Ecology and Economy (in the US)

Irrationality of Continued Fire Suppression : An Avoided Cost Analysis of Fire Hazard Reduction Treatments Versus No Treatment

Gary Snider, P J. Daugherty and D Wood

Abstract

Without large-scale implementation of fire hazard reduction treatments, the costs of uncharacteristic crown fires in southwest forests will continue to increase. Federal policy continues to allocate vastly more funds to suppression than to prefire hazard reduction. We examined the economic rationality of continuing this policy of emphasizing fire suppression activities over restoration-based fire hazard reduction treatments. We compared treatment plus fire suppression costs to the cost of fire suppression without treatments over 40 years for southwestern forests. This avoided-cost analysis estimates the amount one could invest in treatments to avoid the future cost of fire suppression. Using conservative economic values, we found that avoided future costs justifies spending $238-601/ac for hazard reduction treatments in the southwest. We conclude that the policy of underfunding hazard reduction treatments does not represent rational economic behaviour, because funding hazard reduction would pay for itself by lowering future fire suppression costs.

---

The history and culture of forest fires in the United States has been chronicled and described by Pyne (1982), Pyne et al. (1996), and Arno and Allison-Bunnell (2002). In 1905, the Bu­reau of Forestry became the USDA Forest Service with the responsibility of protect­ing newly designated forest reserves. A critical part of its charge was the preven­tion and control of fires. In 1908, Con­gress set up a system, “like an open check­book,” that essentially authorizes deficit, unlimited expenditures for fire suppres­sion (Pyne et al. 1996, Arno and Allison­Bunnell 2002). Subsequent to the fires of 1910 the Forest Service embarked on a campaign to “battle fire to death” (Pyne 2004). Since that time the fire suppression checkbook has never run out of blank checks and remains open to this day.

Although it has become increasingly apparent that an ounce of presuppression ac­tivity is worth a pound of suppression funds (Pyne et al. 1996, Daugherty and Snider 2003) federal land-management agencies continue to allocate vastly more funds to suppression activities than to prefire hazard reduction. To a large degree, the histoty and institutional culture of management agen­cies and a long-conditioned public have per­petuated this policy of investing in the fur­ther disruption of fire cycles and the continued depreciation of forest values. In this article we examine the economic ratio­nality of continuing this policy.

Historical practices starting with European settlers (e.g., overgrazing in the late 1800s, selective harvesting of large trees, and fire exclusion) have created vast areas of un­healthy forest ecosystems in the southwest­ern United States (Covington and Moore 1994). These practices have caused signifi­cant structural and functional changes in southwestern ponderosa pine (PP) and dty mixed-conifer (DMC) forests that include unprecedented tree densities and fuel load­ings and anomalous fire regimes that are in­consistent with the region’s evolutionary history (Covington and Moore 1994, Dahms and Geils 1997). The overly dense conditions, exacerbated by drought, have in­creased bark beetle mortality and the size and frequency of stand-replacing crown fires. These interconnected symptoms warn society of the jeopardy of losing these forest ecosystems. To avoid this loss, we must im­plement large-scale restoration-based thin­ning treatments designed to significantly re­duce tree densities, improve the cover, composition and production of the herba­ceous understory, and reintroduce fire to its normal disturbance regime (Fule et al. 2002), sooner rather than later. Rieman et al. (2003) argued that we need additional study as to the effects of res­toration-based thinning treatments and the use of fire in management on organisms, habitats, watersheds, and ecosystem pro­cesses. Continued study and research is nec­essary if we are to truly implement adaptive management (AM); however, the tired “more research is needed” mantra should not be used as an excuse to forego large-scale restoration-based fuel treatments. Hardy (2005) notes that there is no possible single metric capable of integrating all that must be considered with respect to fire hazard, and Finney (2005) states that the calculations re­quired to determine burn probabilities and fire behavior are computationally and prac­tically impossible because of the nearly infi­nite sequences of weather and ignition tim­ing and location.

The significant gaps in our understand­ing of probabilities and effects are not lim­ited to fire ecology and management. An analogous situation exists in fire economics. Fairbrother and Turnley (2005) have argued that without an understanding of the type, likelihood, and magnitude of changes that will result from fire (uncharacteristic or con­trolled burns) or conversely from the lack of fire, one can not quantify the risks, costs, and benefits associated with fire manage­ment activities. Kline (2004) describes the typical steps in determining the expected value of present net benefits and the infor­mation that would be required to make these calculations. First, one must know the prob­ability of an uncharacteristic wildfire occur­ring at any given point in space and time with and without treatment. This informa­tion is not going to be available any time soon. He further states that any attempt to evaluate the net benefits of fuel treatments via cost-benefit analysis would require such information on everything from potential changes in fire severity and suppression costs to effects on forest conditions and ecological functions. This information must be suffi­cient to meet prevailing demands for scien­tific quality in forest management and poli­cymaking. He concludes that currently there is insufficient information and analysis to determine whether or not investment in haz­ardous fuel treatments represents a rational economic choice. Conversely, the same lack of information and analysis could have been used to argue that we can not determine that continuing the current policy represents a rational choice, implying that a change in policy requires rationality, while continua­tion does not.

Although empirical precision and theo­retical rigor improve the accuracy of cost­benefit analysis, obviously unattainable pre­cision is unnecessary for policymaking. Given the current conditions of PP and DMC forests of the arid southwest and the inevitability of future uncharacteristic large­scale crown fires, simple and relevant is pref­erable to elegant and inexpedient. Although we understand that the issue is not a simple dichotomy between fire suppression and treatment, we argue that waiting for a com­plete understanding of the social and ecolog­ical complexities before taking significant action is folly.

A number of recent studies have asked the question, How much are fire hazard re­duction treatments going to cost and can so­ciety afford to pay for restoration-based treatments? These studies examined the fi­nancial feasibility of implementing these types of treatments (Lynch et al. 2000, Fied­ler et al. 2002, Larson and Mirth 2004). In general, these studies found it difficult to justify treatment expenditures based on the value of the wood fiber removed by the treat­ments. This conclusion suggests that resto­ration-based fuel treatments must be consid­ered an investment and would need to be justified by the marginal benefits/value pro­vided by a healthy ecosystem. However, it is possible we have been asking the wrong question. Perhaps the question we should be asking is, Can we afford not to implement res­toration-based hazardous fuel reduction/thin­ning treatment? In other words: will continu­ing the current policy of emphasizing fire suppression cost society more than the alter­native of investing in hazard reduction.

The ever-increasing cost of not doing restoration-based hazard reduction treat­ments suggests an approach for determining the level of investment that avoids the afore­mentioned described difficulties with cost­benefit analysis. One can set a conservative lower bound on the amount to be invested by setting hazard reduction funding at least equal to the direct costs, such as fire suppres­sion, that would be avoided by invest­ing-in other words, each dollar invested in restoration-based hazard reduction would save at least a dollar in suppression. This study presents results of an economic analy­sis comparing the cost of implementing res­toration-based hazard reduction treatments to no treatment for areas identified at high risk for crown fire. The with- and without­treatment comparison focuses on PP and D MC forest ecosystems in Arizona and New Mexico (Forest Service Region 3). The anal­ysis provides a conservative estimate of the potential economic losses due to a continu­ation of policies that emphasize suppression over proactive restoration-based hazard re­duction treatments.

Methods

We developed a relatively simple eco­nomic analysis that compares the costs of restoration-based hazard reduction treat­ments to the costs of fire suppression and rehabilitation expected without treatment. This avoided-cost approach answers the question of how much can we invest in pre­vention to avoid the continued cost of fire suppression and rehabilitation. The ap­proach assumes that the cost avoided (fire suppression) by investment is otherwise un­avoidable, which is clearly the case with wildfire in the wildland-urban interface.

We used a 40-year time horizon for an­alyzing costs and a 4% discount rate for con­verting costs to PV. Annual suppression and rehabilitation costs were discounted to the present and summed to yield the present value (PV) of costs without treatments. Haz­ard reduction treatments were assumed to have an initial cost (e.g., restoration thin­ning and burning) plus maintenance costs (e.g., prescribed fire) that occurred every 10 years after the initial treatment. The with­treatment PV was calculated as the sum of the discounted costs of initial and mainte­nance treatments and the discounted costs of reduced fire suppression and rehabilita­tion. The net PV (NPV) of treatment was determined using a “with” minus “without” procedure: the PV without restoration­based treatments was subtracted from the PV with treatment to calculate the net value of the treatment. This difference (NPV) be­tween projected with- and without-treat­ment costs represents a partial measure of the economic value of these treatments, i.e., the net avoided cost due to treatment. We determined break-even values by raising or lowering the initial treatment cost per acre until the NPV equaled zero. The break-even value indicates the dollar per acre that could be spent on treatment such that the PV of the treatment costs is just covered by the PV of avoided fire suppression and rehabilitation costs.

Study Area. The PP and DMC forests of Arizona and New Mexico provide a good application area for the analysis. These for­ests have undergone important changes, with two of the most important being sim­plification in structure and unprecedented increased tree densities (from about 50 trees/ac to upward of a 1,000 trees/ac). These conditions have resulted in increased fuel and the existence of a continuous fuel ladder. In 2002, there were 9,887,181 ac of PP and DMC forests in Arizona and New Mexico (Table 1). Over 90% of these forests are at high (condition 3) or moderate (con­dition class 2) risk of losing key ecosystem components (Schmidt et al. 2002). It is not possible to know exactly when, where, or the size of a future fire. We do know that given the current condition of the PP and DMC forests of the arid southwest, those areas at high risk are going to burn.

Table l. Acres and percentage of Ponderosa pine (PP) and dry mixed-conifer (DMC) by fire regime condition class for Arizona and New Mexico combined.

	Class I	Class 2	Class 3	Total
PP	475,184	5,805,000	3,190,625	9,470,808
DMC	246,858	6,425	163,090	416,373
Total	722,042	5,811,424	3,353,714	9,887,181
Class percent	7.3	58.8	33.9	100.0

Source: Based on data from Schmidt et al. (2002)

Data and Mode/Assumptions.

We set the initial expected acres burned per year without treatment at 443,307 ac. This figure was based on the average annual fire acreage for fires greater than 100 ac on state and federal lands in Arizona and New Mexico during the period 1993-2002 (National In­teragency Fire Center [NIFC] 2003). The 443,307 ac represent approximately 4.5% of the total acreage of PP and DMC forests in Arizona and New Mexico. We assumed that future fire occurrence will be similar to that experienced during that decade.

We computed the costs for suppres­sion and rehabilitation as the average per acre expenditure for large fires (greater than 100 ac) on public lands for which the Forest Service provided suppression sup­port during the period 1993-2002 (Ge­bert 2003). All costs were converted into 2002 dollars. We only included variable costs directly associated with large fires. We did not include fixed costs of pre­paredness. During this period Forest Ser­vice Region 3 spent an average of $377/ac per year on suppression for large fires. The $3 77 I ac fire suppression cost is slightly less than the $387 /ac figure reported by Donovan and Noordijk (2005). They spent an additional $22/ac per year on emergency rehabilitation, which only cov­ers emergency measures to control ero­sion/ sedimentation and only occurs on a limited number of acres. Rehabilitation does not include reforestation activities. This approach assumes that Forest Service costs are adequate estimators of costs on all public lands.

We assumed that restoration-based fire hazard reduction treatments can even­tually eliminate large-scale, uncharacteris­tic crown fires and associated variable costs of large-fire suppression and rehabilitation incurred in the recent past. We assumed that a given amount of acres required treatment and that the number of large­fire acres would diminish in proportion to the amount of required acreage shifted from high or moderate risk to low risk.We assumed that the restoration-based hazard reduction treatment would shift forest­land to low risk. We also assumed that large fires also shifted forestland to low risk. The reduction in large-fire acreage was assumed to occur for both causes of shifts to low risk. In other words, both the with- and without-treatment scenarios re­sults in reduction of large fires, with the difference being that treatment can accel­erate the reduction in large fires. As large­scale, uncharacteristic crown fires are di­minished, the associated variable costs of fire suppression and rehabilitation will di­minish also.

We kept the model simple by only in­cluding direct costs in the comparison of treatment versus no treatment. We did not account for losses and damages to structures, private land value, and other infrastructure associated with the wildland-urban inter­face. We also did not include nonmarket val­ues associated with ecological goods and ser­vices. These nonmarket values include recreation, wildlife habitat, watershed con­dition, and water quality. These values gen­erally are considered to be improved by res­toration-based fuel treatments (Mason et al. 2006). Finally, we did not include lost tim­ber harvest values. These values could be substantial, or not, depending on the resolu­tion of such issues as the imposition of di­ameter caps and the successful development of a private sector smallwood utilization in­dustry. Given that the only benefits ac­counted for in this analysis are the fire sup­pression and rehabilitation costs avoided, the total economic value of the restoration­based fuel treatments are significantly un­derestimated.

Analytical Scenarios.

We developed analytical scenarios by varying the required number of acres to be treated and the rate at which they are treated (Table 2). We ini­tially set the required number of acres to be treated at either 33 or 50% of the total acre­age of PP and DMC forests in Arizona and New Mexico (3,262,770 or 4,943,590 ac). We used three treatment rates: treating 5, 10, and 15% of the required acres per year. We set the initial treatment cost to $200/ac, and used $50/ac for the maintenance treat­ments that occur every 10 years after initial treatment. We then varied the initial treat­ment costs to determine the break-even val­ues. For these scenarios, we applied the as­sumption that reduction in large-fire occurrence would be proportional to the amount of required acres treated.

To test the sensitivity of results to the amount of acres requiring treatment, we conducted a series of sensitivity analyses (SEN) that varied the required amount from 30 to 65% of the forest (2,966,154 to 6,426,667 ac). For these scenarios we used annual treatment rates of 5 and 15% of re­quired acres.

To test the sensitivity of the analysis to the proportional reduction assumption, we developed an additional scenario that incor­porates a time lag before reductions in fire suppression and rehabilitation costs occur. This lag scenario (LAG) addresses the con­cept that a certain amount of strategically located treatment must occur before gaining any significant reduction in large-fire occur­rence and associated fire suppression costs. The LAG scenario assumed no reductions in fire suppression and rehabilitation costs un­til treatment had occurred on 35% of the required acreage. We conducted the LAG scenario using 33% required treatment acre­age and a 5% annual treatment rate.

Results.

With an initial treatment cost of $200/ ac, all treatment scenarios had positive NPVs and break-even values (Table 3). The no-treatment scenario’s (A) costs for large­fire suppression and rehabilitation resulted in a PV of negative $2.0 billion. Treatment scenario costs resulted in PVs from -$1.1 billion to -$1.9 billion. The NPV of treat­ment scenarios ranged from $128 to 894 million. Break-even values for treatment sce­narios ranged from $238 to 601/ac.

Table 2. Scenario variables-area requiring treatment and rate of treatment.

Scenario variables
Scenario	Area requiring treatment		Rate of treatment
(code)	(% total ac)	(ac)	(%/yr)	(ac/yr)
A	N/A	N/A	N/A	N/A
B	33	3,262,770	5	163,138
C	33	3,262,770	10	326,277
D	33	3,262,770	15	489,416
E	50	4,943,590	5	247,180
F	50	4,943,590	10	494,359
G	50	4,943,590	15	741,539
SEN	30-65	2,966, 154-6,426,667	5, 15	148,305-964,000
LAG	33	3,262,770	5	163,138

Table 3. Results of analytical scenarios-NPVs, with minus without values and break-even values.

	Treatment variables			Results
Scenario	Required (%ac)	Rate (%/yr)	Rate (ac/yr)	PV costs ($billion)	NPV* ($million)	Break-even ($/ac)
A	N/A	N/A	N/A	-2.0	N/A	N/A
B	33	5	163,138	-1.4	604	530
C	33	10	326,277	-1.2	797	535
D	33	15	489,416	-1.1	894	538
E	50	5	247,180	-1.7	351	326
F	50	10	494,359	-1.6	464	329
G	50	15	741,539	-1.5	524	331
SEN	30-65	5, 15	Variable	-1.4 to -l.9	649-128	601-238
LAG	33	5	163,138	-1.7	344	388

*NPV calculated as difference between PV of with-treatment scenarios (B through LAG) minus PV of without-treatment scenarios (A).

With-treatment NPV increased with increases in the annual rate of treatment and decreased with increases in the required amount of treatment. The increase in NPV with faster treatment rates occurs because the faster treatment rate results in a faster decrease in acres of large fires and associated costs. The earlier decreases in these costs off­set the higher up-front treatment cost asso­ciated with faster rates. Although the NPVs increase with treatment rate, the break-even values show virtually no change. The faster treatment rate restores more acres and the increase in treated acres keeps the per-acre break-even values constant even though NPV increases. With the required treatment level at 33%, the 10% treatment rate (sce­nario C) treats 16% more acres than the 5% rate (scenario B). The 5% increase in treat­ment rate results in a larger percent increase in treatment acres, because the faster treat­ment rate also reduces the number of acres burned, making these acres available for treatment. The faster treatment rate also sig­nificantly decreases the annual amount of acres burned. Figure 1 illustrates this effect using scenarios B and C. The greater the number of acres treated per year, the fewer the number of acres burned by wildfire.

Figure 1. The effect of annual treatment rates on the amount of acres burned over time.

Figure 2. Effect of amount of acres requiring treatment on break-even values.

Increasing the number of acres requir­ing treatment significantly lowers the NPVs and break-even values. Increasing the re­quired acres from 33 to 50% decreased the NPVs by 40% and break-even values by 38%. Figure 2 illustrates the result of the sensitivity analysis (scenarios SEN) on the effect of increasing acres requiring treatment on break-even values. Break-even values range from $601/ac if 30% of acres require treatment to $238/ac if 65% of acres need treatment. The rate of decrease in break­even values slows as the acres requiring treat­ment increases. Even if 100% of the forest needs treatment (not shown), the break­even value remains positive at $93/ac.

The LAG scenario tested the effect of a delay in large-fire reduction and suppression costs, using a 33% required treatment and a 5% treatment rate. The LAG assumption of no reduction in fire suppression or rehabili­tation costs until 35% of required treatment has been accomplished significantly lowers the NPV and break-even value. When com­pared with its no LAG equivalent (scenario B), the LAG scenario’s NPV drops by 43% to $344 million, and the break-even value drops by 27% to $388/ac.

Discussion

The results of all scenarios suggest that continuing the current policy of favoring fire suppression over prevention does not repre­sent a rational economic choice. All with-­treatment scenarios had lower PV of costs than the no-treatment alternative. The amount that could be invested in restora­tion-based hazard reduction treatments ranged from $238 to 601/ac.

We believe these scenario results are very conservative for several reasons. Our es­timates of future fire occurrence use the av­erage acres burned from 1992 to 2002. Given the worsening forest conditions, the assumption of no increase in large fires over the average of 10-year period is conservative. Our estimates of per-acre fire suppression cost did not include all charges for national contracts, such as many aviation or catering services. These costs averaged $96/ac for the period of 1993-2002 (Gebert 2003). The analysis only included variable fire suppres­sion costs. Presuppression and other fixed costs are likely to decline as the occurrence of large fires declines. We assumed that treat­ment costs are net of any value recovered from the fuel treatment. If the treatment generated $200/ac in wood products, the to­tal break-even investment value increases by this amount. We did not include losses and damages associated with structures, private land value, and other infrastructure associ­ated with the wildland-urban interface in the avoided costs. We did not include changes in ecological and social values asso­ciated with restoration-based treatments. Al­though not market valued, the changes gen­erally are considered to be positive and significant in magnitude. We essentially as­sumed that there is no difference between the value of a burned and restored forest.

The analysis supports the idea that the current pace and scale of implementation of restoration-based treatments that would be­gin healing degraded forests and reduce the threat of unnatural wildfire remains woe­fully inadequate. Inadequate federal funding for restoration-based fuel treatments and the reprogramming of treatment dollars to sup­pression activities perpetuates the problem. Although society demands suppression of fires, it is extremely expensive and fails to solve the underlying problem leading to fires. By underfunding restoration-based hazard reduction treatments, we ensure the continued need for suppression by under­mining the rational approach to solving the wildfire problem.

The results indicate that the amount of acres requiring treatment significantly de­creases NPVs and break-even values. This finding points out the importance of effi­cient spatial strategies and fuel treatment de­sign that produce genuine reductions in the probability of uncharacteristic crown fire. Random or arbitra1y fuel treatment patterns will likely be ineffective and increase the to­tal number of acres requiring treatment thus reducing the economically rational expendi­ture level.

There are several forces that prevent im­plementation of large-scale restoration­based fuel treatments in the forests of the arid southwest. Given current forest condi­tions we can not simply cease fire suppres­sion efforts, especially in the wildland-urban interface. Thus, in the short run, costs will increase as we will be paying for both sup­pression and restoration. Elections are al­ways just around the corner and the federal and private wildland fire industrial complex is heavily vested, politically and financially, in perpetuating the annual war on wildfire. Revenues are increasingly scarce and restora­tion of the health and resilience of the for­ested ecosystems of the arid southwest does not appear to be very high on the priority list. This situation is analogous to other funding paradoxes, e.g., funding emergency over preventative health care, funding flood disaster relief over levee maintenance. Clearly, it will be challenging to generate the long-term financial and managerial commit­ment for an activity for which ecological and economic benefits may not be evident for a number of years into the future-beyond the next election, budget crisis, or tenure of a forest supervisor.

Another significant force limiting im­plementation has been the near demise of the region’s wood harvesting and utilization infrastructure over the past 10-15 years. This loss stands as a critical impediment to the implementation of restoration-based fuel reduction treatments. Although there are initial signs of emerging small-scale op­erations, the development of a competitive market for the wood fiber removed by resto­ration-based treatment remains elusive. Commercial use of the wood fiber removed, especially small-diameter logs, could reduce treatment costs that will otherwise be borne by taxpayers; but the current limited scale of fire hazard reduction work in the Southwest presents a barrier to market development. Manufacturing firms want a reasonable ex­pectation of a raw material supply through­out their planning horizon. Currently, the reasonable expectation is that the supply of wood fiber from southwestern restoration projects will remain intermittent and vari­able, because of litigation or the threat of litigation by groups skeptical of the influ­ence that a profit-driven economic system can have on ecological systems. This con­cern that for-profit enterprises will corrupt or distort restoration-based treatment activ­ities is perhaps the most vexing socioeco­nomic issue in the restoration of southwest­ern ponderosa pine forests. Although genuine, this concern prevents innovation by local entrepreneurs, who could create new uses for the raw materials (Daugherty and Snider 2003).

The latest management paradigm pro­moted by federal land-management agencies was ecosystem management, which has since evolved into AM, or learning by doing. We suspect that all the ecological and economic arguments for the implementation of land­scape-scale restoration-based fuel treatments will avail us little for many of the same rea­sons given for our failure to implement AM. These include strong opposition to experi­mental policies and management strategies by persons protecting various self-interests; management agencies trapped by cumber­some, inflexible, formalized processes and narrow interpretations of legal mandates; and demands by various organizations and interest groups for spurious certitude (Walters 1997, Gunderson 1999). The fact that we will continue to learn is a given. The fact that we will do so intentionally and in time to avoid the odious and in some cases irreversible (Savage and Mast 2005) conse­quences of anomalous stand-replacing crown fires is not.

Conclusions

There are no risk-free management ac­tions. Indeed, under present forest condi­tions, the no-action or go-slow alternative may very well be the most risky of all. As pointed out by Dambeck et al. (2004), to avoid risk in wildfire management today is to advocate even larger uncharacteristic wildfires and destruction of property and ecosystems. Our results indicate that the ev­er-increasing ecological and economic costs resulting from high-severity, ecosystem­scale fires in the ponderosa pine and dty mixed-conifer forests of the southwest far exceed the cost to society of proactive resto­ration-based thinning treatments. The cur­rent sociopolitical condition of continuing to spend dollars on fire suppression while implementing limited treatment of high­risk forest areas represents an irrational eco­logical and economic decision.

We no longer face the question of whether society will spend the money or not. We are going to pay, one way or another, unless we make the unlikely choice to no longer spend money trying to fight and con­tain unnatural crown fires. We now face the choice of how we are going to spend the money and what are we likely to obtain from that expenditure. If we invest in restoration­ based hazardous fuel treatments, we invest in the future; we invest in healthy, sustain­able ecosystems for our children and grand­children. By not investing in restoration­based fuel treatments, we continue the depreciation of our forests, increasing the risk of radical shifts in their structure and function because of uncharacteristic crown fire. Given these choices, it makes a great deal of economic sense to conduct forest res­toration on a large scale today to retain fu­ture ecological and economic values. Our analysis shows that the fire suppression costs that can be avoided in the future are suffi­ciently large by themselves to justify restora­tion-based fuel treatment expenditures to­day.