Nfp



Nfp.2019.doc

9/232/05

INVESTIGATING NATIONAL FIRE PLAN IMPLEMENTATION

IN NORTHERN NEW MEXICO*

Curt Shepherd

Department of Economics, University of New Mexico, Albuquerque, NM (USA)

John Talberth

Ecology and Law Institute, Santa Fe, NM (USA)

Joseph M. Little

Department of Economics, University of Alaska—Fairbanks, Fairbanks, AK (USA)

Robert P. Berrens

Department of Economics, University of New Mexico, Albuquerque, NM (USA)

____________________

* Research support was provided by the USDA Forest Service, Rocky Mountain Research Station, Flagstaff, AZ (Carl Edminster) under Research Joint Venture Agreement (02-JV-11221615-039). The authors would like to thank the Forest Service personnel who spent considerable effort filling our data request; Susan Lee (Fire Coordinator, Southwestern Region), Paul Fink (Region 3, Forester), Tom Johnston (Fuels Specialist, Santa Fe National Forest), and Thomas Marks (Timber Management Officer, Cibola National Forest). All errors and opinions are solely those of the authors. Address correspondence to C. Shepherd, Department of Economics, University of New Mexico, Albuquerque, NM 87131; Email: shepherd@unm.edu.

Title: Investigating National Fire Plan Implementation in Northern New Mexico.

Abstract: To combat the threat of wildfire to the Wildland Urban Interface (WUI), U.S. federal land management agencies have implemented a number of forest restoration and wildfire risk reduction programs. In the spirit of econometrically-based revealed preference analyses, the objective of this study is to investigate the pattern and determinants of U.S. Forest Service expenditures under the National Fire Plan for wildfire risk reduction projects in northern New Mexico. In general, results are consistent with risk reduction hypotheses, but also raise issues that hinge on how risk reduction, and its component parts, should be defined for a region defined by large areas of chronic poverty.

1. Introduction

To deal with the growing threat of catastrophic wildfires throughout the western United States, federal agencies such as the U.S. Forest Service (USFS), Bureau of Land Management (BLM), and National Park Service (NPS) sponsor a variety of programs designed to reduce fire risk on millions of acres of federal, state, and private lands. These programs are operated under the auspices of the National Fire Plan (NFP) and, most recently, the Healthy Forests Initiative (HFI).[1]

The size and scope of NFP and HFI programs relating to forest restoration and wildfire risk reduction is extensive, and growing. Between 2000 and 2002, United States Department of Agriculture and Department of the Interior (USDA and USDI) expenditures have risen from just over $1.5 to $3.2 billion (USDA/USDI 2002). In 2002, hazardous fuels reduction treatments were undertaken on 2.3 million acres; an additional 2.7 acres were treated in 2003. Like other regions of the west, wildfire threatens the physical and economic well-being of communities within and bordering forested lands in northern New Mexico.[2] Many of these communities have strong cultural and historical ties to adjacent forested lands. Besides reducing the risk to life, property and other capital assets, NFP/HFI programs also have the potential to generate economic benefits for forest-dependent communities throughout northern New Mexico. Many of which have long been afflicted by chronic poverty.[3] The confluence of high fire danger, rural poverty and traditional cultural reliance on small diameter wood products in the region underscores a need for developing a greater awareness of how federal agencies distribute NFP/HFI funds for the purpose of fire risk reduction (Gunderson 2001).

This analysis follows in the spirit of previous studies that econometrically analyze the revealed preferences of a government agency implementing a particular program. Most basically, a revealed preference analysis attempts to answer the question of whether a pattern of agency (or program) actions or expenditures is consistent with the expressed goals, or with some alternate objectives. There are a wide variety of econometric or statistically-based studies that follow from the initial revealed preference investigations of McFadden (1975 and 1976). Recent applications in resource and environmental contexts include, Berrens et al. (1999) to underground storage tank cleanup in New Mexico; and, Fernandez (2004) to water treatment on the US-Mexico border. While not based on econometric models, but of topical relevance, Mosely and Toth (2004) recently examined the conformity of NFP contracts with an implied preference for local job creation; they found evidence in the Pacific Northwestern region of the U.S. that consideration of local employment was a key determinant in the decision process. Here, the focus is on examining the pattern of NFP-related projects and associated expenditures in a unique geographic region.

Two avenues are taken to examine agency behavior in northern New Mexico within the context of the NFP and HFI. First, and most basically, a geographic information system (GIS) analysis is used to determine the distributional correlation between NFP/HFI project locations and census blocks with: (1) a predominately high fire risk rating; (2) high dependence on firewood; and (3) high incidence of poverty. Second, the significance and magnitude of factors influencing the pattern of NFP/HFI expenditures are analyzed econometrically. The primary purpose of this analysis is to determine the statistical relationships between NFP/HFI project expenditures in an area or site and wildfire risk reduction, while controlling for other socio-demographic factors in the surrounding area, and project types (with differences in underlying cost structures).

2. Background: Forest Policy Challenges in Northern New Mexico

As a geographical focus, the northern New Mexico study region includes the southern tip of the Rocky Mountains, and is a region of unique historical and social traditions, as well as stark economic contrasts. The region includes the metropolitan areas of Albuquerque and Santa Fe as well as rural towns, Indian pueblos, and Hispano villages. The mountainous and forested terrain of the region provides a common focal point for the citizens of the region; the threat that wildfire poses to the region’s forests provides another.

High wildfire danger presents the most significant danger in the wildland urban interface (WUI). The WUI is defined as the “line, area, or zone where structures and other human development meet or intermingle with undeveloped wildland or vegetative fuels” (USDA/USDI 1995). Due to the pattern of settlement in this region, the WUI is extensive in northern New Mexico, encompassing just under 1 million acres (Haskins 2004). New Mexico’s State Forestry Division has identified 85 communities at high risk from wildfire within these WUI zones (EMNRD 2004).

Many small rural communities rely heavily on the surrounding forest lands.[4] For example, Hispano villagers rely heavily on a variety of small diameter wood products such as poles, posts, vigas, and firewood to build and maintain their homes and farms and provide basic heating and cooking fuel (Gunderson 2001). Forestlands are also critical to Hispano villagers as reliable sources of water to maintain the traditional network of acequias that irrigate their croplands, and as sources of quality forage (Green 1998; Raish 2003). Pueblo Indians rely on nearby forests for many of the same uses but also rely on the biological diversity of nearby forestlands for a wide array of wild foods, medicines, and materials (Dunmire and Tierney 1995). A number of Native American ceremonial areas and religious shrines are located on forestlands near the Pueblos (Hurst 1972). The Forest Service has previously recognized the unique status of these lands (Hurst 1972).

Implementation of the NFP and HFI in northern New Mexico is further complicated by high rates of rural poverty. Northern New Mexico has the unfortunate status as one of the regions in the United States afflicted by high to severe rural poverty conditions which have persisted for decades (Nord 1997; ERS 2004). Moreover, it is one of the few regions where rural poverty is getting worse (Nord 1997). According to the U.S. Census Bureau, high rural poverty areas are defined as non-metro regions with a poverty rate of 20 percent or more based on 1999 incomes reported in the 2000 Census (ERS 2004). Roughly 50% of the land area of northern New Mexico is included in census blocks where poverty rates exceed this threshold (Haskins 2004). The presence of persistent rural poverty can intensify the negative impacts of wildland fire on communities that lack adequate resources to reduce or mitigate risk exposure (Niemi and Lee 2001). Poverty can constrain participation in risk averting activities and, as a result, increases risk exposure. Evidence from both survey-based contingent valuation and experimental settings has established that income is a strong predictor of whether or not a household engages in risk averting activity at all, as well as willingness to pay for such activities (Talberth et al. 2004). The threat of wildfire is compounded when one considers that impoverished communities are more likely to rely heavily on nearby forestlands for fuel wood for heating. This is certainly the case in many Hispano and Native American communities in northern New Mexico (Raish 2000). When fire occurs, the communities of northern New Mexico not only suffer disproportionate losses to market resources such as homes and structures but to non-market resources that help support their livelihoods.

3. A Brief Review of NFP/HFI Programs

Implementation of the NFP and HFI in northern New Mexico has been ongoing since 2001. Until recently, the NFP and HFI were operated as administrative level programs, with no overarching national legislative guidance. That changed in 2003, when President Bush signed the Healthy Forests Restoration Act (HFRA) into law.[5] The HFRA authorizes just over $810 million per year nationwide between 2004 and 2008 for fire risk reduction programs (e.g., hazardous fuels treatments) (HFRA 2003).

Title I of the HFRA gives explicit direction to prioritize hazardous fuels reduction treatments near WUI communities identified as being at high risk from wildfire as determined by the Secretaries of Agriculture and Interior in a list published in 2001. With respect to traditional cultural uses, there are special provisions of the HFRA designed to enhance forest stewardship, watershed protection, and restoration needs on tribal and pueblo lands.[6] In particular, the HFRA emphasizes the protection of water supply systems critical to Pueblos and Hispano communities who rely on small ditches, reservoirs, or canals for drinking and irrigation water.[7]

Against this regulatory backdrop, federal agencies in northern New Mexico are implementing the NFP and HFI by directly funding projects on federal lands and by providing grants to the State of New Mexico, local public agencies, tribal agencies, rural or urban fire departments, community organizations, non-profit organizations, and small businesses. NFP/HFI related projects are designed to reduce hazardous fuels, assess and reduce wildfire risk, build firefighting capacity, develop markets for small diameter wood products and biomass, and revitalize rural communities.[8]

4. Descriptive Distributional Analysis of Projects

To gain a better understanding of where, in the region, NFP/HFI projects were planned between 2000 and 2004 a cursory distributional analysis was conducted. The U.S. Forest Service provided the NFP/HFI project data, which includes: (1) a list of all projects funded under the NFP and affiliated programs (e.g., Rural Fire Assistance, Rural Community Assistance, and Forest Land Enhancement) from 2000 through 2004, (2) brief descriptions of the nature of each of the projects (i.e. thinning, prescribed burning, hazard assessment, small diameter utilization, restoration), the project’s size (acres), and general project location (latitude and longitude), (3) a financial summary of each project indicating the total expenditure (NFP and other sources) on each project to date.[9] Expenditure information was not available for a select number of the listed projects. A listed and approved project may simply not have had any expenditure as of 2004; e.g., due to timing, or prescribed actions (burns, etc.) that were planned may not have been carried out due to unfavorable conditions. In total, 455 projects are listed from 2000 through 2004; with expenditure data available for 373 of the projects. Over the five-year period NFP/HFI related project expenditures in the study area totaled approximately $30 million.

To carry out the distributional analysis, project coordinates were matched to the census block in which the project was located. Of the 455 identified NFP/HFI projects that were planned from 2000 through 2004, coordinates (latitude/longitude) were provided for 438 observations. These projects were, accordingly, matched to a census block. The analysis compares the percentage distribution of planned NFP/HFI projects in a census block to fire risk, incidence of poverty, and firewood dependency. The topics of comparison coincide with the primary objectives of the HFRA (e.g., fire risk reduction, supply of small diameter wood products, economic development).

Fire risk data were obtained from the U.S. Department of Agriculture General Technical Report RMRS-87 titled “Development of Coarse-Scale Spatial Data for Wildland Fire and Fuel Management” which was prepared for the Departments of Agriculture and Interior (Schmidt et al. 2000).[10] The Coarse-Scale Fire Risk Map classifies WUI locations into three distinct risk zones. Low risk zones are characterized by relatively sparse numbers of small trees and little ground fuel. The composition of the fuel types in these zones constrains the intensity of wildfires (Schmidt et al. 2000). Moderate (or medium) risk zones are characterized by higher densities of small trees and brush with considerable ground fuel. The composition and abundance of fuels in moderate risk zones sometimes lead to higher intensity burns in some areas (Schmidt et al. 2000). High risk zones are characterized by heavy fuel loads which consist of substantial amounts of dead material and high densities of smaller trees which extend into the canopies of larger, older age trees. The composition and abundance of fuels in the area lead to high intensity, rapid crown fires (Schmidt et al. 2000).

Besides providing descriptions of fire risk, the Coarse-Scale Fire Risk Map Data sheds light on the composition of forest fuels in the study area. In particular, the Coarse-Scale Fire Risk Map indicates that the study area contains all three risk zones, which means that the fuel types in the region extend from small trees and grass to older and larger mature timber.

Data on the degree to which households depend on firewood for a primary heating source were obtained from the 2000 U.S. Census. Firewood dependency is used as a proxy for dependency on small diameter wood products.[11] Information on the incidence of poverty was also obtained from the 2000 U.S. Census. The Census Bureau classifies a census block as having a high incidence of poverty if 20% or more of the households in that block fell below the federally defined poverty line (USCB 2000).

To ease analysis, the census blocks have been categorized into three strata high, medium, and low. The classification depends upon the composition of the census block with respect to the topic of interest. Figures 1 through 3 provide a visible description of the number of NFP/HFI projects by strata. The distributional analysis is presented in Table 1.

Figure 1 illustrates the distribution of NFP/HFI projects with respect to fire risk. Census blocks were classified into three strata based on the fire risk class – high, moderate, or low – that comprise the bulk of the land area within a particular block. As shown in Figure 1, there were 213 projects in the high fire risk stratum, 161 in the moderate fire risk stratum, and 66 in the low fire risk stratum. In Table 1 we can see that the high fire risk stratum represents 16.8% of the northern New Mexico landscape, the moderate risk stratum 58% and the low risk stratum 20.6%. The distribution of planned NFP/HFI projects between 2000 and 2004 is consistent with emphasizing the high fire risk stratum. In particular, there were 30% more NFP/HFI projects located in the high fire risk strata than would be expected under a proportional distribution, 22.6% less in the moderate risk stratum, and 6.1% less in the low risk stratum.

Figure 2 illustrates the distribution of planned NFP/HFI projects with respect to the incidence of poverty. Census blocks were classified into three strata based on the percentage of households in each census block whose per capita income fell below the poverty line in the year 2000. The strata corresponding to the following percentages of a census block that fall below the defined poverty level. The high poverty stratum consists of census blocks where 20% or greater of all households fell below the poverty line. The medium stratum is comprised of census block where 10% to 19% of all households fell below the poverty line. The low stratum consists of census blocks where less than 10% of all households fell below the poverty line. A total of 186 planned projects were identified to fall in the high poverty incidence stratum, 214 in the middle poverty incidence stratum, and 38 in the low poverty incidence stratum. The high poverty incidence stratum represents 49.8% of the northern New Mexico landscape, the middle stratum 35.3% and the lowest stratum 31.1%. The distribution of NFP/HFI projects from 2000 through 2004 has tended to emphasize the middle poverty incidence stratum and de-emphasize census blocks that fall into either the high or low poverty incidence strata. In particular, there were 9.1% more NFP/HFI projects located in the middle poverty incidence strata than would be expected under a proportional distribution, 7.3% less in the high poverty incidence stratum, and 1.8% less in the low poverty incidence stratum.

Figure 3 illustrates the distribution of NFP/HFI projects with respect to firewood dependency. In the 2000 U.S. Census Bureau data, a household is classified as relying on firewood if it is the type of fuel most often used to heat the home (USCB 2000). Census blocks were classified into three strata based on the percentage of households, which indicated that they primarily rely on firewood. If 30% or more of households in a census block depend on firewood, the block was grouped in the highest stratum. If dependency fell between 16 and 29%, the census block was grouped in the middle stratum. The census block was placed in the lowest stratum if dependency on firewood was 15% or less. The ranges chosen insured that, across the study area, there were an equal number of households in each stratum. As shown in Figure 3, there were 176 planned NFP/HFI projects that were identified to fall in the highest firewood dependency stratum, 135 in the middle stratum and 127 in the lowest stratum. From Table 1 we can see that census blocks in the highest dependency stratum represent 30.1% of the northern New Mexico landscape, the middle stratum 29.1% and the lowest stratum 40.8%. A proportional distribution would, then, more or less follow this breakdown. The distribution of NFP/HFI projects from 2000 through 2004 has tended to emphasize census blocks with a high dependency on firewood and de-emphasize census blocks with a low dependency. In particular, there were 10.1% more NFP/HFI projects located in the high dependency census blocks than would be expected under a proportional distribution and 11.8% less in census blocks with low dependency.

Taken together, the distributional analysis suggests that NFP/HFI projects were distributed in a manner consistent which strongly emphasizes census blocks that are predominately high fire risk. In addition, the analysis suggests that the planned projects placed less emphasis on census blocks that exhibit a high dependency on firewood and other small diameter wood products. The distribution of NFP/HFI projects has not strongly emphasized census blocks with the highest incidence of poverty. With the bulk of planned NFP/HFI projects being directed at census blocks that fell into the medium stratum. Although some insights can be drawn as to where NFP/HFI projects have been distributed, the analysis provides no understanding of how project expenditures are allocated. For this we turn to the econometric analysis.

5. Econometric Analysis of Project Expenditures

Although the distributional analysis helps to answer the question of where NFP/HFI projects have been directed; it cannot answer the question of how spending decisions are made.

In particular, it is important to identify if the federal agencies responsible for NFP/NFI implementation have behaved in a fashion consistent with the stated objectives of reducing exposure to risk from wildland fire, improving economic opportunity in communities that suffer from persistent poverty, and/or promoting market uses for small diameter wood products.

In administering NFP/HFI projects the agency of interest is hypothesized to prefer allocating funds in a manner that reduces exposure to risks from wildland fires. We postulate that the preferences of the agency with respect to allocating NFP/HFI funds for projects at site i can be represented by the following preference or utility function, Ui:

Ui = Ui(RISKi, Zi) (1)

Arguments in U(·) fall into two broad categories. The vector RISK would include the agency’s measure of wildfire risk and expected losses. The vector Z would include control variables such as socio-demographics in the surrounding area (e.g., poverty or small diameter wood dependency), size of the related area (e.g., census block), and any information on project type (e.g., burn-related), where underlying cost structures may vary. A reduced form equation on NFP/HFI project expenditures for any given location or area would appear as (subscripts are suppressed for simplicity):

Y = Y(RISK, Z) (2)

Conceptually, risk exposure in the WUI can be thought of in terms of expected losses.[12] The vector RISK includes variables that are indicators of this risk exposure in the WUI. Given the objective of risk reduction, we expect that the pattern of project expenditures should be positively related to risk indicators. More specifically, the vector RISK includes the variables PHIGH, VAL, and OHSG, where: PHIGH is the percentage of census block acres rated high fire risk, VAL is the median value of owner occupied homes in a census block, and OHSG which is the number of occupied housing units in the census block. Thus, we take PHIGH, VAL, and OHSG (or DEN, where DEN is defined as OHSG per census block acre) as positively-related indicators of risk (i.e., as they increase in level, then expected loss increases).[13]

We test the general hypothesis of a significant positive relationship against the null of no relationship. Thus, the specific risk reduction hypotheses are:

[pic]

The caveat on H1c is that past some threshold on the rural-urban continuum in the WUI the risk of a damaging wildfire event may decrease. Thus, we have the expectation of an inverted U-type relationship (e.g., [pic]).

The NFP/HFI project expenditure data are broken down by fiscal year into five panels. A total of 373 NFP/HFI projects were identified as having expenditures on record.[14] Of this number, 366 projects could be linked to census data. Across the five panels, the number of expenditure observations varies with a minimum of five identified projects in 2000 and a maximum of 184 projects in 2004. The continuous variables EXP and AEXP are defined as the total expenditures per project, and expenditures per project per census block acre, respectively. The indicator variables BURN and THIN are used to control for project type.[15] Other biomass removal methods (e.g., mastication, mowing, chipping, etc.) are used as the baseline of comparison. The variable ACRES controls for variations in the size of the census blocks (i.e., the number of acres in the census block). In addition to the primary risk-related and project variables, data from the 2000 Census was used to construct the variables PPOV, which is the percent of individuals in the census block that live below the federal poverty line; and PWOOD, which is the percent of occupied homes in a census block that rely on firewood for heat. The quadratic terms OHSGSQ and HDENSQ are included to identify if there is an inverted U-shape relationship between NFP/HFI expenditures and housing densities, as hypothesized.

At the most disaggregated level of individual projects (in individual years 2000-2004), descriptive statistics of the variables used in the expenditure analysis are found in Table 2.

Of note, mean project expenditures are approximately $81,000. In terms of the census blocks where individual projects are located, the mean of the median home values is $114,000, the mean level of households in poverty is 19 percent, with an average of 28 percent of the homes dependent on firewood as the primary source of heat. There is considerable variability in both housing density and the size of census blocks. Given the “unbalanced” nature of the panels the expenditure data were also aggregated up to the census block level for the entire period (2000-2004). There were 64 identified census blocks over which the 373 projects were distributed. Due to the level of aggregation the project type indicator variables BURN and THIN are not included in the analysis. Descriptive statistics for the aggregated data are provided in Table 3.

The expenditure equations that were estimated follow the reduced form expenditure model presented in equation two (2). It should be noted that a number of the explanatory variables are highly correlated. The variable LNVAL is highly and negatively correlated with the variables PPOV and PWOOD. There is also a strong positive correlation between the variables PPOV and PWOOD. That is, areas with relatively high average home values have low percentage rates of poverty and of dependence on firewood as a primary source of heating. The presence of strong correlations (either positive or negative) suggests that multi-collinearity may pose a problem if the correlated variables were estimated in the same equation. To circumvent the problem of collinear variables, none of the models were estimated with the variables LNVAL, PPOV, and PWOOD together. Rather, we provide a variety of specifications for comparisons.

Alternate specifications of the expenditure models were estimated using both the disaggregated and aggregated data. For the disaggregated models, two sets of equations were estimated using the log of individual NFP/HFI project expenditures (LNEXP) and the log of individual NFP/HFI project expenditures per census block acre (LNAEXP) as the dependent variable.[16] Estimation results from these models are presented in Table 4 (LNEXP) and Table 5 (LNAEXP). Estimated expenditure equations drawing on the aggregated expenditure data used the variables LNAGGEXP and LNAAGGEXP as the dependent variables. Results from the aggregate models are presented in Table 6 and Table 7.

In general, the estimated results are broadly consistent with the proposed hypotheses on risk reduction. However, the evidence supporting any inferences on the effect of the PHIGH risk variable by itself is weak. In only four of the 30 model specifications presented in Tables 4-7 was the estimated coefficient on PHIGH positive and significant. In no case was the estimated coefficient negative and significant. Generally, in the better fitting models, with extended sets of variables, the estimated coefficient on PHIGH was not significantly different from zero. In contrast, the estimated coefficients on the variable LNVAL (natural log of VAL) are positive and significant across all specifications.[17] Thus, NFP/HFI expenditures and expenditures per acre are both estimated to increase as median home values rise. The relationship between median home values and NFP/HFI expenditures and expenditures per acre can also be directly interpreted in terms of expenditure elasticity. For a 10% increase in the median home value, NFP/HFI expenditures are estimated to increase anywhere from 6.5% to 7.1%. The same change in median home value will result in an increase of 3.4% to 7.1% in NFP/HFI expenditures per acre. Thus, while NFP/HFI expenditures are positive and significantly related to changes in housing values, they are relatively inelastic (i.e., not highly responsive in proportional terms).

Although the positive and significant relationship between LNVAL and both LNEXP and LNAEXP is consistent with the hypothesized behavior, there is an implicit interpretation that suggests that project expenditures may not be directed towards impoverished areas. This interpretation is important to keep in mind, especially when we consider that the variable LNVAL had a strong negative correlation with the variable PPOV. In fact, of the four models in which the PPOV variable was included, the estimated coefficients are negative and significant in two (0.05 and 0.10 level). While collinearity prevented the estimation of models with both the LNVAL and PPOV variables there does appear to be a case in support of NFP/HFI expenditures being directed at higher income census blocks.

The estimated coefficients on HDEN are negatively and significantly related to both individual project expenditures (LNEXP) and individual project expenditures per acre (LNAEXP) at the 0.01 level. For the analysis using the aggregated data HDEN is estimated to have a positive and significant relationship with aggregate project expenditures per acre (LNAGGAEXP). Although insignificant, the estimated relationship between OHSG and both expenditures and expenditures per acre is positive. Together, these findings are consistent with hypothesized behavior and suggest that project expenditures are being directed towards census blocks with increasing housing density. Over some range of the rural-urban continuum, these census blocks could most certainly be considered a part of the WUI. Additionally, the housing densities would tend to increase near the periphery of the major metropolitan areas of Albuquerque and Santa Fe. We should also expect that housing values would also increase near the metropolitan areas found in the study area. Thus, the argument that NFP/HFI expenditures are directed towards higher income areas is strengthened when housing densities are controlled for in the analysis. Consistent with our hypothesis of a threshold effect, HDENSQ is negatively and significantly related (0.01 levels) to both individual expenditures and expenditures per acre. The threshold, or turning point, where the relationship between project expenditures and housing density turns from positive to negative, occurs at an average of approximately 1.7 houses per acre (e.g., from Model 7, Table 4) The implication is that areas approaching multiple houses per acres have crossed from the WUI into more high density urban or suburban form.

Risk reduction variables pertain to the benefit side of the agency preference function and the reduced form expenditure function; i.e., what benefits are generated out of project expenditures? Where available, it is also important to differentiate by project type in a revealed preference analysis, since differing underlying cost structures will affect the level of expenditures (e.g., Berrens et al., 1999; Fernandez, 2004). If nothing else, these are important statistical controls for interpreting the results on our risk reduction hypotheses. Thus, turning our attention to project type in the dis-aggregate (project-specific) analyses of Table 4, the coefficient on the THIN binary variable is not significant in either of the models in which it appears, the coefficient for the BURN variable is consistently negative and significant. These results indicate that expenditures on burning-related projects, like prescribed fires, decrease relative to other fuel reduction methods (i.e., mastication, mowing, etc.). These results hold true for expenditures per acre (LNAEXP) as well (Table 5). While we do not endeavor to analyze the methods for reducing risk of wildland fires, the NFP does call for the development of “…cost-effective fire protection among all administrative boundaries (USDA/USDI 2000, pg. 8).” While it is assumed that many factors are considered when choosing a method of eliminating small diameter wood from a given area, burning may be the most cost effective where it is a practical and safe method. More basically, our results are consistent with the available evidence that burn-related projects cost less (see Donovan and Brown, 2005).

6. Discussion and Conclusions

In general, the findings of the revealed preference analysis are consistent with the risk reduction hypotheses proposed to describe agency (U.S. Forest Service) actions. The cursory distributional analysis supports such an argument. The econometric analysis showed only weak support for a positive relationship between expenditures and census blocks with a high percentage of high risk-rated acreage; however, the high rating of risk from the Coarse-Scale Fire Risk Map (Schmidt et al. 2000) is only one component of risk exposure. Evidence suggests that the number and density of houses exposed (with a rural-urban threshold effect), and the relative value of homes are significant positive determinants of NFP/HFI expenditures. Thus, in summary, the findings of the econometric analysis on the pattern of agency expenditures on NFP/HFI projects from 2000-2004 in northern NM are consistent with the primary objective, and general hypothesis, of risk reduction from wildfire in the WUI. Future research might be extended to: (1) similar applications to different geographical area and scales; (2) measuring the actual success of program expenditures in reducing wildfire risk and other physical indicator of forest restoration over time; and (3) more sophisticated econometric modeling (e.g., handling possible spatial correlation issues) as additional data is collected.

But the results also raise some interesting issues that hinge on how risk reduction, and its component parts, should be defined or interpreted. Poverty is a critical concern in northern New Mexico, and there is certainly a philosophical issue about whether the average market value of homes should continue to be a significant empirical determinant of risk reduction expenditures in an area.[18] For example, an argument might be that continuing a pattern determined by market values for housing may be counter to other cultural or community assets that are primarily non-market in nature (e.g., which might be reflected in a location’s dependence on firewood or some other indictors). Taken together, the findings of our models with respect to the variables PPOV, PWOOD, and LNVAL do not support any claim that the pattern of NFP/HFI expenditures are consistent with poverty reduction, at least in terms of locational concentration by census blocks. Rather, when the strong negative correlation between LNVAL and PPOV is considered, the evidence supports the contention that NFP/HFI expenditures are directed towards census blocks with high homes values (positively correlated with income).

However, we caution that no definitive conclusions on the relationship between the NFP/HFI and poverty and rural development can be drawn without looking at the related pattern of expenditures effects on employment. This would include both direct local employment from the projects themselves, and any direct or indirect markets effects (e.g., from purchase of local goods and services, and improved capacity or utilization of small diameter wood products from thinning or restoration efforts). If expenditures aren’t concentrated toward high poverty (and firewood dependent) communities or census tracts, then aiding rural development through employment becomes a critical research question for northern New Mexico. Possible avenues of future research include analysis of changes in employment levels in SIC categories related to the type of work created by NFP/HFI projects, which would provide a limited view of employment changes within these particular categories. Following recent work by Mosely and Toth (2004) in the Pacific Northwest of the U.S., one option would be to survey NFP/HFI contractors in order to obtain information on local hiring practices. Mosely and Toth (2004) provide empirical support for a preference for local hiring, and there is at least anecdotal evidence of important steps toward that objective in northern New Mexico.[19] Further, regional economic modeling (e.g., input-output analysis, or general equilibrium modeling) could be conducted at different geographic scales (counties and sub-regions) to measure aggregate and distributional effects of project expenditures on employment, income and tax revenues.

Given continued population growth, concomitant urban sprawl and the critical fuels build-up in WUI zones across much of the western U.S., it is likely that wildfire management and forest restoration policy will continue to evolve for some time. Thus, investigating the pattern of social, economic and community affects from forest restoration efforts generally, and the NFP/HFRI specifically, will continue to be an important issue in the western U.S., and we hope that this research spurs additional investigations.

Appendix A: Components of the Coarse-Scale Fire risk Map (Schmidt et al 2000)

The ratings are based upon an integration of population density, fuels and weather spatial data. Appendix G, pages 34 - 41 of the General Technical Report RMRS-87 (referred to as the Coarse Scale Spatial Data for Wildland Fire and Fuels Management), includes a series of maps of the contiguous United States that map much of the data used to create the ratings. The maps include: (1) “Potential Natural Vegetation Groups” – Which, in concert with Appendix A page 23 of RMRS-87, lists the natural vegetation groups that could exist in the United States; (2) “Current Cover Types” – Which in concert with Appendix B page 28, maps the type of ground cover that existed prior to 2000. This map includes Agriculture, Water and Urban cover which are used to define areas with no risk of wildland fire; (3) “Historical Natural Fire Regimes”: - which reflect the historical frequency and severity of wildland fires; (4) “Fire Regime Current Condition Classes” – Maps the four ratings listed above, and has been reproduced for the study area as Figure 1 in this paper; (5) “Fire Regime Current Condition Classes by Historical Fire Frequency” - Combines information from maps 3 and 4 above; (6) “National Fire Occurrence” - Maps the location of wildland fires from 1986 – 1996 and the size of those fires; (7) “Potential Fire Characteristics” - Maps the maximum number of days per annum in which the potential flame length of a wildland fire exceeds 8 feet; (8) “Wildland Fire Risk to Flammable Structures” - Compiles data about weather conditions similar to conditions which resulted in multiple structure damage, classification of vegetation types and housing density.

References

Alcantara, A., K. Kargacin, et al, 1997. Poverty in New Mexico: Who are the Poor? Albuquerque, Bureau of Business and Economic Research, University of New Mexico.

Berrens, Robert P., Alok K. Bohara, Amy Baker and Ken Baker, 1999. Revealed Preferences of a State Bureau: Case of New Mexico’s Underground Storage Tank Program. Journal of Policy Analysis and Management. 18, 2, 303–326.

Debuys, W., 1985. Enchantment and Exploitation: The Life and Hard Times of a New Mexico Mountain Range. Albuquerque, University of New Mexico Press.

Donovan, G., and T. Brown. 2005. An Alternative Incentive Structure for Wildfire Management on National Forest Land. Forest Science, In Press.

Dunmire, W. and G. Tierney, 1995. Wild Plants of the Pueblo Province, Exploring Ancient and Enduring Uses. Santa Fe, Museum of New Mexico Press.

Dutton, B. P., 1983. American Indians of the Southwest. Albuquerque, University of New Mexico Press.

EMNRD, 2004. New Mexico Communities at Risk Assessment Plan. Santa Fe, Energy, Minerals, and Natural Resources Department, Forestry Division. Available at: emnrd.state.nm.us/forestry/NMFIREPLAN/docs/NMCOMMRISKASSESSPLAN.pdf.

ERS, 2004. Rural Income, Poverty, and Welfare. Washington, D.C., U.S. Department of Agriculture, Economic Research Service. Available at: www/ers/Briefing/IncomePovertyWelfare/.

Fernandez, Linda, 2004. Revealed Preferences of an International Trade and Environment Institution. Land Economics. 80, 2, 224–238.

Gunderson, G., 2001. Distribution of Timber Sales on Northern New Mexico National Forests 1992 - 1999: Are Small and Medium Sized Businesses Getting Their Share? Southwest Region Working Papers. Santa Fe, Southwest Community Forestry Research Center, Forest Trust.

Haskins, B., 2004. National Fire Plan and Rural Poverty in Northern New Mexico: A GIS Analysis. Santa Fe, Ecology and Law Institute.

Hurst, W. D., 1972. Region 3 Policy on Managing National Forest Land in Northern New Mexico. Albuquerque, USDA Forest Service, Southwestern Region.

McFadden, Daniel, 1975. The Revealed Preferences of a Government Bureaucracy: Theory. Bell Journal of Economics. 6, 2, 401–416.

McFadden, Daniel, 1976. The Revealed Preferences of a Government Bureaucracy: Empirical Evidence. Bell Journal of Economics. 7, 1, 55-72.

Moseley, Cassandra and Nancy A. Toth, 2004. Fire Hazard Reduction and Economic Opportunity: How Are the Benefits of the National Fire Plan Distributed? Society and Natural Resources. 17, 701–716.

Niemi, E. and K. Lee, 2001. Wildfire and Poverty. Eugene, Oregon, EcoNorthwest, Center for Watershed and Community Health.

Nord, M., 1997. Overcoming Persistent Poverty - And Sinking Into It: Income Trends in Persistent Poverty and Other High-Poverty Rural Counties, 1989-1994. Rural Development Perspectives. 12, 2-10.

Raish, C., 2000. Environmentalism, the Forest Service, and the Hispano Communities of Northern New Mexico. Society and Natural Resources. 13, 489-508.

Raish, C., 2003. Economic, Social, and Cultural Aspects of Livestock Ranching on the Espanola and Canjilon Ranger Districts of the Santa Fe and Carson National Forests: A Pilot Study. Gen. Tech. Rpt. RMRS-GTR-113. Ft. Collins, USDA Forest Service, Rocky Mountain Research Station.

Schmidt, K. M., James P. Menakis, Colin C. Hardy, David L. Bunnell, Neil Sampson and Jack Cohen, 2000. Development of Coarse Scale Spatial Data for Wildland Fire and Fuels Management. GTR-CD-000. Ogden, UT. USDA Forest Service, Rocky Mountain Research Station.

Talberth, J., R. P. Berrens, et al, 2004. Averting and Insurance Decisions in the Wildland Urban Interface - Implications of Survey and Experimental Data for Wildfire Risk Reduction Policy. Unpublished Manuscript, in review.

USCB, 2000. American Factfinder: Summary Table SF - 3. Washington, D.C., U.S. Census Bureau.

USDA/USDI, 1995. Federal Wildland Fire Management: Policy and Program Review. Washington, D.C., U.S. Department of the Interior, U.S. Department of Agriculture.

USDA/USDI, 2001. A Collaborative Approach for Reducing Wildland Fire Risks to Communities and the Environment: 10-Year Comprehensive Strategy. Washington, D.C., U.S. Department of the Interior, U.S. Department of Agriculture.

USDA/USDI, 2002. National Fire Plan: FY 2002 Performance Report. Washington, D.C., U.S. Department of Agriculture, U.S. Department of Interior.

USFS, 2000. Protecting People and Sustaining Resources in Fire-Adapted Ecosytems: The Forest Service's Management Response to the General Accounting Office Report GAO/RCED-99-65. Washington, D.C., USDA Forest Service.

USFS, 2001. Biological Assessment: R-3 Wildland Urban Interface Amendment. Albuquerque, USDA Forest Service Region 3.

Table 1: Class Distributions for Fire Risk Rating, Incidence of Poverty, and Dependence on Small Diameter Wood for Heat vs. Actual NFP Project Distributions.

|Fire Risk Rating |

| |High |Moderate |Low |Unspecified |

|Distribution of class |16.8 |58.0 |20.6 |4.6 |

|Distribution of projects |46.8 |35.4 |14.5 |3.3 |

|Departure |+30.0 |-22.6 |-6.1 |-1.3 |

| |

|Percent of Households Within Census Block Living Below Federal Poverty Line |

| |High |Moderate |Low | |

| |(20% +) |(10-19%) |(0-9%) | |

|Distribution of class |49.8 |39.8 |10.4 | |

|Distribution of projects |42.5 |48.9 |8.6 | |

|Departure |-7.3 |+9.1 |-1.8 | |

| |

|Percent of Occupied Homes Within Census Block That Rely On Wood For Heat |

| |High |Moderate |Low | |

| |(30% +) |(16-29%) |(0-15%) | |

|Distribution of class |30.1 |29.1 |40.8 | |

|Distribution of projects |40.2 |30.8 |29.0 | |

|Departure |+10.1 |+1.7 |-11.8 | |

Table 2: Descriptive Statistics of Individual Projects within Northern New Mexico

(Fiscal Years 2000 – 2004)

|Variable |Description |Obs. |Mean |Std. |

| | | | |Dev. |

|EXP |Project Expenditures ($) |373 |80,975.79 |196,552.50 |

|AEXP |Project expenditures ($) per census block acre |373 |4.43 |34.15 |

|PHIGH |Percent of census block acres rated high fire risk by |373 |0.38 |0.22 |

| |Coarse Scale Map | | | |

|PWOOD |Percent of homes where firewood primary source of heat |366 |0.28 |0.17 |

|PPOV |Percent of population in census track with income below|366 |0.19 |0.11 |

| |Federal poverty level | | | |

|VAL |Median value of owner occupied homes in census block |366 |113,923.80 |67,934.35 |

|BURN |NFP project described as a burn |373 |0.31 |0.46 |

| |(1 if yes, 0 else) | | | |

|THIN |NFP project described as |373 |0.51 |0.50 |

| |thinning | | | |

| |(1 if yes, 0 else) | | | |

|ACRES |Number of acres within a census block |373 |236,340.20 |187,586.60 |

|OHSG |Number of occupied housing units in census block |366 |542.98 |323.63 |

|HDEN |Number of occupied housing units per census block acre |366 |0.02 |0.18 |

Notes: Census block (identified as B.G.I.D. or Block Group Identification) is a subdivision of a census tract and is the smallest area for which project information is available. Fire risk rating taken from the Coarse Scale Fire Risk mapping project managed by the U.S. Departments of Agriculture and Interior (Schmidt et al. 2000).

Table 3: Descriptive Statistics of NFP/HFI Expenditures Aggregated by Census Block

(Fiscal Years 2000 – 2004)

|Variable |Description |Obs. |Mean |Std. |

| | | | |Dev. |

|AGGEXP |Aggregate project expenditures |64 |471,937.00 |62,0247.40 |

| |($) for a census block | | | |

|AGGAEXP |Aggregate project expenditures |64 |30.40 |89.51 |

| |($) | | | |

| |per census block acre | | | |

|PHIGH |Percent of census block acres |64 |29.57 |24.73 |

| |rated high fire risk | | | |

|PWOOD |Percent of homes w/primary |62 |20.62 |16.20 |

| |source of heat = wood | | | |

|PPOV |Percent of population with |62 |17.91 |10.66 |

| |income below Federal poverty | | | |

| |level | | | |

|VALUE |Median value of owner occupied |62 |120,314.50 |98,129.33 |

| |homes in census block | | | |

|ACRES |Number of acres within a census|64 |179,246.80 |229,597.50 |

| |block | | | |

|OHSG |Number of occupied housing |62 |464.66 |276.30 |

| |units in census block | | | |

|HDEN |OHSG per census block acre |62 |0.11 |0.47 |

Notes: Census block (identified as B.G.I.D. or Block Group Identification) is a subdivision of a census tract and is the smallest area for which project information is available. Fire risk ratings taken from the coarse scale fire risk mapping project managed by the U.S. Departments of Agriculture and Interior (Schmidt et al. 2000).

Table 4: Population Averaged Estimation of NFP/HFI Expenditure Equation (Dependent Variable = LNEXP)

|Variable |Model 1 |Model 2 |Model 3 |Model 4 |Model 5 |Model 6 |Model 7 |

| |(n = 373) |(n = 366) |(n = 366) |(n = 366) |(n = 366) |(n = 366) |(n = 366) |

|PHIGH |1.50 |1.39 |1.50 |0.79 |0.80 |0.61 |1.08 |

| |(2.06)** |(1.82)* |(2.02)** |(1.03) |(1.05) |(0.80) |(1.32) |

|PWOOD |--- |-0.01 |--- |--- |--- |--- |--- |

| | |(-0.01) | | | | | |

|PPOV |--- |--- |1.68 |--- |--- |--- |--- |

| | | |(1.06) | | | | |

|LNVAL |--- |--- |--- |0.68 |0.80 |0.65 |0.77 |

| | | | |(2.23)** |(2.40)** |(1.79)** |(2.26)** |

|ACRES |--- |--- |--- |--- |0.77 |0.27 |0.71 |

| | | | | |(0.91) |(0.31) |(0.82) |

|HDEN |--- |--- |--- |--- |--- |-0.78 |--- |

| | | | | | |(-0.54) | |

|HDENSQ a |--- |--- |--- |--- |--- |0.00 |--- |

| | | | | | |(0.00) | |

|OHSG |--- |--- |--- |--- |--- |--- |-2.37 |

| | | | | | | |(-1.00) |

|OHSGSQ |--- |--- |--- |--- |--- |--- |1.65 |

| | | | | | | |(0.82) |

|CONSTANT |11.80 |11.86 |11.51 |4.27 |2.74 |4.71 |3.65 |

| |(42.27)*** |(34.51)*** |(26.79)*** |(1.25) |(0.72) |(1.13) |(0.92) |

|lnL |-112.52 |-109.25 |-108.66 |-106.75 |-106.31 |-104.31 |-105.64 |

|AIC |3.58 |3.62 |3.60 |3.54 |3.56 |3.56 |3.60 |

Notes: ***, ** and * implies significance at 0.01, 0.05 and 0.10 levels respectively; Corresponding t-statistics in parentheses;

AIC = [-2*(log likelihood) + 2*(# of coefficients)]/n. ACRES scaled by dividing by 107. OHSG scaled by dividing by 103. OHSGSQ scaled by dividing by 106. a- estimated coefficient = 0.0024.

Table 7: Generalized Linear Estimation of NFP/HFI Expenditure Equation using Aggregated Data

(Dependent Variable = LNAAGGEXP)

|Variable |Model 1 |Model 2 |Model 3 |Model 4 |Model 5 |Model 6 |Model 7 |

| |(n = 64) |(n = 62) |(n = 62) |(n = 62) |(n = 62) |(n = 62) |(n = 62) |

|PH |0.48 |1.10 |0.16 |-0.93 |-0.98 |-0.45 |-1.08 |

| |(0.60) |(1.51) |(0.20) |(-1.27) |(-1.52) |(-0.82) |(-1.56) |

|PWOOD |--- |-4.90 |--- |--- |--- |--- |--- |

| | |(-4.36)*** | | | | | |

|PPOV |--- |--- |-2.87 |--- |--- |--- |--- |

| | | |(-1.68)* | | | | |

|LNVAL |--- |--- |--- |1.44 |0.95 |0.90 |0.99 |

| | | | |(4.91)*** |(3.40)*** |(3.42)*** |(3.44)*** |

|ACRES |--- |--- |--- |--- |-3.12 |-2.38 |-3.07 |

| | | | | |(-4.34)*** |(-3.74)*** |(-4.19)*** |

|HDEN |--- |--- |--- |--- |--- |4.13 |--- |

| | | | | | |(3.98)*** | |

|HDENSQ |--- |--- |--- |--- |--- |-1.09 |--- |

| | | | | | |(-2.97)*** | |

|OHSG |--- |--- |--- |--- |--- |--- |0.95 |

| | | | | | | |(0.47) |

|OHSGSQ |--- |--- |--- |--- |--- |--- |-1.05 |

| | | | | | | |(-0.62) |

|CONSTANT |1.78 |2.62 |2.42 |-14.30 |-8.15 |-8.01 |-8.70 |

| |(5.79)*** |(7.97)*** |(5.19)*** |(-4.34)*** |(-2.53)** |(-2.66)*** |(-2.57)*** |

|lnL |-118.78 |-106.45 |-113.67 |-104.51 |-95.79 |-84.31 |-95.46 |

|AIC |3.77 |3.53 |3.76 |3.47 |3.22 |2.91 |3.27 |

Notes: ***, ** and * implies significance at 0.01, 0.05 and 0.10 levels respectively; Corresponding t-statistics in parentheses;

AIC = [-2*(log likelihood) + 2*(# of coefficients)]/n. ACRES scaled by dividing by 107. OHSG scaled by dividing by 103. OHSGSQ scaled by dividing by 106.

Figure 1

[pic]

Figure 2

[pic]

Figure 3

[pic]

-----------------------

[1] The National Fire Plan is a framework for coordinating the firefighting and risk reduction efforts of federal agencies, states, local governments, and tribes. Details are available at . Site accessed June 28, 2005. The Healthy Forests Initiative is a program operated by the USFS and BLM. Key elements of the HFI were signed into law as the Healthy Forests Restoration Act (H.R. 1904 December 3rd, 2000) (HFRA). Details are available at . Site accessed June 28, 2005

[2] For the purposes of this research, the study area is defined as the southern terminus of the Rocky Mountains in northern New Mexico, which encompasses Bernalillo, Cibola, Colfax, Los Alamos, McKinley, Mora, Rio Arriba, San Miguel, Sandoval, Santa Fe, Taos and Torrance counties. Areas of northern New Mexico not included in the study do not contain substantial areas of forestation. DA detailed maps of the areas of interest areis provided in figures 1 through 3. The study area encompasses 24.8 million acres including all of the Carson and Santa Fe National Forests and a significant portion of the Cibola National Forest (Haskins 2004). According to the U.S. Census (2000), the non-metropolitan population of this region is 310,000.

[3] For a map detailing areas of persistent poverty in the U.S., please see the U.S. Department of Agriculture, Economic Research Service web-based briefing room report on, “rural income, poverty, and welfare: rural poverty”: Site accessed 9/19/05

[4] Northern New Mexico has been a “stronghold” of Hispano culture since colonization in 1598 (Raish 2000). Much of the rural Hispano population lives in dozens of small villages scattered along the forested lower elevations of the Sangre de Cristo, Jemez, and San Juan Mountain ranges. The Pueblo Indians of northern New Mexico have occupied their current locations since at least 1000 A.D. when they migrated away from the plains to occupy upland forest areas that promised a greater abundance of wild foods and water (Dutton 1983).

[5] Additionally, the federal Community Forestry Restoration Act (CFRA) of 2000 establishes cost share grants to stakeholders for experimental forest restoration projects in New Mexico that are designed through a collaborative process and allocates $5 million annually for these purposes. Both the HFRA and CFRA provide federal program managers with the authority, flexibility and funding to configure NFP/HFI projects in northern New Mexico in a manner that simultaneously reduces high fire risk, benefits traditional cultural uses, and alleviates rural poverty.

[6] H.R. 1904, Title III, Section 303.

[7] H.R. 1904, Title I, Section 102, Title III, generally; H.R. 2389

[8] For a detailed description of NFP and HFI grant programs in the Southwest, see . Site accessed June 28, 2005.

[9] To obtain digital information about the location of NFP projects in New Mexico, as well as links to a database providing brief summary statistics about these projects please go to: . Site accessed June 28, 2005.

[10] A more detailed description of this project is provided in Appendix A. It should be noted that, fire risk is depicted for all vegetation types; with fire risk generally increasing as vegetation transitions from lowland scrub to higher elevation woodlands and forests (Schmidt et al. 2000).

[11] Many of the northern New Mexico households that depend on firewood are located in traditional communities that also rely on many other small diameter wood products such as posts, poles, vigas, and latillas for traditional building materials (Gunderson 2001).

[12] Here, we are considering risk in the statistical/economic sense, where Expected Loss = PF*LOSS, where PF is the probability of a damaging wildfire and LOSS is the potential loss of life and property. More specifically, PF could be further broken down (e.g., see Yoder, 2004) into the joint probability PI*PE, where PI is the probability of ignition in the larger area and PE is the probability that a fire “escapes” into a potentially catastrophic fire in the WUI. Further, LOSS could be broken down into the total value of the exposed capital assets multiplied by the conditional probability of loss (e.g., proportion of assets lost, given that a fire burns through an area). Such distinctions may be important in differentiating the potential impacts of mitigation, suppression, and insurance.

[13] The risk variable, PHIGH, is derived from the Coarse Scale Fire Map. Data on the variables VAL and OHSG (and DEN) was taken from the 2000 U.S. Census.

[14] Recall that there were 455 projects identified in the study area. Expenditures were recorded for 373 of the projects.

[15] Of the 373 observations, 115 were described as broadcast burn, hand pile burn or machine pile burn. These burns were funded with $4.6 million of the $30.2 million dollars of the total funds spent on the NFP/HFI projects in northern New Mexico. Thinning described 165 of the 373 observations. These projects used $17.5 million of the total. The rest of the projects, which include, hand pile, chipping, lop & scatter, machine pile, mastication & mowing, monitoring and preparation make up the remaining 93 observations and account for $8.1 million of the total. The overall goal of these activities is not to eliminate fires but to insure that when fires do occur, that they are of a size and intensity commensurate with long term forest ecosystem health and conducive to effective suppression and management efforts (USFS, 2001).

[16] Models were estimated using population-averaged random effects clustered on the fiscal year.

[17] The estimated coefficients on LNVAL were significant at the 0.01 levels in the models using the disaggregated data; and significant at the 0.05 and 0.01 levels in the models using the aggregated data.

[18] To wit, in a related presentation on wildfire risk exposure analysis in Eldorado County by the California Department of Forestry (Fire and Resource Assessment Program) they simply incorporate the number of homes in a given geographical area or zone in their conditional loss definition, and avoid any inclusion of housing values. See: . Site accessed June 28, 2005. However, in the context of a revealed preference analysis, to do so would skirt the empirical question on whether or not housing values are a significant determinant of risk reduction expenditures.

[19] For discussion of New Mexico examples, and related issues within the broader Southwestern regional context, see the Southwest Community Forest Caucus website, . Site accessed June 28, 2005.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download