Choice Modeling Externalities: A Conjoint Analysis of Transportation Fuel Preferences Matthew Winden and T.C. Haab, Ph.D. Agricultural, Environmental, and Development Economics The Ohio State University
Outline Motivation Methodology Results Conclusions
Motivation Transportation Fuel Consumption Creates Large Externalities Market Pricing Mechanism Has Failed -Public Goods Nature of Externalities Government Correction Has Failed -Regressive Nature of Price Correction -Lack of Political Will Power
Motivation Correct price is necessary to achieve efficiency So, What are the optimal levels (costs) of externalities to society? Knowing allows internalization (MSC=MPC)
Motivation Are externality types valued differently? Impacts on: (1)Human Health Risk Vs (2) Natural Resource Depletion Vs (3) Environmental Damage
Motivation Attribute Examples of Attribute Components Env. Damage: Fish and Animal Populations Levels of Air and Water Pollution Nat. Res. Use: Extraction Rates and Stocks for Ores, Minerals, Oil, Natural Gas Hum. Health Risk: Incidence Rate of Asthma & Cancers
Motivation Goals: 1.) Establish Willingness-To-Pay estimates for reductions in damages 2.) Establish Marginal Price estimates for externality classes
Methodology: Conjoint Analysis Estimates the structure of preferences Specify attributes & bundle into alternatives Respondent chooses preferred alternative Resultant choices allow for statistical inference
Methodology: Conjoint Analysis Each alternative represents potential fuel profile (i.e. mix of fuel types used) Different profiles embody different levels of externalities (attributes) imposed on society Impacts of profile measureable and capable of aggregation into an index for each externality
Methodology: Conjoint Analysis Attribute Levels of Attribute Components Env. Damage 37.5, 45, 50, 55, 62.5 Nat. Res Use37.5, 45, 50, 55, 62.5 Hum. Health Risk37.5, 45, 50, 55, 62.5 Price ($/gallon) -10%, -5%, 0%, 5%, 10%
Methodology: Conjoint Analysis Based in RUM Framework Respondent chooses 1 of 3 alternatives Attributes: Environmental Damage Natural Resource Usage Human Health Risk Price
Methodology: Conjoint Analysis
$[GASPRICE] per gallon Fuel Mix A
Methodology: Conjoint Analysis RUM framework V i j = V(x i j, β) + ε i j i = individual j = alternative x = vector of attributes and characteristics ε = stochastic error term
Methodology: Conjoint Analysis RUM Formalized: Linear and IID V i j = β 0 + x i j β 1 + (M i - p i j ) β 2 + ε i j M = Income p = price
Methodology: Conjoint Analysis Probability of K chosen over j, for all j≠k Pr(dV i j >0) = ϑ (Δ(x) β 1 – Δ(p) β 2 ) (See Kanninen 2007)
Results Survey Representative Sample of 857 Ohio Adults Completed by 537 (62.5%), 532 useable; met criteria of (1) Adult Resident of Ohio (2) Estimate Vehicle MPG (3) Estimate price of fuel at last fill-up
Results Homeowner, Older, and Driver (more likely) Price (self-reported) mean = $1.88 min = $1.00 max = $2.99 Attribute means 49.9(ED), 50.2(NR), 50.3(HH)
Results AttributeConditional Logit Parameter Estimates Price-1.722* Env. Damage Nat. Res. Use-0.427* Hum. Health Risk (Environmental Damage) (Nat. Res. Use) * (Hum. Health Risk) * EnvDam × NatRes NatRes × HumHea HumHea × EnvDam EnvDam×NatRes×HumHea
Results Alternative (Difference from Current) WTP ($/Alternative) 10% Reduction in Each Attribute$0.84/gal 25% Reduction in Each Attribute$2.98/gal AttributeMP ($/Alternative) Environmental Damage Reduction$0.030/gal Natural Resource Use Reduction$0.035/gal Human Health Risk Reduction$0.036/gal
Conclusions Demand (WTP) for reduction in externalities related to transportation fuel usage exists Current (baseline situation) reveals one class of externality is not viewed as more important Starting point for policy discussions
Limitations Price increase still necessary (political will) Less impact, result in more driving? Do respondents accurately understand and value indexes? Accurate measurement and combination of attribute components into indexes Uncertainty of externality impacts
Future Research Income element of utility function may be non-linear Fatigue/Learning Effects Exploration of demographic differences (mixed logit) Relaxation of IIA (multinomial probit)
Special Thanks National Science Foundation Agricultural, Environmental, and Development Economics: The Ohio State University Wisconsin Economic Association