia, W.D., R.W. Emmons, V.F. Newhouse and R.E. Peters, "Arbovitus-Vector Studies in the Central Valley of California, 1969." Mosquito News, Vol. 31, No. 2, June, 1971, pp. 160-168. 71 r U.S. Department of Agriculture, The Pesticide Review. Agricultural Stabilization and Conservation Service, Washington, D.C., 1954-1973. U.S. Department of Commerce, Climatological Data, California. National Oceanic and Atmospheric Administration, Environ- mental Data Service, Washington, D.C., 1973. U.S. Department of Commerce, Climatological Data, California. Weather Bureau, in Cooperation with Department of Water Resoruces, State of California, Government Printing Office, Washington, D.C., 1971. Wellman, R.H. , in Chemical & Engineering News. Vol. 47, No. 24, June 9, 1969, pp. 22-24. Womeldorf, D.J. , P. A. Gillies and K.E. White, "Insecticide Susceptibility of Mosquitoes in California: Illustrated Distribution of Organophosphorus Resistance in Larval Aedes nigromaaulis and Culex tavealiS'" Proceedings and Papers, 40th Annual Conference of the California Mosquito Control Association, Inc., 1972, pp. 17-21. Womeldorf, D.J. , P. A. Gillies and R.F. Peters, "Insecticide Susceptibility of Mosquitoes in California: Effects of Resistance Upon Operation, Support and Research." Proceedings of the 25th Annual Meeting of the Utah Mosquito Abatement Association, October 2-3, 1972, Salt Lake City, Utah, pp. 8-11. Davis Enterprise, "Malaria Epidemic." August 15, 1974, p. 2. 72 APPENDIX A Appendix Table A-1 Synthetic Organic Pesticide Production and Sales in the U.S., 1954-1972i/ Production Sales b/ (Domestic & Exports) Change from Change from Quantity Previous Value P TP VT nil Q Year 1000 pounds Year - % 1000 $ Year - % 1954 419,274 _ _ 124,501 1955 506,376 20.8 152,772 22.7 1956 569,927 12.6 172,908 13.2 1957 511,552 -10.2 178,039 3.0 1958 539,396 5.4 196,149 10.2 1959 585,446 8.5 225,469 14.9 1960 647,795 10.6 261,789 16.1 1961 699,699 8.0 302,955 15.7 1962 729,718 4.3 346,301 14.3 1963 763,477 4.6 369,140 6.6 1964 782,749 2.5 427,111 15.7 1965 877,197 12.1 497,066 16.4 1966 1,013,110 15.5 583,802 17.4 1967 1,049,663 3.6 787,043 34.8 1968 1,192,360 13.6 849,240 7.9 1969 1,104,381 -7.4 851,166 .2 1970 1,034,075 -6.4 870,314 2.2 1971 1,135,717 9.8 979,083 12.5 1972 1,157,698 1.9 1,091,708 11.5 aj Includes a small quantity of soil conditions. _b/ Value of sales is not equal to value of production since it is assumed that not all production is sold in the same calendar year. The values in the table are nominal; infla- tion is not taken into account. Source: The Pesticide Review [USDA, 1973], 1963-64, 1971 and 1973. United States Department of Agriculture, Agricultural Stabilization and Conservation Service. 73 A ppendix Table A-2 The Chemical Industry's Time and Cost Distributions for Three Firm Sizes Time from Discovery to Marketing Dis- Total Cost tribution Statistics Distribution Statistics^./ Time Basei' Moan Wn ml^Q T* of Years Elapsed Standard Deviation Condi- tion Mean Standard Deviation T 04? nnn and low time 4.66 .166 M 4,365,000 1,667,000 requirements H 6,112,963 1,420,000 • oma±x X J. Lmb T Li 1 nA9 nnn and higher time 6.08 .500 M 4,365,000 1,667,000 requirements H 6,112,963 1,420,000 ^ti'- Medium firms L $3,505,000 916,800 and low time 5.42 .667 M 5,479,000 1,250,200 requirements H 6,112,963 1,420,000 Mt2 • Medium firms L $3,505,000 916,800 and higher time 6.75 .556 M 5,479,000 1,250,200 requirements H 6,112,963 1,420,000 L^i: Large firms L $4,071,000 600,100 and low time 4.75 .333 M 6,112,963 1,420,000 requirements H 7,285,000 1,417,000 • Large firms L $4,071,000 600,100 and higher time 6.50 .667 M 6,112,963 1,420,000 requirements H 7,285,000 1,417,000 a^/ A time base designated by the subscript "tl" corresponds to conditions with less governmental regulation. A time base designated by the subscript "t2" corresponds to conditions under more regulation. W "L," "M" and "H" represent low, medium and high cost distribu- tions, respectively. cj Note from the flow chart in Figure 6 the distribution from which the total undercounted cost is drawn is conditional on the number of years from discovery to marketing (X*). After both these parameters are established the annual cost (and thus the present value of total cost for a particular sized firm) is calculated. 74 Appendix Table A-3 Firm Size and Time Requirement from Discovery to Marketing Expected Net Present Values of Investment in a Narrow-Spectrum Pesticide, with and without Subsidy, for Three Chemical Firm Sizes under 17- and 20-Year Patent Rights and a 6 Percent Discount Rate Patent Rights = 17 Years Patent Rights = 20 Years a/ Expected a/ Expected NPV When Expected NPV When Subsidy~ Is: NPV When Expected NPV When Subsidy" Is: Subsidy = 0 $75.471 $303,328 $2,829.129 Subsidy = 0 $75,471 $303,328 $2,829,129 (millions) (millions) (millions) (millions) (millions) (millions) (millions) (millions) Small firms and lower time Small firms and higher time Medium firms and lower time Medium firms and higher time Large firms and lower time Large firms and higher time - $ 2.8 - $ 2.7 - $ 2.5 + $ .017 - $ 2.7 - $ 3.4 - $ 3.3 - $ 3.1 - $ .57 - $ 3.2 - $ 3.1 - $ 3.1 - $ 2.9 - $ .325 - $ 3.0 - $ 3.3 - $ 3.2 - $ 3.0 - $ .445 - $ 3.0 - $ 2.8 - $ 2.7 - $ 2.5 - $ 0.00 - $ 2.7 - $ 3.6 - $ 3.5 - $ 3.2 - $ .733 - $ 3.3 - $ 2.6 - $ 2.3 + $ .185 - $ 3.1 - $ 2.9 - $ .357 - $ 2.9 - $ 2.7 - $ .140 - $ 2.9 - $ 2.7 - $ .212 - $ 2.6 - $ 2.4 + $ .155 - $ 3.2 - $ 3.0 - $ .057 a/ These values are lump-sum subsidies and are equal to the discounted (at 6%) sum which the mosquito control ~ agencies would be willing to pay for an effective pesticide to replace materials which are ineffective. Appendix Table A-4 Expected Net Present Values of Investment in a Narrow-Spectrum Pesticide, with and without Subsidy, for Three Chemical Firm Sizes under 17- and 20-Year Patent Rights and an 8 Percent Discount Rate Firm Size and Patent Rights = 17 Years Patent Rights = 20 Years Time Requirement from Discovery Expected NPV When Expected NPV When Subsidy" / Is: Expected NPV When a/ Expected NPV When Subsidy" Is: to Marketing Subsidy = 0 $68,522 $275,400 $2,568,651 Subsidy = 0 $68,522 $275,400 $2,568,651 (millions ) (millions) (millions) (millions) (millions) (millions) (millions ) (millions) Small firms and lower time - $ 3.1 - $ 3.0 - $ 2.8 - $ .470 - $ 2.9 - ^ 2.9 - . uu Rights = 0 .01 .00 -5.875 Rights = 0 01 fin —9 S7 20 Years 1 .02 .01 -5.375 * 20 Years 1 X . ux —9 A9 •k 3 .05 .03 -5. 125 *** J. m . ux — 9 T7'\ "k 2 .07 .02 -4.875 ** 2 n? . — 9 1 9 •kk Subsidy = 2 .09 .02 -4.625 ** -a Oft . U J — 1 H7 i. o/ J ieieit zero 3 .12 .03 -4.375 *** $2 829 129 4 1 7 — 1 A9 X . DZ3 kickk 6 .18 .06 -4.125 ****** 5 17 OS — 1 "^7 S ■kk-kick 8 .26 .08 -3.875 ******** ft . uo X . iZ 3 f\ n f\ f\ *\ ^ 11 .37 .11 -3.625 *********** fi • Ox . Uo — n 87 =^ f\ /* /\ j\ r\ 1 .38 .01 -3.375 * 7 . UD — n A9 U . DZ J JL (1. J, J. i\ *\ f\ n n 9 .47 .09 -3. 125 ********* Q O /l AQ . Uo n Q 7 t; -U. j/5 n .58 .11 -2.875 *********** J-X . jd 1 1 • IX ri IOC — U. iZD 4 .62 .04 -2.625 **** 3 S9 n'^ n 1 9 U. XZ J AAA 5 .67 .05 -2.375 ***** 7 . DO . U / n 17 •I* J> W « Mean = 2 .69 .02 -2.125 ** J AQ ni . U J n A 9 U . DZ J AAA -$2,777,503 7 .76 .07 -1.875 ******* A 7 n/i n Q7 U. o/ J 1^ 1^ 2 .78 .02 -1 6? 5 ** J 7 A . Uj 1 IOC 1. iz5 3\ X « 2 .80 .02 -1.375 ** A Hfl . OU n/i 1 Q7 1. J/J **** 2 .82 .02 -1.125 ** 1 . 81 m • \J X. 1 fi9S * 4 .86 .04 -0.875 **** 4 .85 .04 1.875 **** Standard 3 .89 .03 -0.625 *** Standard 3 .88 .03 2.125 *** Deviation = 1 .90 .01 -0.375 * Deviation = 1 .89 .01 2.375 * $1,595,240 1 .91 .01 -0.125 * $1,595,240 1 .90 .01 2.625 * 1 .92 .01 0.125 * 2 .92 .02 2.875 ** 3 .95 .03 0.375 *** 1 .93 .01 3.125 * 3 .98 .03 0.625 *** 4 .97 .04 3.375 **** Expected NPV = 0 .98 .00 0.875 Expected NPV = 1 .98 .01 3.625 * -$2,662,500 1 .99 .01 1.125 * $ 185,000 0 .98 .00 3.875 1 1.00 .01 1.375 * 2 1.00 .02 4.125 ** Appendix Table A-7 NPV Frequency Distributions for Small Firms and Higher Time Requirements at a 6 Percent Discount Rate Freq CRF RF Mid Pt Freq CRF RF Mid Pt Patent 1 .01 .01 -6.375 * Patent 1 .01 .01 -3.625 * Rights = 0 .01 .00 -6. 125 Rights = 0 .01 .00 -3.375 17 Years 0 .01 .00 -5.875 17 Years 0 .01 .00 -3. 125 2 .03 .02 -5.625 0 .01 .00 -2.875 1 .04 .01 -5.375 * 3 .04 .03 -2.625 *** Subsidy = 3 .07 .03 -5. 125 *** Subsidy = 2 .06 .02 -2.375 ** zero 3 .10 .03 -4.875 *** $2,829,129 4 .10 .04 -2. 125 **** 5 .15 .05 -4.625 ***** 3 .13 .03 -1.875 *** 7 .22 .07 -4.375 ******* 7 .20 .07 -1.625 ******* 8 .30 .08 -4. 125 ******** 8 .28 .08 -1.375 ******** 12 .42 .12 -3.875 ************ 8 .36 .08 -1.125 ******** 3 .45 .03 -3.625 *** 8 .44 .08 -0.875 ******** 10 .55 .10 -3.375 ********** 14 .66 .14 -0.375 ************** 13 .68 .13 -3. 125 ************* 4 .70 .04 -0.125 **** Mean = 4 .72 .04 -2.875 **** Mean = 8 .78 .08 0.125 ******** -$3,517,382 7 .79 .07 -2.625 ******* -$ 688,253 4 .82 .04 0.375 **** 5 .84 .05 -2,375 ***** 5 .87 .05 0.625 ***** 6 .90 .06 -2.125 ****** 6 .93 .06 0.875 ****** Standard 2 .95 .02 -1.625 ** Standard 1 .94 .01 1.125 ****** Deviation 1 .96 .01 -1.375 * Deviation = 2 .96 .02 1.375 ** $1,127,967 2 .98 .02 -1.125 ** $1,136,307 1 .97 .01 1.625 * 1 .99 .01 -0.875 * 2 .99 .02 1.875 ** 0 .99 .00 -0.625 0 .99 .00 2.125 Expected NPV = 0 .99 .00 -0.375 Expected NPV 0 .99 .00 2.375 -$3,392,500 0 .99 .00 -0.125 -$ 567,000 0 .99 .00 2.625 1 1.00 .01 0.125 * 1 1.00 .01 2.875 * NPV Frequency Distributions for Freq CRF RF Patent 1 .01 .01 Rights = 0 .01 .00 20 Years 0 .01 .00 2 .03 .02 1 .04 .01 Subsidy = 4 .08 .04 zero 3 .11 .03 4 .15 .04 7 .22 .07 9 .31 .09 13 .44 .13 2 .46 .02 12 .58 .12 10 .68 .10 Mean = 4 .72 .04 -$3,303,905 8 .80 .08 7 .87 .07 3 .90 .03 Standard 3 .93 .03 Deviation 2 .95 .02 $1,128,657 1 .96 .01 2 .98 .02 1 .99 .01 Expected NPV = 0 .99 .00 -$3,177,500 0 .99 .00 1 1.00 .01 Appendix Table A-8 Small Firms and Higher Time Requirements Mid Pt Freq -6. 125 Patent 1 -5.875 Rights = 0 -5.625 20 Years 0 -5.375 ick 1 -5. 125 ic 2 -4.875 Subsidy = 3 -4.625 ickie $2,829,129 o -4.375 **** 3 -4.125 ******* 9 -3.875 ********* 6 -3.625 ************* 10 -3.375 ** 7 -3.125 ************ 9 -2.875 ********** 12 -2.625 **** Mean = 6 -2.375 ******** -$ 474,776 6 -2.125 ******* 5 -1.875 *** 7 -1.625 *** Standard 3 -1.375 ** Deviation = 2 -1.125 * $1,128,657 1 -0.875 ** 2 -0.625 * 1 -0.375 Expected NPV 0 -0.125 -$ 357,500 0 0.125 * 0 1 a 6 Percent Discount Rate CRF RF Mid Pt .01 . 01 -3. 375 .01 .00 -3. 125 . 01 . 00 -2.875 .02 .01 -2. 625 .04 . 02 -2. J75 .07 .03 -2. 125 .10 . 03 -1.875 JU J> . 13 .03 -1. 625 .22 .09 -1.375 .« «. .». .1. .t. ■■■ -■- -1. .1- .28 .06 -1.125 •k'k'k-k'k'k .38 .10 -0.875 .45 .07 -0. 625 *. ». «. «■ ■■■ .1. xxxxxxx .54 .09 -0. 375 ■ f ■ ■!■ iti .!■ iti il. ill ■ li- XXXXXXXXX .66 .12 -0. 125 xxxxxxxxxxxx ,11 .06 0. 125 ****** .78 .06 0.375 ****** .83 .05 0.625 ***** .90 .07 0.875 ******* .93 .03 1.125 *** .95 .02 1.375 ** .96 .01 1.625 * .98 .02 1.875 ** .99 .01 2.125 * .99 .00 2.375 .99 .00 2.625 .99 .00 2.875 1.00 .01 3.125 * APPENDIX B Derivation of Pesticides' Effectiveness Indices The pesticide effectiveness indices were included in the mosquito abatement annual-data models for Delta VCD and Kern MAD. The variables were Kg, K^i and K14 in Kern MAD's model. The average effectiveness functions were used in this study in lieu of resistance functions because of the difficulty in quantifying several variables which affect the re- sistance of mosquitoes to pesticides. The effectiveness indices which were generated and used as indepen- dent variables in the estimation of the effectiveness functions were based on several simplified assumptions and each generated index represen- ted the average effectiveness of all pesticides used in the district during any year. The indices are measured as the average field percentage control attained by all pesticides rather than the usual laboratory LD5Q used by entomologists for individual chemicals.^/ The field-percentage measure of pesticides' control is used by the districts studied as a measure of the degree of mosquito resistance (or the effectiveness of pesticides). The percentage of control (or kill) in the field is defined in this study as that number of out of 100 acres sprayed or 100 locations treated which does not require respraying or retreatment. For example, if 10 percent of the acres sprayed must be retreated with pesticides, then the effectiveness of one application of pesticides is 90 percent. \J It should be recognized that although there is a correlation between the average field percentage measure of effectiveness and the LD50 measurement used in laboratory tests, they are not necessarily the same. 81 Since data on field percentage control were not kept by the districts for the entire period of this study, it was necessary to generate obser- vations of these measurements. Therefore, the following assumptions were made: (1) Pesticides applied against one species simultaneously affect other species present in the area. (2) As the number of mosquito generations subjected to pesti- cides increases, the average effectiveness of pesticides declines, i.e., resistance to pesticides increases. (3) The average number of generations per year (March - November) is assumed to be: Mar Apr May June July Aug Sep Oct Nov A. nigvomaaulis 15 11223321 i./ generations C. tavsalis 12 11222211 £/ generations C.p. quinque- 13 11122222 ±/ fasciatus generations aj If a pesticide treatment was reported in November, it was assumed to affect one additional generation. (4) When a pesticide has been used during a month it is assumed that all mosquito generations of that month have been subjected to its effect. (5) For each class of pesticides, the average time of use before mosquitoes show signs of resistance will be higher for the first product in the class than for those which follow. (This assumption is accounted for by cross resistance. ) 82 (6) A new pesticide has an effectiveness index of 100, i.e., it is 100 percent effective at the first application. (7) A pesticide is replaced when repeated field observations show that its effectiveness is less than or equal to 80 percent. (8) The effectiveness index for a pesticide (e^) 100 if ta where t is the number of generations affected by the pesticide , a is the number of generations elapsed when resistance (or a decline in effectiveness) appears (this number will differ from one pesticide to another) and b is the amount of decline in the effectiveness per period as the number of generations treated with pesti- cides exceeds the critical number "a". (b will differ from one pesticide to another and from one class to another. ) (9) b is estimated as follows: b = 20 T total number of generations affected by the pesticide - the number of generations elapsed when re- sistance is first observed. Therefore, b, the decline per period, is the same for each period but the rate of decline is increasing. For example, a one-unit decline from an original effective- ness of 95 percent indicates a rate of decline equal to 1.05 percent, but the rate is 1.1 percent if the original effectiveness level was 90 percent. This implies that resistance develops faster as the number of mosquito gen- erations subjected to pesticide selection pressure increases. 83 (10) The generated average effectiveness index is weighted average of effectiveness of each class of pesticide used dur- ing the season. Effectiveness of each class is in turn a weighted average of effectiveness of each product in that class. The following table summarizes the parameters used in cal- culating the effectiveness index for each pesticide. Appendix Table B-1 Calculation of Effectiveness Index Class Pesticide a/ (a) Number of generations before resistance is observed (c) Maximum number of generations treated or estimated to be treated before replacement 20 c-a Kern MAD Kern MAD Chlorinated hydro- carbons: DDT Toxaphene 40 32 b/ y b/ w Chlordane 32 Organophos- Parathion 96 300 .0980 phorus : EPN 96 300 .0980 M. Parathion 48 142 .2128 Malathion 48 136 .2273 Dibrom 48 300 .0794 Baytex 48 300 .0794 Dursban 48 300 .0794 a/ Baygon and Altosid were assumed to be 100 percent effective during the period of this study, b/ No use reported or data available during the study period. 84 The information shown in the Table^/ was used to calculate the amount of per-generation decline in pesticide effectiveness for each species after the initial 100 percent effectiveness has begun to decline, 20 i.e., b = c-a. It should be noted that the number of generations in any year differs from species to species, so the time when one species develops resistance to a pesticide will not necessarily be the same for all species. For example, A. nigvomaoulis will be more likely to develop resistance before C. tarsalis . Also, the total decline in effectiveness in one season depends on the number of generations subjected to pesticides' selection pressure, so we expect that the decline or deterioration in ef- fectiveness will be faster for A. nigvomaoulis than for other species. The above information was then used to calculate annual effectiveness indices for each species and for each pesticide in each class of pesticides. The resulting figures were weighted by the values of each pesticide in the total class (a^, 3i, Yi, o^) to obtain the average class index, EC. The average class indices were then weighted by each class value in the total pesticides during the year (a,e,o). An average annual index for pesticides, ET, was then obtained and used as a dependent variable in equations (8), (9) and (10) of the abatement models. (See Figure B-1. ) 2_/ This information was based on data obtained from the control districts' reports and from: Mulla, Mir S., "Solution to the Phosphate Resistance Problem," papers and procedures of the 33rd Annual Conference of the California Mosquito Control Association, Inc., 1966, pp. 73-76. 85 Appendix Figure B-1 Summary of the Calculations of the Effectiveness Index for Each Species Pesticides Used in Month t 00 ON Proportions in total pesticides Pesticides Proportions in class Pesticide effectiveness index CH (chlorinated hydrocarbons ) (1) (2) (3) ... (n) a ^ICH ^2CH ^3CH •••^nCH OP (organophosphorus) (1) (2) (3) ... (f) 62 B3 ^lOP ^20P ^30P "••®fOP CM (carbamate) I Y (1) (2) Yi Y2 ^ICM ^2CM HR (hormone regulator) i (1) O ^IHR Class effectiveness index Effectiveness index for all pesticides "1 eicH + 2 e2CH "3 e3CH + 4 ^^CR as escH + ••• A aA + B IB + C YC + D = EC oD = ET where n = number of pesticides in CH class and f = number of pesticides in OP clas APPENDIX C Appendix Figure C-1 Expected Coefficient Slgn^ - Kern County Hosquito Abatement District (bUVD) (Annual Model) 00 Normalized Endogenous Variables Kl KlO Kl2 K8 Kg "11 Ki4 Expected alRn of : a/ b/ J^2 K3 K, K5 K6 K7 Ka K9 K^p Ku K^ K,, K^e" K^a K^, K20 Vi^-l ^t-l ^5t-l ^10,-1 ^Ut + + or + or + or + + + + + + or + or + or + or + or + + or + + + + a/ Positive up to a apecle specific lindt. then negative if the number of daya above 100' F. is excessive. lid ;hiJ:::phy?'''""" P-"^- - -^-^ve depending on the MAD practices - Jhf ^"L"na«eL'St phUosophy!""' activities depend on the extent of the accumulated source reduction and APPENDIX D Appendix Table D-1 Kern MAD Monthly-Data Model: Empirical Estimation Results Number of Mosquitoes A. nigvomaaulis Equa. 17 (k^) C. tarsalis Equa. 18 (kg) C.p. quvnque fasaiatus Equa. 19 (k^o) Constant term: A. nigvomaaulis numbers C tavsalis numbers (7. p. quinque- fasciatus numbers Acres treated with pesticides Locations treated with pesticides (kj^) (kg) (kio) (k4) (k5) 11.3760 3.450*** (5.14) - .1156 (-1.03) Endogenous Variables -27.2010 - .-^367 .227 ( .50) - .0982 (1.17) .0071** (1.78) Acres treated with pesticides Equa. 20 (k4) 1.0142 - .0095 ( .11)^ .1116 (1.21) Locations treated with pesticides Equa. 21 (k5) 25.3062 - .3923 ( 1.09) .4796 ( .79) - 8.2138 (-1.03) Appendix Table D-1— Continued Number of Mosquitoes A. nigvomasulis Equa. 17 (ki) C. tarsalis Equa. 18 (kg C.p. quinque fasaiatus Equa. 19 (k^o) Acres treated with pesticides Equa. 20 (k4) A. nigromaaulis numbers in the (l^lt-l) previous month C. tavsatis numbers in ^^9t-l^ previous month C.p. quinque fasaia- tus numbers in (^lOt-l) the previous month .1318* (1.59) Predetermined Variables .4233*** (4.04) .5070*** (6.77) Locations treated with pesticides Equa. 21 (ks) Temperature Rainfall Irrigation water River flow Fills constructed in the previous month (k2) - .4047 (-1.10) (k3) 1.4387 (.47) .0344 (kfe) (1.28) -.1631*** (ky) (-4.36) .0832 (ks) (.20) .4452** (2.11) 1.5900 (.90) .00149 (.084) .00054 (.021) -.2922* (-1.30) .0043 (.42) .0231 (.21) .0125** (2.27) .0367*** (6.53) Appendix Table D-1 — Continued Number of Mosquitoes A. nigromaaulis Equa. 17 (k^) C. tarsalis Equa. 18 (kg C.p. qmnque fasaiatus Equa. 19 (k^o) Acres treated with pesticides Equa. 20 (k4) Acres treated with pesticides in (^At-l^ the previous month Locations treated with pesticides (^St-l) in the previous month Predetermined Variables .3710***^ (3.93) 561.6*** (6.31) Locations treated with pesticides Equa. 21 (k5) a/ t-ratios are given in parentheses below the respective coefficient estimate. - ** and * designate the level of significance equal to 1%, 5% and 10%, respectively. t-ratio is defined as t = bi/standard error of bi_, where bi's are the estimated coefficients.