Animal Disease Pre Event Preparedness Vs. Post Event Response
Transcript Of Animal Disease Pre Event Preparedness Vs. Post Event Response
ANIMAL DISEASE PRE EVENT PREPAREDNESS VS. POST EVENT RESPONSE: WHEN IS IT ECONOMIC TO PROTECT?
LEVAN ELBAKIDZE Assistant Research Professor Department of Agricultural Economics
Texas A&M University College Station, TX, 77843-2124
(979)845 3225 [email protected]
and BRUCE A. MCCARL
Regents Professor Department of Agricultural Economics
Texas A&M University College Station, TX, 77843-2124
(979) 845 1706 [email protected]
This research was supported in part through the Department of Homeland Security National Center for Foreign Animal and Zoonotic Disease Defense at Texas A&M University. The conclusions are those of the author and not necessarily the sponsor.
ANIMAL DISEASE PRE EVENT PREPAREDNESS VS. POST EVENT RESPONSE: WHEN IS IT ECONOMIC TO PROTECT?
Abstract We examine the economic tradeoff between the costs of pre event preparedness and post event response to potential introduction of an infectious animal disease. In a simplified case study setting we examine the conditions for optimality of an enhanced pre event detection system considering various characteristics of a potential infectious cattle disease outbreak, costs of program implementation, severity of the disease outbreak, and relative effectiveness of post event response actions. We show that the decision to invest in pre event preparedness activities depends on such factors as probability of disease introduction, disease spread rate, relative costs, ancillary benefits and effectiveness of mitigation strategies.
Key words: Animal disease, mitigation strategies, economic balance, preparedness, response JEL Classifications: Q16, Q18, Q19
ANIMAL DISEASE PRE EVENT PREPAREDNESS VS. POST EVENT RESPONSE: WHEN IS IT ECONOMIC TO PROTECT?
Introduction
Possible intentional or unintentional introductions of contagious animal diseases could result in substantial economic losses as seen during the UK, US and Canada BSE (mad cow) events or the European Foot and Mouth Disease events (Thompson et al., 2003; Mangen and Barrell 2003; Henson and Mazzocchi 2002, Khan et al. 2001). Events with major consequences raise the specter of preventative/protective actions. Many calls have been issued for such actions in the post 9/11 world. However the cost of following all of the protection/prevention actions that have been called for is far in excess of any practically available budget.
Many issues can be raised about animal disease management and the need for protection. One such issue involves the balance between pre event investments in prevention/protection/response capability versus the post event costs of the event and associated disease management efforts. A key economic point in the context of this balance is the distinction between pre event and post event costs. Pre event actions impose costs regardless of event occurrence, while post event costs are only incurred when an incident occurs and thus are multiplied by the probability of the event when computing expected annual costs. For example, the costs of setting up and operating a continuing animal health surveillance system are encountered whether or not an outbreak ever takes place. However, the costs of diseased animal slaughter, reduced market supply, disinfection and event enhanced detection arise only in the event of disease
1
introduction. This paper reports on an investigation of the above mentioned balance problem
addressing how disease event characteristics and mitigation options affect the desirability of pre event investments versus post event response. In carrying out this investigation we first present and analyze a theoretical model of the balance problem. Subsequently we conduct an empirical case study motivated by data representing Foot and Mouth Disease (FMD).
A model of pre and post event decision making
Decisions in the context of an animal disease event can be categorized into 6 basic categories. These are
• Anticipation actions –things undertaken to improve the forecast of event likelihood and consequences such as intelligence gathering. These are largely pre event actions.
• Prevention actions – things undertaken to avoid event introduction or mitigate event implications upon introduction such as changes in sanitary or feeding practices along with the use of vaccinations. These are largely pre event actions.
• Detection actions – things undertaken to screen for precursors to an outbreak that speed detection and allow rapid treatment such as inspection for sick animals. These can be pre or post event actions. In a post event setting they are reflective of enhanced detection to help avoid disease spread and or avoid entry of contaminated products into the food chain.
2
• Installation actions– facilities installed to allow more rapid or effective disease detection and management. For example installation of sensors, construction of veterinary laboratories, training of first responders or stocking of vaccines. These are largely pre event actions.
• Response actions –disease management activities undertaken to halt the spread of the event such as slaughter of infected animals, carcass disposal, vaccination of animals in proximity to an event etc. These are post event actions.
• Recovery actions –things undertaken to reestablish productive capacity and market demand post event such as decontamination of production and processing facilities or advertising to increase consumer confidence. These are post event actions.
There are a number of important characteristics of decision making in this pre/post event type of situation. These include,
• Irreversibility – when a pre event action has not been undertaken, once an event has occurred it is generally not possible or at least very expensive to put it in place.
• Conditional response –certain response options can be used only if certain pre event actions are undertaken. One cannot use detection equipment that was not previously acquired and installed.
• Fixed cost versus probabilistic variable costs – in total cost accounting the pre event costs are always present; the post event costs are only encountered when
3
an event occurs. • Large span of possible events -- there is an enormous span of possible events
that can never practically be enumerated. Thus we will only deal with sample and abstract events below. Furthermore these events differ in nature and severity. • Probabilities – event probability is difficult to anticipate and in the case of deliberate actions is likely to be modified by pre event actions. This leads us to a restatement of the balance problem as the establishment of the optimal tradeoff between the cost of pre event actions and occasional post event damages including response and/recovery costs. In such a setting, the best strategy would depend on many factors including pre event action costs, disease management costs, potential damages, and event probability being a balance between these factors. Formal model development This problem can be addressed more formally. Consider the decision tree situation that depicts occurrence or non occurrence of a single event (Figure 1). Here we have a simple two stage decision process. The first stage is pre event and the second stage post event but allows for no event to have occurred (event occurrence probability is P, and no event (1-P)). In stage one decision makers have the option to invest in pre event actions, such as anticipation, prevention, installation and detection, as well as doing nothing. In stage two, there is a probabilistic possibility of an event - introduction of infectious cattle disease - or of no event. At the second stage decision makers can either initiate post event response actions with knowledge of an event taking place, or do
4
nothing. The post event response actions for animal disease management generally involve slaughter, vaccination, and quarantine strategies that are chosen so as to minimize disease induced economic losses. If there is no event then industry activities continue under normal conditions, although the costs of pre event actions implemented in the first stage will be incurred.
Under the context considered in this work, mitigation costs are composed of the pre event set of actions (s) with per unit cost ws, and the post event set of actions (r) with per unit cost wr. Lets assume that the event damages L(δ,s,r) are a function of pre event actions and post event response actions along with an incident severity parameter (δ). Denoting probability of event occurrence as P we can write average cost as:
(1) C = P ⋅ (L(s, r,δ ) + ws ⋅ s + wr ⋅ r) + (1 − P)ws s
Comparative statics analysis We adopt an expected cost minimization approach to investigate the relationship
between pre event preparedness and post event response mechanisms. Now suppose we study the optimal amount of pre and post event action and how it is influenced by
• the probability of the event • the costs of the pre and post event actions • the severity of the event. First order condition for the optimality of pre event and post event actions are as follows:
(2) PLs (δ,r,s) + ws = 0 (3) Lr (δ,r,s) + wr = 0
5
Comparative static analysis can be used to examine the balance between pre and post event actions under variations in disease severity (δ), pre and post event action costs (ws and ws), and probability of event occurrence (P). The total differential arising from equations (2) and (3) is given in (4). By applying Cramer’s rule we get (5 through 11), which permit examination for comparative static results.
( 4)
⎡PLss ⎢
PLsr
⎤⎡ds⎤ ⎥⎢ ⎥
=
⎡− ⎢
dws
−
Ls dP
−
PLsδ dδ⎤ ⎥
⎣ Lrs Lrr ⎦⎣dr⎦ ⎣ − dwr − Lrδ dδ ⎦
(5)
ds =
− Lrr
dw s P(L ss Lrr − Lsr 2 )
(6) dr =
Lrs
dws P(Lss Lrr − Lsr 2 )
(7) ds =
Lsr
(8)
dr = − Lss
dwr (Lss Lrr − Lsr 2 )
dwr (Lss Lrr − Lsr 2 )
(9) ds = − Lsδ Lrr + Lsr Lrδ dδ (Lss Lrr − Lsr 2 )
(10)
dr = − Lrδ Lss + Lrs Lsδ dδ (Lss Lrr − Lsr 2 )
(11) ds = − Ls Lrr dp P(Lss Lrr − Lsr 2 )
Assume the L function is convex in r, s and δ. In turn the above equations reveal information on the sensitivity of the optimal balance between pre and post event actions relative to the other model parameters. Namely,
• Equation 5 can be signed as negative indicating downward sloping demand for pre event actions, namely the higher the per-unit cost of pre event action, the less of that activity is used.
• Equation 8 similarly indicates downward sloping demand for post event actions. 6
• Equation 11 can be signed to be positive sign since L is decreasing in s and convex in r indicating that pre event actions increase with increasing probability of event occurrence.
• The signs of the terms within (6) and (7) are determined by the sign of Lrs and Lsr and when negative indicate complementarity between pre event preparedness and post event response, while positive signs imply they are substitutes.
• Equations (9) and (10) are not readily signed being dependent on the signs and relative magnitudes of Lsr, Lsδ, and Lrδ and thus remain ambiguous.
Empirical investigation using FMD motivated data
Our ability to sign some but not all of the terms above combined with the somewhat abstract nature of the pre and post event actions makes it desirable to do a case study. Thus, we empirically investigate the optimal combination of pre event preparedness and post event response strategies in an empirical setting using data drawn from the FMD literature in the context of possible introduction into Texas. Case study background
Although the US has been FMD free since 1929 (McCauley et al. 1979), disease introduction has been shown to have substantial implications elsewhere. For example, Great Britain experienced an FMD outbreak in 2001 where associated total losses are estimated to be £5.8-8.5 billion (Thompson et al. 2003 p. 25, Mangen and Barrell, 2003 p. 126). Given such large risks FMD is a priority area of concern within USDA and DHS.
7
Analysis of FMD related decision-making has been the topic of numerous studies (e.g. Bates et al. 2001, 2003a,b; Garner and Lack 1995, Schoenbaum and Disney, 2003; Berentsen et al.1992; McCauley et al. 1979; Ferguson et al. 2001; Keiling et al. 2001). These studies mainly concentrate on decision-making once an outbreak has occurred largely addressing post outbreak disease spread management with vaccination and slaughter.
Less attention has been devoted to pre event decision-making. Issues have been raised regarding surveillance systems (Bates et al. 2003b; Akhtar and White 2003, Ekboir 1999), but we cannot find empirical investigations that address the economic balance that might be drawn between pre event preparedness and post event response actions. We will address this issue focusing on the balance in a limited setting focusing on the installation/operation of surveillance and detection systems versus post event slaughter actions. In particular, we will examine the balance between initiation and operation of a farm level periodic animal testing program versus slaughter.
A major decision in this setting involves the level of pre event investment in the animal testing program. We examine the reliance within an optimal cost minimizing plan on pre event periodic animal health testing, versus sole reliance on post event response measures. Empirical Model Setup
Modeling of this situation requires a modeling formulation that depicts the two stage decision making process in Figure 1. Namely, decision making has to be represented in multiple stages with decisions to install and operate the pre event animal inspection procedure at the first stage and second stage decisions conditional on both
8
LEVAN ELBAKIDZE Assistant Research Professor Department of Agricultural Economics
Texas A&M University College Station, TX, 77843-2124
(979)845 3225 [email protected]
and BRUCE A. MCCARL
Regents Professor Department of Agricultural Economics
Texas A&M University College Station, TX, 77843-2124
(979) 845 1706 [email protected]
This research was supported in part through the Department of Homeland Security National Center for Foreign Animal and Zoonotic Disease Defense at Texas A&M University. The conclusions are those of the author and not necessarily the sponsor.
ANIMAL DISEASE PRE EVENT PREPAREDNESS VS. POST EVENT RESPONSE: WHEN IS IT ECONOMIC TO PROTECT?
Abstract We examine the economic tradeoff between the costs of pre event preparedness and post event response to potential introduction of an infectious animal disease. In a simplified case study setting we examine the conditions for optimality of an enhanced pre event detection system considering various characteristics of a potential infectious cattle disease outbreak, costs of program implementation, severity of the disease outbreak, and relative effectiveness of post event response actions. We show that the decision to invest in pre event preparedness activities depends on such factors as probability of disease introduction, disease spread rate, relative costs, ancillary benefits and effectiveness of mitigation strategies.
Key words: Animal disease, mitigation strategies, economic balance, preparedness, response JEL Classifications: Q16, Q18, Q19
ANIMAL DISEASE PRE EVENT PREPAREDNESS VS. POST EVENT RESPONSE: WHEN IS IT ECONOMIC TO PROTECT?
Introduction
Possible intentional or unintentional introductions of contagious animal diseases could result in substantial economic losses as seen during the UK, US and Canada BSE (mad cow) events or the European Foot and Mouth Disease events (Thompson et al., 2003; Mangen and Barrell 2003; Henson and Mazzocchi 2002, Khan et al. 2001). Events with major consequences raise the specter of preventative/protective actions. Many calls have been issued for such actions in the post 9/11 world. However the cost of following all of the protection/prevention actions that have been called for is far in excess of any practically available budget.
Many issues can be raised about animal disease management and the need for protection. One such issue involves the balance between pre event investments in prevention/protection/response capability versus the post event costs of the event and associated disease management efforts. A key economic point in the context of this balance is the distinction between pre event and post event costs. Pre event actions impose costs regardless of event occurrence, while post event costs are only incurred when an incident occurs and thus are multiplied by the probability of the event when computing expected annual costs. For example, the costs of setting up and operating a continuing animal health surveillance system are encountered whether or not an outbreak ever takes place. However, the costs of diseased animal slaughter, reduced market supply, disinfection and event enhanced detection arise only in the event of disease
1
introduction. This paper reports on an investigation of the above mentioned balance problem
addressing how disease event characteristics and mitigation options affect the desirability of pre event investments versus post event response. In carrying out this investigation we first present and analyze a theoretical model of the balance problem. Subsequently we conduct an empirical case study motivated by data representing Foot and Mouth Disease (FMD).
A model of pre and post event decision making
Decisions in the context of an animal disease event can be categorized into 6 basic categories. These are
• Anticipation actions –things undertaken to improve the forecast of event likelihood and consequences such as intelligence gathering. These are largely pre event actions.
• Prevention actions – things undertaken to avoid event introduction or mitigate event implications upon introduction such as changes in sanitary or feeding practices along with the use of vaccinations. These are largely pre event actions.
• Detection actions – things undertaken to screen for precursors to an outbreak that speed detection and allow rapid treatment such as inspection for sick animals. These can be pre or post event actions. In a post event setting they are reflective of enhanced detection to help avoid disease spread and or avoid entry of contaminated products into the food chain.
2
• Installation actions– facilities installed to allow more rapid or effective disease detection and management. For example installation of sensors, construction of veterinary laboratories, training of first responders or stocking of vaccines. These are largely pre event actions.
• Response actions –disease management activities undertaken to halt the spread of the event such as slaughter of infected animals, carcass disposal, vaccination of animals in proximity to an event etc. These are post event actions.
• Recovery actions –things undertaken to reestablish productive capacity and market demand post event such as decontamination of production and processing facilities or advertising to increase consumer confidence. These are post event actions.
There are a number of important characteristics of decision making in this pre/post event type of situation. These include,
• Irreversibility – when a pre event action has not been undertaken, once an event has occurred it is generally not possible or at least very expensive to put it in place.
• Conditional response –certain response options can be used only if certain pre event actions are undertaken. One cannot use detection equipment that was not previously acquired and installed.
• Fixed cost versus probabilistic variable costs – in total cost accounting the pre event costs are always present; the post event costs are only encountered when
3
an event occurs. • Large span of possible events -- there is an enormous span of possible events
that can never practically be enumerated. Thus we will only deal with sample and abstract events below. Furthermore these events differ in nature and severity. • Probabilities – event probability is difficult to anticipate and in the case of deliberate actions is likely to be modified by pre event actions. This leads us to a restatement of the balance problem as the establishment of the optimal tradeoff between the cost of pre event actions and occasional post event damages including response and/recovery costs. In such a setting, the best strategy would depend on many factors including pre event action costs, disease management costs, potential damages, and event probability being a balance between these factors. Formal model development This problem can be addressed more formally. Consider the decision tree situation that depicts occurrence or non occurrence of a single event (Figure 1). Here we have a simple two stage decision process. The first stage is pre event and the second stage post event but allows for no event to have occurred (event occurrence probability is P, and no event (1-P)). In stage one decision makers have the option to invest in pre event actions, such as anticipation, prevention, installation and detection, as well as doing nothing. In stage two, there is a probabilistic possibility of an event - introduction of infectious cattle disease - or of no event. At the second stage decision makers can either initiate post event response actions with knowledge of an event taking place, or do
4
nothing. The post event response actions for animal disease management generally involve slaughter, vaccination, and quarantine strategies that are chosen so as to minimize disease induced economic losses. If there is no event then industry activities continue under normal conditions, although the costs of pre event actions implemented in the first stage will be incurred.
Under the context considered in this work, mitigation costs are composed of the pre event set of actions (s) with per unit cost ws, and the post event set of actions (r) with per unit cost wr. Lets assume that the event damages L(δ,s,r) are a function of pre event actions and post event response actions along with an incident severity parameter (δ). Denoting probability of event occurrence as P we can write average cost as:
(1) C = P ⋅ (L(s, r,δ ) + ws ⋅ s + wr ⋅ r) + (1 − P)ws s
Comparative statics analysis We adopt an expected cost minimization approach to investigate the relationship
between pre event preparedness and post event response mechanisms. Now suppose we study the optimal amount of pre and post event action and how it is influenced by
• the probability of the event • the costs of the pre and post event actions • the severity of the event. First order condition for the optimality of pre event and post event actions are as follows:
(2) PLs (δ,r,s) + ws = 0 (3) Lr (δ,r,s) + wr = 0
5
Comparative static analysis can be used to examine the balance between pre and post event actions under variations in disease severity (δ), pre and post event action costs (ws and ws), and probability of event occurrence (P). The total differential arising from equations (2) and (3) is given in (4). By applying Cramer’s rule we get (5 through 11), which permit examination for comparative static results.
( 4)
⎡PLss ⎢
PLsr
⎤⎡ds⎤ ⎥⎢ ⎥
=
⎡− ⎢
dws
−
Ls dP
−
PLsδ dδ⎤ ⎥
⎣ Lrs Lrr ⎦⎣dr⎦ ⎣ − dwr − Lrδ dδ ⎦
(5)
ds =
− Lrr
dw s P(L ss Lrr − Lsr 2 )
(6) dr =
Lrs
dws P(Lss Lrr − Lsr 2 )
(7) ds =
Lsr
(8)
dr = − Lss
dwr (Lss Lrr − Lsr 2 )
dwr (Lss Lrr − Lsr 2 )
(9) ds = − Lsδ Lrr + Lsr Lrδ dδ (Lss Lrr − Lsr 2 )
(10)
dr = − Lrδ Lss + Lrs Lsδ dδ (Lss Lrr − Lsr 2 )
(11) ds = − Ls Lrr dp P(Lss Lrr − Lsr 2 )
Assume the L function is convex in r, s and δ. In turn the above equations reveal information on the sensitivity of the optimal balance between pre and post event actions relative to the other model parameters. Namely,
• Equation 5 can be signed as negative indicating downward sloping demand for pre event actions, namely the higher the per-unit cost of pre event action, the less of that activity is used.
• Equation 8 similarly indicates downward sloping demand for post event actions. 6
• Equation 11 can be signed to be positive sign since L is decreasing in s and convex in r indicating that pre event actions increase with increasing probability of event occurrence.
• The signs of the terms within (6) and (7) are determined by the sign of Lrs and Lsr and when negative indicate complementarity between pre event preparedness and post event response, while positive signs imply they are substitutes.
• Equations (9) and (10) are not readily signed being dependent on the signs and relative magnitudes of Lsr, Lsδ, and Lrδ and thus remain ambiguous.
Empirical investigation using FMD motivated data
Our ability to sign some but not all of the terms above combined with the somewhat abstract nature of the pre and post event actions makes it desirable to do a case study. Thus, we empirically investigate the optimal combination of pre event preparedness and post event response strategies in an empirical setting using data drawn from the FMD literature in the context of possible introduction into Texas. Case study background
Although the US has been FMD free since 1929 (McCauley et al. 1979), disease introduction has been shown to have substantial implications elsewhere. For example, Great Britain experienced an FMD outbreak in 2001 where associated total losses are estimated to be £5.8-8.5 billion (Thompson et al. 2003 p. 25, Mangen and Barrell, 2003 p. 126). Given such large risks FMD is a priority area of concern within USDA and DHS.
7
Analysis of FMD related decision-making has been the topic of numerous studies (e.g. Bates et al. 2001, 2003a,b; Garner and Lack 1995, Schoenbaum and Disney, 2003; Berentsen et al.1992; McCauley et al. 1979; Ferguson et al. 2001; Keiling et al. 2001). These studies mainly concentrate on decision-making once an outbreak has occurred largely addressing post outbreak disease spread management with vaccination and slaughter.
Less attention has been devoted to pre event decision-making. Issues have been raised regarding surveillance systems (Bates et al. 2003b; Akhtar and White 2003, Ekboir 1999), but we cannot find empirical investigations that address the economic balance that might be drawn between pre event preparedness and post event response actions. We will address this issue focusing on the balance in a limited setting focusing on the installation/operation of surveillance and detection systems versus post event slaughter actions. In particular, we will examine the balance between initiation and operation of a farm level periodic animal testing program versus slaughter.
A major decision in this setting involves the level of pre event investment in the animal testing program. We examine the reliance within an optimal cost minimizing plan on pre event periodic animal health testing, versus sole reliance on post event response measures. Empirical Model Setup
Modeling of this situation requires a modeling formulation that depicts the two stage decision making process in Figure 1. Namely, decision making has to be represented in multiple stages with decisions to install and operate the pre event animal inspection procedure at the first stage and second stage decisions conditional on both
8