|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 ECONOMICS OF FOREST BIOMASS BASED BIOENERGY IN THE SOUTHERN UNITED STATES By ANDRES IGNACIO SUSAETA LARRAIN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIR EMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2 2009 Andres Ignacio Susaeta Larrain
3 To my family : Toyi, Ignacio, Cristobal, D oquinha, Vivi a nd Juan cr i I know that someda y we all will be together again .. T o my own private Gainesville experience
4 ACKNOWLEDGMENTS First and foremost I would like to thank my advisor Dr. Janaki Alavalapati for his intellectual support and constructive criticism throughout my doc toral program. I am also grateful to the rest of my committee members Dr. Douglas Carter, Dr. Sherry Larkin, Dr. Laila Racevskis and Dr. Siva Srinivasan, f or their guidance and encouragement. Special thanks go to Dr. Alfonso Flores Lagunes for sharing his expertise in econometrics. Than ks go to Puneet Dwivedi, Pankaj Lal, Ming Yuan Huang, Tyler Nesbit, Sid Kukrety and Rao Matta for their views and suggestions in my research. Thanks also go to the United States Department of Agriculture (USDA) and the Depar tment of Energy (DOE) for funding my doctoral studies. Probably there are no enough words to acknowledge all the people that were involved in this tremendous challenge. I want to thank several people for their support and friendship: Carlos, Claudia, Crist obal, Cynnamon, Pato Dawn, Fede, Jose, Bernardo, Francisco, David, Sebastian and Fernando. I also have to acknowledge my football (the real one, with a round ball) teammates for sharing the love and joy for this sport especially those ones that cheer for My friends from the Master group of the Gator Swim Club also deserve recognition for stimulating to train with confidence and passion. I am grateful to a very special person, Raissa, for her tenderness and suppor t, and for addin g a different way to understand life. Finally, I w ant to thank a person who always encouraged me to go further and never give up Thanks, mama.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ........... 4 LIST OF TABLES ................................ ................................ ................................ ...................... 7 LIST OF FIGURES ................................ ................................ ................................ .................... 8 LIST OF ABBREVIATIONS ................................ ................................ ................................ ...... 9 ABSTRACT ................................ ................................ ................................ ............................. 11 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ............. 13 Overview ................................ ................................ ................................ ........................... 13 United States Policies for Biofuels ................................ ................................ ..................... 14 Ethanol Production: Corn versus Cellulosic Biomass ................................ ......................... 15 Current Situation of Nonindustrial Private Forest Lando wners ................................ ........... 16 Previous Research on Economics of Forest Biomass Based Energy ................................ .... 17 Research Plan ................................ ................................ ................................ ..................... 20 2 MODELING IMPACTS OF BIOENERGY MARKETS ON NONINDUSTRIAL PRIVATE FOREST MANAGEMENT ................................ ................................ .............. 22 Introduction ................................ ................................ ................................ ........................ 22 Model Specific ation ................................ ................................ ................................ ........... 23 The Model Application to Slash Pine Stands ................................ ................................ ...... 30 Results and Discussion ................................ ................................ ................................ ....... 32 Conclusions ................................ ................................ ................................ ........................ 36 3 ASSESING PUBLIC PREFERENCES FOR FOREST BIOMASS BASED BIOENERGY ................................ ................................ ................................ .................... 45 Introduction ................................ ................................ ................................ ........................ 45 Study Design and Data Collection ................................ ................................ ...................... 46 Contingent Valuation Questionnaire ................................ ................................ ............ 46 Welfare Estimates ................................ ................................ ................................ ....... 54 Results and Discussion ................................ ................................ ................................ ....... 54 Attributes ................................ ................................ ................................ .................... 55 Socioeconomic Variables ................................ ................................ ............................ 56 Willingness to Pay ................................ ................................ ................................ ....... 59 Conclusions ................................ ................................ ................................ ........................ 61
6 4 MODELING EFFECTS OF BIOENERGY MARKETS ON TRAD ITIONAL FOREST PRODUCT SECTORS ................................ ................................ ................................ ....... 75 Introduction ................................ ................................ ................................ ........................ 75 Model and Econometric Specification ................................ ................................ ................ 77 Data Construction ................................ ................................ ................................ .............. 84 Traditional Forest Product Sector ................................ ................................ ................ 84 Biomass for Bioenergy ................................ ................................ ................................ 86 Results and Discussion ................................ ................................ ................................ ....... 87 Coefficients and Elasticities of the Residues Only Biomass (ROB) Model .................. 89 Calibrat ion of the Residues Pulpwood Biomass (RPB) Model ................................ ..... 91 Policy Simulation ................................ ................................ ................................ ........ 93 Conclusions ................................ ................................ ................................ ........................ 95 5 SUMMARY AND CONCLUSIONS ................................ ................................ ............... 107 Introduction ................................ ................................ ................................ ...................... 107 Results from Modeling Impacts of Bioenergy Markets on Nonindustrial Priv ate Forest Management ................................ ................................ ................................ ................. 108 Results from Assessing Public Preferences for Forest Biomass based Bioenergy .............. 109 Results from Modeling Effec ts of Bioenergy Markets on Traditional Forest Product Sectors ................................ ................................ ................................ .......................... 110 Policy Implications and Further Research ................................ ................................ ......... 111 APPENDIX A LIST OF VAR IABLES ................................ ................................ ................................ .... 115 B SURVEY INSTRUMENT ................................ ................................ ............................... 118 C COMPLETE HISTORICAL DATA SET ................................ ................................ ......... 121 D V OLUMES AND PRICES OF SAWTIMBER, PULPWOOD AND BIOMASS FOR BIOENERGY 1970 2006 ................................ ................................ ................................ 124 LIST OF REFERENCES ................................ ................................ ................................ ........ 126 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ... 140
7 LIST OF TABLES Table page 2 1 LEV ($ acre 1 ) for the three scenarios at different levels of risk and salvage ................... 39 2 2 LEV ($ acre 1 ) for the thinning scenario for bioenergy for different bioenergy prices ..... 40 3 1 Description of the attributes and levels ................................ ................................ .......... 64 3 2 Description of the choice situation ................................ ................................ ................. 65 3 3 Socioeconomic variables ................................ ................................ ............................... 66 3 4 Descriptive statistics for socioe conomic variables, E10 ................................ .................. 67 3 5 Descriptive statistics for socioeconomic variables, E85 sample ................................ ...... 68 3 6 Probabilities of using biofuels fo r level of environmental attributes ............................... 69 3 7 Probit model results for E10 sample ................................ ................................ .............. 70 3 8 Probit model results for E85 sample ................................ ................................ .............. 71 3 9 WTP ($ gallon 1 ) for biofuels at state level ................................ ................................ ..... 72 4 1 Descriptive statistics of the data set ................................ ................................ ................ 98 4 2 Rank and redundancy tests for all instruments ................................ ............................... 99 4 3 Coefficients, standard errors, p values and serial correlation of the system of equations ................................ ................................ ................................ ..................... 100 4 4 Rank, overidentifying restriction, model comparison, symmetry and contemporaneous correlation tests ................................ ................................ .............. 101 4 5 Estimates and standard errors (SE) of the short r un supply and demand elasticities of the ROB model ................................ ................................ ................................ ............ 102 4 6 Coefficients, standard errors, p values and serial correlation of the RPB model ............ 103 4 7 Estimates and standard errors (SE) of the short run supply and demand elasticities of the RPB model ................................ ................................ ................................ ............. 104 4 8 Baseline and policy scenario results for ROB and RPB models ................................ ..... 105 4 9 Total timber sales values for ROB and RPB models (million US$) ............................... 106 C 1 Yearly dataset ................................ ................................ ................................ .............. 121
8 LIST OF FIGURES Figure page 2 1 LEVs for both the no thinning scenario and thinning scenario for bioenergy when the salvageable portion is zero ................................ ................................ ....................... 41 2 3 LEVs for both the no thinning scenario and thinning scenario for pulpwood when the salvageable portion is zero. ................................ ................................ ............................ 43 2 4 LEVs for both the no thinning scenario and thinning scenario f or pulpwood when the salvageable portion is 0.8. ................................ ................................ .............................. 44 3 1 Percentage of yes responses for E10 ................................ ................................ .............. 73 3 2 Percentage of yes responses for E85 ................................ ................................ .............. 74 D 1 Volume (Million m 3 ) of sawtimber, pulpwood and biomass for bioenergy 1970 2006 124 D 2 Stumpage Price ($ m 3 ) of sawtimber, pu lpwood and biomass for bioenergy 1970 2006 ................................ ................................ ................................ ............................ 125
9 LIST OF ABBREVIATIONS ACES American Clean Energy Security Act AR Arkansas BS Black Scholes CPS Consumer Population Survey CV Contingent Valuation DW Durbin Watson EFI GTM E uropean Forest Institute Global Trade Model EIA Energy Information Agency EHF Enzymatic Hydrolosis Fermentation ENFA European Non Food and Agriculture E10 Ethanol 10 E85 Ethanol 85 FASOM Forest and Agriculture Sector Optimization Model FL Florida FR Fusif orm Rust GAPPS Georgia Pine Plantation Simulator GF Gasification Fermentation GFPM Global Forest Products Model GHG Green h ouse Gas GJ Gigajoule GLPF Generalized Leontief profit function IRR Internal Rate of Return KN Knowledge Networks LEV Land Expectatio n Value
10 RES Renewable Electricity Standards RFS Renewable Fuel Standards NIPF Nonindustrial Private Forest Landowner NPV Net Present Value OLS Ordinary Least Squares OPEC Organization of the Petroleum Exporting Countries SPB Southern Pine Beetle TE Total E xpenditures TMS Timber Mart South 3SLS Three Stage Least Squares 2SLS Three Stage Least Squares TWh Terawatt hour U.S United States VA Virginia WTP Willingness to Pay
11 Abstract of Dissertation Presented to the Graduate School of the Universi ty of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ECONOMICS OF FOREST BIOMASS BASED BIOENERGY IN THE SOUTHERN UNITED STATES By Andres Ignacio Susaeta Larrain December 2009 Chair: Janaki Alav alapati Major: Forest Resources and Conservation Concerns about greenhouse gas (GHG) emission and dependency on foreign oil have prompted policy makers to develo p environmentally friendly sources of energy. Forest biomass, a carbon dioxide (CO 2 ) neutral s ource of energy, is thought to provide higher energy ratios, better environmental benefits regarding GHG reduction and be potentially cost competitive compared to agricultural crops. Further more the use of forest biomass for energy can decrease the risk o f wildfires and increase the profitability of forestlands. This dissertation explores the economics of using forest biomass for energy in the southern United States. First, the effects of bioenergy markets on nonindustrial private forest management are mod eled using a Black Scholes formula integrated with a modified Hartman model. This combined model assumes that stumpage prices are stochastic and forests are affected by catastrophic disturbances su ch as wildfires pest outbreaks. Second public perceptions and views about using forest biomass biofuels are analyzed A contingent valuation survey was conducted in Arkansas, Florida and Virginia to estimate the willingn ess to pay for biofuels to assess the effect of socioeconomic variables on the choice for rene wable energy. Third, the effe cts of bioenergy markets on traditional forest product sectors are estimated Assuming a Leontief profit
12 function, both supp ly and demand functions for different components of the forest sector are derived. In addition, the eff ects of an increase demand for biomass for bioenergy on the sawtimber and pulpwood markets were simulated Results of this research show that bioenergy markets would increase the profitability of forest stands. Forestl and values are shown to be greater whe n thinning material is us ed for bioenergy production relative to pulpwood production Further more, revenues are shown to increase as the rate of catastrophic disturbances is reduced On average, bioenergy production increase d the forest land value around 11 .6% compared to pulpwood production As bioenergy markets evolve it is expected that profitability of forestlands will increase. Public perception study indicate s that people tend to pay for biofuels to realize environmental and social benefits Although people show heterogeneous preferences in terms of environmental attributes, their willingness to pay is greater when higher reduction of CO 2 and improvements of biodiversity are offered. For an ethanol blend of 10% (E10), people were willing to pay an extr a $0.56 gallon 1 $0.58 gallon 1 and $0.50 gallon 1 in Arkansas, Florida and Virginia, respectively. For an ethanol blend of 85% (E85), the extra willingness to pay was $0.82 gallon 1 $1.17 gallon 1 and $1.06 gallon 1 for the same states. Finally, an i nc rease in the demand for biomass for bioenergy causes an increase in the price of biomass for bioenergy pulpwood and sawtimber and t he quantity of pulpwood and sawtimber are shown to be decreased. Price of biomass for bioenergy, pulpwood and sawtimber in creased 52%, 104% and 6.5%, respectively. Quantity of pulpwood and sawtimber decreased 20% and 11%, respectively. On the other hand, forest landowners and bioenergy sector s would benefit from bioenergy production while pulp and sawmill sectors are shown to contract with an increase in the demand for biomass for bioenergy
13 CHAPTER 1 INTRODUCTION Overview Currently, around 26% of the total energy used in the United States (U.S.) is imported, and 84% of the imports are represented by crude oil and petroleum p roducts (EIA, 2009). Further more 46% of the petroleum imports come from the Organization of the Petroleum Exporting Countries (OPEC) (EIA, 2009). The transportation sector was the largest consuming sector of petroleum in 2008, standing at 13.7 million bar rels day 1 (70% of all petroleum used), and motor gasoline was the single largest petroleum product consumed (64% of all petroleum consumption) (EIA, 2009). On the other hand, the continued increase in anthropogenic emissions of greenhouse gases (GHG), pre dominantly carbon dioxide (CO 2 ), is expected to have dramatic impacts on climate, such as glacier shrinkage, worldwide sea level rise, and threats to biodiversity. This strong dependency on foreign markets, particularly from volatile Middle Eastern count ries, together with concerns about the effects of greenhouse gas (GHG) emissions, has prompted policy makers to find alternative renewable energy sources. Currently, biofuels represent 19% of the total consumption from renewable energy sources, which accou nt for only 7% of the total energy consumption in the U.S. Further, 53% of renewable energy consumption comes from biomass (EIA, 2009). In spite of the low share of renewable energy in total energy consumption, the U.S has the potential to displace 30% of current petroleum consumption with biofuels by 2030, providing a sustainable supply of biomass of more than 1 b illion dry tons (Patzek, 2005).
14 United States Policies for Biofuels Liquid biofuels such as ethanol have been strongly encouraged since the late 1970s, although the first policies were energy security oriented. The Energy Tax Act of 1978 provided a $0.40 gallon 1 exemption from the federal gasoline excise tax for a blend with at least 10% ethanol, which increased to $0.60 gallon 1 when the Tax Ref orm Act of 1984 was enacted. It was reduced to the current level of $0.51 gallon 1 by the 1998 Transportation Equity Act of the 21st Century T he American Jobs Creation Act of 2004 replaced the excise tax exemption with a volumetric ethanol excise tax cred it of $0.51 gallon 1 until 2010 Further, a tariff of $0.54 gallon 1 was imposed on imported ethanol under the purview of the Omnibus Reconciliation Tax Act of 1980 to stimulate domestic industry. Several other federal policies have been adopted to addres s environmental concerns about the use of fossil fuels. For example, the Clean Air Act amendment of 1990 established an oxygenated gasoline program to create a new, balanced strategy to address the problem of urban smog. The Energy Policy Act of 1992 exten ded the tax exemption to include blends of 7.7% and 5.7% ethanol. The Energy Policy Act of 2005 introduced the concept of a Renewable Fuel Standard (RFS) requiring that a minimum amount of renewable fuel production must be met, starting with 4 billion gall ons in 2006 and achieving 7.5 billion gallons in 2012. After 2012, renewable fuel and gasoline production will grow at the same rate (Duffield and Collins 2006). The Energy Independence and Security Act of 2007 requires a RFS of 36 billion gallons by 2022, of which 21 billion gallons must come from cellulosic products. The 2002 and 2008 Farm Bill biobased products to support development of biorefineries and assistance to farmers and ranchers
15 (RES) of 6% in 2012, which can be gradually incre ased to 25% by 2025. Ethanol Production: Corn versus Cellulosic Biomass The U.S. became the main producer of ethanol worldwide in 2006 (Hettinga et al., 2009), and around 95% of it came from corn (Solomon et al., 2007). Corn based ethanol production has be en criticized for its effect on food security, and consequently, the increasing prices of related products such as milk, meat, and eggs (Pimentel and Patzek, 2005). Additionally, several environmental impacts and low (even negative) net energy balance rati os have been claimed (Pimentel and Patzek, 2005; Hill et al., 2006; Solomon et al., 2007). However, it has also been argued that food prices will remain high along with higher energy prices ( Renewable Fuels Association [RFA], 2008 ). Urbanchuk ( 2007 ) argues that the increase in food prices due to corn prices is expected to have half the impact of the same percentage increase in energy prices on the CPI for food. Further, as ethanol production expands and new technologies and ethanol feedstocks materialize, a ny increase in food prices will be offset by lower energy prices ( Evans 1998 ). Cellulosic biomass 1 for ethanol production, on the other hand, has a higher net energy balance ratio, provides more environmental benefits in term of GHG reduction, and is pote ntially cost competitive compared to food based biofuels (Hill et al., 2006). In addition, the use of forest biomass for cellulosic ethanol production could make non commercial components such as harvest residues marketable and reduce flammable materials l eft in the field as waste. Such a development would also reduce the risk of wildfires (Neary and Zieroth, 2007; Polagye et al., 1 Cellulosic feedstock is comprised of cellulose, hemicellulose, lignin, and solvent extractives (D wivedi et al., 2009). Percent range of constituents in wood differs from other cellulosic feedstocks such as agricultural waste, grass and municipal solid waste (Olsson and Hagerdal, 1996).
16 2007 ; Nicholls et al., 2009 ) and pest outbreaks (Roser et al., 2006; Evans, 2007) besides improving the profitability of forest landowners and stimulating employment (Faaij and Domac, 2006). On the other hand, the use of renewable energy coming from woody biomass, assuming no land use changes, ensures a neutral carbon dioxide source of energy (Richardson et al., 2002) representing a diversification strategy to reduce dependence on foreign oil. However, careful considerations must be taken when managing forests intensively for bioenergy. Without appropriate planning and incorporating optimal harvest systems and maintaining the conne ctivity of habitat networks, the production of bioenergy could lead to loss of biodiversity (Cook et al., 1991). Extensively degraded lands can be considered for bioenergy plantations trees or other energy crops to reduce erosion, restore ecosystems, an d provide shelter to communities (Riso, 2003), but these benefits would only be realized under clear land use regulations, especially in places where forests are at risk of conversion to other land uses (Worldwatch Institute, 2007) Current Situation of No nindustrial Private Forest Landowners The southern United States (U.S.) contains 214 million acres of forestland. Out of this, 204 million acres (95.3%) are categorized as timberland. About 145 million acres of timberland (71.8%) are owned by nonindustrial private forest (NIPF) landowners, also called family forests. NIPF landowners contribute 68% (7.7 million cubic feet) of annual growth and 68% of annual removals ( Smith et al ., 2009). Thus, private forests, and particularly NIPF landowners, are significan t contributors to the southern U.S. forest sector. Currently, southeastern U.S. NIPF landowners managing slash pine ( Pinus elliotti ) plantations face several challenges, including catastrophic risk events, increased offshore competition (Paun and Jackson, 2000), and low pulpwood prices. The combined effect of fire suppression, lack of prescribed burning, and high planting densities has been that extensive areas
17 are overstocked and are susceptible to wildfires and pest attacks ( Graham et a l., 1999; Le V an Gr een and Livingstone, 2003; Polagye et al 2007). For example, the average annual number of wildfires in the state of Florida is as high as 5,550 incidents burning 220,000 acres each year ( Florida Department of Community Affairs and Florida Department of A g riculture and Consumer Services Division of Forestry 2004). In order to prevent and manage wildfires, $58 million was allocated for fiscal year 2007 08. Pest outbreaks such as fusiform rust ( Cronartium quercuum [ Berk .] Miy abe ex Shirai f. sp. fusiforme ) (FR) and southern pine beetle ( Dendroctonus frontalis Zimmermann) (SPB) have caused significant damage to NIPF landowners. For example, FR caused an annual economic loss of $35 million in five southern states and $8 million in Florida (Schmidt, 1998); SPB caused southwide damages of $1.5 billion between 1970 and 1996 ( Price et al., 1998). Foreign competition is also of concern. Costs for producing fiber in the southern hemisphere are lower compared to the U.S. due to lower labor and other input costs. W ear et al. (2007) found delivered cost differentials of 24% and 27% in Brazil and Chile as compared to the southern U.S., respectively. Further, low pulpwood prices have resulted from a contraction of domestic pulp and paper demand, as evidenced by declin ing pulp mill capacity and expanded use of recycled material. S oftwood pulpwood prices in the southern U.S. decreased by 36% between 1998 and 2005 (Timber Mart South, 2007), while p ulping capacity decreased by 11% between 1997 and 2006 (Johnson et al., 200 8). Timber inventory reductions, however, have not kept pace with reduced demand, further exacerbating the problem. Previous Research on Economics of Forest Biomass Based Energy Several authors have investigated the potential availability of forest based biomass for energy Smeets and Faaij (2006) explored the biomass potential from forestry for the year 2050. The authors estimated that forest s can become a major source of bioenergy without threatening
18 the supply of traditional forest product s and without deforestation. However, regional shortages of industrial roundwood might occur in South Asia the Middle East and North Africa. Supporting this, in a previous study conducted by Yamamoto et al. (2001), the authors approached bioenergy potential with a mu ltiregional global land use energy model. They concluded that the total area of global forests will remain stable although mature forest area will decrease because of increased population growth and demand for biomass in developing countries especially in Central Asia, the Middle East North Africa and South Asia. Perlack et al. (2005) predicted a 30% replacement of current U.S petroleum consumption with biofuels by 2030 based on agricultural and forestland biomass potential around 1.3 billion dry tons per year 368 million dry tons would come from forestlands, including 52 million tons of fuelwood, 145 million tons from wood processing mills 47 million tons from urban wood residues, 64 million tons from logging and site clearing operations and 60 mill ion tons from fuel treatment operations. Galik et al. (2009) showed that in spite of large amount s of forest residues within the states of North Carolina, South Carolina and Virginia they would not be sufficient to meet the long term biomass electricity production requirements imposed by a hypothetical national R enewable Portfolio Standard (RPS) and RFS. Walsh (1998) determined a marginal price for bioenergy crops at the farm gate of $24 dry ton 1 and $ 26 dry ton 1 at national level for 2010 and 2020 con sidering total bioenergy crops quantities of 50 million dry ton s and 110 million dry ton s, respectively. It was also assumed that two thirds of total biomass quantity w ould be supplied by switchgrass and one third w ould be supplied by hybrid poplar and hyb rid willow in 2010. By 20 20 switchgrass would provide half of the biomass quantity and the other half w ould be supplied by woody crops. Gan and Smith (2006) explore d the availability of logging resid u es and their potential for electricity production
19 in th e U.S. T hey calculated that recoverable residues might generate 67.5 TWh electricity per year displacing 17.6 million tons of carbon emitted from coal generated electricity at a cost ranging from US$60 ton 1 to $ 80 ton of carbon The economics of cellulosic biomass ha ve been widely explored in the li terature. Piccolo and Bezzo (2009 ) investigated the economics of the enzymatic hydrolosis and fermentation (EHF) process and the gasification and fermentation (GF) process for bioethanol production. The Net Present Value (NPV) of both the EHF and GF process es were shown to be only positive for a 5 year investment payback period but as a result the ethanol selling price needed to be 38% and 106% higher respectively, than the present market value of fuel grade ethanol. Additionally, a low internal rate of return (IRR) of less than 15% was found for both processes implying a poorly profitable investment choice. Hamelic et al. (2005) concluded that with a higher EHF efficien cy, lower capital investments, increased scale s of operations and lower biomass stocks might reduce ethanol production costs by 40% in a 10 to 15 year time scale and even by 90% over 20 or more years. Similarly, Hess et al. (2007) claimed that the competi tiveness of cellulosic ethanol is highly dependent on biomass stock, accounting for 35 50% of the total ethanol production cost. On the other hand, Solomon et al. (2007) found an encouraging cost of producing cellulosic ethanol of $2.16 gallon 1 $0.24 gal lon 1 lower than the price of grain ethanol and $0.08 gallon 1 below the gasoline price. Other factors such as the use of cheap residues for biomass feedstocks, low cost mechanism s to finance debt, and the integration with biorefineries platform might redu ce the cost of production of ethanol even further. Economic analysis of silvicultural intervention based bioenergy has also been a subject of research Polagye et al. (2007) explored the economic feasibility of utilizing thinning of overstocked forests to produce bioenergy and reduce wildfires in the western U.S. Findings
20 suggested cofiring of thinnings with coal as the most viable option for transportation distances of less than 500 km. Other conversion pathways such as pelletization, fast pyrolosis and m ethanol synthesis became cost competitive for different ranges of thinning yields (50 500 km 2 of annual thinned area) and duration (1 15 years) beyond 300 km of transportation distance. Ahtikoskia et al. (2008) investigated the economic viability of using thinning based energy from young stands in Finland. They concluded that thinning based energy would be economically viable if removal exceeded 42 m 3 ha 1 with an average ethanol volume per stem larger than 15 l. However, without government intervention th e use of thinning for bioenergy would be unprofitable. Research Plan So far, the main concerns that the U.S. is facing in terms of energy security and GHG emissions have been outlined. Further, corn based ethanol production has raised the challenge of low energy ratios, food security, and increasing food prices. On the other hand, NIPF landowners have had to confront a detriment in the profitability of forest stands due to risk events, offshore competition, and low pulpwood prices. Forest biomass, a renewab le and more carbon neutral energy resource, could provide a potential solution to GHG emissions, high energy prices, and dependency on foreign oil. Additionally, cellulosic ethanol can be a viable alternative to food based biofuels, increasing profitabilit y of forestlands and stimulating rural employment. Previous research on forest biomass based bioenergy suggests that a research gap exists with regard to the integrated effect of bioenergy markets with the risk of current catast rophic events on profitabil ity of nonindustrial private forests assuming that stumpage prices follow a stochastic approach. Furthermore, information is required on public preferences for the use of bioenergy rel ative to traditional fossil fuels considering the impacts of bioenergy on reducing the risk of wildfires and pest outbreaks in southern forestlands. Similarly, existing research has not accounted for the tradeoffs between bioenergy markets and conventional forest product sectors.
21 The aim of this dissertation is to fulfill th e identified research gap by explor ing some aspects of the economics of forest biomass based energy in the southern U.S. Chapter 2 assesses the potential impacts of fore st biomass markets for energy on slash pine plantations by applying a model that combin es the stochastic condition of timber prices (Black Scholes formula) and the risk of natural disturbances such as wildfire and pest outbreaks, assuming that catastrophic events follow a Poisson distribution (modified Hartman model). It analyzes the impact of bioenergy markets on the profitability of the forest stand, the effect on profitability of reducing the risk of natural disturbances through thinnings, and change s in price and price volatility. Chapter 3 explores public preferences for forest biomass b ased transportation biofuels, particularly for blends of 10% and 85% ethanol, by conducting a contingent valuation (CV) study in the southern U.S. A discrete choice model is run and the willingness to pay (WTP) and total expenditures (TE) for both blends a re calculated in Arkansas, Florida, and Virginia. Further, socioeconomic conditions and their effects on the probability of choosing a particular blend are discussed. Chapter 4 assesses the dynamic effect s of an increased demand for woody biofuels on the t raditional forest product s sector. We apply the model developed by Just, Hueth and Schmitz (2004) along the lines proposed by Brannlund and Kristrom (1996) and Arkanhem et al. (1999). Specifically we determine the partial equilibrium model for the suppl y and demand for sawtimber, p ulpwood, and biomass for bioenergy and their respective cross price elasticities. Also, we simulate an increased dema nd for biomass for bioenergy and estimate the effect on equilibrium quantities and prices of traditional fores t products. Finally, results and policy implications are summarized in Chapter 5, as well as future directions for research.
22 CHAPTER 2 MODELING IMPACTS OF BIOENE RGY MARKETS ON NONIN DUSTRIAL PRI VATE FOREST MANAGEMENT Introduction The main utilization of fo rest biomass for bioenergy purposes has been the generation of steam or electricity for the forest products industry (Guo et al ., 2007). However, with further advancements in cellulosic, enzymatic, and thermochemical technologies forest biomass based bioen ergy could open up new opportunities for nonindustrial private forest (NIPF) landowners. At present the traditional pulpwood market is suppressed thus landowners are less inclined to undertake thinnings (Mason et al ., 2006). On the other hand, f orest s can be affected by catastrophic events such as wildfires and pest outbreaks. In general, catastrophic disturbance rates in forests are around 1% annually, ranging fro m 0.5% to 2% (Runkle, 1985). Several studies have demonstrated the effect of catastrophic ri sk on forest management. For example, Reed (1984 ) extended the Hartman model by incorporating the probability of a stand of being affected by catastrophic events. A similar approach was recently followe d by Stainback and Alavalapati ( 2004 ) to model the eff ect of catastrophic risk on carbon sequestration in slash pine forests. Furthermore, Englin et al ( 2000 ) explored the optimal rotation age in a multiple use forest in the presence of fire risk. All these studies showed that the presence of risk of catastr ophic of stand destruction decreases the optimal rotation age and the land expected values. In general, catastrophic events can yield several economic implications that can affect all timber market participants. Prestemon and Holmes (2000, 2004) and Preste mon et al (2006) illustrated the short and long run timber price dynamics after a natural catastrophe. In the short run inventory is reduced and salvaged timber gluts the market. Prices fall and decrease producer welfare while at the same ti me causing an increase in consumer welfare In the long run thi s
23 situation might be reversed: prices increase due to losses of standing inventory and contracted supply (time of salvageable exhaustion) improving producer and reducing consumer benefits. Silvicultural pr actices such as stand thinnings are commonly used to extract small diameter wood and reduce excessive amounts of forest biomass, which enhances residual stand growth as well as lowers wildfire and pest risk. Research suggests that thinning from below is mo re effective in reducing crown fire compared to crown and selection thinnings (Graham et al 1999; Peterson et al ., 2003). Further, it is well known that maintaining an appropriate stand density is an effective way to reduce southern pine beetle (SPB) da mage. Overstocked stands reduce tree vigor making them more susceptible to SPB attack (Cameron and Billings, 1988; Belanger et al ., 1993). Thus, the use of forest biomass for energy purposes can provide additional avenues for small diameter trees and help out NPIF landowners to meet management goals and increase the profitability of their forest stands. M odel Specification The stochastic condition of forest stumpage prices has been extensively explored in the literature. It has been shown that the expected value of the stand increases when stochastic prices are incorporated (Haight, 199 1; Haight and Smith 1991 ; Plantinga 1998 ; Lohmander, 2000; Lu and Long 2003 ) The Black Scholes (BS) formula ( 1973 ), which considers the stochastic natur e of prices, was primarily developed to value options but has been widely used in forestry analyses as well. Shaffer (1984) proposed the option pricing methodology for valuing long term timber cutting contracts. Zinkhan (1991) applied the option pricing t heory to the problem of valuing the land use conversion option. Thomson ( 1992 ) explored the binomial option pricing model to determine the optimal forest rotation age. Yin and Ne wman ( 199 6) studied the effect of catastrophic risk on forest investment decis ions following the option approach under investment uncertainty. Plantinga ( 1998 ) analyzed the rotation age problem
24 highlighting the role of the option value in determining the optimal timing of harvest by assuming that stumpage prices follow a random walk or an autoregressive process. Hughes (2000) used the Black Scholes option formula to value the forestry co Yap ( 2004 ) modeled the Philippine forest plantation lease as an option considering market uncertainty and irreversible sun k establishment costs. The BS formula is extended to include the probability of risk of natural disturbances and integrate with the Hartman model. The BS formula assumes that prices follow a diffusion process (random walk with a drift or geometric Brownian motion) which can be represented as: = + (2 1) Where is the stock price, is the drift rate of the stock price, d t is the time increment is the volatility of stock price and is the increment of a Weiner process defined as = where is a normally distributed random variable with E( ) = 0, E( 2 ) = 1 and E( ) = 0 for al ; is independent and normally distributed with mean zero and variance dt Equation (1) states that a change in P depends on a deterministic component and stochastic term T he other underlying ass umptions of the BS formula are that the stock pays no dividends, the option is exercised at the time of expiration there are no transaction costs, and there are no penalties to short selling ( Black Scholes, 1973). The BS formula is represented as: = 1 2 ( 2 2 )
25 W here C is the value of an option (premium) S is the market or stock price N(d) represents the cumulative normal density function X is the strike or exercise price r stands for the risk free interest rate and d1 and d2 show a relationship among the stock price, risk free interest rate, strike price, volatility and current and maturity date; d1 and d2 are represented as follows: 1 = ln S X + ( r + 2 / 2 ) ( ) ( 2 3) 2 = 1 ( 2 4) = 1 2 = 1 1 ( 2 5) represents the average stock price, n is the horizon time in which volatility is calculated, t and T are the current and maturity date, respectively, N(d1) and N(d2) stand for the probabilities that a normal variable take s on values less than or eq ual to d1 or d2 respectively. N(d1) is also known as the option delta ; the degree to which an option value will change given a small change in the stock price N(d2) is the probability that the option will be exercised or the change of the stock price at expiration time N(d1) is always larger than N(d2 ) because d1 is greater than d2 by Thus the difference between N(d1) and N(d2) will be greater for higher stock volatilities and/or long dated options. SN(d1) reflects the benefits of acquiring the option while XN(d2) represents the price of paying the option at expiration time. Following Hughes ( 2000 ) S is the stumpage price volatility (the standard deviation of the natural logarithm of stumpage prices) and X is the cumulated exercise forest costs per unit of merchantable volume at time T T he exercise cost has
26 to be interpreted as the option of the forest landowner of holding the forest stock (stumpage) and incurring costs associated with certain activities such as site preparation, planting, fertilization, weed control, management, etc. or s elling the stumpage. Revenues from thinnings are considered as a negative cost (Hughes, 2000). The decision to sell the stumpage will depend on whether the payoffs from doing so are greater than the value of waiting (Plantinga 1998). If the value of the t imber exceeds the cumulated cost incurred by the forest landowner, the stumpage will be sold otherwise the sale will b e put off. The expected value and net expected value of the timber can be represented respectively as : = ( ) [ 1 ] ( 2 6 ) = [ 1 2 ] ( 2 7 ) W here V(T) is the total merchantable volume at time T Contrar y to financial options where the time of expiration of the option is fixed, the harvest date T fo r forest options can be variable (Hughes, 2000). The expected net present value of the timber for the first rotation can be expressed as : = ( 2 8 ) W here is the discount rate. If the land is ass umed to be used for timber production in perpetuity the land expectation value can be modeled as :
27 = ( 1 ) (2 9) W here LEV(T) is the BS formula based Land Expectation Value Starting from bare land ( t =0 ) and simulating harvest dates T 1 T 2 T 3 etc., the time T that maximizes LEV(T) is the expected optimal rotation age. The BS formula is integrated with a modified Hartman model (1976) accounting for catastrophic disturbance rates in forests. It is assum ed that these catastrophic events follow a Poison process which means per unit of time. Thus the Poisson parameter represents the average rate of a catastrophic event. The second assumption is that the waiting time between successive catastrophic events is al so a random variable. Following Reed ( 1984 ) the time between each successive destructions of the stand are denoted by x 1 x 2 ,.. x n Further occurs every year and x follows the exponential distribution ( 1 x ) The probability density function of x be fore reaching the optimal rotation age ( x < T ) age is given by ( x ) At the optimal rotation age ( x = T ) the probability density function is Therefore the probability of a stand being destroyed by a catastrophic event before the time of rotation age T and the probability of the stand reaching the rotation age T are, respectively : < = 1 ; = = 1 < = ( 2 10 ) The net return will depend on both the timing of the catastrophic even ts and the timing of It is also considered that some portion of the stand is salvageable on a proportion k after a catastrophic event If a
28 catastrophic event occurs the landowner will h arvest any salvageable timber and replants to start a new rotation. Thus the value of one rotation can be represented for the following two states : = if x = k if x < ( 2 11 ) If a catastrophic event happens at time ( x < T ) the landowner salvages a proportion of the stand and incurs in the exercise costs associated with the development of a new forest stand The net rent at time x is given by k However, if the stand reaches the optimal rotation age without being affected by a catastrophic event ( x = T ) the landowner harvests all the timber and incurs in the exercise costs associated with the devel opment of a new forest stand. The net rent obtained at time T is On the other hand, Reed (1984) showed that when risk is present the LEV can be modeled as follows : = 1 + 2 + + = 1 ( 2 12 ) Furthermore, because of the independence of the variables x n E quation 2 1 2 can be rewritten as : ( ) = [ 1 + 2 + + 1 = 1 ] (2 13) = 1 = 1 = 1
29 = E 1 In addition, = 0 ( 2 1 4 ) = x 0 + = + + + Using E quations 2 10 2 11 and 2 14 we obtain an expression that incorporates the outcomes represented in E quation 2 12 The left hand side of E quation 2 15 represents the expected value of a single rotation that can be expressed as the sum of: the stand being affected by a catastrophic event with a salvageable portion harvested before reaching the optimal rotation (first term of the right hand side) and the stand being harvested at the optima l rotation age (second term of the left hand side). Both expressions are multiplied by their probabilities of occurring and discounted to year zero. Thus : = k ( ) x 0 + ( ) ( 2 1 5 ) Using equations ( 2 15) and ( 2 14 ) and substituting them into Equation 2 1 3 the LEV can be redefined :
30 = + ( 1 + T ) ( + ) ( ) + ( + ) k ( ) 0 ( 2 1 6 ) Again, the time T that maximizes the LEV is the optimal rotation age. Recall that i set to zero, E quation 2 16 reve rts to Equation 2 9. A numerical solution with slash pine will be presented next to facilita t e the model comprehension. The M odel A pplication to Slash Pine S tands Slash pine ( Pinus elliottii ) is one of main commercial timber species in the southern U.S., occupying around 10 million acres. It is a fast growing species that yields good quality fiber and lumber (Barnett and Sheffield, 2002). The software GaPPS 4.20 (Georgia Pine Plantation Simulator, Bailey and Zhou, 1997) was used to obtain the growth and yield data. Three scenario sets were conside status quo no thinning scenario thinning scenario for pulpwood and The thinning age was set at year 16 and the percentage of trees left was 70%. Slash pine stands are typically thinned between years 12 to 18 or when the total tree height reaches 40 feet (Dickens and Will, 2002). The site index and stand density at year 5 were assumed to be 70 and 585 trees acre 1 respectively. Four product classes were defined: sawtimber ( st ), chip and saw ( cs ), pulpwood ( pw ) and forest biomass for bioenergy ( fbb ). It was assumed that the small end diameter for st cs pw and fbb was 10, 6, 3 and 0.1 inches respectively while the minimum length for st cs pw and fbb was 8, 8, 5 and 0.1 feet respectively. St cs and pw we re assumed to be obtained under no thinning scenario while st cs pw and fbb were thinning scenario for bioenergy
31 The nominal stumpage prices for st cs pw were obtained from Timber Mart South (TMS, 2006). TMS has been one of the main sources of prices and trends for forest products in c analyses such as Munn et al. (2002), Stainback and Alavalapati (2004 ), Newman (1987), Prestemon and Holmes (2000) and Washburn and Binkley (1990), among others. The nominal prices were deflated by using the lumber Producer Price Index (PPI) (base year=2005) provided by the U.S. Department of Labor, Bureau of Labor Statisti cs (2007). Thus the real stumpage prices for st cs pw were $42.2 ton 1 $25.75 ton 1 and $7.46 ton 1 respectively. While there is no formal market for forest biomass to utilize for bioenergy, we assumed the real price for fbb to be $3 ton 1 The histo st cs pw ) st cs pw were 19%, 24% fbb = 15%) was calculated using the 1970 2004 deflated time series of biomass based electricity industrial prices for five southeastern U.S. states: Florida, Georgia, Alabama, Arkansas and Virginia (EIA, 2007). The risk free interest rate and the real discount rate were set to 3% and 5%, respectively. The common costs associated with the silvicultural activities for the three scenarios were based on Smidt (2005). Costs of $205 acre 1 and $58 acre 1 were assumed for mechanical site preparation (shear, pile, rake and bed) and mechani cal planting, respectively. Weed control and fertilization costs were assumed to be $78 acre 1 and $55 acre 1 respectively. Fertilization was and was considered in year 16 after thinning. In addition, a timber marking cost of $14.6 acre 1 before thinning (year 16) was considered. Annual forest management costs such as taxes, general fire protection and management plans were set to $6
32 acre 1 In both scenarios where thinning was undertaken, the thinning cost is reflected by the stumpage price that is paid to the landowner. For this case fbb was multiplied by a factor of 0.9. an thinning scenarios are expected to have different rates of catastrophic risk. The former was modeled with a risk of 3% while both were modeled with risk levels from 0 to 3%. Outcalt and Wade ( 2000 ) found that the highest tree mortality rate of southern pines after a catastrophic even such as fire occurred when prescribed burning had not been used since plantation establishment, averaging a mortality of 89% Thus, two situations concerning the salvageable portion after a catastr ophic event for the three scenarios were considered: the stand being completely destroyed (k=0) and 80% the stand is salvageable (k=0.8). Results and Discussion The maximum LEVs for the thr ee scenarios are shown in Table 2 1 With a positive salvageable po rtion (k=0.8), the LEV for the thinning scenario for pulpwood and the thinning scenario for bioenergy was greater than the no thinning scenario at all risk levels 4.6% and 5.8% higher LEVs respectively. When salvage is zero (k=0), the land value for the no thinning scenario 2.6%, respectively (for thinning for pulpwood and thinning for bioenergy ). The LEV for the thinning scenario for bioenergy was greater than the thinning scenario for pul pwood at all comparable risk/salvage levels, exceeding the latter by 11.2%, 11.4%, 11.6% k=0 the difference between LEVs was steady at 11.7% for all levels of risk A one percent reduction in risk increased LEVs by between 9% (k=0.8) and 19% (k=0). Thus, land values are impacted less by increased risk levels when salvage is possible. The optimal rotation for the thinning scenarios was longer than the no thinning scenario 27 versus 21 years, respectively.
33 The incorporation of thinnings for bioenergy increased the profitability of slash pine forestry over the no thinning scenario when salvage was possible, but not when salvage was not possible at a risk level of 3%. In fact, if the stumpage price for fbb was set equal to $0 ton 1 the thinning scenario for bioenergy was still greater ($659.17 acre 1 ) than the no thinning scenario, the reason being the high revenues obtained by producing more proportion of sawtimber and less proportion of pulpwood. The breakeven point for stumpage price for fbb was $2.1 ton 1 thinning scenario for bioenergy was compared to the thinning scenario for pulpwood At th is price level the land value was the same for both thinning scenarios, $679.2 acre 1 For k= 0, the breakeven stumpage price for fbb was slightly greater, $2.2 ton 1 Regardless of the risk level, and consistent with finding s of Stainback and Alavalapati ( 2004 ) the LEV was greater when the salvageable portion increased f or the three scenarios (Figures 2 1 2 2, 2 3 and 2 4 ) Further, as risk decreased for both thinning scenarios, the relative difference in LEVs when k=0 as compared to when k=0.8 also de creased. The increase in LEV for a stand partially salvaged as compared to a stand completely destroyed under both thinning scenarios was consistent across thinning scenarios at 26.4%, 16.6%, 7.8% and 0% for risk levels of 0.03, 0.02, 0.01 and 0, respecti vely. When the risk continuously drops, the probability of selling the stumpage and replanting due to a catastrophic event declines as such the difference decreases. The decrease in merchantable volume due to thinnings caused the land values to drop in yea r 16 (Figures 2 1 2 2, 2 3 and 2 4). Although the total volume before the thinning was the same for the three scenarios the land value for the no thinning scenario was lower than the thinning scenario for bioenergy due to the inclusion of biomass for bioe nergy that could be
34 harvested at age 16. In the no thinning scenario only sw cs and pw were included as merchantable volume. However in the thinning scenario for bioenergy fbb was included along with these other three products. This difference became gr regardless of the salvageable portion (Figures 2 1 and 2 2 ). The results (except for the impacts of risk) did not extend to the thinning scenarios for pulpwood (Figures 2 3 and 2 4) since fbb was not a factor. With regard to the thinning scenario for pulpwood there was no difference in LEV compared to the no thinning scenario before the year 16 for the same level of salvageable portion and catastrophic risk. For both scenarios the merchantable volume was the same without i ncluding fbb LEV for the thinning scenario for pulpwood became greater than the status quo (Figure 2 3 and 2 4). In general the difference between LEVs for the no thinning scenario because the rate at which the LEV was discounted became higher. At year 16 the LEV s for the thinning scenarios fell strongly due to the 30% of the tree removal and the cost of thinning and marking. This drastic decline of the land value can be explained by the fact that the percentage of tree removal represented around 22% of total merchantable volume. Before thinning nearly 50% and 57% of the total merchantable volume was pulpwood for the thinning scenario for bioenergy and thinning scenario for pulpwood r espectively. Further, the inclusion of the cost of thinning and marking increased the cumulated silvicultural exercise cost. On average, the LEVs for the thinning scenario for bioenergy decreased by 31.4% and 33.1% for a k=0.8 and k=0 respectively. The LEV s for the thinning scenario for pulpwood decreased by 28.3% and 30% for a k=0.8 and k=0 respectively. After thinning the LEV started increasing because a larger proportion of sw and lesser proportion of pw growing LEV trend was accelerated by
35 increasing the discount rate. The timing when the LEVs for the thinning scenarios started exceeding the no thinning scenario the break even age occurred earlier comparing with the thinning scenario for bioenergy the break even age was 24 years compared 2 1 and 2 2). On the other hand regardless of k the break even ag thinning scenario for bioenergy (Figure 2 2 ). Thus when positive environmental effects are considered once thinnings are carr ied out landowners would financially benefit as the land value peaks faster exceeding the value when thinnings are not undertaken. Although the inclusion of thinning s increased LEVs the difference between the thinning scenarios versus the no thinning scen arios was relatively small for the same risk level As forest biomass based bioenergy markets continue to expand it is plausible that prices paid for woody biomass may increase, as well as their volatility. We simulated independently two impacts: increas ed price ( $5 ton 1 $10 ton 1 and $15 ton 1 ) and increased price volatility (0.2, 0.25, and 0.3) for fbb Table 2 2 shows the LEVs for different level of prices and volatility. From Table 2 2 and consistent with expectations the LEV increased as fbb price increased. This increase was slightly higher when salvageable was possible. For example, on average the increase when fbb price changed from $5 ton 1 to 10 ton 1 and $10 ton 1 to $15 ton 1 was 6. 4 % and 6. 1 % for a k=0.8 a nd k=0 respectively. Furthermore wh en comparing to the original scenario for bioenergy, the LEV increased by 9.6 % and 9% for k=0.8 and k=0 respectively In addition, the profitability of the forest stand was greater when comparing to the no thinning scenario and thinning scenario for pulpwo od Compared to the former the LEV
36 increased on average by 34% and 41 % for k=0.8 and k=0 respectively when fbb price was increased. With regard to the latter, the LEV increased by 10.8 % and 10.2% for k=0.8 and k=0 respectively. Regarding volatility, the i ncrease of the LEV was lower compared to the increase of the LEV when price was increased T he average increase in the LEV when volatility changed from 0.2 to 0.25 and 0.25 to 0.3 was 0.6% steady for all levels of salvageable portions With regard to the n o thinning scenario the land value was on average 24.4% and 31% higher for k=0.8 and k=0 respectively. The difference with regard to the previous bioenergy scenario and the thinning scenario for pulp wood accounted for 1.4% and 1.2 % and 2.5% and 2.4% for k =0.8 and k=0 respectively. Thus as bioenergy prices are expected to rise and consequently their volatility, this combined effect will result in greater returns to forest landowners. C onclusions The incorporation of thinnings increased forestland values re gardless of the risk level when the salvageable value of the stand was positive. Results suggested that once pulpwood or forest biomass for bioenergy is incorporated and the whole stand became commercially marketable, the revenues obtained due to stochasti c price variation could offset the cost of performing silvicultural activities such as thinning. However, when the landowner was not allowed to salvage any portion of the stand and the risk level was assumed to be 0.03 the land value for the no thinning sc enario was higher than for the thinning scenarios. Under these conditions, the revenues associated with the increase of the volumetric growth after thinning for greater value added product and the low price for forest biomass based bioenergy were not enoug h to cover the loss of volume and consequently the profits at the time of thinning Bioenergy price of $5 ton 1 broke even the land value for the bioenergy scenario with regard to the status quo when stand was completely destroyed and risk was 0.03
37 Includ ing thinnings for bioenergy increased the land value around 11.6% compared to the thinning for pulpwood scenario. As expected, increased risk decreased land values for all salvage levels, dropping greater when salvage was zero. On average, the increase o f the land value when risk was decreased by 1% was 10% and 19% for both thinning scenarios when the salvageable portion was 0.8 and 0, respectively the higher risk damage being proportionally more compensated by revenues from salvage. On average, salvage increased land values 17%. Thus policies that help landowners mitigate risk through silvicultural interventions to reduce the size of the damage would have a positive impact on the profitability of forest stands. Although the inclusion of thinnings inc reased the land value of a forest stand the difference with the status quo scenario might be consider small. Furthermore landowners would have to wait longer to harvest. However it is expected as the supply and demand of bioenergy increases bioenergy price s will also increases. Thus by increasing stumpage price and volatility for bioenergy and consistent with features of the Black Scholes formula the land value increased. The impact on the land value was higher when price was increased: the increase in the land value with regard to the origina l scenario for bioenergy was 9.6 % and 9% and when the salvageable portion was 0.8 and 0 respectively while it was 1.4% and 1.2% for the same salvageable portions when volatility was increased Increase in land values due to thinning and bioenergy markets will benefit landowners. Current NIPF landowners will become more competitive and future landowners can be influenced to undertake thinning and even switch from other land uses to forestry. Thinnings will concentrate growth on fewer large trees which will bring higher stumpage prices. Furthermore as bioenergy markets continue evolving small diameter wood for bioenergy purposes will become a competitor for other uses for this type product, for example pulpwood, raising their prices.
38 However, current fluctuating pulpwood markets and the lack of a formal market for forest biomass based bioenergy could be a threat for NPIF landowners to undertake thinnings. In this study, the thinning age was set at year 16. Further resear ch is needed to set an optimal thinning age maximizing the amount of forest biomass to be thinned and the benefit cost of this silvicultural practice. In addition, the incorporation of thinnings will also benefit society. Other commercial activities such a s the possibility of silvopastoral use will be allowed. Due to a decrease of risk and the intensity of the catastrophic event, positive externalities will arise. Forest health and wildlife habitat will be improved because of a reduction of pest outbreaks a nd wildfires respectively. Dependency on external markets for oil and concerns about greenhouse emissions on can be alleviated. In addition other environmental services such as landscape and recreation values will be enhanced. Thus, more factors can be req uested an incorporated in order to assess the profitability of southern pines.
39 Table 2 1. LEV ($ acre 1 ) for the three scenarios at diffe rent levels of risk and salvage Scenarios No thinning scenario Thinning scenario for pulpwood Thinning scena rio for bioenergy Relative increase in LEV for a 1% reduction in risk Relative increase in LEV for a 1% reduction in risk Risk \ salvage k=0.8 k=0 k=0.8 k=0 k=0.8 k=0 k=0.8 k=0 k=0.8 k=0 901.1 901.1 911. 7 911. 7 822. 3 762.3 831.8 771. 3 1.09 1.18 1.09 1.18 748. 3 641. 5 756. 8 649.0 1.09 1.19 1.09 1.19 649. 3 556. 2 679. 2 537. 1 686.8 543. 4 1.10 1.19 1.10 1.19
40 Table 2 2 LEV ($ acre 1 ) for the thinning scenario for bioenergy for different bioenerg y p rices and levels of volatility Fbb price $ ton 1 Scenarios Thinning scenario for bioenergy Fbb volatility Scenarios Thinning scenario for bioenergy 5 Risk \ salvage k=0.8 k=0 0.2 Risk \ salvage k=0.8 k=0 934. 9 934. 9 917 .4 917.4 853.5 790.9 837. 4 776.1 777. 1 665.5 762. 2 65 7 0 705. 8 557. 2 692. 1 546. 9 10 Risk \ salvage k=0.8 k=0 0.25 Risk \ salvage k=0.8 k=0 993. 3 993. 3 922. 9 922. 9 908 3 840. 3 842. 7 780. 7 828. 3 707. 1 767. 4 707. 1 753. 6 59 2 0 697. 1 550.0 15 Risk \ salvage k=0.8 k=0 0.3 Risk \ salvage k=0.8 k=0 1052.2 1052.2 928.5 928.5 963. 7 890.2 848 2 785.5 880. 3 749. 1 772. 7 661.0 802.2 627.1 702.2 553. 4
41 Figure 2 1 LEVs for both the no thinning scenario and thinning scenario for bioenergy when the salvageable portion is zero 0 100 200 300 400 500 600 700 800 900 1000 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 LEV US$ acre 1 Age Land value thinning scenario for bioenergy Land value thinning scenario for bieoenergy Land value thinning scenario for bioenergy Land value thinning scenario for bioenergy Land value no thinning scenario
42 Figure 2 2. LEVs for both the no thinning scenario and thinning scenario for bioenergy when the s alvageable portion is 0.8. 0 100 200 300 400 500 600 700 800 900 1000 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 LEV US$ acre 1 Age Land value thinning scenario for bieoenergy Land value thinning scenario for bioenergy Land value thinning scenario for bioenergy Land value thinning scenario for bioenergy Land value no thinning scenario
43 F igure 2 3. LEVs for both the no thinning scenario and thinning scenario for pulpwood when the salvageable portion is zero. 0 100 200 300 400 500 600 700 800 900 1000 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 LEV US$ acre 1 Age Land value thinning scenario for pulpwood Land value thinning scenario for pulpwood Land value thinning scenario for pulpwood Land value thinning scenario for pulpwood Land value no thinning scenario
44 Figure 2 4. LEVs for both the no thinning scenario and thinning scenario for pulpwood when the salvageable portion is 0.8. 0 100 200 300 400 500 600 700 800 900 1000 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 LEV US$ acre 1 Age Land value thinning scenario for pulpwood Land value thinning scenario for pulpwood Land value thinning scenario for pulpwood Land value thinning scenario for pulpwood Land value no thinning scenario
45 CHAPTER 3 ASSESING PUBLIC PREF ERENCES FOR FOREST BIOMASS BASED BIOENERGY Introduction Ethanol blends are the most widely used l iquid biofuels in the U.S. transportation sector, accounting for 95% of the total biofuel consumption (EIA, 2009). Blends of 10% ( E10 ) and 85% ( E85 ) are currently found on the market; the former can be run in any vehicle, while the latter can be accommodat ed only by Flexible Fuel Vehicles. The main disadvantages of these blends are their low energy content: E10 and E85 have 3.3% and 24.7% less energy content per gallon, respectively, compared to gasoline. This implies that around 1.03 gallons of E10 and 1.3 3 gallons of E85 are required for a vehicle to cover the same distance that it would cover with 1 gallon of gasoline (EIA, 2007). However, the use of ethanol blends has the inherent benefits of contributing to environment and energy security by decreasing the use of petroleum and reducing GHG emissions (Wang et al., 1999). Greenhouse gas emissions per mile traveled are reduced by around 2% and 25% by using corn based E10 and E85 respectively. These reduction levels increase to 10% and 90% for cellulosic ba sed E10 and E85 (Wang et al., 1999). Besides the environmental benefits already described we expect that developing new bioenergy markets will improve forest sustainability by increasing rural employment and financial returns to landowners. In addition to the traditional environmental benefits related to reduction of GHG emission we also considered the benefits associated to biodiversity such as reduction of wildfires and pest outbreaks. Despite several environmental and social benefits and favorable incen tives for potential producers of cellulosic biofuels, the arising question is does the public care for the environmental benefits associated with this bioenergy use? If they do, how much of a premium is the public willing to pay for this bioenergy?
46 Tradit ionally stated preference techniques have been applied to value renewable energy. A vast number of studies that explored this issue (e.g., see Menegaki, 2007) mainly focused on the generation of green elect ricity. For example, Batley et al. (2001) found th at respondents would pay 16.6% extra for electricity generated from renewable resources in the Uni ted Kingdom (UK). Roe et al. (2001) projected a median willingness to pay ranging between $0.38 and $5.66 year 1 tion and affiliation with an environmental organization, for decreasing 1% of GHG emissions using green electricity in the U.S. Nomura and Akai (2004) estimated a willingness to pay of 2000 yens month 1 household 1 for green electricity in Japan. Bergmann et al. (2006) claimed that respondents would be willing to pay an additional £14.03 year 1 household 1 in Scotland for having renewable energy projects that do not increase air pollution, compared to a program which results in a slight increase in pollutio n. Solomon and Johnson (2009) conducted a study more directly related to our research. They used through estimating willingness to pay for cellulosic ethanol in the upp er Midwestern U.S. They applied a contingent valuation method and a fair share method finding a mean total willingness to pay of $556 per capita year 1 and a fair share of $472 per capita year 1 Study Design and Data Collection C ontingent V aluation Q uest ionnaire We applied C ontingent V aluation (CV) a survey based method, to elicit public preferences for public goods. We conducted a survey at the household level in three southern U.S. states: Arkansas (AR), Florida (FL), and Virginia (VA), to understand p ublic attitudes towards transportation biofuels. The online questionnaire was administered and hosted by Knowledge Networks (KN). KN was founded in 1998 seeking to develop online research methodologies, and established the first online research panel Kno wledgePanel based on
47 probability sampling covering online and offline populations in the U.S. KN selects households using random digit dialing (RDD), and households are provided with access to the internet and hardware if needed. Once a person decides to join the panel, she/he is sent a survey by email. Thus, KN surveys are not limited only to web users or computer owners. KN sample design is an equal probability sample that is self weighting. However, there are some inherent deviations from an equal prob ability sample design such as oversampling of minorities or household s with access to the internet and subsampling of telephone numbers without an address, among others. Adjustments addressing geographic frame and language are incorporated in t o the base w eights. Thus, all base weights are adjusted with a post stratification weight obtained from the Current Population Survey (CPS) d emograp hic and geographic distribution benchmarks for all adults older than 18 years of age. Web based surveys have strongly em erge d during the last decade (Champ, 2003) due to their low cost, speed, and accuracy in stated preferences studies ( Berrens et al., 2004; Banzhaf et al., 2006; Marta Pedroso et al., 2007). Although web based surveys provide similar welfare estimates compa red to traditional m ail surveys (Fleming and Bowden, 2007), the main criticisms are related to the sample frame selection and non response bias ( Lozar Manfreda 2001). KN takes several complementary measures in order to minimize non response bias. KN encou rages participation through incentives, newsletters, and other techniques (e.g., a toll free helpline for providing assistance with survey questions ). Further, KN follows up and contacts non respondents. Wiebe and et al. (2001) conducted a key study of non response bias using the KN methodology, finding that the inclusion of data from non response follow up of panel recruitment non responders did not affect the statistical estimates, concluding that non response bias was operating at a low level. Lastly, fi nal data are subject to a post stratification using
48 current demographic distribution as a benchmark to adjust for non response bias and non coverage (Huggins et al., 2002). All of this is reflected in h igher response rates, around 65 % in our study, which r educed chances of non response bias significantly. A random sample of 630 households was drawn from the KN online research panel that met the criteria of being in the general population and over 18 years old. The questionnaire was administered to these hou seholds. The questionnaire contained two parts. The first part comprised the CV section. The CV section, respondents were asked to choose between two alternative pl ans: Plan A and Plan B. Plan A was described in terms of the attributes associated with using fuel ethanol. The attributes were developed based on a literature review regarding fuel ethanol and discussions with stakeholders and experts specializing in biom ass forestry research. Additionally, two steps were carried out to determine the attributes and levels: two focus groups of 12 people each randomly selected and contacted by phone were conducted at the University of Florida. Furthermore, a pilot survey was conducted in the three study states. Plan B reflected the status quo of not using fuel ethanol. Another similar stated preference method is conjoint analysis 1 (CJ) popularized in the marketing literature. Individuals make choices about different stat es of the world they prefer based on the attributes and their respective level s Typical CJ formats are contingent rating, contingent ranking and the binary response format (closely related to contingent ranking). The binary response option is mathematica lly similar to the dichotomous choice CV method (Adamowicz et al., 1994; Roe et al., 2001 ). Although Plan A is described in 1 Choice modeling (CM) is another popular technique that has arisen from conjoint analysis. Generally, individuals are asked to choose a particular option according to levels of each attribute (Adamow icz et al., 1998). However, CM has been also applied to ra nk and rank attribute based alte rnatives (Hanley et al., 2001)
49 terms of attributes and their levels in our study we have used the CV terminology because we do not focus on valuing attributes but on a particular improvement of environmental quality methods. When the CV questionnaire was introduced to respondents, they were informed about the purpose of usin g forest biomass as feedstock for ethanol. The expected benefits of using these biofuels were described regarding the reduction of GHG emissions and improvement of biodiversity through reduction of wildfires and pest outbreaks. In order to avoid the typica l problem of stated preference experiments (i.e., the difference between stated and actual behavior), the design of a cheap talk script, suggested initi ally by Cummings and Taylor (199 9) an d used in several studies ( Cummings and Taylor 19 9 9; List 2001; Men ges et al 2005; Carlsson et al 2005), was included. The attributes and their respective levels were explained to respondents, and an example was provided to facilitate comprehension. Respondents were then asked to provide their views about bioenergy an d outline their stated preferences. Respondents were provided only one questionnaire, regarding the use of either E85 or E10 The attributes chosen were (1) reduction of CO 2 emission per mile traveled, (2) improvement of biodiversity by decreasing wildfir es and pest outbreaks, and (3) increase in the price of fuel at the pump. A brief description of the attributes and their levels is given in Table 3 1. The levels of percentage reduction depend on the energy and chemical usage intensity of biomass farming ethanol yield per dry ton of biomass, and electricity credits in cellulosic ethanol plants (Wang et al., 1999). In order to facilitate understanding of the study, we linked each level to a non numerical category: low, medium or high reduction. C atastroph ic disturbance rates in forests are generally around 1% annually, ranging from 0.5% to 2% (Runkle, 1985). The
50 levels of reduction of pest outbreaks and wildfires wer e assumed following Susaeta et al. (2009). Again, each level of reduction was linked to a n on numerical category: low, medium, or high reduction. We assumed a higher premium level for E85 based on its better benefits and its lesser energy content per gallon. Because the decision to pay a premium for biofuels depends on current market fuel prices a reference price of gasoline was provided to facilitate the gasoline prices were $3.13 gallon 1 $3.07 gallon 1 and $3.69 gallon 1 respectively (http://e85prices.com/archive.php). The three attribut es and their respective levels provided 36 possible combinations (3 2 x 4 1 ) for Plan A, achieving a 100% A efficiency. Because it is practically unfeasible for an individual to answer 36 different CV questions, the orthogonal full factorial experiment desi gn followed in this experiment was blocked into six different questionnaire versions, each having six pair wise alternative plans. The SAS 9.1 %MKTRuns and %MktEx macros were used to determine the number of alternative plan sets and the linear design (Kuhf eld, 2005). This was in accordance with previous CV studies, which have considered the number of alternative plan sets to range betwee n 4 and 12 (Carlsson et al., 2003; Shresta and Alavalapati, 2004; Mogas et al., 2006) without violating the assumption of stability of preferences (Hanemann 1984). E ach respondent answered six sets of questions, each consist ing of two plans, Plan A and Plan B representing six different observations. Table 3 2 presents an example of the alternative plan situation. The valuat extra $0.60 per gallon at the pump for reducing the CO 2 emissions between 61 70% (medium reduction) and improving the biodiversity between 1 25% (low improvement) (Plan A) or not to pa y a premium at all without having any changes in CO 2 emissions and biodiversity
51 Table 3 3. An example of the survey is included in Appendix B. Econometric M odel We applied a probit model to the CV. The theoretical framework to analyze the CV method is the random utility model developed by McFadden (1974). Under this framework, the indirect utility of an individual results from the sum of a deterministic part and a s tochastic element. Formally: = + ( 3 1) Where is the utility for each respondent i to choose among different j alternatives, is the deterministic part of the utility and reflects the uncertainty or unobservable influences on respondent choice. In the case of CV studies, alternatives are reduced to two; therefore, the individual has the option to choose alternative j which reflects an improved state, over alternative k (status quo) if the utility associated with alternative j exceeds the utility of alternative k Because a random component is involved, only probability statements about either option can be made. Thus, the probability that individual i will choose alternative j over k c an be formally expressed as: = + > + ( 3 2) By using this framework, Hanemann (1984) conceptualized dichotomous CV responses and designed a framework to obtain welfare estimates. Follow ing Haab and McConnell hereafter
52 (2002), and assuming a linear utility function in income and covariates, the deterministic part of the indirect utility function for an individual i can be written as: ( ) = + ( 3 3) Where is the income of individual i and is the matrix of attributes and socioeconomic characteristics of individual i and and are the multidimensional vector and t he marginal utility of income of alternative j respectively. The dichotomous question requires each individual to choose between alternative j paying an amount and the status quo. Thus, the deterministic parts of a utility function for alternatives j and k are: ( ) = + ( ) ( 3 4) ( ) = + ( ) ( 3 5) Replacing Equations 3 4 and 3 5 into Equation 3 2 and rearranging, we obtain the following expressions: ( ) = + ( ) + > + ( ) + ( 3 6) ( ) = ) + ( ) ( ) + > 0 ( 3 7)
53 Assuming that the marginal utility of income is constant and denoting = and = the probability of a yes response is: ( ) = + + > 0 ( 3 8) We assumed that ~ ( 0 2 ) and by conver ting the errors to a standard normal, we obtained the probit model: ( ) = ( 3 9) Estimates for the parameters are obtained by maximizing the likelihood function. In the case of a probi t model, the log likelihood function takes the following form: Ln | = = 1 + 1 1 ( 3 10) Where T is the sample size and = 1 if individual i answers yes.
54 Welfare E stimates Two measures of central tendency were developed by Hanemann (1984), the expected willingness to pay (E[WTP]) a nd the median willingness to pay (Md[WTP]), which are equal under the assumption of a linear utility function. Thus, = = ( 3 11) Where is the mean of attributes and socioeconomic characteristics. Results and Di scussion The online questionnaire was administered during March and April 2008. A total of 40 8 questionnaires were completely a nswered (65% response rate), 20 1 questionnaires regarding E10 (56 in AR, 76 in FL, and 69 in VA) and 207 questionnaires about E85 (53 in AR, 79 in FL, and 74 in VA). We used STATA 9.0 to estimate the probit models for E10 and E85 Tables 3 4 and 3 5 present the descriptive statistics for the socioeconomic variables of the E10 and E85 samples, respectively. Generally, respondents wer e not part of any environmental organization. In addition, most of them had achieved one of the two highest levels of education and owned an automobile. Further, respondents belonged mainly to the middle income category, with the exception of E10 responden ts in VA. Non automobile owners, even though they were few in number, were included in the survey because they might also be interested in purchasing biofuels. However, the findings showed that preferences of automobile owners and non automobile owners we re dissimilar. The percentages of non automobile owners who chose to use E10 were 100% in AR, 50% in FL, and 64% in VA. These percentages were lower for E85 in the
55 cases of AR and FL (17% and 25%, respectively), while in VA 100% of non automobile owners ch ose to use biofuels. Consistent with expectations, the sample showed that respondents were less likely to accept the premium wh en the bid was increased (Figures 3 1 and 3 2). Regardless of the premium level, the average relative decrease for a yes response was around 10% in each state for E10 In the case of E8 5 the average relative decrease amounted to 15.1%, 9%, and 8.6% for AR, FL, and VA, respectively. The majority of the respondents were willing to pay a premium for both blends in FL and for only E85 in AR. In VA, the majority of the respondents tended not to be willing to pay a premium for either E10 or E85 Attributes Tables 3 7 and 3 8 show the coefficients, p values, and standard deviations of the probit model in AR, FL, and VA for E10 and E85 respectively. The log likelihood ratios ( p < 0.001) suggested that the overall models for both blends were statistically significant in all three states. STATA routines dropped variables that perfectly predicted success or failure in the dependent variable For E10 a ll respondents owned a car in AR ; thus, this dummy variable only took the value of 1, showing no variation across the sample. In the case of educational variables, Less high was not included in the model to avoid the dummy variable trap. Howeve r, this variable only took the value of 0 in AR as all respondents had some high school level. Thus, College was dropped to avoid collinearity. For E85 Member in AR and Ownership in VA were dropped. Almost none of the respondents were members of an enviro nmental organization, and almost all owned an automobile. Regarding E10 2 ( Reco2 ) statistically significant only in FL and VA was positive in all three states. Biomp ) not statistically significant in
56 any of the states was positive in AR and FL. This indicates that the probability of paying a premium increased as the reduction of CO 2 increased and biodiversity conditions improve d. Along with improvements in environmental conditions at higher rates, respondents from AR and VA were less likely to use E85 The variable Reco2 which was significant only in AR, had negative coefficients in AR and VA, while the variable Biomp had nega tive coefficients in all three states. Interestingly, Biomp was not significant in any of the states. Except for the variable Reco2 in VA, the results for E85 fuel were consistent with the probabilities of the model for the environmental attributes (Table 3 6). The probability of paying a premium for E10 increased in all three states as either the reduction of CO 2 increased or biodiversity improved. These results were consistent with previous studies (Roe and et al., 2001; Bergmann et al., 2004) which found that when environmental quality improved, the utility of respondents increased, and therefore they were willing to pay more for green electricity. However, the same trend was not observed in the E85 scenario, as respondents did not intend to pay a premium for reducing CO 2 in AR or improving biodiversity in all states at higher rates. Consistent with economic theory under the assumption of a negative price elasticity of demand, the utility of individuals decreased as the premium increased for either blend. Further, this attribute was statistically significant in all three states. Socioeconomic Variables Member Member ) nor having Knowled ge ) was statistically significant in any of the states for E10 For E85 while Member was not significant in FL or VA, Knowledge was significant only in VA. Further, the likelihood of switching to biofuels would increase if respondents were members of an e nvironmental organization in AR and VA for E10 and in VA for E85 Similar behav i or was observed for E85 respondents in AR if they were aware
57 Ownership ( Ownership ) this proba bility increased only for E85 and was significant in AR and FL. The variable ( Miles week ) was found to be significant when deciding whether to switch to biofuels. For E10 the probability of switching to biofuels increased in FL an d decreased in VA. However, the results showed that Miles week was not significant in any of the states for E85 Educational background showed different trends in the three states. The variable High ) was st atistically significant and positive, Bachelor ) was not statistically significant for E10 in AR. In the same state, the likelihood of choosing E85 increased for individuals who had higher educati on levels compared to individuals with less than a high school education. The same situation was observed in FL for E10 In VA, no education variables were degre es were not likely to pay a premium, but individuals with some college education would tend to use the E10 blend Individuals with high school in FL and some college education in VA showed no intention of using E85 Income was another variable that showed Hincome ) was statistically significant for both biofuels. Contrary to expectations, the probability of paying a premium decreased as an individual had grea ter earnings. Although this is inconsistent with economic theory, respondents from Arkansas might have considered exogenous factors such as the unfavorable economic situation prevailing in the country during the time the survey was conducted.
58 In FL, the va Mincome ) and Hincome were statistically significant as compared to the lowest income level for E10 Although this situation was not observed for E85 the utility of choosing any of the biofuel blends increased when individuals came from middle income and high income households. In VA, neither income category provided statistically significant results for E10 However, for E85 the middle income category showed significant differences compared to low income individual s. Age was significant only in AR for E10 and E85 Further as AR respondents aged they would only be likely to choose E10 On the other hand, Gender was statistically significant in all three states for E10 and in FL and in VA for E85 The probability th at females would choose either blend was lower compared to males in FL. The variable Respondent was working ( Work ) was not significant in any of the states for E10 and significant only in AR for E85 It was also found that individuals would continue usin g cheaper fuel if they were unemployed in AR for E10 and FL and VA for E85 The ( Size ) was statistically significant in AR and VA for E10 Consistent with expectations, as the number of people in the household increased, individuals were less likely to use biofuel blends. For E85 Size was statistically significant only in AR, where the results showed that an increase in number of people in the household would decrease Head ) was significant in FL for E10 and AR and FL for E85 Respondents were not likely to choose E85 in AR and FL or E10 in VA. Finally, the effect of unobservable influences was statistically significant in all three states for E10 and in VA and AR for E85 Further, respondents had a natural tendency not to pay a premium for E85 in AR and in FL for E10
59 Willingness to P ay The extra WTP estimates are shown in Table 3 9. The greatest WTP for E10 was in FL ($0.58 gallon 1 ), followed c losely by AR. VA was the state with the lowest WTP ($0.50 gallon 1 ). Similar findings were obtained by Bh attacharjee et al. (2008) who calculated a mean WTP of $0.49 gallon 1 for E10 in a U.S. nationwide study. The WTP for E85 was greater than the WTP for E10 The increase in the WTP for E85 accounted for 1.46, 2.00 and 2.1 times the WTP for E10 in AR, FL and VA, respectively Three multiple comparison tests (Bonferroni, Scheffe and Sidak tests) were performed to detect differences in WTP among the three states for both blends. For E10 there were no significant differences in WTP among t he three states. For E85 there were significant differences in the WTP between AR and FL and between AR and V A. However, there were no significant differences in the WTP be tween FL and VA. The average prices of E85 when the questionnaire was administered were $2.52 gallon 1 $3.00 gallon 1 and $3.07 gallon 1 in AR, FL, and VA, respectively ( www.e85prices.com ). As noted earlier, average gasoline prices were $3.13 gallon 1 $3.21 gallon 1 and $3.69 gallon 1 for the same states. Thus, the ratios of WTP to actual E85 price were 1.57, 1.46, and 1.54 in AR, FL, and VA, respectively, averaging 1.52. On the other hand, assuming the current price of gasoline as a proxy for E10 the ratios were much lower: 1.18, 1.18, and 1.13 in AR, FL, and VA, respectively, averaging 1.16 The ratios for both blends might be higher in the future, as market prices for gasoline and E85 are expected to increase Gasoline is predicted to have an annua l increase of 1.4% reaching $4 gallon 1 (2007 dollars) in 2030 while the annual price increase of E85 will be 0.5% over the same period reaching less than $3 gallon 1 (EIA, 2008a). It was interesting to note that respondents were willing to pay more for biofuels once the proposed change offered better conditions for the environment. In VA, the respondents might have considered that the change in environmental conditions would not compensate for the
60 premium to be paid for E10 ; thus, the state had the lowes t WTP. The opposite happened for E85 for which 1 ). Although the percentages of rejection of the premiums were almost the same for both blends in VA (Figs. 1 and 2), the percentages of rejectio n of the higher premiums were greater. For example, the rejection percentages for premiums of $0.75 gallon 1 and $1 gallon 1 were 65% and 67%, respectively, for E10 whereas in the case of E85 the rejection percentages for premiums of $1 gallon 1 and $1.5 gallon 1 were 55% and 62%, respectively. Further, the ratios of no versus yes responses for the premiums described were 1.86 and 2.03 for E10 and 1.22 and 1.63 for E85 T he extra WTP for ethanol was converted into total future expenditures per year (TE e ). Following Solomon and Johnson (2009 ), the total expenditures were calculated by multiplying the total WTP by the quantity of gallons of ethanol (Q e ) consumed as a proportion of total fuel consumption in the next period compared to the previous one. This p roportion is given by the price elasticity of the demand for biofuels (Ed e ) obtained in the model. The average per capita motor gasoline expenditures in 2006 (EIA 2008b) were used to calculate the quantity of gallons of ethanol. The real motor gasoline exp enditures (2007= 100) accounted for $1295, $1200, and $1 373 per capita in AR, FL, and VA respectively. The total WTP is separated into an average price of gasoline (P g ) previously described in this section and the mean WTP for ethanol in each state (WTP e ) Formally: = ( + ) ( 3 12) Equation 3 12 was applied for E10 and E85 in each state. The mean total expenditures for E10 were $585.20, $485.90, and $596.20 per capita year 1 in AR, FL, and VA, respectivel y. For E85 the total expenditures were $919.60, $330.80, and $532.60 per capita year 1 in the same
61 states. With the exception of E85 in AR the results were somewhat similar to those found by Solomon and Johnson (2009). They reported a mean total future ex penditure of WTP of $556 per capita year 1 in Michigan, Minnesota, and Wisconsin. The ratios of total expenditures of E85 to E10 were 1.57, 0.68, and 0.93 in AR, FL, and VA, respectively. The price elasticity of the demand was relatively inelastic ( 1 < E d e < 0) for both blends in the three states. The values for Ed e for E10 were 0.38, 0.34, and 0.38 in AR, FL, and VA, respectively, while those for E85 were 0.56, 0.2, and 0.31. In general, the Ed e for E85 was less elastic compared to the Ed e for E10 with the exception of AR, where the total expenditures for Arkansans were higher. Conclusions This chapter reported the findings of a contingent valuation experiment designed to elicit WTP for E10 and E85 and assess public preferences for these biofuels in AR, FL, and VA. The results indicated that individuals had a positive WTP for both blends and a greater WTP for biofuels that led to environmental improvements. No significant differences were found in the WTP among the three states for E10 For E85 si gnificant differences were found in WTP between AR and FL and between AR and VA. The WTP ratios of E85 to E10 were 1.46, 2.00, and 2.12 for AR, FL, and VA, respectively. Thus, southern U.S. consumers value the environmental benefits obtained from a modifie d transportation fuel. When the WTP was converted into future total expenditures, t he ratios of total expenditures of E85 to E10 were 1.57, 0.68, and 0.93 in AR, FL, and VA, respectively. With the exception of AR, the total future expenditures were slated to be higher for E10 because of a more elastic price elasticity of the demand. The results also showed heterogeneous preferences for environmental attributes. E10 in order to achieve CO 2 reduction in all three states and biodiversity improvement in AR and FL. However,
62 in all three states, respondents stated the opposite for biodiversity conditions for E85 This heterogeneity was also observed in some socioeconomic variables. For exampl e, individuals with higher levels of education would only change to E10 in FL and E85 in AR. The high oil prices at the time of the survey and the higher premium proposed for E85 might explain why individuals from middle and high income households in AR an d VA were reluctant to pay more for that biofuel. We assumed that the attributes and the socioeconomic variables of this discrete choice model were exogenous. It might be argued that the distance driven per week is correlated with the error term (endogenou sly determined). If endogeneity arises for this particular case, the estimated coefficient of weekly mileage will be upwardly or downwardly biased depending on the direction of the correlation with the error terms. Potential solutions to correct for endoge neity are the use of instrumental variables or the determination of the endogenous variable by equili brium model (Besanko et al., 1998). Correction for endogeneity bias was beyond the purview of this research Understanding present and future individual p references for bioenergy is an important tool for policymakers This positive elicited WTP might support the initiation of a consistency policy instrument such as the Renewable Fuel Standards (RFS) aiming to produce 7.5 billion gallons of renewable fuels by 2012. However it also under scores the need for continuous reinforcement of benefits from federal or state governments for consumers of green energy. Although the findings suggested that individuals were willing to pay a premium for biofuels periodic r evisions of these studies are certainly important to formulate policies based on changing public perceptions and preferences. Further different approaches might be used to allow welfare measures to be to be adjusted for different policy contexts. For inst ance the use of
63 meta analysis is particularly interesting to validate and explore the systematic and identifiable variation of WTP in order to determine its appropriateness for benefit transfer. Another situation considered for this study was that we ass ume d that people were homogen e ous within each state. However a more specific level of aggregation or an incorporation of spatial variation could be a plausible extension of this study
64 Table 3 1. Des cription of the attributes and levels Attribute Description Level E10 E85 Reco2 Percentage reduction of CO 2 emissions (per mile traveled) 1 3 % (low) 4 7% (medium) 8 10% (high) 1 60% (low) 61 70% (medium) 71 90% (high) Biomp Percentage improvement of biodiversity by reducing wildfire risk & improving forest health 1 20% (low) 21 40% (medium) 41 60% (high) 1 25% (low) 26 50% (medium) 51 75% (high) Prem Increase of the price of fuel at the pump per gallon $0.2, $0.5, $0.75, $1 $0.3, $0.6, $1, $1.5
65 Table 3 2. Description of the choice situation Please choose Plan A Plan B Reco2 Reduction of CO 2 between 61 70% per mile traveled No reduction (0%) Biomp Improvement of biodive rsity between 1 25% No improvement (0%) Prem Additional payment of $0.60 per gallon at the pump No extra payment ($0)
66 Table 3 3. Socioeconomic variables Variable Description Member Membership in an environmental organization: 1 if respondent is a member and 0 otherwise Knowledge Knowledge of other natural resources based energy: 1 if respondent knows and 0 otherwise Ownership Ownership of an automobile: 1 if respondent owns and 0 otherwise Age Years Miles week Distance driven w eekly (miles) Education Less high : 1 if respondent has exclusively less than high school level and 0 otherwise High : 1 if respondent has exclusively high school level and 0 otherwise Some college : 1 if respondent has exclusively some college level an d 0 otherwise Bachelor : 1 if respondent has exclusively bachelor degree or higher level and 0 otherwise Income Lincome : 1 if household Annual Income is less than $24,9999 and 0 otherwise Mincome : 1 if household Annual Income is between $25,000 $ 74,999 and 0 otherwise Hincome : 1 if Household Annual Income is greater than $75,000 and 0 otherwise Size Number of people in the household Work 1 if respondent is working and 0 otherwise Gender 1 if respondent is male and 0 otherwise Head 1 if res pondent is the household head and 0 otherwise
67 Table 3 4 Descriptive statistics for socioeconomic variables, E10 Variable AR FL VA Mean Std Min Max Mean Std Min Max Mean Std Min Max Member 0.05 0.23 0 1 0.11 0.31 0 1 0.12 0.32 0 1 Knowledge 0. 46 0.50 0 1 0.54 0.50 0 1 0.49 0.50 0 1 Ownership 0.98 0.13 0 1 0.89 0.31 0 1 0.90 0.30 0 1 Miles Week 169.3 142.7 0 750 121 102.4 0 420 168.4 179.7 0 1100 Age 51.9 12.8 22 76 52.3 17 18 81 46.7 15.4 20 89 Less high 0 0 0 1 0.08 0.27 0 1 0.06 0.23 0 1 High 0.21 0.41 0 1 0.26 0.44 0 1 0.30 0.46 0 1 College 0.34 0.47 0 1 0.34 0.47 0 1 0.32 0.47 0 1 Bachelor 0.45 0.50 0 1 0.32 0.47 0 1 0.32 0.47 0 1 Gender 0.41 0.49 0 1 0.45 0.50 0 1 0.46 0.50 0 1 Head 0.93 0.26 0 1 0.92 0.27 0 1 0.87 0.34 0 1 Lincom e 0.14 0.35 0 1 0.21 0.41 0 1 0.17 0.38 0 1 Mincome 0.63 0.48 0 1 0.51 0.50 0 1 0.35 0.48 0 1 Hincome 0.23 0.42 0 1 0.28 0.45 0 1 0.48 0.50 0 1 Size 2.39 1.31 1 6 2.30 1.40 1 9 2.65 1.26 1 6 Work 0.64 0.48 0 1 0.43 0.50 0 1 0.67 0.47 0 1 Number of obs ervations 330 456 414
68 Table 3 5 Descriptive statistics for socioeconomic variables, E85 sample Variable AR FL VA Mean Std Min Max Mean Std Min Max Mean Std Min Max Member 0.04 0.19 0 1 0.06 0.24 0 1 0.07 0.25 0 1 Knowledge 0.43 0.50 0 1 0.43 0.50 0 1 0.39 0.49 0 1 Ownership 0.96 0.19 0 1 0.92 0.27 0 1 0.99 0.12 0 1 Miles week 128.8 127.3 0 580 132.3 126.4 0 500 158.5 146.4 0 750 Age 53.1 14.6 21 87 51.1 18.9 19 89 45.4 15.1 18 81 Less high 0.02 0.14 0 1 0.10 0.30 0 1 0.07 0.25 0 1 High 0.13 0.34 0 1 0.24 0.43 0 1 0.31 0.46 0 1 College 0.42 0.49 0 1 0.37 0.48 0 1 0.26 0.44 0 1 Bachelor 0.43 0.50 0 1 0.29 0.45 0 1 0.36 0.48 0 1 Gender 0.36 0.48 0 1 0.82 0.38 0 1 0.45 0.50 0 1 Head 0.94 0.23 0 1 0.47 0.50 0 1 0.92 0.27 0 1 Lincome 0.21 0.41 0 1 0.27 0.44 0 1 0.12 0.33 0 1 Mincome 0.53 0.50 0 1 0.43 0.50 0 1 0.53 0.50 0 1 Hincome 0.26 0.44 0 1 0.30 0.46 0 1 0.35 0.48 0 1 Size 2.30 1.19 1 5 2.30 1.37 1 6 2.43 1.30 1 7 Work 0.53 0.50 0 1 0.55 0.50 0 1 0.68 0.47 0 1 Number of obse rvations 306 474 444
69 Table 3 6 Probabilities of using biofuels for level of environmental attributes Attribute E10 E85 Reco2 Level AR FL VA Level AR FL VA 1 3% 0.476 0.382 0.367 1 60% 0.667 0.534 0.573 4 7% 0.491 0.484 0.468 61 70% 0.520 0.5 63 0.577 8 10% 0.506 0.588 0.573 71 90% 0.370 0.592 0.580 Biomp 1 20% 0.429 0.481 0.453 1 25% 0.530 0.585 0.583 21 40% 0.490 0.489 0.449 26 50% 0.520 0.564 0.577 41 60% 0.551 0.497 0.445 51 75% 0.510 0.543 0.570
70 Table 3 7 Probit model results for E10 sample Variable AR FL VA Coefficients Std Coefficients Std Coefficients Std Reco2 0.020 0.102 0.261 *,**,*** 0.092 0.262 *,**,*** 0.095 Biomp 0.159 0.107 0.020 0.093 0.009 0.01 Prem 0.973 *,**,*** 0.322 0.883 *,**,*** 0.26 7 0.962 *,**,*** 0.310 Member 0.465 0.430 0.180 0.266 0.409 0.277 Knowledge 0.238 0.211 0.191 0.142 0.223 0.179 Ownership n.a n.a 1.146 *,**,*** 0.315 0.320 0.305 Miles week 0.0002 0.001 0.002 *,**,*** 0.001 0.002 *,**,*** 0.000 Age 0.017 **,*** 0.009 0.000 0.006 0.010 0.007 High 0.718 *,**,*** 0.239 1.267 *,**,*** 0.445 0.341 0.315 College n.a n.a 1.705 *,**,*** 0.391 0.570 0.377 Bachelor 0.050 0.269 2.095 *,**,*** 0.427 0.134 0.396 Gender 0.528 *,** 0.207 0.57 2 *,**,*** 0.151 0.355 0.210 Head 0.302 0.351 1.767 *,**,*** 0.344 0.423 0.267 Mincome 0.093 0.278 0.409 0.230 0.214 0.241 Hincome 0.763 *,** 0.368 0.660 *,**,*** 0.239 0.188 0.282 Size 0.205 *,** (0.014) 0.083 0.118 (0.114) 0.074 0.236 ,**,*** 0.074 Work 0.140 0.209 0.001 0.189 0.029 0.252 Intercept 1.393 0.829 2.848 *,**,*** 0.774 1.352 0.775 Number of observations 330 456 414 Log likelihood 186.7 227.3 237.84 Log likelihood ratio 67.34 129.9 86.47 p value 0.00 0.00 0. 00 Pseudo R 2 0.178 0.278 0.172 Notes : Significant at < 0.1; ** S i gnificant at < 0.05; *** Significant at < 0.0 1.
71 Table 3 8 Probit model results for E85 sample Variable AR FL VA Coefficients Std Coefficients Std Coefficients Std Reco2 0.441 *,**,*** 0.156 0.073 0.087 0.005 0.082 Biomp 0.017 0 .161 0.053 0.090 0.012 0.082 Prem 1.431 *,**,*** 0.227 0.508 *,**,*** 0.163 0.803 *,**,*** 0.153 Member n.a n.a 0.521 0.335 0.293 0.295 Knowledge 0.310 0.194 0.004 0.170 0.316 0.167 Ownership 2.936 *,**,*** 0.738 1.532 *,**,*** 0.309 n.a n.a Miles week 4.14E 06 0.0008 0.0003 0.0006 3.2E 05 0.0005 Age 0.055 *,**,*** 0.010 0.007 0.005 0.006 0.006 High 3.281 *,**,*** 0.636 0.037 0.287173 0.181 0.289 College 1.614 *,**,*** 0.493 0.179 0.289 0.056 0.290 Bachelor 1.870 *,**,*** 0.516 0.404 0.285 0. 0148 0.304 Gender 0.320 0.213 0.465 *,**,*** 0.155 0.287 *,** 0.143 Head 1.501 *,**,*** 0.414 0.733 *,**,*** 0.206 0.400 0.315 Mincome 0.187 0.309 0.258 0.193 0.463 0.245 Hincome 0.702 0.386 0.266 0.211 0.285 0.252 Size 0.349 *,**,*** 0.099 0.081 0.059 0.073 0.058 Work 0.569 *,** 0.257 0.139 0.167 0.07 0.190 Intercept 5.234 *,**,*** 1.141 0.559 0.540 1.055 *,** 0.512 Number of observations 306 474 438 Log likelihood 144.1 275.8 272.8 Log likelihood ratio 114. 4 88.8 43.36 p value 0.00 0.0 0 0.00 Pseudo R 2 0.316 0.16 0.09 Notes : Significant at < 0.1; ** S ignificant at < 0.05; *** Significant at < 0.0 1.
72 Table 3 9 WTP ($ gallon 1 ) for biofuels at state level Blend AR FL VA E10 0.56 0.58 0.50 E85 0.82 1.17 1.06
73 Figu re 3 1. Percentage of yes responses for E10 0 10 20 30 40 50 60 70 0.2 0.5 0.75 1 % Participation Premium $ gallon 1 AK FL VA
74 Figure 3 2. Percentage of yes responses for E85 0 10 20 30 40 50 60 70 80 0.3 0.6 1 1.5 % Participation Premium $ gallon 1 AK FL VA
75 CHAPTER 4 MODELING EFFECTS OF BIOENERGY MARKETS ON TRADITIONAL FOREST PRODUCT SECTORS Introduction Southern U.S. forests, c omprising 27% of the total forestland nationwide and being about 86% privately owned (Smith et al., 2009), can potentially account for a large portion of the annual national estimate of 348 million dry tons of woody materials that can be diverted from the U.S. fore sts for bioenergy purposes (Perlack et al., 2005). However, agricultural or even forest bioenergy markets might adversely impact traditional forest industry. Bioenergy production m ight lead to land use competition between food based biofuels and forestry ( A u l is i et al., 2007) C onversion of forest lands to agricultural lands would be likely if energy policies favor food based biofuels (Malmsheimer et al., 2008). C onverting forest lands and using them to produce food crop based biofuels releases more carbon dioxide into the atmosphere as compared to the greenhouse gas reduction provided by using these biofuels for energy production ( Fargione et al., 2008 ; Searchinger et al., 2008 ) Degraded lands can be considered for bioenergy plantations trees or other ene rgy crops to reduce erosion, restore ecosystems, and provide shelter to communities (Riso, 2003), but these benefits would only be realized under clear land use regulations, especially in places where forests are at risk of conversion to other land uses ( Worldwatch Institute, 2007) On the other hand, diverting wood for bioenergy production would increase competition among wood users (FAO, 2008). Increased demand of wood for energy implies that bioenergy plants would likely face competition with the energy industry for low quality fiber (Hillring, 2006). Aulisi et al. (2007) suggest that the pulp and paper industry and panel industry are more likely to be hurt by the emerging wood based energy industry. This study reports that the panel industry would face more competition because of absence of secondary products (sawdust, slabs,
76 and chips) to be provided to the energy sector. Although chemical pulp mills would also face competition and increased prices for fiber, they might opt for manufacturing high value products and position themselves as integrated forest biorefineries (Aulisi et al., 2007). High value solid wood product markets, on the other hand, are not likely to be affected by new bioenergy markets because they have no competition with woody biomass for energy (Scott and Tiarks, 2007). In fact, sawmills might benefit from bioenergy development due to the higher prices for secondary products such as sawdust and chips demanded by bioenergy markets (Bolskesjo et al., 2006; Aulisi et al., 2007). Differen t studies have explored the economic impact of emerging bioenergy markets on the agricultural sectors in the U.S Gen erally, these studies point towards an increase in crop and livestock prices farm income, and empl o yment. Westcott (2007) claimed that cor n prices will increase by more than 50 cents a bushel by 2016. Walsh et al. (2003) determined that crop prices and annual farm income will increase by 9% to 14% above a baseline farm gate price of $2.44 GJ 1 and $6 billion above baseline. Birur et al. (200 8) explored the impact of biofuels in world agricultural markets. Their finding revealed increases of 9%, 10%, and 11% in the prices of coarse grains, oilseeds and sugarcane in the U.S., European Union (E.U.), and Brazil, respectively. Although an extensiv e number of international economic models of forest and agricultural sectors have been developed that can include bioenergy trade assessment, such as the European Forest Institute Global Trade Model (EFI GTM), Global Forest Products Model (GFPM), Forest an d Agriculture Sector Optimization Model (FASOM), and European Non Food and Agriculture (ENFA) (Solberg at al., 2007), to the best of our knowledge, the implications of emerging woody bioenergy markets on the U.S. equilibrium supply and demand for other for est product markets have not been deeply researched. Thus, we define an econometric model that
77 accounts for the participation of forest bioenergy markets along with traditional forest product sectors. Further, we quantify the effects on the traditional for est product sectors by simulating an increase in the demand for forest biomass for bioenergy. Model and Econometric Specification The model consists of four sectors, namely forest landowners, sawmills, pulpmills, and the bioenergy sector. The supply side o f the model is represented by the forest landowners who require labor and capital to produce sawtimber, pulpwood, and biomass for bioenergy production. The demand side of the model is represented by sawmills, pulpmills, and bioenergy producing firms. Sawmi lls demand sawtimber, energy, labor, and capital to produce lumber. Pulpmills use pulpwood, energy, labor, and capital to produce pulp. Finally, bioenergy producing firms require biomass, energy, labor, and capital to generate electricity for household use For pulpmills or energy from bioenergy firms to either sawmills or pulpmills, are not considered. Furthermore, flows within sectors such as the self supply of en ergy are ruled out. Other assumptions are: firms maximize their profits and only output and input variables adjust when prices change, i.e., capital remains fixed in the short run. Given perfect competition in the four sectors, t he dual restricted profit function is defined as ( ) where is the short run profit function, and are vectors of output and input prices respectively with = ( ) and is the fixed vector of fixed inputs and outputs The profit function satisfies the followi ng properties: (i) non negativity; (ii) non increasing in ; (iii) non decreasing in ; (iv) convex and continuous function in ( ) ; (v) homogenous of degree 1 in ( ) ; and (vi) differentiable in and (Chambers, 1988). The last condition (Hot lemma) allows for obtaining functional forms for supply and demand. Thus:
78 ( p w v ) = (4 1) ( p w v ) = (4 2) Where and are the supply function and demand function for output and input respectively The conditions for the supply and demand functions are the following: (i) supply is increasing in ; (ii) demand is non increasing in ; (iii) supply and demand are homogenous of degree zero in ( ) ; (iv) cross price effects are reciprocal i n nature (Chambers, 1988). Flexible functional forms have been widely adopted for econometric analysis of supply and demand (Christensen et al., 1971; Die we rt, 1973). Flexibility refers to a process by which consumer preferences can be represented without imposing prior restrictions at a base point (Caves and Christenhen, 2009). Contrary to traditional forms such as Cobb Douglas and Leontief, in which the elasticities of substitution are 1 and 0, res pectively, the elasticity of substitution depends on data and varies across the sample (Chambers, 1988). However, this flexibility has its limitations. Flexible functional forms are well behaved over a limited range of points (Despotakis, 1986). Certain is sues such as multicollinearity and lack of requisite degrees of freedom forestall the process of approximating the underlying structural relation due to the limited number of observations in practical applications (Mountain and Hsiao, 1989). Further, these functional forms are not flexible while representing separable technologies (Chambers, 1988). For our study we chose the restricted generalized Leontief profit function (GLPF) widely used in the forest sector (Newman and Wear, 1993; Hardie and Parks, 199 6; Brannlund and Kristrom, 1996; Williamson et al., 2004). GLPF has some advantages over other flexible
79 functional forms such as the Translog and the Normalized quadratic. Morrison (1988) claims that due to the nonlinear logarithm form of the Translog it i s difficult to obtain numerical convergence for long run elasticities. Further, the use of the Normalized Quadratic functional form implies an arbitrary normalization using a numeraire input price, leading to an asymmetry of demand equations and lack of in variance of empirical results (Morrison, 1988). On the other hand, the primary limitation of GLPF is that it provides a poor regional approximation if the true technology allows easy input substitution or output transformation (Behrman et al., 1992), i.e., GLPF tends to underestimate price and substitution elasticities (Williamson et al., 2004). Following Diewert (1973), the restricted GLPF must: (i) be linear in parameters; (ii) contain precisely the number of parameters needed to provide a second order ap proximation to an arbitrary twice differentiable profit function; (iii) have a functional form that satisfies the appropriate regularity conditions over a range of values for the independent variables, given a simple set of inequalities restrictions on the unknown parameters; (iv) be homogenous of degree 1 in v (constant return to scale in all factors). The GLPFs for the forest landowners, sawmills, pulpmills and bioenergy sector are defined as follows: , = 1 2 1 2 + (4 3) = , = 1 2 1 2 + (4 4) =
80 ( , ) = 1 2 1 2 + (4 5 ) = ( , ) = 1 2 1 2 + (4 6) = Where are the short run profit functions for the forest landowner sector, sawmill industry, pulpmill industry, and bioenergy sector, respectively; are the output prices of sawtimber, pulpwood, and biomass for bioenergy production respectively, for the forest landowner sector; is the wage rate in forestry and is the forest capital. For the sawmills, is the price of lumber, denote the input prices of sawtimber, energy and wa ge rate and is the capital stock. For the pulpmills, P p is the price of pulp, and refer to the input prices of pulpwood, energy, and wage rate, respectively, and is the capital stock. For the bioenergy sector, is the price of electricity in the residential sector. The input prices of biomass, energy, and wage rate are denoted by respectively, and represents the capital stock. Recall that = = and = Symmetry is imposed on the system, i.e., = = = and = for all and 3, we obtain the supply of sawtimber, pulpwood, and biomass for bioenergy, and the demand for labor in forestry. Likewise, applying 3, 4 4, 4 5 and 4 6, we obtain the supply function of lumber, pulp, and electricity, and the demand function for inpu ts in the sawmill, pulpmill, and bioenergy sector, respectively. In our study we are particularly interested in deriving the forest landowner
81 supply functions and sawmill, pulpmill, and bioenergy sector demand functions for sawtimber, pulpwood, and biomass for bioenergy. Equations 4 7 to 4 12 represent our system of equations to be estimated. = 1 2 + = (4 7) = 1 2 + = (4 8) = 1 2 + = (4 9) = 1 2 + = (4 10) = 1 2 + = (4 11) = 1 2 + = (4 12) Where and derived from Equations 4 7, 4 8, and 4 9 are the supply of sawtimber, pulpwood, and biomass for bioenergy, respectively. On the other hand, an d from Equations 4 10, 4 11, and 4 12 represent the demand for sawtimber, pulpwood, and biomass for bioenergy, respectively. The market equilibrium conditions for sawtimber,
82 pulpwood, and biomass for bioenergy are obtained by equating supply wit h demand, i.e., = = and = The system of equations so arrived at becomes an econometric model by adding a disturbance term to each supply and demand equation. A salient feature of this simultaneou s equation system is that quantities of sawtimber, pulpwood, and biomass for bioenergy and their respective prices are jointly determined with the input and output factors. The right hand side endogenous variables from Equations 4 7 to 4 12 are correlated with the disturbance terms. The use of traditional ordinary least squares (OLS) in such situations will produce biased and inconsistent estimators (Wooldridge, 200 2 ). Thus, an instrumental variable method is required, where the instruments are the exogenou s variables correlated with endogenous variables but not correlated with the disturbance term. The estimation technique proposed is the two stage least squares (2SLS), which provides a consistent estimator and produces the most efficient instrumental varia ble estimator under the absence of heterokedasticity and autocorrelation (Greene, 2003). Following Baum (200 7 ), the 2SLS approach can be represented as = + where and represent 1 matrices of the dependent variable and the disturbance term, respectively, and where is the number of observations. is the matrix of independent variables. can be divided in 1 2 wi th 1 regressors 1 assumed to be endogenous i.e., E[ ] 0 and 1 exogenous regressors. exogenous instrumental variables are considered i.e., E[ ] = 0, forming a matrix The instruments can also be partit ioned into 1 2 where 1 instruments 1 are excluded instruments and 1 instruments 2 are included instruments. Recall that 2 is identical to 2 Summarizing, we have regressor = 1 2 = 1 2 = [Endo genous Exogenous] and instruments = 1 2 = [Excluded and Included]. The first stage of the 2SLS approach requires regressing 1 on performing OLS. The second stage
83 implies replacing 1 with their fitted values 1 and performing OL S of on 1 A sufficient condition to solve the structural equations is to meet the rank condition, i.e., the 1 matrix 11 from first stage, 1 = 1 2 [ 11 12 ] + has to have full column rank 1 If th e rank is < 1 the equation is underidentified. Another condition that is necessary but not sufficient is the order condition for identification. There must be at least as many excluded instruments 1 as there are endogenous regressors 1 If 1 = 1 or 1 1 the equation is identified or overidentified, respectively. We also derive the short run supply and demand elasticities. The own price and cross price elasticities of the system can be determined as: = 1 2 1 / 2 = ; (4 13) = 1 2 1 2 = ; = (4 14) = ; = = ; = = 1 2 1 / 2 = ; (4 15)
84 = 1 2 1 / 2 = ; = (4 16) = ; = = ; = Where and are the own price elasticities of the supply and demand functions respectively; and denote the cross price elasticity of the supply and demand; [ ] represents the parameters for = respectively. Data C onstruction This study uses annual input and output prices and capital data for the southern U.S between 1970 and 2006. The data used for this study were primarily derived from official statistical reports of the forest and energy industries. A total of 37 annual sets of observations were obtained. In cases where part of the information was not available for some years, data were generated through a method of linear interpolation between time periods. To facilitate comparison among different year datasets, all prices were deflated to 1997 dollars. Traditional F orest Product Se ctor Annual saw log production information was used for historical quantity of sawtimber The saw log production information was obtained from United States Timber Industry An Assessment of Timber Product and Use, 1996 (Johnson, 2001). This report provided discontinued saw log production data for the following years: 1970, 1976, 1986, 1991 and 1996. Linear interpolation using the rate of growth of historical softwood lumber production as per Howard (2007) was used to arrive at values between those given years to complete the dataset through 2006. Pulpwood production was procured from Trends in the Southern Pulpwood
85 Production, 1953 2006 (Johnson et al., 2008). The roundwood pulpwood production information was used to reflect quantity of annual pulpwood production for our study Prices for sawtimber and pulpwood stumpage were procured from U.S Timber Production, Trade, Consumption, and Price Statistics 1965 2005 (Howard, 2007) Softwood lumber and Southern Bleached Softwood Kraft (SBSK) pulp were used for the output price data of sawmills and the pulp industry respectively Annual data on ear nings for logging camps and contrac tors, lumber and wood products except furniture and paper and allied products, were used as labor prices for the forest landowner sector, sawmills, and pulp and paper industry, respectively. Output and labor prices for t he sawmills and pulp industry were procured from U.S. Timber Production, Trade, Consumption, and Price Statistics 1965 2005 (Howard, 2007) and Kinnucan and Zhang (2005). Data regarding electricity prices were obtained from the EIA State Energy Data System database. For the sawmills and pulp industry, the average electricity price in the industrial sector was used as the input price for energy. Standing inventory information was utilized as a proxy for capital in the forest landowner sector. Inventory data was sourced from the Forest Resources of the United States 2007 report (Smith et al., 2009) and the 2005 RPA Timber Assessment Update (Haynes et al., 2007). For the periods in which inventory data were not available, a linear interpolation technique was ap plied to generate information. S outhern sawmill and pulp mill capacities were used as prox ies for capital in the sawmill sector and pulp sector, respectively. Two sources were used to construct the time series information for the sawmill capacity. Data from 1995 onwards have been surveyed by Spelter (2007) and published in Profile 2007: Softwood Sawmills in the United States and Canada Profiles 1980) were recreated by accessing available yearly issues of the Directory of the Fores t Products Industry published by Miller
86 Freeman. Data gaps for certain years were filled by linear interpolation. Southern pulpmill capacity information was procured from Trends in the Southern Pulpwood Production, 1953 2006 (Johnson et al., 2008). B ioma ss for Bioenergy Biomass for bioenergy production information was obtained from Trends in the Southern Pulpwood Production, 1953 2006 (Johnson et al., 2008). Total Southern pu lpwood production was divided into the sum of the production of round wood pulpwoo d and wood residues. T he former was used for quantifying pulpwood production. The wood residue production information was utilized as a proxy for biomass for bioenergy production, as this production is a relatively recent phenomen on and information regardi ng it is not available in historical time series dating back to 1970. Pine pulp chip prices at the consuming mills were used as a proxy for biomass for bioenergy prices. These data were obtained from Timber Mart South (TMS) However, TMS only maintained pr ice series from 1976 onwards. For previous time periods the average growth of the price pulp chip series information was used to extrapolate data back to 1970. For the bioenergy sector the output price was represented by the total electricity average pri ce and the input price was reflected by coal expenditure s at the industry level (EIA State Energy Data System database) The reasoning behind this allocation was that coal production is mainly used up by the electricity s ector. 61% of coal production was consumed by the electric power sector in 1970 whereas this increased to 93% in 2008 averaging 83% for this period as a whole (EIA, 2009). Labor price information was procured from the U.S Department of Labor, Bureau of Labor Statistics Current Employmen t Statistics (CES) program. Southern installed generating capacity was chosen as a proxy for capital in the bioenergy sector and was taken from the U.S. Census Bureau Statistical Abstracts of the United States from
87 1970 to 2006. Table 4 1 shows the descri ptive statistics of the data set. The complete data set can be found in Appendix C. Results and Discussion STATA 9.2 software was used to estimate the econometric model. Initially, all the exogenous variables ( , ) in the SES were employed as instruments for estimation. The Kleibegen Paap rk LM statistic test was performed for each equation to test the rank condition under the null that the equation is underidentified ( 0 = 1 1 ) and distributed Chi square with L K+ 1 degrees of freedom. The Chi square (6) for and accounted for 10.62 and 9.86, respectively, failing to reject the null of underidentification at = 0.05 level of significance. However, the Chi square (6) for was 13.53, rejecting the null of underidentification at = 0.05 but failing to reject it at = 0.01 level of significance level (Table 4 2). The redundancy of certain instrumental variables not likely to be correlated with the endogenous variable was also tested. For example, we postulated that wages, input energy price, and capital in the energy sector were not highly correlated with sawtimber and pulpwood prices. Likewise, we assumed that wages, input price of energy, and capital of the sawmill i ndustry had little explanatory power on the biomass for bioenergy price. The redundancy test was performed under the null that n times the sum of the square canonical correlations between the potential redundant instrumental variables and the endogenous re gressors was zero. It is distributed Chi square with degrees of freedom equal to the number of endogenous variables times the number of instruments being tested. Thus, the Chi square (18) for the instruments already mentioned was 14.51, 21.48, and 28.43 fo r and respectively, failing to reject the null of the instruments being redundant at = 0.05 level of significance (Table 4 2). Thus, a new set of
88 instrumental variables was chosen, eliminating the exogenous variables that pr oved to be redundant. Table 4 3 shows the coefficients, robust standard errors, and values of this new model. Serial correlation was analyzed by the Durbin Watson (DW) statistic. The DW statistic for and was inconclusive, whereas it showed no evidence of positive autocorrelation for The rank condition was again tested. The Chi square (3) for and Chi square (4) for were 9.91, 8.90 and 8.17, respectively, rejecting the null hypotheses of underid entification at = 0.05 for the first two and rejecting it at = 0.1 level of significance for the third (Table 4 3). Thus, there is strong evidence that the instruments are adequate to identify the equations. As the number of excluded instruments is lar ger than the endogenous regressors, the Hansen test of overidentifying restrictions was also conducted in order to ascertain the validity of the instruments. Under the null hypotheses, the instruments are deemed to be valid, i.e., independent from the dist urbances. The Hansen statistic is distributed Chi square with L K overidentifying restrictions. The Chi square (2) for was 13.80, 5.36, respectively. The Chi square (3) for was 11.61. We failed to reject the null hypotheses of valid instruments for the endogenous variables related to price of pulpwood at = 0.05. This cast a slight doubt on t he orthogonality of the instruments for the endogenous variables of equations and Symmetry was tested using a Chi square statistic under the null that cross coefficients are equal. Results showed that the null was rejected at = 0.05 (Table 4 4). An F test was conducted to compare the model with constrained parameters (restricted model) versus the model without imposed symmetry (unrestricted model). The F ( 2 28) values were extremely low, and we failed to reject the null that both mod els were identical. Thus, the symmetry imposed model was preferred over the unrestricted model. We also tested the contemporaneous correlation of the
89 residuals, performing a Breusch Pagan test under the null of independent residuals (Table 4 4). The Chi sq uare (3) statistic showed that the null was rejected at = 0.03, indicating some evidence of correlation between the disturbances. This might suggest using 3SLS for solving the simultaneous equation system. Under the evidence of contemporaneous correlatio n of the residuals and overidentification in the system, the three stage least squares (3SLS) approach would result in a more efficient estimator than 2SLS (Wooldridge, 2002). However, 3SLS requires the equations to be correctly specified, and if that cann ot be assured, a single equation procedure such as 2SLS is more robust (Wooldridge, 2002). Further, 2SLS performs better than 3SLS in small sample sizes (Heij et al., 2004). Thus, because of the small number of observations (37) and the lack of certainty r egarding specification of the model, we preferred a conservative approach of following 2SLS, trading off efficiency for robustness. Coefficients and Elasticities of the Residues Only Biomass (ROB) Model Our data construction stated that wood residues are o nly used as a proxy for reflecting the production of biomass for bioenergy. As the coefficient values of the restricted model (hereafter called ROB model ) do not provide intuitive interpretations, we have focused on the significance of the parameters. 33% of the parameters were significant at = 0.1 or lower level of significance. Regarding the forest landowner sector, only the parameters representing wages for the supply of sawtimber as well as wages, sawtimber, and pulpwood price were significant at = 0.05. Turning to the sawmill sector, parameters representing energy, wages, and capital were significant at = 0.05. Only capital in the pulp industry and output electricity price in the bioenergy sector were significant at = 0.1. The short run supply an d demand elasticities valuated at the mean values and their standard errors are presented in Table 4 5. Practically all the elasticities were significant at =
90 0.05. As expected, the own price elasticities of the supply of sawtimber and pulpwood are posit ive. However, the own price elasticity of the supply of biomass for bioenergy is negative, which is not consistent with the economic theory. The price elasticity of the supply of sawtimber proved to be elastic, while the supply of pulpwood and biomass for bioenergy were inelastic, showing less influence of changes in price on supply of the latter two products. Cross price elasticities of the supply showed dissimilar results. The elasticity of the supply of sawtimber and pulpwood with respect to labor had t he expected sign and a great influence on the supply of both products. The cross price elasticity of sawtimber with respect to the price of pulpwood was positive, indicating that sawtimber and pulpwood are complements. Likewise, biomass for bioenergy and p ulpwood seemed to be complements. Sawtimber and biomass for bioenergy, on the other hand, proved to be substitute products. Dissimilar results have been found regarding short run elasticities in timber market models. Newman and Wear ( 1993) and Wear and New man (1991) claimed negative cross price between pulpwood and sawtimber in the short run Ankarhem et al. (1999) claimed that pulpwood and biomass for bioenergy are complements in a study about modeling the effects of a rise in the use of forest resources f or energy generation in Sweden. In a similar study, Ankarhem (2005) found that sawtimber, pulpwood and biomas s for bioenergy are substitutes for each other. Turning to the demand elasticities, the own price elasticity of the demand for sawtimber and pul pwood were negative, as expected. The own price elasticity of the demand for biomass for bioenergy was positive, which seems quite implausible. The response of demand to changes in price was elastic in the case of sawtimber and inelastic for pulpwood and b iomass for bioenergy. As expected, labor and sawtimber, as well as input energy and biomass for bioenergy, came out as complements. The price of lumber had a positive effect on demand for sawtimber.
91 Smaller cross price effects occurred in the pulp and bioe nergy industry. With the exception of energy in the pulp industry and output electricity price in the bioenergy industry, the demand elasticities were all less than one. This implies that a small percentage in any of the inputs for both industries results in a change less than proportional to the demand for pulpwood and biomass for bioenergy. The signs of some elasticities are not consistent with economic theory. For example, the positive cross price elasticity of the demand for sawtimber with respect to th e price of energy suggested that energy and sawtimber are substitutes in production. Other implausible implications from the cross price elasticities are that energy and pulpwood, labor and pulpwood, and labor and biomass for bioenergy seemed to be substit utes in production. Output prices of pulp and electricity have a negative impact on the demand for pulpwood and biomass for bioenergy, respectively. Thus, the results and implications outlined here should be considered with a degree of caution Calibratio n of the Residues Pulpwood Biomass (RPB) Model We found that pulpwood and biomass for bioenergy were complements implying that an increase in price of either pulpwood or biomass for bioenergy would lead to an increase in the supply of both forest products This degree of complementarity is consistent as per our data construction. Wood residue production used as a p roxy for biomass for bioenergy is a part of total s outhern pulpwood production which has increased along with softwood pulpwood production (J ohnson et al., 2008) However, competition for wood with biorefineries and pulpmills could lead to higher prices for forest products. Thus, under the scenario what would have been if part of the production of pulpwood was considered for production of bioma ss for bioenergy, we calibrated a new model called RP B model Additionally, substitutability regarding
92 the effect of labor on the supply of biomass for bioenergy and the price of electricity on the demand for biomass for bioenergy were also imposed. We art ificially modeled an increase in the share of wood residues over pulpwood production over time to reflect competition between these two forest products. We assumed that 1% of the pulpwood production would go for biomass for bioenergy production each year. Consistent with changes in production, it is also expected that pulpwood and biomass for bioenergy prices change over time. Available elasticities provided by the forestry literature were used to account for changes in prices. A short run own price supply elasticity of 0.48 wa s reported by Newman and Wear (1993) and was used for pulpwood. We also assumed a lower degree of market power for forest biomass for bioenergy. Thus, a supply elasticity of 0.3 found by Galik et al. (2009) for pulpwood was used as a p roxy for biomass for bioenergy The cross price elasticity between sawtimber and pulpwood proved to be complements in the ROB model We maintained this assumption, thus, a cross price elasticity of pulpwood with respect to sawtimber of 0.15 (Wear and Newma n, 1991) was chosen to calculate the sawtimber price. An own price elasticity of 0.55 (Newman, 1991) was considered for sawtimber. Because bioenergy production is a recent activity, historical data from 1990 until 2006 was used to simulate the elasticities of the RP B model Table 4 6 shows the coefficients, robust standard error s, and values of the RPB model The DW statistic for and were show n to be inconclusive Table 4 7 shows the elasticities of the RP B model All elasticities were significant at = 0.05 The cross price elasticity of pulpwood with respect to the biomass for bioenergy was negative, confirming the substitutability between the products. This implies that an increase in the price of biomass for bioenergy will decrease the production of pulpwood, redirecting the production toward biomas s
93 for bioenergy. Further, consistent with economic theory, the own price elasticity of the supply and price elasticity of the demand for the three forest products were positive and negative, respectively. Contrary to the ROB model the demand equations sho wed that biomass for bioenergy and energy are substitutes, and biomass for bioenergy and labor are complements. Policy Simulation The Energy Independence Policy Act of 2007 and American Clean Energy Security draft Bill of 2009 set renewable fuel and electr icity standards. An induced government policy scenario would favor the production of bioenergy. Thus, the demand for biomass from the bioenergy sector is increased to produce renewable energy. For simulation purposes, we have considered a scenario in which the demand for biomass for bioenergy is increased by 15% for the ROB and RP B model The simulation procedure was carried out first, developing a baseline for quantities and prices of sawtimber, pulpwood, and biomass for bioenergy using the average data o f labor, energy, capital, and output prices of each sector between 1990 and 2006 The estimated equations, then, were used to predict the output baseline for sawtimber, pulpwood, and biomass bioenergy. For our policy simulation, we increased the demand for bio mass for bioenergy by 15% for both models. T he estimated supply and demand equations were adjust ed for the new equilibrium quantities and prices for sawtimber, pulpwood and biomass for bioenergy Finally, the new equilibrium was compared and contrasted with baseline information to discern the percent change. Table 4 8 shows the baseline and the policy scenario results for the ROB and RPB model For the ROB mode l the increase in the demand for biomass for bioenergy by 15% impl ied an increase of 4.5 % i n the quantity of sawtimber while the sawtimber price decreased by 2.9 %. The p ulpwood price increased by 91.3 % while the quantity decreased by 27.3 %. Price of biomass for
94 bioenergy decreased by 4.9 %. The demand for s awtimber and biomass for bioenergy mov ed in opposite directions supporting the evidence of substitutability found in the cross price elasticities. Given the negative own supply elasticity of biomass for bioenergy, the increased demand for biomass for bioenergy resulted in a lower price for bi omass for bioenergy for the forest landowner sector The negative cross price elasticity of the sawtimber with respect to biomass for bioenergy implied expanding supply of sawtimber. The demand for sawtimber facing an increased level of production adjust ed by reducing the equilibrium sawtimber price The reduced price of sawtimber and biomass for bioenergy, in turn, led to an inward shift of the supply of pulpwood. Thus, the demand for pulpwood adjusted to meet the contracted production increasing the eq uilibrium price of pulpwood. With regard to the RPB model the increased demand for biomass for bioenergy caused an increase in the price of biomass for bioenergy, stimulating the forest landowner to expand the supply. Price of biomass for bioenergy increa sed by 51.9%. The substitutability between biomass for bioenergy and pulpwood resulted in an inward shift of the supply of pulpwood. The demand for pulpwood adjusted to a lower level of production a fall of 20% increasing the equilibrium price by 103.5%. Cross price elasticity between sawtimber and biomass for bioenergy showed that both products were substitutes and had a magnitude higher than that between sawtimber and pulpwood. Thus, the supply of sawtimber would be expected to contract. The overall eff ect was a 11.1% decrease. The demand for sawtimber adjusted in terms of a rightward shift, reflecting a 6.5% increase in equilibrium price. In sum mary for the ROB model the increased demand for biomass for bioenergy increased the price of pulpwood and de creased the prices of biomass for bioenergy and
95 sawtimber. Quantity of pulpwood was reduced while quantity of sawtimber expanded In case of the RPB model prices of biomass for bioenergy, pulpwood, and sawtimber increased. Quantities of pulpwood and sawti mber supplied decreased. The total value of the demand for forest resources increased in the case of both models. For the ROB model the values of the equilibrium quantities price times quantity increased by 1.5% 38.9%, and 9.2% for sawtimber, pulpwood and biomass for bioenergy, respectively, in case of a 15% increase in demand for biomass for bioenergy (Table 4 9). The gain of the quantity values for the forest landowners accounted for 1 0.9% (US$ 862 .5 million). The forest landowners, in turn, sell less pulpwood at a higher price and more biomass for bioenergy and sawtimber. Therefore, the timber sales offset the losses incurred by selling more sawtimber and biomass for bioenergy at a lower price and less pulpwood. With regard to the RPB model the value s of the equilibrium quantities increased for pulpwood and biomass for bioenergy, accounting for a rise of 8.9% (US$ 565.3 million) of the total value. On the other hand, the pulpmill sector was adversely affected in both models. Pulpmills would have to pa y more for pulpwood and acquire less pulpwood from landowners. The bioenergy sector benefited as it could afford buying more biomass for bioenergy in both models. Sawmills in the ROB model also benefited buying more sawtimber at a lower price. However, t his sector was negatively affected in the RPB model buying less sawtimber at a higher price. The r esults of this study are in line with the findings of Ankarhem (2005) which claimed a positive effect of an increased demand for biofuels on the forest land owner and energy sector. Conclusions An econometric model was constructed to represent the interactions and impacts of wood bioenergy markets in the southern U.S. Four sectors were identified: the forest landowners, who supply sawtimber, pulpwood and bioma ss for bioenergy; the sawmill sector, which demands
96 sawtimber to produce lumber; the pulpwood sector, which uses pulpwood to produce pulp and paper; and the bioenergy sector, which requires biomass to produce electricity. First, we estimated the partial eq uilibrium model; then, elasticities were calculated and we termed it the ROB model In addition, we calibrated a new model under the assumption that production of biomass for bioenergy reduced the production of pulpwood. A policy scenario was simulated in which the demand for biomass for bioenergy was increased by 15%. The effect of increased demand for biomass for bioenergy on the other sectors was assessed for both models. The short run supply and demand elasticities calculated at the mean values showed d issimilar results in the two models. In the ROB model o wn supply price elasticities were positive for sawtimber and pulpwood but the own elasticity supply of biomass for bioenergy was negative. The own supply price elasticities were all positive in the R PB model In general, sawtimber supply proved to be more sensitive to changes in its own price compared to pulpwood and biomass for bioenergy supply functions. Evidence showed that sawtimber and pulpwood and sawtimber and biomass for bioenergy are compleme nts and substitutes in the ROB and RPP model respectively. However pulpwood and biomass for bioenergy came out as complements in the ROB model b ut substitutes in the RP B model Consistent with expectations, the own price elasticities of the demand for s awtimber and pulpwood in both models came out as negative. We only found inconsistency in the positive sign of the own price elasticity of the demand for biomass for bioenergy in the ROB model The own price demand elastic ity was elastic for sawtimber and inelastic in the cases of p ulpwood and biomass for bioenergy. In the ROB model when biomass for bioenergy and pulpwood were complementary products, the increase in the demand for biomass for bioenergy resulted in an increase of the
97 equilibrium price of p ulpwood and a decrease in the equilibrium price of biomass for bioenergy. Further, the quantity of pulpwood was reduced. We are aware that the inconsistency regarding the sign of the own price elasticity of biomass for bioenergy implied a decrease in the p rice of this forest product. The price of sawtimber was reduced, while the quantity of sawtimber increased. The results of the RPB model concluded that an increase in the demand for bioenergy indicated an increase in the price of pulpwood and biomass for b ioenergy and a decrease in the quantity of pulpwood. On the other hand, the quantity of sawtimber decreased while the price increased. The inclusion of this policy scenario resulted in a greater value of the demand for forest resources in turn benefiting the forest landowners. The bioenergy sector could afford to buy more biomass to generate bioenergy. The sawmill sector benefited by t he increase in the demand for biomass for bioenergy in the ROB mode l However, this sector was financially damaged regardin g the RPB model The pulp mill sector was adversely impacted in both models, having lower amount s of pulpwood available at a higher price. Our study itself could be improved in different ways. Time series data could be used to forecast future production of biomass for bioenergy and its tradeoffs with other forest products Incorporation of cross sectional data in light of data availability would provide a more complete assessment of the impacts of future bioenergy markets. Another plausible option would be to assume that capital stocks adjust over time to see the long run responses regarding change in input and output prices. Under this assumption, it is expected that elasticities become larger in magnitude due to the Chatelier principle. Finally, the paper industry has been historically concentrated over time, showing a high degree of market power (Mei and Sum, 2008). Thus, a reasonable assumption would be to model the pulp markets as an oligopsony.
98 T able 4 1. Descriptive statistics of the data set Sector Variable Unit Mean Std Min Max Forest landowners Million m 3 61.28 19.57 29.84 91.01 Million m 3 61.33 5.74 49.52 71.69 Million m 3 26.84 4.54 12.97 32.12 $ m 3 58.54 17.92 31.57 92.1 $ m 3 29.28 2.46 24.77 36.16 $ m 3 6.39 1.22 4.46 9.48 $ hour 1 9.91 3.39 3.43 15.56 Million m 3 2917.59 161.21 2524.63 3295.57 Sawmill sector $ KWh 1 0.048 0.0067 0.039 0.06 $ hour 1 8.19 3.0 6 2.97 13.32 $ m 3 6 5.47 23.07 21.53 100.01 Million ton year 1 34.65 8.19 13.9 48.6 Pulpmill sector $ KWh 1 0.048 0.006707 0.039 0.06 $ hour 1 10.97 4.59 3.51 18.36 $ ton 1 73.91 31.78 22.33 125.15 Million ton year 1 43.7 1 5.97 30.79 51.32 Bioenergy sector $ KWh 1 0.0048 0.000695 0.00321 0.00588 $ hour 1 19.65 2.83 15.27 24.85 $ KWh 1 0.066 0.00445 0.06 0.075 Million KWh year 1 283.98 77.40 127.4 429.01
99 Tab le 4 2. Rank and redundancy tests for all instruments Equation Kleibergen Paap rk LM statistic Chi square (6) value LM test of redundancy Chi square (18) value instruments test 10.62 0.1 14.51 0.69 9.86 0.13 21.48 0.25 13.53 0.03 28.43 0.06
100 Table 4 3 Coefficients, standard errors, p values and serial cor relation of the system of equation s Sawtimber supply Coefficients robust SE value Sawtimber demand Coefficients robust SE value 288.41 526.13 0.292 30.77 137.27 0.411 13.99 63.21 0.824 1358.62 633.47 0.032 13.77 236.06 0.953 8568.66 2030.68 0.000 862.21 430.08 0.045 41.73 41.79 0.318 0.24 0.11 0.267 8.058 1.57 0.010 Pulpwood supply Coefficients robust SE value Pulpwood d emand Coefficients robust SE value 114.84 328.05 0.363 130.72 87.27 0.067 13.99 10.79 0.195 210.02 185.92 0.258 27.09 51.28 0.597 616.56 1063.61 0.562 224.33 169.55 0.185 9.176 19.21 0.633 0.015 0.031 0.814 1.7171 0.47 0.067 Biomass for bioenergy supply Coefficients robust SE Value Biomass for bioenergy demand Coefficients robust SE value 157.39 150.63 0.148 112.27 31.51 0.000 13.77 3.88 0.000 1.485 13.67 0.913 27.09 9.27 0.003 1907.46 1383.73 0.168 62.88 18.62 0.000 789.6 2 418.30 0.059 0.030 0.025 0.541 0.19 0.075 0.209 Equation F value 42.72 11.51 18.63 value 0.00 0.00 0.00 R square 0.92 0.75 0.84 Durbin Watson 1.48 1.37 2.12
101 Table 4 4 Rank, overidentifying restriction, model comparison, symmetry and contemporaneous correlation tests Equation Rank test Kleibe r gen Paap rk LM statistic 9.91 8.90 8.17 Chi square (3) value for C hi square (4) value for 0.019 0.03 0.08 Overidentifying restriction test Hansen test 13.80 5.36 11.61 Chi square(2) value Chi square (2) value for Model comparison Restricted vs. unrestricted model F ( 2 28) 0.0008 0.0014 0.0007 Symmetry test = = = Chi square (1) value 0.11 0.88 0.13 Contemporaneous correlation Breusch Pagan test 8.99 Chi square (3) value 0.03
102 Table 4 5 Estimates and standard errors (SE) of the short run supply and demand elasticities of the ROB model Sawtimber supply Pulpwood supply Biomass for bioenergy supply Est imates SE Estimates SE Estimates SE 2.851 1.316 *,* 0.161 0.022 *,** ,*** 0.776 0.181 *,** ,*** 0.081 0.044 0.800 0.177 *,** ,*** 1.080 0.227 *,** ,*** 0.037 0.024 0.103 0.016 *,** ,*** 1.762 0.217 *,** ,*** 2.894 1.333 *,** 1.064 0.181 *,** ,*** 1.458 0.258 *,** ,*** Sawtimber demand Pulpwood demand Biomass for bioenergy demand Est imates SE Est imates SE Est imates SE 2.617 0.982 *,** 0.969 0.142 *,** ,*** 0.577 0.216 *,** 4.147 1.771 *,** 1.048 0.205 *,** ,*** 0.048 0.013 *,** ,*** 2.013 1.271 0.205 0.033 *,** ,*** 0.976 0.130 *,** ,*** 0.483 0.196 *,** 0.284 0.057 *,** ,*** 1.504 0.324 *,** ,*** Elasticities valued at the mean values Notes: Significant at <0.1; ** Significant at <0.05; *** Significant at <0.01
103 Table 4 6 Coefficients, standard errors, p values and serial correlation of the RPB mo del Sawtimber supply Coefficients robust SE value Sawtimber demand Coefficients robust SE value 5.84 548.93 0.49 6 156.35 280.30 0.288 4.24 55.85 0.469 1575.86 904.07 0.04 0 7.37 208.00 0.48 6 7903.36 1970.73 0 .000 1139.65 609.96 0.031 67.21 40.49 0.048 0.18 0.09 0.021 7.12 1.86 0 .000 Pulpwood supply Coefficients robust SE Value Pulpwo od demand Coefficients robust SE value 172.14 706.51 0.403 83.16 208.53 0.345 4.24 6.94 0.27 1 33.10 113.90 0.385 20.54 49.79 0.339 1195.24 608.69 0.02 5 110.14 107.30 0.152 2.67 13.05 0.418 0.02 0.03 0.229 1.32 0.41 0 .000 Biomass for bioenergy supply Coefficients robust SE Value Biomass for bioenergy demand Coefficients robust SE value 284.28 1548.47 0.427 213.00 314.78 0.249 7.37 2.16 0 .000 16.88 11.42 0.069 20.54 6.43 0 .000 694.04 1033.83 0.251 179.82 16.68 0 .000 162 7.23 305.14 0 .000 0.05 0.02 0 .000 0.23 0.06 0 .000 Equation F value 35.48 10.06 102.08 value 0.00 0.00 0.00 R square 0.91 0.73 0.96 Durbin Watson 1.37 1.43 2.34
104 Table 4 7 Estimates and standard errors (SE) of the short run supply and demand elasticities of the RPB model Sawtimber supply Pulpwood supply Biomass for bioenergy supply Estimates SE Estimates SE Estimates SE 3.989 0.739 *,**,*** 0.083 0.021 *,**,*** 0.125 0.010 *,**,*** 0.019 0.004 *,**,*** 1.171 0.332 *,**,*** 0.191 0.026 *,**,*** 0.048 0.008 *,**,*** 0.323 0.077 *,**,*** 1.628 0.143 *,**,*** 3. 960 0.734 *,**,*** 0.931 0.275 *,**,*** 1.312 0.139 *,**,*** Sawtimber demand Pulpwood demand Biomass for bioenergy demand Estimates SE Estimates SE Estimates SE 4.307 0.687 *,**,*** 0.735 0.176 *,**,*** 0.895 0.084 *,**,*** 5.077 0.896 *,**,*** 0.306 0.089 *,**,*** 0.158 0.014 *,**,*** 1.586 0.341 *,**,*** 0.584 0.132 *,**,*** 0.099 0.010 *,**,*** 0.817 0.094 *,**,*** 0.154 0.04 4 *,**,*** 0.836 0.081 *,**,*** Elasticities valued at the mean values Notes: Significant at <0.1; ** Significant at <0.05; *** Significant at <0.01
105 Table 4 8 Baseline and policy scenario results for ROB and RPB m odel s Variable Baseline ROB model 15 % increase demand for biomass % change Baseline RPB model 15 % increase demand for biomass % change 72.64 70.55 2 87 68.29 72.72 6.48 29.63 56.68 91 27 14.52 29.56 103.56 5.82 5.53 4.98 15.13 22.94 51.93 79.67 83.24 4.47 74.36 66.06 11.16 66.03 47.96 27.35 46.95 37.54 20.03 28.38 32.64 15.00 36.56 42.05 15.00
106 Table 4 9. Total timber sales values for ROB and RP B mod e ls (million US$) Forest product Baseline ROB model Total value 15% biomass demand increase % change Baseline RPB model Total value 15% biomass demand increase % change Sawtimber 5788.15 5873.26 1.47 5078.91 4804.57 5.40 Pulpwood 1956.78 2718.94 38. 94 681. 956 1110.06 62.77 Bio mass for Bioenergy 165.23 180.44 9.23 553.41 964.91 74.35 Total 7910.18 8772.65 10. 9 6314.28 6879.55 8.95
107 CHAPTER 5 SUMMARY AND CONCLUSIONS Introduction Ongoing environmental issues such as global w arming have prompted policy makers to promote new environmentally friendly technologies and energy systems to meet human requirements. Southern f orests can play a d ual role in fulfill ing energy requirements decreasing the dependency on foreign oil, and se rving as a neutral source of CO 2 Forests sequester carbon from the atmosphere storing it in biomass and produc e biofuels replac ing fossil fuels. In addition, production of forest biomass bioenergy might increase the currently depressed profitability of southern forestlands. Potential policies that enhance the production of energy from the forest to reduce GHG emissions, reduce energy dependence, and stimulate forestry sector are of national interest. Thus, it is necessary to evaluate the role that fore st biomass bioenergy markets will have for family forests in terms of management and economic returns. Also, information about human behavior and its perspectives regarding the use of green energy is needed. Further, it is necessary to identify the implica tions of growing forest bioenergy markets on the traditional forest products due to policies favoring the development of new sources of bioenergy. Insights from these analyses are important to evaluating the efficiency and effectiveness of current and alte rnative potential policies that stimulate the use of bioenergy. Following this rationale, the aim of this dissertation was to explore some economic aspects of utilizing forest biomass based bioenergy in the southern U.S.
108 Results from Modeling Impacts of Bioenergy Markets on Nonindustrial Private Forest Management In general terms, the inclusion of thinnings increased the profitability of slash pine forest stands regardless of the catastrophic disturbance risk levels. Assuming a positive salvageable portio n, the profitability of forest stands measured as land expectation value ( LEV ) was higher when thinnings were performed for either pulpwood or bioenergy purposes 4.6% and 5.8%, respectively than the LEV when no thinnings were carried out. Thus, when the w hole forest stand becomes commercially marketable, the revenues obtained offset the costs of thinning. However, when salvage accounted for zero, the LEV for no thinning scenario exceeded both of the thinning scenarios by 3.5% and 2.6%, respectively. In the case of the thinning scenario for bioenergy the revenues associated with the increase of the volumetric growth after thinning for greater value added product and the low price for forest biomass based bioenergy were not enough to cover the loss of volume and consequently the profits at the time of thinning A b ioenergy price of $5 ton 1 made the land value break even f or the bioenergy scenario compared to the status quo when the stand was completely destroyed Devoting part of the forest stand to bioenerg y purposes was more profitable than the scenario of producing pulpwood. In fact, the LEV for the thinning scenario for bioenergy was higher than the LEV for the thinning scenario for pulpwood at all comparable salvage/risk levels, due to the utilization of non commercial forest biomass for bioenergy production. On average, the LEV was greater by 11.5% and 11.7% for salvage levels of k= 0.8 and k= 0, respectively. Further, when risk was decreased by 1%, the LEV increased by 9% and 19% for salvage levels of k = 0.8 and k= 0, respectively. Thus, policies that help landowners mitigate risk through silvicultural interventions to reduce the size of the damage would have a positive impact on the profitability of forest stands.
109 We also simulated an independent increa se in bioenergy prices and volatility prices assuming that forest biomass based bioenergy markets continue to expand. As expected, the LEV increased as price and volatility increased. When price increased, the LEV increased by 6.4% and 6.1% for levels of k = 0.8 and k= 0, respectively. The LEV was greater than the original thinning scenario for bioenergy thinning scenario for pulpwood and no thinning scenario by 9.4%, 10.6% and 16% respectively, for a salvage portion of k= 0.8. When the forest stand was com pletely destroyed, the differences were 9%, 10.2% and 6.4%, respectively. On the other hand, the increase in the LEV was much lower as volatility prices increased. On average, the increase was 0.6% for all levels of salvageable portions. Compared to the or iginal thinning scenario for bioenergy thinning scenario for pulpwood and no thinning scenario the LEV increased by 1.4%, 2.5% and 1.07%, respectively, for a level of k = 0.8. For k = 0, the increase was 1.2%, 2.4%, and 1%, respectively. Thus, the combin ed effect of increasing prices and volatilities will be greater returns to forest landowners. Results from Assessing Public Preferences for Forest Biomass based Bioenergy The results indicated that individuals had a positive extra willingness to pay (WTP) for E10 and E85 and a greater WTP for biofuels that led to environmental improvements. For E10 the WTP was $0.56 gallon 1 $0.58 gallon 1 and $0.50 gallon 1 in AR, FL, and VA, respectively. For E85 the WTP was $0.82 gallon 1 $1.17 gallon 1 and $1.06 ga llon 1 in AR, FL, and VA, respectively. Thus, southern consumers valued the environmental benefits obtained from a modified transportation fuel and were willing to pay more for biofuels once the proposed change offered better c onditions for the environmen t. The extra WTP for ethanol was converted into total future expenditures per year. The mean total expenditures for E10 were $585.20, $485.90, and $596.20 per capita year 1 in AR, FL, and VA, respectively. For E85 the total expenditures were $919.60, $330. 80, and $532.60 per capita year 1 in the same states. The price elasticity of
110 the demand was relatively inelastic and for E85 it was less inelastic compared to the price elasticity of the demand for E10 with the exception of AR. Heterogeneous preferences for environmental attributes were also a characteristic of the respondents. Respondents were willing to pay a premium for E10 in order to achieve CO 2 reduction in all three states, and for biodiversity improvement in AR and FL. However, respondents were n ot willing to pay a premium for E85 for an increased reduction of CO 2 in AR or for biodiversity improvement in any of the three states. This heterogeneity was also observed in some socioeconomic variables. For example, educational background showed differe nt trends in the three states. Individuals with higher levels of education would only change to E10 in FL and E85 in degree were not likely to switch to E10 but individuals with some college education would tend to pay a premium for E10 Income was another variable that showed heterogeneity in In AR, the probability of paying a premium decreased as an individual had greater earnings. In FL, individuals that came from high or middle income households were likely to pay a premium and change to both blends. Results f rom Modeling Effects of B ioenergy Markets on Traditional Forest Product Sectors Two models were calibrated, namely RO B and RPB model Estimation of the supply functions indicated that sawtimber and pulpwood showed positive own price elasticities and a negative own price elasticity of the supply of biomass for bioenergy for the ROB model For the RPB model all own price elastic ities had the expected sign. The supply of sawtimber proved to be more sensitive than either the supply of pulpwood or biofuels with respect to changes in their own prices. The cross price elasticity of sawtimber with respect to the price of pulpwood was p ositive, indicating that sawtimber and pulpwood are complements. Likewise, evidence showed that biomass for bioenergy and pulpwood are complements for the ROB model In the case of the
111 RPB model biomass for bioenergy and pulpwood were substitutes. Sawtimb er and biomass for bioenergy, on the other hand, proved to be substitute products for both models. Turning to the demand functions, for the ROB model the own price elasticities of the demand for sawtimber and pulpwood were negative, and it was positive fo r the demand for biomass for bioenergy. For the RPB model all own price elasticities had the expected sign. The response of demand to changes in price was more sensitive in the case of sawtimber compared to pulpwood and biomass for bioenergy. On the other hand, the demand for biomass for bioenergy was increased by 15%. For the ROB model and RPB mode l the pric e of pulpwood increased by 91.3 % and 103.5%, respectively. The price of biomass for bioenergy decreased by 4.9 % for the ROB model but increased by 5 1.9% for the RPB model The quan tity of pulpwood decreased by 27 .3% and 20%, respectively. The price of sawtimber decreased by 2.9 % and increased by 6.5% in the ROB model and RP B model respectively. The quanti ty of sawtimber increased by 4.5 % in the ROB m odel but it decreased by 11.1% in the RPB model As a result of the policy, the total value of the demand for forest resources was also increased. In total, the gain of the quantity values for the fore st landowners accounted for 10.9 % and 8.9% for the ROB model and RPB model respectively. The forest landowner and bioenergy sectors were net winners in both models. In the ROB mode l sawmills were also benefited. However, sawmills were financially damaged in the RPB model Pulpmills were negatively affected by the inc rease in the demand for biomass for bioenergy in both models. Policy Implications and Further Research Production of bioenergy showed to increase the profitability of forest stands and larger socioeconomic benefits are expected as bioenergy mark ets evolve. However, lower prices for forest biomass for bioenergy and high cost of thinnings and transportation might become a barrier for bioenergy production. Policy incentives for biomass production would stimulate forest
112 landowner s to encourage t he pr oduction of bioenergy. However, careful considerations must be taken when managing fo rests intensively for bioenergy thus policy incentives must also aim sustainable silvicultural activities. Although the introduction of bioenergy production would increase the profitability of forest stands and reduce the risk of catastrophic disturbances, the unsustainable use of forest biomass for bioenergy can lead to negative ecological impacts, increase GHG emissions and adverse effects on communities. Lal et al. (2009 ) proposed a set of indicators such as land use change, biodiversity conservation, soil and water quality, profitability and community benefits for a sustainable production of forest biomass for bioenergy. On the other hand, f orest landowners would be bett er positioned as certification systems such as the American Tree Farm System (ATFS), the Sustain able Forestry Initiative (SFI), or schemes recognized by the Programme for the Endorsement of Forest Certification (PEFC) and the Forest Stewardship Cou ncil (FS C) develop standards or incorporate specific guidelines for bioenergy production. This research can be extended in several ways to address different issues. The incorporation of certain silvicultural activities to produce bioenergy allows other commercial activities to be undertaken. For example, silvopastoral activities can be developed and environmental services such as recreation can be incorporated. These factors can increase further the profitability of forest stands. On the o ther hand, the optimal th inning age was arbitrarily set at year 16 for the bioenergy market model regarding nonindustrial private forest management. A dynamic optimization model such as the reservation price approach is a plausible extension of the bioenergy market model to determ ine the optimal thinning age Understanding present and future individual preferences for bioenergy is an important tool for policymakers F indings suggested that individuals were willing to pay a premium for
113 biofuels reflecting environmental and social b enefits f rom using forest biomass for bioenergy production These findings would provide a scientific basis to formulate policies that stimulate biomass production for energy Thus, the implications of assessing public preferences for bioenergy are in conc ordance with policy incentives for forest landowners to undertake sustainable forest practices for bioenergy production. In addition, this positive willingness for bioenergy would facilitate the development of appropriate educational programs that en hance the public support for forest biomass for energy. Particularly, education campaigns might be carried out in certain areas where renewable energy systems are not internalized by the community or fossil fuels industry is a key support for the economy and peo ple are not likely to switch to bioenergy Thus, clean energy communities might play a significant role in convincing people to choose renewable energy systems. P eriodic revisions of these studies are needed to evaluate the validity and consistency of the results and formulate improved policies based on changing publ ic perceptions and preferences. In addition, the use of meta analysis is a possible appro ach to explore variation of willingness to pay to transfer information to another location or context. An other extension of Current U.S. energy policies aim to decrease the depen dency on foreign oil and the greenhouse gas emissions. The search for environmentally friendly sources of energy and improvement of efficiency of current technologies would be needed to meet this goal. We have pointed out the use of forest biomass as a pla usible option to switch to green energy Our dynamic partial equilibrium model showed that an increased demand for biomass for bioenergy would benefit bioenergy firms and forest landowners. A subsidy for the bioenergy sector that
114 induces an increased consu mption of forest biomass would reduce the cost of bioenergy production. This might allow for further advances in research and development to find more efficient energy conversion techniques. T he other conventional sectors, pulp and sawmill would be adverse ly affected. However, because of the absence of data, our results have to be taken with a certain degree of caution. Pulp and sawmills can create synergy acting strategically. Sawmills might establish joint ventures with bioenergy sector by selling low val ue sawmill residues at competitive prices. On the other hand, potential competition between pulpmills and bioenergy sector might be reduced pulpmills position themselves as integrated forest biorefineries. F urthermore bioenergy firms might also sell elect ricity to pulpmills at competitive prices. This dynamic model could be further improved by in corporati ng of cross sectional data in light of data availability would provide a more complete assessment of the impacts of bioenergy markets. Further, the assump tion of capital stocks adjusting over time to explore the long run responses regarding change in input and output prices also deserves further research.
115 APPENDIX A LIST OF VARIABLES Variables Definition S stumpage price ($ ton 1 ) V(T) merchantable volume at time T (ton) N(d) cumulative normal density function X forest exercise cost ($ ton 1 ) R risk free interest rate T harvest date N(d1) probability that a normal variable takes on values less than or equal to d1 Also known as the option delta ; the degree to which an option value will change given a small change in the price N(d2) probability that a normal variable takes on values less than or equal to d2 Probability that the option will be exercised (forest will be harvested) or the change of the price at expiration time expected value of the timber ($ acre 1 ) net expected value of the timber ($ acre 1 ) expected net present value of the timber for the first rotation ($ acre 1 ) discount rate land expectation value at time T ($ acre 1 ) net land rent time ($ acre 1 ) K salvageable portion expected value of a single rotation average probability per unit of time of a catastrophic event that fo llows a Poisson distribution S t s awtimber P w pulpwood C s chip and saw F bb forest biomass for bioenergy st volatility for sawtimber cs volatility for chip and saw pw volatility for pulpwood fbb volatility for forest biomass for bioenergy utility for each respondent i to choose among different j alternatives deterministic part of the utility of respondent i when choosing among different j alternatives disturbance term of respondent i when choosing among different j alternatives
116 probability that individual i will choose alternative j over k alternatives L likelihood function Reco2 percentage reduction of CO2 Biomp percentage improvement of biodiversity Prem additional payment at the pump ($ gallon 1 ) Member membership of an environmental organization Knowledge knowledge of other sources of renewable energy Ownership ownership of a car Age age of respondent Miles week miles driven weekly Less high condition of a respondent with exclusively les s than high school level High condition of a respondent with exclusively high school level Some college condition of a respondent with exclusively some college level Bachelor condition of respondent with exclusively bachelor degree or higher level Lin come household Annual Income less than $24,9999 Mincome household Annual Income between $25,000 $74,999 Hincome household Annual Income greater than $75,000 Gender gender of respondent Head head of the household Size number of people in the househ old Work work of respondent WTP willingness to pay ($ gallon 1 ) TE total expenditures ($ year 1 ) average price of gasoline ($ gallon 1 ) price elasticity of the demand for biofuels quantity of gallons of ethanol profit function for the forest landowners ($ year 1 ) profit function for the sawmill sector ($ ye ar 1 ) profit function for the pulpmill sector ($ year 1 ) profit function for the energy sector price of sawtimber ($ m 3 ) p rice of pulpwood ($ m 3 ) p rice of biofuels ($ m 3 ) labor wage in for estry ($ hour 1 ) capital in forestry (million m 3 ) labor wage in sawmills ($ hour 1 )
117 price of energy in sawmills ( $ KWh 1 ) price of lumber ($ m 3 ) capital in sawmill sector (million ton year 1 ) lab or wage in pulpmills ($ hour 1 ) price of energy in pulpmills ( $ KWh 1 ) price of sbk pulp ($ ton 1 ) capital in pulpmills (million ton year 1 ) labor wage in energy sector ($ hour 1 ) price of input energy in energy sector ( $ KWh 1 ) price of electricity ( $ KWh 1 ) capita l in energy industry ( m illion KWh year 1 ) quantity of sawtimber (million m 3 ) quantity of pulpwood (million m 3 ) quantity of biofuels (million m 3 ) own price elasticity of the supply own price elasticity of the demand cross price elasticity of the supply cross price elasticity of the demand
118 APPENDIX B SURVEY INSTRUMENT The University of Florida School of Forest Resources and Conservation, with the support of the U.S. Department of Agriculture, U.S Department of Energy, Virginia Polytechnic Institute and State University and the University of Arkansas is conducting a survey a t household level about the use of energy coming from renewable resources. You have been selected to receive the following questionnaire in order to know your preferences about this type of energy. To understand public preferences for the use of renewable resources based energy better, we would like to ask you some questions about your opinion of using forest biomass (wood, branches and bark) to generate energy (liquid biofuels and electricity). The questionnaire is not difficult to answer and should only take about 10 minutes to complete. Your responses will be very helpful in determining the best way to understand the potential of bioenergy. Your participation in this survey is voluntary, but we sincerely hope that you will help us with this study. You a re not required to answer any question that you do not wish to answer. Your answers will be kept entirely confidential. We will not release any information that can be used to identify any individuals participating in this survey. There is no direct benefi t or compensation to you for participating in this study, and returning the completed questionnaire will be interpreted as your consent to participate in the survey. There is no risk to any human beings, animals or the environment from this questionnaire. If you have any questions concerning your rights as a survey participant, please feel free to contact the UFIRB office, PO Box 112250, University of Florida, G ainesville, F l 32 611 2250; (352) 392 0433. Forests are not only a source of wood products. In add ition, forest biomass (branches, wood and bark) can also be utilized to produce energy: fuel ethanol which is blended with fossil fuels (typically gasoline) to be used by different types of automobiles. For example a "10%
119 blend" (E10) is a mixture of 10% o f ethanol and 90% of gasoline (one tenth of the gallon is ethanol and nine tenths of the gallon are gasoline); a "85% blend" (E85) is a mixture of 85% ethanol and 15% gasoline and so on. The greater the blend, the lesser the emission of carbon dioxide in t he atmosphere. Growing trees to generate forest biomass based energy would absorb CO 2 and reduce the global warming caused by human activities such as the use of fossil fuels and deforestation use so can provide environmental benefits such as reduction in global warming and wildfire occurrence risk and improvements in air quality and biodiversity which are explained on the following screens. Loss of habitat because of wildfires or pest outbreaks could result in a decrease or the extinction of some species that live in the forest. Overstocked forest areas are susceptible to wildfires and also show a lack of vigor (poor forest health). However, silvicultural practices such as thinnings have been used to reduce the excessive amount of biomass and, thus, risk o f wildfire and pest infestations. This practice also produces small diameter wood that has an additional market value. The price of ethanol will be variable and highly dependent of the location of the ethanol plant, the availability of woody biomass and th e local forestry infrastructure. This price is likely to be passed onto you in the form of an increase of the price of gasoline you currently use at the pump. In the next pages several alternative impl ementation plans are described to you and you will be asked to evaluate them as if you were voting in a local referendum Please go through them carefully and vote for the plan that best reflects your preference for the use of fuel ethanol.
120 In the next sec tion we want you to evaluate 6 different scenarios. These scenarios consist of several plans for implementing forest biomass based production of fuel ethanol, in this particular case, a 85% blend or E85. The plans differ based on the extent of changes in t he following: REDUCE CO 2 EMISSIONS (50 60%, 61 70%, 71 90% reduction of CO 2 per mile traveled) IMPROVE BIODIVERSITY (1 25%, 26 50%, 51 75% improvement of biodiversity by reducing wildfire risk & improving forest health) INCREASE PRICE OF THE FUEL AT THE PUMP (an additional extra payment of $0.3, $0.6, $1 and $1.5 per gallon). Each scenario describes 2 alternative plans (A and B). Please indicate whether you choose Plan A or B and how certain you are with that choice (i.e., how easily it was to mak e). Plans A differ between scenarios, so please read them carefully Plan B will always reflect the status quo, which is the current situation of having no change regarding the emissions of CO 2 and biodiversity improvement. Which of the plans would you c hoose? Please choose Plan A Plan B Reco2 Reduction of CO 2 between 61 70% per mile traveled No reduction (0%) Biomp Improvement of biodiversity between 1 25% No improvement (0%) Prem Additional payment of $0.60 per gallon at the pump No extra payment ($0)
121 APPENDIX C COMPLETE HISTORICAL DATA SET Table C 1 Yearly dataset Year million m 3 million m 3 m illion m 3 $ m 3 $ m 3 $ m 3 $ hour 1 m illion m 3 1970 39.64 57.41 12.97 35.42 28.90 9.48 3.43 2524.63 1971 43.50 53.65 16.12 41.40 28.98 8.33 3.77 2573.38 1972 44.05 54.60 17.57 46.92 29.06 7.56 4.11 2622.14 1973 44.05 55.59 20.36 52.71 30.04 6.26 4.41 2670.89 1974 38.54 57.34 20.69 47.86 31.78 6.55 4.75 2719.65 19 75 38.54 49.52 19.74 39.36 30.09 7.07 5.11 2768.40 1976 45.31 52.59 24.88 46.61 29.94 5.86 5.83 2817.15 1977 48.43 52.59 27.57 52.04 29.30 4.95 6.36 2865.91 1978 48.91 53.76 27.42 62.94 29.70 4.46 7.01 2868.45 1979 47.96 58.17 28.89 75.70 29.74 4.52 7. 71 2870.99 1980 40.05 62.07 26.88 59.35 27.73 5.88 8.36 2873.53 1981 37.65 61.64 26.35 53.17 27.10 6.95 8.81 2876.07 1982 29.84 58.18 26.26 40.73 28.21 7.58 9.47 2878.61 1983 34.93 58.74 29.97 44.68 28.23 7.43 9.84 2881.15 1984 36.28 59.75 31.13 43.13 24.77 7.68 10.35 2883.69 1985 34.59 56.38 31.08 32.26 28.64 7.90 10.56 2919.34 1986 62.30 60.54 31.12 31.57 28.84 7.31 10.46 2954.99 1987 65.49 62.67 32.12 40.36 29.60 6.86 10.33 2990.64 1988 67.62 61.38 30.57 42.42 29.37 7.32 10.43 2971.63 1989 65.4 9 60.39 29.06 42.45 29.55 7.62 10.76 2952.61 1990 67.09 68.34 28.14 44.23 28.74 7.69 10.85 2933.60 1991 67.96 70.00 27.52 46.98 31.48 7.98 10.70 2914.58 1992 73.02 68.27 30.30 53.50 32.87 6.92 10.80 2923.63 1993 86.50 64.16 29.26 64.75 33.43 5.45 11.00 2932.68 1994 89.87 66.26 30.24 77.32 28.72 5.02 11.06 2941.73 1995 86.01 68.08 29.22 88.00 32.73 5.95 11.26 2950.77 1996 76.46 65.45 27.66 76.00 29.43 5.42 11.37 2959.82 1997 76.80 71.39 29.47 91.04 34.04 5.26 11.76 2968.87 1998 76.82 71.69 30.60 92. 10 36.16 5.63 12.07 2986.85 1999 81.09 68.14 29.84 82.75 30.93 4.83 12.80 3004.84 2000 79.67 61.29 30.19 83.28 26.76 5.67 13.25 3022.83 2001 77.33 58.10 29.01 73.70 26.70 5.57 13.93 3040.82 2002 80.13 59.31 29.02 79.16 27.80 5.63 14.27 3058.80 2003 81 .86 64.10 24.52 71.07 26.86 6.06 14.46 3118.00 2004 87.38 67.15 24.76 71.50 25.56 4.83 14.54 3177.19 2005 91.01 64.87 26.17 68.79 25.80 5.31 15.22 3236.38 2006 75.52 65.93 26.58 70.79 25.78 5.75 15.56 3295.58
122 Table C 1. Cont inued Year $ KWh 1 $ hour 1 $ ton 1 million ton year 1 $ KWh 1 $ ton 1 1970 0.050 3.51 127.26 30.79 2.97 38.77 1971 0.049 3.68 129.23 32.32 3.10 45.34 1972 0.048 3.85 133.67 32.38 3.23 51.28 1973 0.049 4.09 143.53 33.98 3.50 63.53 1974 0.051 4.42 178.56 35.00 3.77 62.39 1975 0.054 4.88 200.26 35.57 4.13 59.36 1976 0.055 5.33 211.11 36.13 4.58 73.50 1977 0.055 5.81 219.49 37.15 4.95 88.78 1978 0.057 6.36 229.85 37.97 5.43 103.06 1979 0.058 6.95 257.97 4 0.50 5.89 109.75 1980 0.057 7.64 292.99 41.18 6.36 100.66 1981 0.059 8.38 322.09 42.15 6.78 98.76 1982 0.060 9.08 339.85 43.07 7.21 96.74 1983 0.060 9.68 350.70 44.07 7.57 108.23 1984 0.055 10.15 374.87 44.31 7.79 108.23 1985 0.055 10.56 384.73 45.05 7.98 101.79 1986 0.053 10.90 394.60 44.36 8.09 101.54 1987 0.052 11.14 413.83 45.59 8.15 110.38 1988 0.052 11.40 442.94 46.71 8.35 108.74 1989 0.050 11.66 468.09 47.44 8.58 104.57 1990 0.047 12.00 479.44 47.82 8.81 107.60 1991 0.044 12.40 485.35 49. 33 8.97 107.48 1992 0.043 12.74 493.25 49.38 9.16 126.29 1993 0.044 13.08 500.15 49.23 9.32 163.30 1994 0.043 13.42 517.91 51.06 9.55 176.56 1995 0.041 13.87 584.99 50.85 9.82 161.53 1996 0.041 14.31 573.15 50.06 10.13 171.89 1997 0.040 14.68 570.00 51.16 10.44 180.08 1998 0.040 15.12 580.06 51.32 10.77 171.20 1999 0.040 15.57 591.40 49.06 11.12 179.71 2000 0.041 15.84 624.30 47.57 11.58 155.81 2001 0.041 16.44 628.04 44.93 11.90 147.58 2002 0.039 17.06 632.12 46.40 12.13 140.52 2003 0.041 16.88 645.71 46.50 12.33 140.71 2004 0.043 17.45 665.08 45.69 12.64 172.84 2005 0.045 17.53 688.53 45.47 12.77 167.74 2006 0.044 18.36 713.34 45.66 13.32 156.32
123 Table C 1. Cont inued Year million ton year 1 $ hour 1 $ ton 1 million KWh year 1 $ KWh 1 1970 13.90 22.59 0.00346 0.070 127.40 1971 17.23 22.03 0.00347 0.068 139.53 1972 20.56 24.85 0.00342 0.068 156.22 1973 23.89 24.65 0.00321 0 .068 172.91 1974 25.25 20.79 0.00403 0.067 189.60 1975 26.60 18.95 0.00389 0.070 202.80 1976 27.96 19.07 0.00428 0.070 210.50 1977 28.94 18.22 0.00467 0.070 222.50 1978 29.91 18.10 0.00481 0.071 233.30 1979 30.89 17.96 0.00507 0.073 238.10 1980 31.8 6 16.18 0.00545 0.071 237.40 1981 32.22 15.61 0.00586 0.074 250.00 1982 32.57 15.27 0.00583 0.075 266.70 1983 32.93 15.76 0.00588 0.075 261.40 1984 33.28 15.84 0.00576 0.071 267.60 1985 33.64 16.10 0.00560 0.071 278.10 1986 34.00 16.61 0.00537 0.069 283.00 1987 34.35 17.38 0.00541 0.069 285.70 1988 34.71 17.86 0.00535 0.069 290.10 1989 35.06 17.81 0.00529 0.068 295.39 1990 35.42 18.05 0.00524 0.066 295.39 1991 35.78 17.67 0.00526 0.064 296.84 1992 36.13 17.88 0.00527 0.063 300.76 1993 36.49 18. 35 0.00514 0.063 304.79 1994 36.84 19.07 0.00492 0.063 309.68 1995 37.20 19.28 0.00487 0.061 312.99 1996 39.40 19.76 0.00474 0.061 318.95 1997 40.30 20.59 0.00458 0.060 320.39 1998 42.30 21.74 0.00466 0.061 322.42 1999 42.70 22.53 0.00470 0.061 325.9 6 2000 43.30 22.89 0.00483 0.062 342.72 2001 43.90 22.39 0.00465 0.062 361.09 2002 44.40 23.03 0.00452 0.060 392.85 2003 45.20 23.08 0.00462 0.062 414.07 2004 46.40 23.52 0.00455 0.064 422.94 2005 48.20 23.37 0.00483 0.066 428.46 2006 48.60 22.24 0. 00495 0.067 429.01
124 APPENDIX D VOLUMES AND PRICES OF SAWTIMBER, PULPWO OD AND BIOM ASS FOR BIOENERGY 1970 2006 Figure D 1. Volume (Million m 3 ) of sawtimber, pulpwood and biomass for bioenergy 1970 2006 0 10 20 30 40 50 60 70 80 90 100 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 Million m 3 Year Sawtimber Pulpwood Biomass
125 Figure D 2. Stumpage Price ($ m 3 ) of sa wtimber, pulpwood and biomass for bioenergy 1970 2006 0 10 20 30 40 50 60 70 80 90 100 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 $ m 3 Year Sawtimber Pulpwood Biomass
126 LIST OF REFERENCES Adamowicz, W., Louviere, J., Williams, M., 1994. Combining stated and revealed preference methods for valuing environmental amenities. Journal of Environmental Economics a nd Management 26, 271 292 Adamowicz, W., Boxall, P ., Williams, M., Louviere, J. 1998. State d preference approaches for measuring passive use values: choice experiments versus contingent valuation. American Journal of Agricultural Economics 80, 64 75. A htikoski, A., Heikkila, J., Aleniusa, V., Sirenc, M., 2008. Economic viability of utilizing biomass energy from young stands The case of Finland. Biomass and Bioenergy 32, 988 996. Amacher, G., Malik, A., Haight, R., 2005. Not getting burned: the importan ce of fire prevention in forest management. Land Economics 81 (2), 284 302. Ankarhem, M., Brnnlund R. and Sjstrm M.,1999. Biofuels and the forest sector. In: Yoshimoto, A., Yukutake, K. (Eds.), Global Concerns for Forest Resource Utilization: Sustainab le Use and Management. Kluwer Academic Publishers, Dordrecht. 361 pp. Ankarhem, M., 2005. Effects of increased demand for biofuels: a dynamic model of the swedish forest sector Umea University, Department of Economics Umea Economic Studies 658 Aulisi, A., Sauer, A., Well ington, F., 2007. Trees in the g reenhouse: w hy c limate c hange is t ransforming the f orest p roducts b usiness. World Resources Institute Report. 74 pp. Bailey, R., Zhou, B., 1997. GaPPS 4.20 G eorgi a P ine P lantation S imulator. Forest Biome trics Consulting, Athens, Georgia, United States. Retrieved January 15, 2008, from http://www.bugwood.org/intensive/98 018.html Banzhaf S., Burtraw D., Evans D., Krupnick A., 2006. Valuation of natural resource improvements in the Adirondacks. Land Econo mics 82 (3), 445 464. Barnett, J., Sheffiled, R., 2002. Slash Pine: characteristics, history, status and trends. In: Dickens, E., Barnett, J., Hubbard, W., Jokela, E. (Eds.), Proceedings of the Slash Pine Symposium, Slash Pine: Still Growing and Growing!. Jekyll Island, Georgia, April 23 25, 2002 pp. 1 6. Batley S., Colbourne D., Fleming P., Urwin P., 2001. Citizen versus consumer: challenges in the UK green power market. Energy Policy 29 (6), 479 487. Baum, C., Schaffer, M., Stillman, S., 2007. Enhance d routines for instrumental variables/gmm estimation and testing. Boston Coll ege Economics Working Paper 667, 38 pp. Behrman, J., Knox Lovell, C., Pollak, R., Sickles, R., 1992. The CET CES generalized Leontief variable profit function: an application to Indian agriculture. Oxford Economic Papers No. 44, Oxford University Press, New York. pp. 341 354.
127 Belanger, R., Hedden, R., Lorio, P., 1993. Management strategies to reduce losses from the southern pine beetle. Southern Journal of Applied Forestry 17 (3), 150 154. Bergmann A., Hanley N., Wright R., 2006. Valuing the attributes of renewable energy investments. Energy Policy 34 (9), 1004 1014. Berrens R., Bohara A., Jenkins Smith H., Silva C., Weimer D., 2004. Information and effort in contingent valua tion surveys: application to global climate change using national internet samples. Journal of Environmental Economics and Management 47 (3), 331 363. Besanko D., Gupta S., Jain, D., 1998. Logit demand estimation under competitive pricing behavior: an e quilibrium framework. Management Science 44 (11), 1533 1547. Bhattacharjee S., Petrolia D., Herndon C., 2008. Estimating willingness to pay for E10 fuel: a contingent valuation method. Southern Agricultural Economics Association Annual Meeting, Dallas, Texas 6730 February 2 5, 2008. Birur, D., Hertel, T., Tyner, W., 2008. Impact of b iofuel p roduction on w orld agricultural m arkets: A computable general equilibrium a nalysis Department of Agricultural Economics, Purdue University, GTAP Working Paper No. 53. Black, F., Scholes, M., 1973. The pricing of options and corporate liabilities. Journal of Political Economy 81 (3), 637 654. Bolkesjo, T., Tromborg, E., Solberg, B., 2006. Bioenergy from the forest sector: economic potential and interactions with t imber and forest products markets in Norway. Scandinavian Journal of Forest Research 21 (2), 175 185. Brannlund R., Kristrom B., 1996. Welfare measurement in single and multimarket models: theory and application. American Journal of Agricultural Econom ics 78, 157 165. Brazee, R., Mendelson, R., 1988. Timber harvesting with fluctuating prices Forest Science 34 (2), 359 372. California Energy Commission (CEC), 2004. Ethanol Fuel Incentives Applied in the US. Staff Report P600 04 001, Sacramento, California. Retrieved March 17, 2008, from http://www.energy.ca.gov/reports /2004 02 03_600 04 001.PDF Cameron, R., Billings, R., 1988. Southern pine beetle: factors associated with spot occurrence and spread in young plantations. Southern Journal of Applied Forestry 12 (3), 208 214 Carlsson F., Frykblom P., Liljenstolpe C., 2003. Valuing wetland attributes: an application of choice experiments. Ecological Economics 47 (1), 95 103.
128 Carlsson F., Frykblom P. Lagerkvi st, C., 2005. Using cheap talk as a test of validity in choice experiments. Economics Letters 89 (2), 147 152 Caves, D., Christensen L., 1980. Global properties of flexible functional forms. American Economic Review 70 (3), 422 432 Chambers, R 1988. Applied Production Analysis: A Dual Approach. Cambridge University Press, Cambridge, UK. 352 pp. Champ P., 2003. Collecting survey data for nonmarket valuation. In: Champ P., Boyle K., Brown T. (Eds.), A Primer on Nonmarket Valuation, Kluwer Academic Pu blishers, Dordrecht, pp 59 98. Christensen, L., Jorgenson, D., LAU, L., 1971. Conjugate duality and the transcendental logarithmic production function. Econometrica 39 255 256. Cook J., Beyea J., Keeler K., 1991. Potential impacts of biomass productio n in the United States on biological diversity. Annual Review o f Energy and the Environment 16, 401 431. Cook, J., Beyea, J. 2000. Bioenergy in the United S tates: progress and possibili ties. Biomass and Bioenergy 18, 441 455. Cummings R., Taylor L., 199 9. Unbiased value estimates for environmental goods: A cheap talk design for the contingent valuation method. American Economic Review 89, 649 65. Despotakis, K., 1986. Economic performance of flexible functional forms: Implications for equilibrium modeli ng. European Economic Review 30 (6), 1107 1143. Dickens, D., Will, R., 2002. Planting density impacts on slash pine stand growth, yield, product class distribution, and economics. In: Dickens, D., Barnett, J., Hubbard, W., Jokela, E. (Eds.), Proceedings of the Slash Pine Symposium, Slash Pine: Still Growing and Growing!. Jekyll Isl and, Georgia, April 23 25, 2002, pp. 36 44. Diewert, W., 1973. Functional forms for profit and transformation functions. Journal of Economic Theory 6, 284 316. Dixit, A., Pind yck, R., 1994. Investment Under U ncertainty. Princeton University Press, New Jersey, 468 p p Dwivedi, P., Alavalapati, J., Lal, P., 2009. Cellulosic ethanol production in the United States: Conversion technologies, current production status, economics, an d emerging developments Energy for Sustainable Development 13 (3), 174 182. Ene rgy Information Administration (EIA), 2007a. State Energy, State Energy Consumption, Price and Expenditures Estimates. Retrieved February 17, 2009, from htpp://www.eia.doe.gov/emeu/states/_seds.html.
129 Energy Information Administration (EIA), 2007 b Annual Energy Outlook 2007 with Projections to 2030. US Department of Energy DOE/EIA 0383(2007), Washington, DC. E nergy Information Administration (EIA), 2008a. Annual Energy Review 2007. US Department of Energy DOE/EIA 0384(2007), Washington, DC. Energy Information Administration (EIA), 2008b. State Energy Data 2005: Prices and Expenditures Retrieved May 20, 2008 f rom http://www.eia.doe.gov/emeu/states/sep_sum/html/pdf/rank_pr_mg.pdf. Energy Information Administration (EIA), 2009. Annual Energy Review 2008. US Department of Energy DOE/EIA 0384(2008), Washington, DC. Evans M., 1998. The Economic Impact of the dema nd for Ethanol. Diane Publishing Company, 49 pp. Englin, J., Boxall, P., Hauer, G., 2000. An empirical examination of optimal rotations in a multiple use forest in the presence of fire risk. Journal of Agricultural and Resource Economics 25 (1), 14 27. E 85 Price Archives, (2009). Retrieved March 15, 2009, from http://e85prices.com/archive.php. Fargione, J., Hill, J., Tilman, D., Polasky, S., Hawthorne, P., 2008. Land clearing and the biofuel carbon debt. Science 319 (5867), 1235 1238. Fleming C., Bowde n M., 2007. Web based surveys as an alternative to traditional mail methods. Journal of Environmental Economics and Management 90 (1), 284 292. Faaij, A., Domac, J., 2006. Emerging international bioenergy markets and opportunities for socio economic deve lopment. Energy for Sustainable Development 1, 7 19. Florida Department of Community Affairs and Florida Department of Agriculture and Consumer Services Division of Forestry, 2004. Wildfire Mitigation in Florida: Land Use Planning Strategies and Best Deve lopment Practices. Retrieved November 20 th 2008, from http://www.fl dof.com/wildfire/wf_pdfs/ Wildfire_Mitigation_in_FL.pdf. Galik, C., Abt, R., Wu, Y., 2009. Forest biomass supply in the southeastern United States implications for industrial roundwood and bioenergy Production. Journal of Forestry 107 (2), 60 77 Gan, J., Smith, C., 2006. Availability of logging residues and potential for electricity and carbon displacement in the USA. Biomass and Bioenergy 30 (12), 1011 1020
130 Graham, R., Harvey, A., Jain, T., Tonn, J., 1999. The Effects of Thinning and Similar Stand Treatments on fire Behavior in Western Forests. United States Department of Agriculture Forest Service Pacific Northwest Research Station, United States Department of the Interior Bureau of Land Management, General Technical Report PNW GTR 463. Greene, W., 2003. Econometric Analysis. Prentice Hall, 5 th edition. 1026 pp. Guo, Z., Sun, C., Grebner, D., 2007. Utilization of forest derived biomass for energy production in the U.S.A.: statu s, challenges, and public policies. International Forestry Review 9 (3), 748 758. Haab T., McConnell K., 2002. Valuing Environmental and Natural Resources: The Econometrics of Non Market Valuation. Edward Elgar Publishing, Northampton, Massachusetts, 326 pp. Haight, R., 1991. Stochastic log price, land value, and adaptive stand management: numerical results for California white fir. Forest Science 37 (5), 1224 1238. Haight, R., Smith, W., 1991. Harvesting loblolly pine plantations with hardwood competitio n and stochastic prices. Forest Science 37 (5), 1266 1282 Hamelinck, C., Van Hooijdonk, G., Faaij, A., 2005. Ethanol from lignocellulosic biomass: techno economic performance in short, middle and long term. Biomass and Bioenergy 28, 384 410. Hanemann W. 1984. Welfare evaluations in contingent valuation experiments with discrete responses. American Journal of Agricultural Economics 66 (3), 332 341. Hanley, N., Mourato, S., Wright, R., 2001. Choice modelling approaches: A superior alternative for environ mental valuation?. Journal of Economic Survey 15: 433 60. Hardie, I., Parks, P., 1996. Program enrollment and acreage response to reforestation cost sharing programs. Land Economics 72 (2), 248 260. Hartman, R., 1976. The harvesting decision when a stan ding forest has value. Economic Inquiry 14 (1), 52 58. Haynes, R., Adams, D., Alig, R., Ince, P., John R., Zhou, X., 2007. The 2005 RPA timber assessment update. Gen eral Tech nical Rep ort PNW GTR 699 U .S. Department of Agriculture Forest Service, Pacific Northwest Research Station. 212 pp. Heij, C., D e Boer, P. Hans Franses, P., Kloek, T., Van Dijk, H., 2004. Econometric methods with applications in business and economics Oxford University Press.787 pp. Hess, R., Wright, C., Kenney, K., 2007. Cellulos ic biomass feedstocks and logistics for ethanol production. Biofuels, Bioproducts and Biorefining 1, 181 190.
131 Hettinga, W., Junginger H., Dekker S., Hoogwijk M., McAloon A., Hicks K., 2009. Understanding the reductions in US corn ethanol production costs : An experience curve approach. Energy Policy 37 (1), 190 203. Hill J., Nelson E., Tilman D., Polasky S., Tiffany D., 200. Environmental, economic, and energetic costs and benefits of biodiesel and ethanol biofuels. Proceedings of the National Academy of Sciences 103 (30), 11206 11210. Hillring, B., 2006. World trade in forest products and wood fuel. Biomass and Bioenergy 30 (10), 815 825. Howard, J., 2007. U.S. timber production, trade, consumption and price statistics 1965 to 2005. Research Paper FLP RP 637 U .S. Department of Agriculture Forest Service, Forest Product Laboratory. 91 pp. Huggins V., Dennis M., Seryakova K., 2002. An evaluation of nonresponse bias in internet surveys conducted using the knowledge networks panel. Annual Meeting of th e American Statistical Association, Joint Statistical Meetin g, New York, August 11 15, 2002, pp. 1525 1530. Hughes, W., 2000. Valuing forest as a call option: the sale of forestry corporation of New Zealand. Forest Science 46 (1), 32 39. Johnson, T., 200 1. United States timber industry an assessment of timber product output and use, 1996. General Technical Report SRS 45. U.S. Department of Agriculture Forest Service, Southern Research Station. 145pp. Johnson, T., Steppleton, C., Howell,M., 2008. Trends in southern pulpwood production, 1953 2006. Resource Bulletin SRS 138. U.S. Department of Agriculture Forest Service, Southern Research Station. 53 pp. Just, R., Hueth, D., Schmitz, A., 2004. The Welfare Economics of Public Policy: a Practical Approach to Project and Policy Evaluation. Edward Elgar Publishing, Northampton, MA. 688 pp. Kinnucan, H., Zhang, D., 2005. Incidence of the 1996 Canada U.S. softwood lumber agreement and the optimal export tax. Canadian Journal of Agricultural Economics 52(1), 73 78. Kuhfeld W., 2005. Marketing Research Methods in SAS. SAS Institute Inc, Cary, North Carolina, 857 pp. Retrieved December 18 th 2008, from http://support.sas.com/techsup/technote /ts722.pdf Lal, P., Alavalapati, J., Marinescu, M., Rao Matta, J., Dwi vedi, P., Susaeta, A. 2010. Developing s ustainability indicators for woody biom asss in the United States. Journal of Sustainable Forestry, in press.
132 Le Van Green, S., Livingstone, J., 2003. Uses for small diameter and low value forest thinnings. Ecologica l Restoration 21 (1), 34 38. List J., 2001. Do explicit warnings eliminate the hypothetical Bias in elicitation procedures? Evidence from field auctions for sports cards. American Economic Review 91, 1498 1507. Lohmander, P., 2000. Optimal sequential fore stry decisions under risk. Annals of Operation Research 95 (1 4), 217 228. Lu, F., Gong, P., 2003. Optimal stocking level and final harvest age with stochastic prices. Journal of Forest Economics 9 (2) 119 136. Lozar Manfreda K., 2001. W eb Survey Erro rs. Ph.D. dissertation, University of Ljubljana. Malmsheimer, R., Heffernan, P., Brink, S., Crandall, D., Deneke, F., Galik, C., Gee, E., Helms, J., McClure, N., Mortimer, M., Ruddell, S., Smith, M., Stewart, J., 2008. Forest management solu tions for m it igating climate change in the United States. Journal of Forestry 106 (3), 115 171. Marta Pedroso C., Freitas H., Domingos T., 2007. Testing for the survey mode effect on contingent valuation data quality: A case study of web based versus in person int erviews. Ecological Economics 62 (3 4), 388 398. Mason, L., Lippke, B., Zobrist, Z., Bloxton, T., Ceder, K., Comnick, J., McCarter, J., Rogers, H., 2006. Investment in fuel removals to avoid forest fires result in substantial benefits. Journal of Forestry 104 (1), 27 31. Mcfadden D., 1974. Conditional logit analysis of qualitative choice behavior. In: Zarembka P., (Eds.), Frontiers in Econome trics. Academic Press, New York, pp. 105 142. Mei, B., Sun, C., 2009. Assessing time varying oligopoly and oligops ony power in the U.S. paper industry. Journal of Agricultural and Applied Economics 40 (3), 927 93 9. Menegaki A., 2008. Valuation for renewable energy: a comparative review. Renewable and Sustainable E nergy Reviews 12 (9), 2422 2437. Menges R., Schroeder C., Traub S., 2005. Altruism, warm glow and the willingness to donate for green electricity: an artefactual field experiment. Environmental and Resource Economics 31, 431 458. Miller Freeman Publications, 1970. Directory of Forest Products Industry. 51 st ed. Portland. Miller Freeman Publications, 1973. Directory of Forest Products Industry. 54 th ed. Portland. Miller Freeman Publications, 1976. Directory of Forest Products Industry. 57 th ed. Portland. Miller Freeman Publications, 1977. Directory of Fore st Products Industry. 58 th ed. Portland.
133 Miller Freeman Publications, 1979. Directory of Forest Products Industry. 60 th ed. Portland. Miller Freeman Publications, 1980. Directory of Forest Products Industry. 61 st ed. Portland. Mogas J., Riera P., Bennet J., 2006. A comparison of contingent valuation and choice modelling with second order interactions. Journal of Forest Economics 12 (1), 5 30. Morrison, C. 1988. Quasi fixed inputs in U.S., and Japanese manufacturing: a generalized Leontief restricted cost function approach. Review of Economics and Statistics 70 (2), 275 287. Mountain, D., Hsiao, C., 1989. A combined structural and flexible functional approach for modeling energy substitution. Journal of the American Statistical Association 84, 76 87. Mu in the south central United States. Journal of Environmental Management 64 (1), 65 76. Neary, D., Zieroth., E., 2007. Forest bioenergy system to reduce the hazar d of wildfires: White Mountains, Arizona. Biomass and Bioenergy 31, 638 645. Newman, D., 1987. An econometric analysis of the southern softwood stumpage market: 1950 1980. Forest Science 33 (4), 932 945. Newman, D., Wear D., 1993. Production economics of private forestry: A comparison of industrial and nonindustrial forest owners. American Journal of Agricultural Economics 75, 674 684. Newman, D., 1991. Shifting southern softwood stumpage supply: Implications for welfare estimation from technical change. Forest Science 36 (3), 705 718. Nicholls, D., Monserud, R, Dykstra, D., 2009. International bioenergy synthesis lessons learned and opportunities from the western United States. Forest Ecology and Management 27 (8), 1647 1655 Nomura N., Akai M., 2004. Willingness to pay for green electricity in Japan as estimated through contingent valuation method. Applied Energy 78 (4), 453 463. Olsson L., Hgerdal, H., 1996 The effect of water soluble inhibitors from steam penetrated willow on enzymatic hydrolysi s and ethanol fermentation, Enzyme Microb. Technol. 18 (1996), pp. 312 331. Outcalt, K., Wade, D., 2004. Fuels management reduces tree mortality from wildfires in southeastern United States. Southern Journal of Applied Forestry 28 (1), 28 34. Paun, D., J ackson, G., 2000. Potential for Expanding Small Diameter Timber Market: Assessing Use of Wood Posts in Highway Applications. General Technical R eport 120. Madison, Wisconsin, U.S. Department of Agriculture, Forest Service, Forest Product s Laboratory. 28 pp
134 Perlack, R., Wright, L., Turhollow, A., Graham, R., Strokes, B., Erbach, D., 2005. Biomass as Feedstock for a Bioenergy and Bioproducts industry: The Technical Feasibility of a Billion ton Annual Supply. U S Department of Agriculture and U.S. Department of Energy, Oak Ridge National Laboratory Techn ical Report ORNL/TM 2005/66. Oak Ridge, TN, 59 pp. Peterson, D., Johnson, M., Agee, J., Jain, T., Mckenzie, D., Reinhardt, E., 2003. Fuels planning: managing forest structure to reduce fire hazard. Proceedings of the 2nd International Wildland F ire E cology and Fire Management C ongress, Orlando, Florida, November 16 20, 2003. Piccolo, C., Bezzo, F., 2009. A techno economic comparison between two technologies for bioethanol production from lignocelluloses. Bioma ss and Bioenergy 33, 478 491. Pimentel D., Patzek T., 2005. Ethanol production using corn, switchgrass, and wood; biodiesel production using soybean and sunflower. Natural Resources Research 14 (1), 65 76. Plantinga, A., 1998. The optimal timber rotati on: an option value approach. Forest Science 44 (2), 192 202. Polagye, B., Hodgson, K., Malte, P., 2007. An economic analysis of bio energy options using thinnings from overstocked forests. Biomass and Bioenergy 31 (2 3) 105 125. Prestemon, J., Holmes, T., 2000. Timber price dynamics following a natural catastrophe. American Journal of Agricultural Economics 82 (1), 145 160. Prestemon J., Holmes, T., 2004. Market dynamics and optimal timber salvage after a natural catastrophe. Forest Science 50 (4), 49 5 511. Prestemon, J., Wear, D., Stewart, F., Holmes, T., 2006. Wildfire, timber salvage, and the economics of expediency. Forest Policy and Economics 8 (3), 312 322. Price, T., Doggett, C., Pye, J., Smith, B., 1998. A History of Southern Pine Beetle Out breaks in the Southeastern United States. The Georgia Forestry Commission, Macon, GA, 65 pp. Reed, W., 1984. The effects of the risk on the optimal rotation of a forest. Journal of Environmental Economics and Management 11, 180 190. Renewable Fuel Associ ation, 2008. Will the Plunge in Grain Prices Mean Lower Food prices at the Supermarket? Ren ewable Fuel Association Report. Retrieved July 15, 2009,from http://www.ethanolrfa.org/objects / documents/1945/will_the_plunge_in_grain_prices_mean_low er_food_prices _at_the_supermarket.pdf. Richardson, J., Bjrheden, R., Hakkila, P., Lowe, A., Smith, C. (Eds.), 2002. Bioenergy from Sustainable Forestry: Guiding Principles and Practice. Kluwer Academic Publishers, Dordrecth, Netherlands. 344 pp.
135 Riso (National Labor atory for Sustainable Energy), 2003. New and emerging bioenergy technologies. In: Larsen, H., Kossmann, J., Sonderberg Petersen, L. (Eds.), Ris Energy R eport 2 Roskilde, Denmark, Ris Technical University of Denmark. Roser, D., Pasanen, K., Asikainen, A ., 2006. Decision support program EnerTree for analyzing forest residue recovery options. Biomass and Bioenergy 30 (4), 326 333 Royer, J., 1987. Determinants of deforestation behavior among southern landowners. Forest Science 33 (3), 654 667 Roe B., Teisl M., Levy A., Russell electricity. Energy Policy 29 (11), 917 925. Runkle J., 1 985. Disturbance regimes in temperate forests. In: Pickett S., White P. (Eds.), The Ecology of Natural Disturbances and Patch Dynamics. Academic Press, Orlando, pp. 17 33. Schmidt, R., 1998. Fusiform Rust Disease of Southern Pines: Biology, Ecology and Ma nagement. Technical Bulletin 903, Agricultural Experiment Station, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL. Scott, D., Tiarks A., 2008. Dual cropping l oblolly p ine for biomass e nergy and conventional wood p rodu cts. Southern Journal of Applied Forestry 32 (1), 33 37. Searchinger, T., Heimlich, R., Houghton, R., Dong, F., Elobeid, A., Fabiosa, J., Tokgoz, S., Hayes, D., Yu, T. 2007. Use of U.S. croplands for biofuels increases greenhouse gases through emissions from land use change. Science 319 (5867), 1238 1240. Shaffer, M., 1984. Valuation of certain long term cutting contract. Forest Science 30 (3), 774 787. Shrestha R., Alavalapati J., 2004. Valuing environmental benefits of silvopasture practice: a case study of the Lake Okeechobee Watershed in Florida. Ecological Economics 49 (3), 349 359. Smeets, E., Faaij, A., 2007. Bioenergy potentials from forestry in 2050: an assessment of the drivers that determine the potentials. Climate Change 81 (3 4), 353 390 Smidt, M., Silveira Folegatti, B., Dubois, M., 2005. Costs and cost trends for forestry practices in the South. Forest Landowner 64 (2), 25 31. Smith, B., Miles, P., Vissage J., Pugh S., 200 9 Forest Resources of the United States, 2 007 General Technical Report WO 78 St. Paul, MN: U.S. Department of Agriculture, Forest Service, North Central Research Station Solberg, B., Dornburg, V., Bolkesjo, T., Faaij, A., Junginger, H., Tromborg E., 2007. Bioenergy and B iomass trade: Evaluation of nternational Trade of
136 Biomass and B ioenergy products. Copernicus Institute, Utrecht Univ ersity, Research Report for IEA Bioenergy Task 40. Solomon B., Barnes J., Halvorsen K., 2007. Grain and cellulosic ethanol: history, economics, and energy policy. Biomass and Bioenergy 31 (6), 416 425. Solomon B., Johnson N., 2009. Valuing climate prot ection through willingness to pay for biomass ethanol. Ecological Economics 68(7), 2137 2144 Stainback, A., Alavalapati, J., 2004. Modeling catastrophic risk in economic analysis of forest carbon sequestration. Natural Resource Modeling 17 (3), 299 317. Steierer, F., Fischer Ankern, A., 2007. Wood Energy in Europe and North America: A New Estimate of Volumes and Flows. Joint Wood Energy Enquiry Report. 35 pp. Spelter, H., McKeever, D., Alderman, M., 2007. Profile 2007: softwood sawmills in the United St ates and Canada. United States Department of Agriculture Forest Service forest Products Laboratory, Research Paper FPL 644. 65 pp Susaeta A., Alavalapati J., Carter D., 2009. Modeling impacts of bioenergy markets on nonindustrial private forest manage ment in the southeastern United States. Natural Resource Modeling 22(3), 345 369. Timber Mart South, 2006. Timber Mart South Market Newsletter: Quarterly Stumpage Prices 1976 2005. Athens, Georgia, United States. Thomson, T., 1992. Option pricing and t Land Economics 68 (4), 462 466 United States Bureau of Labor Statistics, 2007. Bureau of Labor Statistics Data. Retrieved June 15, 2008, from http://data.bls.gov/cgi bin/surveymost. United States Bureau of Labor Statistics Current Employment Statistics, 2009. Employment, hours and earnings national. Retrieved July 25, 2009 from http://www.bls.gov/ces/cesbtabs.htm United States Census of Bureau, Statistical A bstract of the United States, 2009. Energy a nd Utilities, Retrieved July 25, 2009 from http://www.census.gov/compendia/statab/past_years.html United States Department of Agriculture Forest Service (USDA) 2004. USDA Forest Service Strategic Plan for Fiscal Years 2004 08. FS 810. Washington, DC: U.S Department of Agriculture, Forest Service. 32 p p U nited S tates House of Representatives, 1978. Energy Tax Act of 1978. Public Law 95 618, 92 Stat. 3174.
137 U nited S tates House of Representatives, 1978. Tax Reform Act of 1984. Public Law 98 369, 98 Stat. 494. U nited S tates House of Representatives 1980. Omnibus Reconciliation Tax Act of 1980. Public. Law 96 499, 94 Stat. 2599. United States H ouse of Representatives, 1990. Clean Air Act Amendments of 1990. Public Law 101 549, 104 Stat. 2399. United Stat es House of Representatives, 1992. Energy Policy Act of 1992. Public Law 102 4864, 106 Stat 2776. U nited S tates House of Representatives, 1998. Transportation Efficiency Act of the 21st Century. Public Law 105 178, 112 Stat. 107. United States House of Representatives, 2000. Biomass Research and Development Act of 2000. Public Law 106 224, 114 Stat. 428. United States House of Representatives, 2002. Farm Security and Rural Investment Act of 2002. Public Law 107 171, 116 Stat. 134. United States House o f Representatives, 2003. Healthy Forests Restoration Act of 2003. Public Law 108 148, 117 Stat. 1887. U nited S tates House of Representatives, 2004. American Job Creation Act of 2004. Public Law 108 357, 118 Stat. 1418. United States House of Representati ves, 2005. Energy Pol icy Act of 2005. Public Law 109 58, 119 Stat. 594. United States House of Representatives, 2007. Energy Independence and Security Act of 2007. Public Law 110 140, 121 Stat. 1492 United States House of Representatives, 2008. Food, Co nservation, and Energy Act of 2008. Public Law 110 234, 122 Stat. 923. United States House of Representatives, 2009. American Clean Energy and Security Act of 2009. Draft Legislation. Urbanchuk J., 2007. The Relative Impact of Corn and Energy Prices in t he Grocery Aisle. LECG Report. Retrieved April 15, 2009, from http://www.ethanol.org/pdf/contentmgmt/ Impact _of_Corn_and_Energy_Prices_in_the_Grocery_Aisle_June_07.pdf. Walsh, M., 1998. U.S. bioenergy crop economic analyses : status and needs. Biomass and B ioenergy 4 (1), 341 350
138 Walsh, M., de la Torre, D., Shapouri, H., Slinsky, S., 2003. Bioenergy crop production in the United States: potential quantities, land use changes, and economic impacts on the agricultural sector. Environmental and Resource Econ omics 24 (4), 313 333. Wang M., Sarick C., Santini D., 1999. Effects of Fuel Ethanol Use on Fuel Cycle Energy and Greenhouse Gas Emissions. Center for Transportation Research, Energy Systems Divison, Argonne National Laborato ry ANL/ESD 38, Argonne, IL Washburn, C., Binkley, C., 1990. Informational efficiency of markets for stumpage. American Journal of Agricultural Economics 72 (2), 394 405. Wear, D., Newman, D., 1991. The structure of forestry production: short run and long run results. Forest Scien ce 37 (2), 540 551 Prospective Analysis of Recent Change. General Technical Report SRS 99. Asheville, NC: U.S. Department of Agriculture, Forest Service, Southern Resear ch Station. 29 p p Westcott P., 2007. Ethanol Expansion in the United States: How Will the Agricultural Sector Adjust? Outlook Report FDS 07D 01. U .S. Department of Agriculture Economic Research Service. 20 pp. Wiebe E., Eyerman J., Loft J., 2001. Eva luating non response in a web enabled survey on health and aging. Presented at the May 2001 Conference of the American Association for Pu blic Opinion Research, Montreal, May 17 20. Williamson, T., Hauer G., Luckert, M., 2004. A restricted Leontief profit function model of the Canadian lumber and chip industry: potential impacts of US countervail and Kyoto ratification Canadian Journal of Forestry Research 34 (9), 1833 1844. Wooldridge, J., 2002 Econometric analysis of cross section and panel data. MI T P ress. 7 52 pp Worldwatch Institute 2007 Biofuels for Transport: Global Potential and Implications for Sustainable Agriculture and Energy in the 21st Century, London, UK, Earthscan. 480 pp. Yamamoto, H., Fujino, J., Yamaji, K., 2001. Evaluation of b ioenergy potential with a multi regional global land use and energy model. Biomass and Bioenergy 21, 185 203 Yap, R., 2004. Option valuation of philippine forest plantation leases. Environment and Development Economics 9 (2), 315 333. Yin, R., Newman, D ., 1996. The effect of catastrophic risk on forest investment decisions. Journal of Environmental Economics and Management 31, 186 197. Yin, R., Caufield, J., 2002. A profile of timber markets in the U.S. southeast. Forest Product Journal 52 (3), 25 34.
139 Z 67 (3), 317 325.
140 B IOGRAPHICAL SKETCH Andres Susaeta was born in Santiago de la Nueva Extremadura aka Santiago, Chile He earned a Forestry Engineering degree with a specialization in Forest M anagement at the University of Chile in 1999. In 2003 and a fter working three years in National Center for the Environment and Faculty of Forest Science s at the University of Chile as a research assistant Andres was awarded the New Zealand Aid International Development (NZAID) Scholarship to pursue a Master of Forestry Science focused on forest economics in the land of the Lord of the Rings. After graduating in 2005 and working for an environmental NGO in Chile, Andres was offered a scholarship to start his Ph.D. at the University of Florida focusing on forest economics and policy.