Ā«Toward More Effective Endangered Species Regulation By Jacob P. Byl Dissertation Submitted to the Faculty of the Graduate School of Vanderbilt ...Ā»
foraging habitat for each RCW colony (U.S. FWS 1985). This can deny the landowner of 100 acres a timber harvest worth $200,000. The land use restrictions imposed by the FWS can continue indefinitely because RCW colonies are very long-lived and the colonies can be passed on to subsequent generations of birds. The agency previously tried to protect 300 acres per colony, but has more recently scaled it back to 100 acres per colony to encourage landowner cooperation by lowering the threat of regulatory restrictions (U.S. FWS 2010).
harbor program in the Sandhills region of North Carolina in 1995. The program was intended to encourage cooperative relationships with landowners. The FWS desired to shed the adversarial approach that required it to expend significant resources on enforcement and resulted in popular and political backlashes to land-use restrictions (U.S.
FWS 2010). To enter a safe-harbor agreement, landowners promise to maintain some suitable RCW habitat in exchange for a promise from the FWS not to impose more onerous restrictions if RCWs move onto the property in the future. The land-use restrictions in the agreements depend largely on the baseline of endangered species Although there are listed plants in the pine forests of North Carolina, such as the sweet pitcher plant, the ESA does not give FWS the same broad jurisdiction to protect plants as it does to protect animals, so private landowners are not threatened with the same onerous land-use restrictions.
residing on the property at the time of negotiation, so an agreement with a baseline of zero RCWs will have fewer obligations than an agreement with a baseline of three RCW colonies. Landowners can unilaterally leave these voluntary agreements at any time, although once they leave the landowner can face other ESA regulations just as landowners who had never entered agreements. Safe-harbor agreements, and some other contracts between FWS and landowners, come with āassurancesā from the FWS that the landowner will not be required to comply with new regulations in the future. Because of this regulatory certainty, safe-harbor agreements can be attractive to landowners as insurance policies for regulatory outcomes.
harbor agreements, the threat of regulation may still have unintended consequences if landowners are destroying potential habitat to prevent the woodpeckers from moving onto their property in the first place. The next section lays out a framework for how a landowner with nearby RCWs may deal with the threat of FWS regulation.
depicted in Figure 1.2 One decision is whether to harvest for timber or let the trees grow another time period. The landowner discounts the future with the real interest rate r and maximizes the expected present value of timber harvests, both the trees currently standing and the trees that can be planted there in the future. For simplicity, the landownerās harvest decision is modeled as binary: either harvest all of the trees (clearThis model is similar to the two-stage decision tree in Lueck and Michael (2003), augmented with optimal-harvest variables and a stage involving safe-harbor agreements.
cut) or let them all grow another period.3 The harvest value of timber on the property at time t is Ļ t. The increase in timber value from time t-1 to time t is given by Ļ t, which measures the marginal benefit of letting the trees grow another period. Young pine trees
are used for saw timber, with saw timber commanding the highest price and pulp the lowest price. As such, Ļ t is large when an extra period of growth allows the trees to enter a higher price category for a more lucrative use.
when there are standing trees on the property, there is a chance that endangered woodpeckers will move in and settle a colony there. If the FWS discovers the presence of the woodpecker colony, it invokes jurisdiction under the ESA to regulate the entire property. When the FWS regulates property, it forbids the harvest of any pine trees for the foreseeable future. In the model, Ī³ is the probability that woodpeckers move in and the FWS discovers their presence and invokes jurisdiction to regulate.
The other decision that landowners make is whether to enter into a safe-harbor agreement or not. If landowners enter into safe-harbor agreements, they limit the amount they can harvest in any period by a factor of ķ µą·°, with 0 ķ µą·°100. In exchange for limiting their harvest, landowners get assurance from regulators that the landowner will be able to continue harvesting even if woodpeckers do move onto the property, so Ī³ becomes immaterial to them if they intend to stay in the agreement. For this paper, ķ µą·° is assumed to be smaller, meaning there are fewer restrictions on harvest, forā¬properties that are in the FWS target zone for landowner incentive programs.
This simple model generates predictions of how changes in parameters will impact the probability of harvest during a given time period by seeing how the attractiveness of different routes in Figure 1 change relative to each other. For example, increases in Ī³, the probability that the FWS regulates because of woodpecker presence, would tend to induce more landowners to harvest in the early period instead of waiting.
ā¬ This occurs because a high Ī³ decreases the right-hand side of (1) but does not change the left-hand side. Intuitively, the greater threat of a logging ban decreases the expected value of future harvest, ā¬ causing landowners to opt for the relative safety of harvesting now, even if that means foregoing increased timber value. Variations in Ī³ can allow for empirical tests of this hypothesis. Property that has many endangered woodpeckers ā¬ nearby is more likely to have woodpeckers move in, so these landowners have higher Ī³
thinning of trees and controlled burns, are more likely to have woodpeckers move in because their land is more attractive to the birds, increasing the Ī³ for their properties.
Prediction 2 is that landowners who are near woodpeckers are less likely to manage land
Landowners who have entered safe-harbor agreements have less incentive to destroy woodpecker habitat and may be more likely to improve habitat for wildlife.
Prediction 3 is that landowners who have lower ķ µą·°, such as those who are actively recruited by the FWS for the program, are more likely to be in safe-harbor agreements and, consequently, less likely to destroy habitat: ķ µą· ķ µą·ķ µą·ķ µą· ķ µą·ķ µą· ķ µą·ķ µą· ķ µą· ķ µą·° 0. Prediction 4 is that
these landowners with lower ķ µą·° are also more likely to improve wildlife habitat:
ķ µą· ķ µą·ķ µą· ķ µą· ķ µą· ķ µą· ķ µą·ķ µą· ķ µą· ķ µą·° 0.
VI. Data on Forest Plots and Woodpeckers Forest plot data come from the U.S. Forest Service, which conducts the Forest Industry and Analysis (FIA) survey annually. The Forest Service uses a random search algorithm to select approximately 500 active forestry sites per county, resulting in over 30,000 sites in the Piedmont and coastal regions of North Carolina. To conduct the survey, Forest Service agents visit the property and count and measure trees, take core samples to determine tree age, and sample soil to determine site productivity. FIA data include variables for whether trees have been harvested from the site in the past five years and whether the site has been managed to improve wildlife habitat. Land is coded as managed to improve habitat for wildlife if there was strategic thinning or controlled burns for the purpose of increasing quality of habitat, usually for things like deer and other game animals. Sites in the analysis were surveyed on a rolling basis between 1982 and 2013. Only FIA sites that are predominately pine, meaning 75% pine and higher, are retained for the analysis.
Data on RCWs come from the North Carolina Department of Environment and Natural Resources, which maintains a database of all known RCW colonies in the state.
There are approximately 1,000 colonies spread across the Piedmont and coastal regions of North Carolina, with a large cluster in the Sandhills area. Forest plots and RCW colonies in the Piedmont and coastal regions of North Carolina are plotted in Figure 2.
Data on safe-harbor agreements were obtained from the FWS through a Freedom of Information Act Request. These data include location of properties that are or have been in safe-harbor agreements, the baseline of RCWs on the property, and the current status of the agreements. There are 145 safe-harbor agreements covering approximately 91,000 acres. Forest plots, RCW colonies, and safe-harbor agreements in the Sandhills and surrounding counties of North Carolina are plotted in Figure 3.
Controlling for market forces requires information on timber prices. Timber Mart South, a nonprofit affiliated with the University of Georgia, maintains timber price trends for the Southeast including North Carolina. These data are paired with information on tree diameter and tonnage in the FIA survey to construct variables that measure the value of timber and its growth. This can control for when the landowner would optimally harvest without woodpeckers. Tree and price data are used to construct a variable for the starting value of the standing timber on the site and a variable for the value of letting the trees grow another year.
After combining FIA site, RCW, and price information, the resulting data are a series of cross sections of forest plots. Summary statistics of the data are presented in Table 1. Panel A includes FIA sites from the Piedmont and coastal regions of North Carolina that were sampled between 2001 and 2013. In the five years prior to the survey, 18% of sites were harvested and 5% of sites were managed to improve wildlife habitat.
The average forest plot has four woodpeckers within ten miles. The starting value of trees averages $290 per acre and the value of letting those trees grow an additional year averages $26 per acre. The majority of sites (69%) have loblolly pine on them. 41% of sites have water features on or near them, 10% have a steep slope, 22% are classified as having highly productive growing conditions, and 1% are more than a mile from an improved road. About a quarter of sites (24%) are lightly forested, meaning that canopy cover does not exceed 75% of the land area.
For empirical tests of the effectiveness of the safe-harbor program in the Sandhills area of North Carolina, the sample is narrowed to the twelve counties in and around the Sandhills area in the southern portion of the state. FIA data going back to 1982 are used to measure the effect of the program with observations from both before and after the programās start in 1995. Summary statistics of two of the key variables of interest for this sample are laid out in Panel B of Table 1.
Table 2 presents summary statistics of sites that have at least one RCW colony within ten miles and sites that have no RCW colonies within ten miles. Although some variables like probability of harvest and improvement are similar across the groups, other variables differ. Trees on sites near RCWs tend to be worth less, the sites are more likely to have water nearby, and are less likely to have steep slopes. These sites near RCWs also tend to have less productive growing conditions for trees. An ideal dataset would include sites that are similar in all aspects other than the presence of RCWs, but without random distribution of RCWs it is not surprising that there are systematic differences. Although regression analysis can control for observed differences in the forest plots, there are likely to be some differences that are not picked up in the forestry data. For example, landownersā tastes for the amenity value of forest and wildlife may differ across areas. A preference for wildlife could influence the probability of RCWs living nearby if landowners encourage RCW habitat, such as by having periodic controlled burns. The same tastes for wildlife would also impact the timber harvest decision of landowners, possibly leading to a biased estimate of the impact of RCWs on the timber harvest decision.
One way to try to deal with the heterogeneity of forest plots is to compare potential RCW habitat with plots that have similar forestry properties but are not at risk of RCWs moving in. Sites that are predominately pine but have hardwoods mixed in are not suitable habitat, as discussed in Section III. The probability of harvest differs substantially across sites depending on proximity to RCWs and mixture of tree types, as presented in Table 3. Pure pine sites that are near RCWs have a high probability of harvest at close to 20%. Pure pine sites that are not near RCWs have a 17% chance of being harvested. Mixed sites near RCW have a 15% chance of being harvested, while 13% of mixed sites not near RCW are harvested. Because RCW should not impact harvests on mixed sites that are not suitable habitats for the birds, the difference in harvest rates between mixed sites that are near RCW and those that are not is evidence that there might be some systematic differences in forest plots. Although the summary statistics in Table 3 suggest a relationship between RCW and tree harvests, regression analysis allows for more robust evidence with controls for confounding factors.
while models to test the effectiveness of the safe-harbor program use data from a longer time period but smaller geographic area. There are three pertinent econometric models to test Predictions 1 and 2 from the conceptual framework. For these models, I use the more recent FIA data on sites that were surveyed after 2000 to get a sense of the recent landowner response to threats of ESA regulation.4 A probit model similar to that used by Lueck and Michael (2003) tests whether nearby RCW colonies increase the probability of harvest in a given time period.
years preceding the survey. In the model, the probability of Harvest for a plot follows a standard normal distribution with the cumulative distribution function Š¤ with the following explanatory variables. The coefficient Īø measures the impact of woodpeckers