Source: UNIVERSITY OF NEVADA submitted to
OPTIMAL DECISION-MAKING IN ECOSYSTEMS CHARACTERIZED BY WILDFIRE AND UNCERTAINTY
 
PROJECT DIRECTOR: Rollins, K. Nalle, D.
 
PERFORMING ORGANIZATION
RESOURCE ECONOMICS
UNIVERSITY OF NEVADA
RENO,NV 89557
 
NON TECHNICAL SUMMARY: Existing decision-support models for wildfire risk management omit the economic implications of irreversibilities, uncertainty and the sequential nature of these decisions. This research estimates the cost to society of omitting these considerations. This project will develop a decision-making model that incorporates the implications of uncertainty and irreversiblities in wildfire management and compare the outcomes, in terms of decisions and value of the forest resource, between the existing and proposed models.
 
OBJECTIVES: The objective of this project is to develop a decision-model that will generate sequential sets of cost-minimizing (or benefit maximizing) decision criteria over time, given uncertainty and irreversibilities associated with catastrophic wildfire events, and the ability of decision-makers to observe and incorporate new information about wildfire risk over time. Rangeland and forest management decisions regarding activities that are aimed at managing the risk and costs associated with wildfires are characterized by uncertainty and potentially irreversible outcomes. Uncertainty arises from imperfect knowledge about environmental conditions, and the inability to predict perfectly the exact outcome of a given management action. A decision is reversible if the outcome of the action does not constrain or limit subsequent decision options. An irreversible decision is one for which subsequent sets of options are constrained by the action. In terms of wildfire risk management, we consider the costs of fuelwood removal, and other management activities, as well as the costs of a catastrophic wildfire. The occurence of a catastrophic wildfire is taken as an irreversible outcome of previous decisions. We consider the on-going accumulation of information about random environmental events as reducing uncertainty, thereby improving the ability of decision-makers to predict the outcome of risk management activities. The decision-maker therefore faces a trade-off: How long does one wait for the accumulation of better information that increases predictive ability, given the delay in implementing expensive management actions may increase the probability of an irreversible outcome? And secondly, is it reasonable to attempt to prevent irreversible outcomes at all costs? The goal of the model is to optimally incorporate the value of on-going information collected from monitoring and assessment programs. The model recognizes that sequential observations of new information can reduce uncertainty. But uncertainty cannot be avoided - decisions must still be made without full information. So the question is the timing of the decision, and the amount of new information that is gained with delays in decision-making. This ultimately boils down to determining the value of waiting and the value of the information gained. When the marginal expected value of new information is equal to the expected marginal value of the increased cost of delay - a decision must be made based on available information at that time. The expected cost of any new information will be greater than its value for managemnt uses, due to the delay in decision-making. Thus, the model results in optimal timing for decisions and valuation of on-going information that reduces uncertainty. In the case of range or forest fire management strategy, the new information is the resulting data from on-going monitoring and assessment programs. Therefore, a by-product of this research is a method to value the information generated by monitoring programs.
 
APPROACH: The value of accumulated information gained by continuous monitoring of random environmental variables can be modeled in two ways. The standard approach is to incorporate the information as a distribution that predicts the probability a random variable will take on a given value, and then solving the decision-problem in a dynamic-stochastic framework (optimal control or dynamic programming) using expected values. This approach results in optimal decision criteria for all time periods. This approach treats uncertainty the same way throughout the time horizon for the problem within a single ex ante solution set. The problem with this approach is that if decision-making is in reality sequential so that decision criteria being updated to reflect new sets of future probabilities as new information is added, the full value of the new information in reducing uncertainty is undervalued in circumstances where irreversibilities occur. Decision-makers at the time of the irreversible outcome are essentially looking at a given set of future probabilistic outcomes, and omitting a set of options that are no longer viable. The single-solution set model would continue to consider the expected values of all possible outcomes, including those that have since been ruled out by the irreversible outcome. The standard approach would undervalue the new information, and the marginal conditions for an optimum would not reflect the reality of the costs and benefits of a decision at that point and at all future points. An alternative approach that has been shown to be appropriate for sequential decisions where irreversible outcomes might occur is the real options approach. This method is also dynamic, but produces decision criteria for each relevant time period, taking the information about the outcomes of random variables for all previous time periods as given. Random events are modeled as stochastic processes that produce a given outcome at each period. Thus, the marginal value of observing additional random outcomes is reflected in reduced uncertainty in predicting future outcomes of a current decision. The difference between the two approaches in how they value the added information from sequential observations could lead to very different decision rules and overall valuation of the resource or asset under consideration. In the financial literature, this second approach is refered to as a real option approach. This research will start with a real options framework, and will strive to extend it by using numerical approximation methods that would allow considerably more complexity to the problem than is typically encountered in existing applications of real options models. The research compare the results of both approaches, calibrating the models with data from the Tahoe National Forest. The results should indicate the cost of using the first approach when in reality the actual decision-making problem is specified as consistent with the second, real options approach.
 
CRIS NUMBER: 0196492 SUBFILE: CRIS
PROJECT NUMBER: NEV05161 SPONSOR AGENCY: CSREES
PROJECT TYPE: HATCH PROJECT STATUS: TERMINATED MULTI-STATE PROJECT NUMBER: (N/A)
START DATE: Jul 1, 2003 TERMINATION DATE: Jun 30, 2006

GRANT PROGRAM: (N/A)
GRANT PROGRAM AREA: (N/A)

CLASSIFICATION
Knowledge Area (KA)Subject (S)Science (F)Objective (G)Percent
605061230106.320%
605061231006.315%
605079031006.310%
609619930103.115%
609619931003.115%
610061230102.115%
610079030102.110%

CLASSIFICATION HEADINGS
KA609 - Economic Theory and Methods
KA610 - Domestic Policy Analysis
KA605 - Natural Resource and Environmental Economics
S6199 - Economy, general/other
S0612 - Conifer forests of the West
S0790 - Rangelands, other
F3010 - Economics
F3100 - Management
G3.1 - Economic Opportunities for Growth
G6.3 - Protect and Manage Forests and Rangelands
G2.1 - Expand Domestic Market Opportunities


RESEARCH EFFORT CATEGORIES
BASIC 70%
APPLIED 20%
DEVELOPMENTAL 10%

KEYWORDS: economics; decision making; uncertainty; forestry; irreversibility; risk management; wildfire; fire management; optimization; ecosystems; range management; rangelands; public policies; policy analysis; costs; forest management; forest fuels; value determination; econometric models; biophysics; land management

PROGRESS: Jul 1, 2003 TO Jun 30, 2006
Outcomes include: (1) Development of an economic approach to take into account the possiblity of irreversible changes caused by fire and disturbances on public lands in Nevada. Standard dynamic models that don't take irreversible changes into account would underestimate the economic value of fuel treatments designed to forestall and/or prevent such changes in a dynamic system. Because fuel management treatments can be costly to perform, it is important to account for the full value of these treatments when deciding a course of action. One issue concerning the role of fire in the Great Basin is the possibility of irreversible shifts of the ecology to permanently accelerated fire cycles in which native vegetation is not able to reproduce naturally. Therefore the use of an economic approach that can take account of these shifts is relevant for future land management decision-making in the region. (2) This economic modeling approach has been incorporated into a 5 year (2006 - 2011) Joint Fire Science Program (JFSP) project, in which ecologists are performing research to quantify more precisely how much disturbance of the native vegetation is sufficient to cause the irreversible shift to be inevitable - a threshold. The project also compares efficacy of alternative fuels treatments. These project outcomes will be used to calibrate an economic model. The expected impact includes the contributions to this new project. (3) Our work included an evaluation of public sector land managers attitudes regarding use of treatments and beliefs about their efficacy. This is being incorporated into University of NEvada Cooperative Extension (UNCE) Natural Resources Education programming. (4) Our work included an evaluation of the general public's attitudes, values and beliefs about the risk of fire and permanent ecological losses. Using economic valuation methods, we quantified the value of non-market environmental losses to the public from irreversible changes in the fire cycle and impacts on vegetation on Nevada Public lands. We measured how these values vary by county, rural and urban residents, age, years of Nevada residence, and other variables that allow for calibration with demographic changes expected in the stat. These results are also being incorporated into UNCE programming. (5) Our work included evaluation of attitudes toward risk and willingness to participate in reducing risk of people who live in the Wildland Urban Interface (WUI) in the most fire-prone areas in Nevada. This is currently being incorpoated into UNCE programming in the Living with Fire program. (6) This project trained 2 undergraduate researchers, 2 graduate student researchers, and involved collaboration between faculty in extension, and among disciplines (including sociology). This training will impact future projects and collaborations. The undergraduate student involvement included 2 students at about 10 hours per week for a total of about 2 years, and 3 person years of half time (20 hours per week) graduate student time. All students will have coauthored publications and presented at scholarly conferences as a direct result.

IMPACT: 2003-07-01 TO 2006-06-30 Impacts: (1) Development of an economic approach to take into account the possiblity of irreversible changes caused by fire and disturbances on public lands in Nevada to correctly value fuel treatment programs. These programs would be undervalued if irreversible consequences of change are ignored. (2) Incorporation of economic models and results into a 5 year Federal Great Basin regional Joint Fire Science Program project. (3) Incorporation of results into University of Nevada Cooperative Extension (UNCE) Natural Resources Education programming. Specific results used by UNCE include fire risk/natural ecosystem vegetation values held by land managers and general public, and evaluation of attitudes toward risk and willingness to participate in reducing risk of people who live in the Wildland Urban Interface (WUI) in fire-prone areas of Nevada. (4) Training of 2 undergraduate researchers and 2 graduate student researchers. Undergraduate student involvement amounted to a total of about 20 hours per week for a total of about 2 years. Graduate student involvement amounted to 3 years at 20 hours per week. All students will have coauthored publications and presented at scholarly conferences as a direct result. (5) Collaboration between faculty in extension and among disciplines at UNR and other institutions (federal and state) will impact future projects and collaborations.

PUBLICATION INFORMATION: 2003-07-01 TO 2006-06-30
Rollins, K and Mariah Evans, Statistical Portrait of Public Opinion on Vegetation Management in Nevada, in the Symposium Public and Rangeland Managers' Opinions About Vegetation Management and Fire, Rollins, K., organizer, Abstracts in the Proceedings for the Society for Rangeland Management Meetings, Reno, NV, February 15, 2007.

PROJECT CONTACT INFORMATION
NAME: Rollins, K. S.
PHONE: 775-784-1677
FAX: 775-784-1342