Prioritizing Invasive Species Threats Under Uncertainty

Agricultural and Resource Economics Review, Apr 2006 by Moffitt, L Joe, Osteen, Craig D

Due to the time and costs of collecting information and analyzing pest threats and response options, agencies often respond to the threats as information becomes available to prevent major crop damage or loss of export markets. So, program managers might make a series of decisions. The sequential decision making process has characteristics of dual control or active learning where decisions are made to gain information (Kendrick 2005). Program managers may respond to new, passively or actively obtained information about exotic pest threats with active learning in order to reduce uncertainty about pest threats or to develop control or other preparedness methods, and later use that information to select appropriate methods for each pest, rather than simultaneously determining optimal strategies for all pests. For example, USDA agencies, such as the Agricultural Research Service and the Cooperative State Research, Education, and Extension Service, have programs to study new pest threats, develop control or preparedness methods, or disseminate information, which can be used by other decision makers. If control or preparedness methods are unknown or poorly understood, decision makers have no choice but to follow such a decision making process. Different decision rules can prioritize learning, preventative, or other response actions, depending on available information.

In some circumstances, decision makers might respond with provisional prevention or preparation measures, as well as active learning, when new, sparse information indicates that an exotic pest threat is highly damaging, irreversible, and imminent. The active learning could better quantify the threat or develop methods to detect or manage the pest, but could also indicate whether the measures should be continued, modified, or terminated to prevent misallocation of resources and complaints from other nations that the measures do not meet the standards of the Agreement on the Application of Sanitary and Phytosanitary Measures (World Trade Organization 1994).

Approaches to Managing Invasive Species Under Uncertainty

Eiswerth and van Kooten (2002) developed a firm-level, stochastic dynamic programming model for control of invasive weeds that recognizes several sources of uncertainty by using fuzzy sets, including lack of data, measurement error, variability in rate of spread, and impact of management measures, as well as the need to use categorical descriptions of uncertain events.

Horan et al. (2002) question the appropriateness of probabilities for representing uncertainty in the case of invasive species, suggesting that expected utility calculations, used in the risk management or Bayesian frameworks, may have limited value for analysis of some invasive species problems, and they identify these reasons: (i) the frequency of invasive species introductions may depend on trade levels and may be low, (ii) probabilities are difficult to assign to one-time events with no historical precedent, such as some invasive species introductions, and (iii) invasive species events may be very costly and irreversible. They developed a model of decision making under ignorance-where neither subjective probabilities nor detailed knowledge of the event space is available-based on Katzner (1998), Shackle (1969), and Vickers (1987). Their model features the notion of potential surprise, a measure of disbelief of potential future outcomes, and recognizes that inconceivable outcomes may occur. Decision makers would not focus on all outcomes and expected utilities, but rather on "the least unbelievable conjectured losses or gains from the activity," or "focus loss," without necessarily accounting for a probability distribution. However, catastrophic events with high potential surprise would not factor into the focus loss. Under this behavioral model, decision makers would take a cautious approach when considering uncertain outcomes so that low probability outcomes that are considered possible-those with low potential surprise-receive more prevention resources than under a risk management approach.