A Selective Review of the Economic Analysis of Animal Health Management

Transcription

1 Journal of Agricultural Economics, Vol. 69, No. 1, 2018, doi: / A Selective Review of the Economic Analysis of Animal Health Management Lovell S. Jarvis and Pablo Valdes-Donoso 1 (Original submitted September 2014, revision received April 2015, accepted July 2015.) Abstract Economists and veterinarians use similar approaches to analyse animal health threats, but veterinarians are concerned primarily with providing practical guidelines to peers and/or policy-makers, while economists focus more on understanding the benefits to society as a whole and often provide only general guidelines to policy-makers and little specific direction to practicing veterinarians. Despite the benefits of working together, differences in perspective and analytical approach often cause economists and veterinarians to struggle in dialogue and to lose some of the mutual gains that could be achieved through collaboration. This article discusses the use of economics in animal health management, identifies several useful literature reviews, and analyses a number of recent studies to explore the advantages and disadvantages of different conceptual and methodological approaches. Keywords: Animal disease control; economics of animal health; economists; veterinarians. JEL classifications: H59, K29, L59, Q18, Q Introduction 2 Livestock production accounts for a large proportion of total agricultural value added, roughly 50% in developed countries and 40% in developing countries, where less meat is consumed on a per capita basis. Animal disease is an important constraint to animal production, with the OIE suggesting that disease may reduce output by 1 Lovell Jarvis is with the Department of Agricultural and Resource Economics and Pablo Valdes-Donoso is with the School of Veterinary Medicine and the Department of Agricultural and Resource Economics, University of California, Davis, CA. lsjarvis@ucdavis.edu for correspondence. 2 A preliminary version of this article was presented at an OECD Conference on Livestock disease policies: Building bridges between animal sciences and economics, Paris, 3 4 June We are grateful to the Editor of the journal and the editors of the Special Edition, to one referee, and to Andres Perez for comments.

2 202 Lovell S. Jarvis and Pablo Valdes-Donoso more than 20% worldwide (OIE, 2014). In addition, animal disease creates significant risks for human health and for animal welfare (OIE, 2014). As a result, increasing effort is being made to reduce animal disease and its impact. Economists and veterinarians 3 play important roles, often in collaboration, particularly in efforts to better understand the nature and magnitude of animal disease impact and to help design public policy and private actions to reduce disease prevalence and damage. Both economists and veterinarians analyse the threat and reality of animal disease with the intent of preventing, controlling or eradicating it. Both professions are aware that such efforts have costs as well as benefits, and each seeks to balance those considerations when deciding on a course of action. However, their perspectives and approaches to disease management may differ. Economics and veterinary medicine are highly specialised fields with distinct disciplinary paradigms, as well as differing knowledge emphases. Practitioners from the two disciplines have relative strengths and weaknesses in understanding their clientele (e.g. animal owners, policy-makers and the broader public) and in communicating with them. Economists and veterinarians have increasingly come to recognise the value of each other s specific knowledge. Economists who analyse animal disease issues must familiarise themselves with the intrinsic epidemiological dynamics of diseases, in addition to how human behaviour affects these dynamics. In turn, veterinarians must understand the economic factors that contribute to disease occurrence and spread, and recognise that public policy and producers efforts to maximise profits can influence disease dynamics. Research results are generally better when the two disciplines work in collaboration. Despite the benefits of working together, economists and veterinarians often struggle to feel comfortable in dialogue with each other. As a result, some of the mutual gains that could be achieved through collaboration are lost. Both disciplines seek to improve welfare, but their approaches often differ in important aspects. While economists focus on increasing human welfare (or utility) irrespective of whether disease is reduced or farmers welfare is improved, veterinarians generally seek to improve human (and animal) welfare strictly by reducing disease. Veterinarians focus on the way animals interact with each other, other animals and humans, on the health effects of disease processes, and on preventing and controlling disease. Economists look at these issues, but focus on how disease affects total economic welfare, considering a variety of economic relationships, the effect of changing incentives, and the implementation of new policies, including animal health policies. For economists, animals are important, but are simply part of the production environment. In the animal health context, infectious diseases generate economic losses. In addition to losses related to animal mortality, morbidity and associated aspects such as barrenness and miscarriage, a reduction in feed conversion efficiency reduces overall system technical efficiency and farm profit. Furthermore, increased costs associated with vaccination, treatments and biosecurity measures to prevent or contain diseases also decrease system profitability and, hence, producer welfare. Finally, the threat of international pandemics is growing as the result of rising animal populations, interactions between animal and human populations and the consequent exchange of zoonosis. The economic analysis of private and social costs and benefits, and consideration of the effects of changing incentives, can help in allocating animal health resources to 3 In this article we refer in particular to veterinarians with a specialisation in epidemiology.

3 Economic Analysis of Animal Health Management 203 avoid, control and eradicate animal disease or otherwise reduce their damage. Individual farmers and policy-makers need to understand the economic impact of different actions in order to choose sensibly among them. Veterinary knowledge is needed to understand the risks of disease contagion, how these are affected by multiple factors, including human behaviour, and how disease affects animals and their productivity. Economic knowledge is needed to assess the economic consequences of different decisions that may be taken to control and/or eradicate disease, taking account of how changing incentives affect human behaviour, and how economic effects may vary across individuals and society as a whole. A key issue for improving collaboration between economists and veterinarians is to develop awareness of and mechanisms to promote the effective exchange of disciplinary knowledge between the two professions when one or both are engaged in analysing animal health management. Economists need to gain and better use knowledge about veterinary medicine in building economic models, and veterinarians need to broaden and improve their knowledge of economics, particularly when using it to clarify decision options for their clientele. To advance this effort, this article provides a brief overview of economic concepts that are frequently of importance in studies of animal disease, emphasising the need to take account of externalities and general equilibrium effects. We then select a number of high quality studies whose efforts to use economics in the analysis of animal health management cover a range of applications. We summarise each study, characterising the disease issue of concern, the analytical approach used, and the broad results. In the process, we attempt to identify methodological strengths and weaknesses and discuss how the studies findings have been or might be used to influence policy. We hope that these summaries will illustrate the main points raised in the following section and be useful to those using economics in future studies. 2. A Quick Overview of Economics and Animal Health Analysis Economics has been described as the science of choice. In general, economic agents must choose between alternatives, each with distinct costs and benefits, in an attempt to maximise welfare or utility. If one course of action is chosen, other choices are effectively discarded. Economics generally proceeds by attempting to determine the net benefit of any action, relative to all such other feasible actions. It can do so by comparing the effects of different actions separately and identifying the one that provides the best result, or a single model can be developed with variable parameters and the model solved to identify the best, or optimal, result. In brief, however, economics can be used to help clarify which decisions improve economic welfare, and whether it is the welfare of an individual producer and/or of society as a whole. For example, in the midst of a disease outbreak, one way to slow and hopefully halt a disease is to cull all infected animals, or even all animals that have had close contact with infected animals. Culling animals may reduce the probability of future infections and associated losses, but at a cost of sacrificing animals that might be healthy and/or that might recover from the disease. In general, culling is only desired if the expected benefits are greater than the expected costs. That decision can be complex, as the decision to cull and, if so, how many animals to cull, is sensitive to the values of animals and the products they produce, the costs of culling and disposing of culled animals, the duration of lost production, the cost of replacing a culled animal, and how retaining an animal will affect the probability that other animals will become infected.

4 204 Lovell S. Jarvis and Pablo Valdes-Donoso Economists generally use monetary units, or prices, to value the gains and losses associated with a given action or set of actions. The use of monetary units is based on the (rather heroic) assumption that prices observed in markets reflect the true (relative) values of the different resources exchanged there. This assumption allows economists to add up gains and losses regardless of the form that they assume (e.g. labour expenditures, pharmaceutical costs or improving the health of an animal). Prices provide the weights that allow the net effect of any action to be calculated, taking account of the different costs and benefits. However, the assumption that market prices reflect true values depends on restrictive conditions regarding how a market works, a fact sometimes missed in the economics analyses performed by animal scientists. Not all markets work well. If markets do not work well, the prices observed in those markets will likely not be the socially correct prices to use in economic analysis. For example, economists note that many actions create externalities, i.e. effects on other economic agents that are not considered by those who undertake the actions. These external effects, which can be positive or negative, are not reflected in market prices, but they can be important and must be considered to determine what actions are most desirable from society s viewpoint, and not just the viewpoint of the individual. A well-known case of externalities occurs in actions to protect against animal disease, such as vaccination. Vaccination of one animal provides benefits for non-vaccinated animals insofar as the vaccinated animal is less likely to become infected and transmit a contagious disease. If many animals are vaccinated, a relatively few unvaccinated animals probably will not become infected. Epidemiologists call this phenomenon mass immunity. As a result, the benefit of vaccination to a group of producers, or society as a whole, may exceed the benefit to the owner of the vaccinated animal and this externality must be considered when formulating the best policy decision, i.e. whether to subsidise or mandate vaccination. Similarly, producers also have different motivations and means to control disease (Rich, 2007). Their incentives will depend on factors such as the nature and severity of the disease, the type of production systems being used, other producers actions and government policy (Hennessy, 2007). For example, a farmer may know that his neighbours have vaccinated, reducing the threat of disease and thus his incentive to vaccinate. If more generalised, numerous farmers could fail to vaccinate, hoping to free-ride on the actions of others (Rasmusen, 2010). The likelihood that free-riding will impair disease control is one reason for mandating that all producers vaccinate. Producer response to incentives, including changing incentives, is often an important aspect of modelling disease control. For example, if policy-makers require vaccination of all animals, some owners may attempt to hide their animals to evade the cost. Thus, a model that attempts to determine the effect of mandatory or voluntary vaccination should include an effort to determine who will attempt to evade, what effect such evasion will have on disease spread, and what can be done to reduce evasion. Economists also recognise the presence of both endogenous and exogenous variables when constructing their models. The difference is important in the econometric estimation of any model because endogenous variables are those that are basically determined simultaneously during estimation of the model. They are not given to the process. Failure to model appropriately the determination of endogenous models is likely to cause bias in the estimated results. A biased result can lead to mistaken understanding of the dynamics of a disease, for example, or an incorrect assessment

5 Economic Analysis of Animal Health Management 205 of how a policy will affect producer and/or social welfare. Attention to the endogeneity of certain variables is a key aspect of most economic studies. The question of how much to include in any model is a key aspect of any study of disease control. Economists sometimes construct elaborate models intended to capture the effects of all the variables deemed important in a particular problem or choice. However, economists often narrow the scope of their analysis to focus on a smaller part of the problem because it is easier and may allow a satisfactory answer more quickly and at significantly lower cost. Economists often speak of partial equilibrium analysis and general equilibrium analysis. The former assumes that a useful answer or solution can be obtained by focusing on a subset of the total economy; the problem confronted may affect other parts of the economy, but in such a small way that they are not essential for inclusion. General equilibrium analysis, by contrast, is used when a larger set of interactions are considered important to the overall result, so that a suitable answer requires broader modelling. For example, with partial equilibrium, an economist might analyse the degree to which an increase in the price of vaccines decreases the number of cattle vaccinated. In a restricted study, this may be sufficient. The economist may decide to disregard longer-term interactions, such as the chance that a reduction in the rate of vaccination will lead to an increased rate of infection, which could then lead to a long-term decrease in the supply of beef and, perhaps, a higher price of beef. If these interactions are thought to be unimportant to the question being asked, they can be ignored. If knowledge of the longer-run interactions is expected to be important, however, the results from the simplified partial equilibrium analysis could be seriously biased. For example, if the goal of the study is to know how the price of vaccination is likely to affect the long-run supply of cattle, and thus meat processors and consumers, the long-range interactions should not be ignored. Similarly, if the economist wants to study whether it is economically attractive to eradicate foot and mouth disease (FMD) in a country with a large cattle herd, an appropriate answer probably cannot be obtained without using a broad model that analyses the effects of policy decisions on farmers, food processors, consumers, and important constituencies linked to international trade and the macro economy. Thus, the design of the analytical model is crucially linked to the answers that are sought, but economists cannot model everything, as complete models are too costly to develop and implement. Although economists attempt to be selectively comprehensive in their analyses and often develop highly-detailed studies, most economic studies still do not attempt to provide precise policy instructions to be used by economic agents or producers. Even fairly detailed models are seen by economists as primarily useful for providing conceptual insights to the problem being analysed. Economists tend to believe that real world problems differ in important ways from any problem that has already been conceptualised and analysed. For example, although a general model may indicate that ring vaccination can reduce the probability of disease spread, the actual cost/benefit relationship of ring vaccination is likely to depend on a variety of circumstances that vary with each disease outbreak and that, accordingly, cannot be known until a specific outbreak in a specific place has occurred. Thus, economists usually believe it is more important for policy-makers to understand the complex elements likely to be involved in a particular type of problem rather than propose a specific recipe for action. The policy-maker needs to be able to devise an appropriate policy when a need arises, e.g. designing and implementing a disease control policy in a rapidly evolving context that is broadly understood and anticipated even if not precisely understood

6 206 Lovell S. Jarvis and Pablo Valdes-Donoso when initiated. It is worthwhile noting that economists, because they focus on wider social welfare when considering policy initiatives, are much less likely than veterinarians to identify desirable activities for producers, let alone individual producers, which, of course, is what veterinarians often seek to offer. Paul Samuelson, one of the great economists of the 20th century, said that economics could be described as a set of parables, which in his case was a short story illustrating a principle or lesson. Characterising economic discoveries as parables says a lot about how economists think about their profession, and provides insights into their likely contributions. As noted, a parable is illustrative, but does not provide specific instructions. Instead, it is a useful way of thinking about a problem an intellectual guide to understanding complex issues and an effort to identify the types of policy responses that may be helpful for improving social welfare. Thus, although economists usually do not provide concrete instruction regarding what to do in a specific context, most economic research is concerned with appropriate policy formulation of a general sort. While the previous effort to characterise economists approach to modelling animal disease management is partial and has doubtlessly omitted important issues, we hope to use it to compare and contrast methods used by veterinarians when they incorporate economics into their analyses. Veterinarians focus on the welfare of producers and seek to understand what producers should do when faced with a given health issue. In recent years, they have increasingly incorporated economics into their analyses. Much of the analysis is fairly straightforward. An effort is made to identify how a particular action is likely to affect disease management, e.g. if a farmer vaccinates, what is the expected gain from reducing the threat of disease and how does that gain compare to the cost of vaccination. Because their focus is more specifically on the owners of animals, as opposed to society as a whole, veterinarians are much more likely to utilise partial equilibrium as opposed to general equilibrium analytical approaches. Similarly, they often seek specific guidelines that can be applied at the producer level, e.g. given the following parameters regarding the nature of the disease threat, prices for inputs and outputs, and available technology, when in this situation, take the following specific action. Their analyses rarely consider the existence of important externalities. Thus, even if the recommendations are potentially beneficial to the producer, the failure to consider externalities may mean that the recommendations are appropriate for individual farmers, but not for groups of farmers or society as a whole. Similarly, studies implemented by veterinarians rarely consider the issue of endogeneity and thus the results can be subject to significant bias. The point of our analysis, however, is not to complain. We are delighted to see veterinarians using economics and want to encourage veterinarians to expand their knowledge of economists and/or work with economists to further develop their analytical approach. In similar fashion, economists must have the knowledge of disease that veterinarians and epidemiologists have to undertake their analyses of animal disease. For example, if the assumptions used by economists in their models do not accurately reflect the infectious behaviour and/or productive impact of a disease, these models will not produce accurate results. Veterinary epidemiologists have traditionally analysed disease dynamics within an animal population based on observed interactions among the host, pathogen and environment. However, epidemiologists have increasingly recognised that animal disease dynamics are often influenced as well by anthropogenic effects, e.g. owners responses to the disease, government intervention, and their interaction.

7 Economic Analysis of Animal Health Management 207 For example, epidemiological research has successfully modelled the effects of changing government policies on disease dynamics by introducing evolving disease risk factors for disease transmission. However, changing risk factors are necessarily associated with changing management practices, which in turn depend on how farmers perceive their options. For example, farmers recognise that vaccinating animals, restricting their movement, reducing herd density, and/or increasing the frequency of laboratory diagnostics will all increase production costs. Understandably, farmers may resist implementing new management practices that appear to reduce their profits unless they are persuaded that they profits will decline even more if they do not protect themselves again disease. Veterinarians have thus tried to provide information about the expected costs and benefits of different management practices to persuade farmers that higher production costs are worth accepting as the means of reducing the potential losses from animal disease when the disease risk is great. Since disease control policies almost inevitably depend on the commitment of producers, the latter must recognise the benefit of any action if it is to be successfully implemented. The role of veterinarians in working with producers to influence their management practices is highly important. Veterinarians have increasingly incorporated additional economic tools to better understand and justify actions or investments that producers should take. Decision trees, benefit/cost analysis and cost-effectiveness analysis are tools frequently used by animal health scientists to help farmers make decisions and also persuade farmers of the importance of accepting and embracing public interventions. For example, decision tree analysis utilises probability information and monetary values to address sequential problems that producers may face. The tree usually is built to reflect a set of decision alternatives, where different possible outcomes result from each decision. Probabilities are associated with each possible outcome and monetary values are associated with the outcomes as well, so that the farmer can observe the expected values associated with each set of sequential decisions. This tool provides an intuitively satisfying analysis from the viewpoint of the individual farmer. In turn, cost/benefit analysis discounts the expected streams of costs and of benefits using a discount rate selected to reflect the opportunity cost of capital, and calculates the ratio of the present values of benefits to costs that is associated with a specific proposed action or intervention. Alternatively, the same stream of costs and benefits can be used to calculate the net present value or an internal rate of return for the same action or intervention. Cost-effectiveness analysis seeks to identify, among a number of alternative actions or policies that are all deemed to achieve the exact same objective, which action achieves the goal at least cost. These tools are fairly easy to implement and can provide valuable insights and prescriptions to producers. However, their focus is narrowly restricted to decision making by the individual producer and the results are sometimes worrisome to the economist because of their restricted perspective. In subsequent reviews, we illustrate some of the resulting problems. 3. Economic Methods Used in Animal Health Management Many books and articles explain the economics paradigm, basic theory, and general methodologies used to identify and analyse economic problems. A number of publications have been written by economists specifically for veterinarians and others interested in animal health. For example, Rushton (2009) provides a highly

8 208 Lovell S. Jarvis and Pablo Valdes-Donoso useful overview of economic principles as applied to livestock disease control. The emphasis is on using a profit maximising framework to assist in making decisions, whether at the farm level or in terms of national policy. The book also notes that risk and uncertainty are especially important in dealing with animal disease management, so that decision makers often must take decisions in situations where a deterministic framework does not adequately characterise the problem. Similarly, the book emphasises the importance of the livestock, economic and, especially, the institutional/political context in determining policy. Numerous chapters, some by other contributors, contain illustrative applications of economics to specific animal health problems. See also the article by Otte and Chilonda (2000), for a much briefer overview that can serve as a basic introduction to the use of economics in animal health management. Rich, alone and with co-authors, has published several articles reviewing some of the most common methodologies used by veterinarians in animal health analysis. Rich, Miller and Winter-Nelson (2005) provide a useful overview of economic methodologies, including quantitative techniques that can be used in conjunction with epidemiological information to analyse animal disease problems. Although different techniques are identified, each attempts to determine the desirability of a specific solution to an identified animal health issue. In that sense, all approaches balance benefits against costs, with the choice of approach determined primarily by the modeller s assessment of what type of analysis is needed to get a useful answer, what data are available, what is known about the disease and disease processes, and also the modeller s facility with different techniques. The approaches discussed include basic cost/ benefit analysis, which is then complemented with discussion of linear and mathematical programming models, partial equilibrium models, multi-market models, input output and social accounting models, and computable general equilibrium models. Some methodologies are intended to be parsimonious in their demand for data and calculations, while others are more demanding in data and calculation, but also incorporate a greater number of economic actors in the analysis and achieve more detail and, hopefully, precision. Some methodologies assume that only first-order effects need to be considered, while others seek to understand additional effects via a wide array of linkages across sectors. Some models attempt simply to determine whether one particular choice is beneficial; others seek the optimum choice among numerous alternatives. In a second article, Rich, Winter-Nelson and Miller (2005) extend the discussion to link economic and epidemiological models and to consider the importance of both time and scale in dynamic models. For example, although simple benefit/cost analysis and cost effectiveness analysis have advantages for evaluating the short-run direct impact of animal disease at the herd level, they cannot assess large-scale or long-run effects. They note that a number of studies attempt to scale up from the analysis of a single herd to a regional level, but in doing so commonly omit certain large scale effects, e.g. changes in the prices of inputs and outputs, and in producer s responses, distorting the results. Rich and Perry (2011) explore the application of value chain analysis to disease control. Although value chain analysis, in and of itself, is not highly useful in developing specific results, its use leads the analyst toward a more complete understanding of the broader context within which animal health issues evolve and helps identify which interactions warrant more detailed examination. This type of analysis is necessary for the development of any broader general equilibrium model.

9 Economic Analysis of Animal Health Management Applications of Economic Tools Despite a growing literature, economics remains underused by animal scientists and economists often inadequately use veterinary knowledge. To illustrate how economics has been used in animal health studies, we review a number of studies that we have selected to demonstrate many of the issues outlined above, attempting to identify strengths and weaknesses in their approaches. We briefly characterise each study, emphasise their conceptual approach, main results and their policy implications. We have placed greater emphasis on views prevalent in the US, with which we are more familiar Potential impact of FMD in California: The role and contribution of animal health surveillance and monitoring services (Ekboir, 1999) Ekboir hypothesised that California s dairy sector was vulnerable to extensive loss if an FMD outbreak were to occur and sought to explore this hypothesis with the intent to influence farmer management practices and public policy. He noticed that the sector was composed mainly of large dairies (more than 500 cows each) that were dependent on external vendors, including veterinarians, and milk pickup, feed delivery and manure removal services that visited multiple dairies on a daily or weekly basis. The entry and exit of multiple, frequent visitors significantly increased both the probability of infection on any one farm, as well as the likelihood that any infection would rapidly spread among other farms. He noted the lack of biocontrols in dairies. Considering the 3-day latency period after an outbreak begins and before clinical signs appear, Ekboir hypothesised that the frequent movement of external providers from one dairy to another would cause any outbreak to become widely diffused before the outbreak had been discovered. Moreover, he found that the disease would continue to spread in the absence of a strict quarantine. The disease is highly contagious, airborne, many dairies are located in close proximity to one another, and many are located within heavily-trafficked automobile corridors. Ekboir s study was undertaken relatively rapidly, and the model and its parameters were somewhat crude. For example, although Ekboir consulted frequently with veterinarians and epidemiologists, he worked without a detailed epidemiological framework. He used a Markov-Chain analysis with approximate disease risk parameters to simulate the rate of disease diffusion under different policy scenarios, estimating the expected economic losses associated with each. The three key policy elements he identified that could influence overall disease impact were: (i) sensitivity to and monitoring of disease presence, i.e. the length of time from initiation of an outbreak until the presence of the disease was recognised, and thus the degree of disease spread before control could be implemented, (ii) the rate at which diseased and exposed herds could be culled, once the disease was identified, and (iii) the effectiveness of quarantine on restricting animal movements and further contagion. Although Ekboir s model was necessarily approximate, the study had two particularly surprising and important results. First, it showed that the expected spread of an FMD outbreak in the California dairy sector was considerably more rapid and extensive than had been previously recognised. Any outbreak would therefore be extremely serious, requiring the culling of a large fraction of the dairy herd and resulting in a sharp reduction of California milk output for a prolonged period. There would also be huge losses to the US beef cattle sector as a result of resulting international trade

10 210 Lovell S. Jarvis and Pablo Valdes-Donoso restrictions (harming ranchers and processors, benefitting US consumers and harming foreign consumers). The principal economic loss to the US economy from an FMD outbreak in the California dairy sector was shown to result from an expected immediate and nearly complete cessation of international trade in beef. This is because primary importing countries principally Japan and South Korea rejected all beef from countries with open FMD outbreaks, even if the outbreak was far from the site of the cattle whose beef might be exported. Moreover, the cessation of imports was to continue for at least 6 months after disease eradication. The expected US loss would therefore be great, even if the outbreak in the dairy sector was eradicated relatively quickly and although California exports very little beef. The economic loss arising from a decline in US milk exports was much smaller, though the expected losses associated with animal culling and lost production on dairy farms were significant. Second, the study found that measures taken to control the outbreak in California would face serious obstacles. Many dairies are located in zones that have become largely urbanised in recent decades. Auto traffic is heavy in these regions, as individuals commute to work and firms ship cargo in all directions. It therefore seemed difficult to impose an effective quarantine on movement within the region, as the economic losses in the broad economy associated with halting traffic would greatly exceed any savings that would result from more quickly controlling the FMD outbreak. Further, Ekboir found that it would be extremely difficult to dispose of the hundreds of thousands of carcasses that the model suggested could result from attempts to stamp out the disease. The carcasses could not be burned because much of the area where dairies are located suffers from severe air pollution, and permits for burning simply would not be provided by the local air quality control boards. In addition, firewood supplies were not considered adequate to permit the incineration of animals as fast as they were culled. Similarly, the carcasses could not be buried because of likely groundwater contamination. Therefore, although US policy called for stamping out in case of an FMD outbreak, this policy was expected to be limited by other factors in much of the region. Ekboir did not propose specific remedies, though alternatives were mentioned and suggested for further analysis. However, the study did awaken California state authorities (specifically, the Secretary of Agriculture and the Head of the State s Animal Health Branch within the Department of Food and Agriculture) to the significant threat of an FMD outbreak. The study also prompted more attention to FMD within the UC Davis School of Veterinary Medicine. Leaders of dairy producer associations began holding conversations immediately, organising meetings throughout California to discuss the implications of a possible FMD outbreak. These meetings were intended to discuss proposed measures that the industry thought useful, including courses for farm workers to bring attention to possible signs of an outbreak. Dairies were urged to improve biosecurity, specific measures were proposed, and the state began contingency planning for how best to deal with a possible outbreak. In short, the Ekboir study was important primarily because it identified FMD as an important disease threat, conceptualised and modelled the problem to highlight the major policy decisions that would be faced if an outbreak occurred, and garnered attention from animal health authorities and producers. Ekboir s study showed the need to consider the various issues that may be faced by efforts to deal with an animal health issue, and to prepare policy-makers for various possible scenarios. As a result of Ekboir s analysis, numerous subsequent studies were conducted to analyse FMD sanitary policy, anti-terrorism policy and the best policies for FMD control in case of

11 Economic Analysis of Animal Health Management 211 an outbreak. These included the feasibility and economic attractiveness of using ring vaccination to control an outbreak. Although these studies might have been developed even if the Ekboir study had not been undertaken, it appears that Ekboir s work significantly heightened attention to this topic. In addition, Ekboir sought partial funding for his study from the School of Veterinary Medicine at the University of California, Davis and carried out the analysis with significant input from university faculty, state veterinary authorities and dairy producers. Participation from veterinary faculty and the producer community significantly improved the quality of the study and brought its results very quickly into the hands of those most interested in animal health policy. Thus, the study demonstrated the advantages of collaboration between economists, veterinarians, farmers and policy-makers. Although necessarily approximate given the knowledge available, the general results clearly demonstrated a need for significantly greater attention to the disease threat from FMD in California Expected utility of voluntary vaccination in the middle of an emergent Bluetongue virus serotype 8 epidemic: A decision analysis parameterised for Dutch circumstances (Sok et al., 2014) Bluetongue virus is a non-contagious, insect-borne disease affecting ruminants. Following an outbreak of Bluetongue virus serotype 8 (BTV-8) in 2007, the EU encouraged its member countries to vaccinate against the disease in 2008 and provided financial incentives for such efforts. Because a mandatory campaign for a different disease had been marred by use of a contaminated vaccine, which had left farmers unhappy, the Netherlands implemented a voluntary campaign for Bluetongue virus vaccination, including a subsidy element to reduce farmer cost. The vaccination coverage achieved was 70 80% in 2008 and new infections in subsequent years were not reported. Based on this successful experience, Sok et al. (2014) used decision analysis, i.e. a decision tree model, to assess farmers willingness to vaccinate against BTV-8, with the intent to better understand the likely effectiveness of subsidies to induce voluntary action by producers. The model was developed assuming farmers maximised expected utility under risk. The decision tree framework assumed the costly vaccine reduced the probability of disease in a farmer s herd, and converted the reduced disease impact to an expected monetary value. Thus, the authors pose the farmer s choice into one between two possible costs: (i) the cost of vaccinating, and (ii) the cost associated with incurring the disease if vaccination does not occur. In the latter case, producers that exported heifers had also to consider that the export of heifers was permitted only if there was no BTV-8 virus circulating or if the farmer had vaccinated his herd. Thus, vaccination also increased the probability that a farmer could export heifers and thus the expected price of heifers. Rather than simply determining whether the expected monetary loss from vaccination was greater or less than the expected loss from disease if the farmer did not vaccinate, the authors converted monetary losses into utility losses using a negative exponential utility function to express farmers risk preferences. The transformation of income into utility allowed the authors to explore the effects of producer risk aversion on their decisions, comparing the expected utility associated if the farmer vaccinated vs. that if the farmer did not vaccinate, given the expected damages from

12 212 Lovell S. Jarvis and Pablo Valdes-Donoso disease. The expectation was that farmers with greater risk aversion would be more likely to vaccinate, given the large expected loss associated with not vaccinating. The authors used Monte-Carlo simulation to estimate the income losses associated with vaccination and no vaccination. The expected net benefit from vaccination in the first year was large, suggesting that nearly all farmers should have been expected to vaccinate voluntarily. In the second year, however, the expected benefit of vaccination varied significantly, depending on whether the farmer was assumed to have vaccinated or experienced the disease, and was sometimes negative, e.g. if the farmer had vaccinated in the first year, as there was a probability of some remaining vaccine effectiveness so that it was not cost effective to vaccinate again. As expected, when the authors assumed that farmers were risk averse, the net utility of vaccination rose. The net utility of vaccination was also higher if farmers produced heifers for export. What is most striking about this analysis is that the authors do not mention the importance of herd immunity or the possibility that producers decisions to vaccinate may be affected by their perceptions about their peer s behaviour. Bluetongue is not contagious, but midges spread the disease from diseased animals to uninfected animals. Thus, on the assumption that midges are present, vaccination can achieve herd immunity by sufficiently reducing the number of infected animals to effectively eliminate the disease reservoir. This appears to have resulted in the Netherlands. When the vaccine campaign was initiated, vaccine manufacturers claimed vaccines would be effective for at least 1 year. In fact, vaccines turned out to be effective for at least 3 years, and vaccine coverage was apparently sufficient to eradicate the disease. Irrespective of these facts, the authors analysed how the level of vaccination in the first year would affect producers perceptions of the probability of infection in the second year and thus their decision to vaccinate in that year, with the decision determined by expected changes in utility that varied with farmer income. In this analysis, the authors used assumed transmission factors that were distinct from those that actually seem to have prevailed. Thus, the analysis is contingent on the assumed parameters, not those that actually prevailed in the Netherlands during the second year. Accordingly, if individual farmers had assumed that a high proportion of their peers were going to vaccinate, they might reasonably have assumed that they could free ride because vaccination by others would achieve herd immunity. Apparently, some 20 30% did free ride. This probably was an individually sensible choice since 70 80% of producers vaccinated during the first year, a number apparently adequate to achieve herd immunity based on a lack of outbreaks in year 2. The authors conclude that Dutch farmers demonstrated a willingness to vaccinate against BTV-8 and seem then to suggest that it might be appropriate for the government to allow Dutch farmers to decide voluntarily whether to vaccinate not only for this disease but for others, noting that farmers decisions might depend on the level of subsidy offered. We find it dubious that vaccination should be left to voluntary decision if the disease is, as in this case, indirectly contagious, as vaccination then necessarily involves an externality. If there is significant interest in achieving herd immunity, it seems better to mandate vaccination, though careful analysis of farmers incentives and behaviours is often needed even then to determine effectiveness. Whether to subsidise vaccination ought to be determined as part of the analysis of farmer behaviour, but if the private gains are large relative to the cost of vaccination, subsidisation does not seem socially attractive. The use of expected utility as an analytical framework also raises an important issue. In all studies, there is a potential trade-off between methodological

13 Economic Analysis of Animal Health Management 213 sophistication and the subsequent communication of results, particularly to farmers and/or policy-makers. In this case, there was some potential analytical benefit from the focus on producer utility instead of on simple farm income, but the authors had no data on actual farmer risk preferences and thus had to use a simple proxy that probably had little relation to actual farmer risk preferences. Thus, the focus on utility demonstrated the use of this technique, but not in a way that could have made the results more credible, while simultaneously making the article less accessible to many readers. In this case, we would have preferred an analysis that simply showed the very large private and social gain that vaccination should have achieved and noted the importance of indirect contagion for a non-contagious disease Emergency vaccination to control FMD: Implications of its inclusion as a US policy option (Hagerman et al., 2011) The Hagerman et al. article re-examined the attractiveness of vaccination as a control strategy, looking both at a possible FMD outbreak in the California dairy sector and in the Texas beef cattle sector. The study noted that USDA policy favours emergency vaccination use only if standard culling practices may not be enough to control spread of the disease. In other words, the study sought to determine whether emergency vaccination might be cost effective in achieving disease control in the US. This study is at least partly influenced by the successful policy introduced in the Netherlands in 2001 that utilised ring vaccination and subsequent slaughter of vaccinated animals as a disease control mechanism. Designed by a team of veterinarians and economists, the Hagerman et al. study used simulation modelling with both epidemiological and economic details to analyse and compare a standard culling approach with the use of standard culling plus several emergency ring vaccination strategies. Under standard culling, herds that are infected with FMD or that are deemed to be dangerous as a result of exposure to FMD, are culled. Remaining herds are monitored and movement restrictions are imposed. The alternative approach includes standard culling and also vaccination of herds that are deemed most at risk and the vaccinated animals are later slaughtered. Vaccination is undertaken as an additional precautionary effort to stop spread of the disease. It should be noted that epidemiological information is used in these models to provide parameters for the economic analysis, but the epidemiological and economic analyses are not directly linked. Rather, results from the epidemiological analyses are used as inputs in the economic analysis. To date, this is standard practice because of the difficulty of integrating the two types of models. The study examined the effects of the different vaccination strategies on both animal loss and on economic welfare. One of the major issues highlighted was the skewed distribution of the estimated benefits and losses among producers and consumers. Indeed, the article notes that if a particular strategy leads to social gains of 60, with farmer losses of 20 and consumer gains of 80, it may be difficult to implement the policy because farmers may seek to avoid the components that are associated with their economic loss. It is important to mention that veterinary studies probably place more emphasis on animal and producer effects, while economists probably place more emphasis on the broader economy, and this difference may occasion some tension. The article also explores the likelihood that an FMD outbreak could become truly massive in scope, and whether a ring vaccination strategy might reduce the probability of such an outbreak even if it comes at a greater expected cost than simple culling.

14 214 Lovell S. Jarvis and Pablo Valdes-Donoso In fact, results from the article s simulations yield an average result in which culling is the most cost-effective strategy for controlling an FMD outbreak. However, under numerous scenarios, if adjusted for risk, the benefits of using vaccination are deemed greater than those arising from simple culling. The study concludes that ring vaccination may be an attractive instrument for reducing the risk of an expanded outbreak. Note that although this study utilises large and complex models, it obtains solutions that are somewhat inconclusive. Nonetheless, its results were sufficiently informative to persuade the authors to recommend that the US look seriously at changing currently policy limitations on the possible use of ring vaccination. The authors do not definitively recommend the use of ring vaccination in case of an outbreak, but recommend that ring vaccination be permitted as a policy option. The value of the study is thus in providing insight regarding the potential gains and losses from ring vaccination in different circumstances which will hopefully beneficially enlighten policymakers facing an actual FMD outbreak but no specific recommendation is given The economics of managing infectious wildlife disease (Horan and Wolf, 2005), and joint management of wildlife and livestock disease (Horan et al., 2007) A number of recent studies examine the dynamics of disease and the optimal path of disease control actions over time. These studies also examine the desirable use of various policy instruments, such as testing, culling, vaccination and/or use of biosecurity and other actions, with the use of each instrument depending on its cost and also on the value of the animal resources in question. The studies have shown that in some cases it is economically optimal to eradicate a disease while in other cases it is economically optimal to simply control the disease without eradication. Focusing on control rather than on eradication in these studies has often depended on the existence of a wild animal reservoir of infected animals. However, a similar result can occur if policy-makers in one region cannot eradicate a disease in another region, and thus prevent the re-entrance of disease from that region, e.g. Rich and Winter-Nelson (2007). These two articles consider the potential spread of infectious disease between wild and domesticated animals, in which wildlife acts as a disease reservoir and a vector of disease transmission. Eradication of a disease in wildlife is usually more difficult and costly than eradication in domestic animals. If wildlife are infected, the disease can remain present in wildlife even if it is eradicated in domesticated animals and continued contacts between (infected) wildlife and domesticated animals are then likely to re-introduce disease. Bicknell et al. (1999) advanced analysis in this area using a bioeconomic dynamic model to analyse the control of bovine tuberculosis spread by Australian brush-tailed possums to dairy herds. While most policies had previously focused on urging farmers to regularly test cows and cull those infected and exposed, Bicknell et al. also recommend hunting possums to reduce the wildlife reservoir and the likelihood of contacts. Horan and Wolf (2005) utilised a similar, but more complex model to analyse the management of infectious bovine tuberculosis within a four-county area in the US state of Michigan, where the disease is also endemic among white-tailed deer. In this case, disease is regularly spread between cattle and deer. However, the authors show that simply reducing the deer population is not the best strategy in this context. Although the infected deer create risks for cattle, white-tailed deer are valuable as hunting trophies. Indeed, Horan and Wolf assert that deer hunting is arguably the