Project Topics
Get Complete Project Material File(s) Now! »
Heterogenous other-regarding behaviors and structure of preferences
While individuals may differ according to their context at the individual level, they may also behave in their own way regarding others, especially when faced with a social dilemma. The study of interactions between individuals is of great interest in experimental economics. In the public good game setting, individuals are assigned to a group of several individuals. They have the choice to invest either in their private account or a public account which benefits them and all the members of their group. Various types of other-regarding behavior have been explored. Individuals may satisfy the Homo-economicus definition i.e. behave in a purely self-interested manner. But it is often not the case. In the most common type of public good game (the linear game), only 33% of individuals behave self-interestedly (Fischbacher et al., 2001). Behav-iors which depart from the Homo-economicus definition mainly consist of reciprocity (Croson, 2008), inequality aversion (Fehr and Schmidt, 1999; Dannenberg et al., 2007), altruism and warm-glow (Andreoni, 1995). While the previous challenge tackles the issue of context differ-ences potentially at the origin of different structures of preferences across individuals, it seems worthwhile studying the interaction between individuals with different preference structures.
Only heterogeneity in terms of endowments or marginal value of the public or the private good has been studied so far in the public good game literature. Most of these studies build upon a linear structure e.g. Reuben and Riedl (2013), or a non-linear but additive structure of pref-erences, e.g. Willinger and Ziegelmeyer (2001), Bracha et al. (2011), which underlies perfect substitutability between the private good and the public good. Yet, individuals may experi-ence utility based on different interactions between goods. And these different structures of preferences may yield different (degrees of) other-regarding behaviors.
An illustration through biofuel production projects
One of the main objectives of biofuel production is to fight against climate change, the oth-ers being securing employment and energy supply. Mostly, biofuels are an alternative to oil in the transport sector. Oil combustion generates large releases of greenhouse gases (GHG) to the atmosphere. Biofuels are biomass-based fuels, which implies that the carbon captured by the biomass during plant growth is released by vehicles after combustion. This results in a carbon-neutral process contrary to oil combustion. This statement is partly true however. The production of biofuels necessitates several steps (agricultural practices, extraction of oil from the plant, drying of the plant, etc) which require energy thus involves GHG emissions. Generally, after an analysis of global warming impacts related to production and consump-tion, biofuels are still better than oil combustion. This incentivized policies to foster biofuel production.
The European Renewable Energy Directive (RED) adopted in 2009 particularly aims at reducing greenhouse gases (GHG) emissions by 20% vs. 1990 by 2020. This objective encap-sulates a minimum incorporation of 10% of renewable energy in the transport sector for each of the European Union members including first and second generations of biofuels. Although biofuels are considered as an alternative to fossil fuels, it also constitutes a source of land use change (LUC) (IPCC, 2007) which should be accounted for in the analysis of global warming impacts.
Changes in land use are the most important environmental impact of biofuels production (Van Stappen et al., 2011). LUC results in carbon stock changes after a land is converted to a new use. The carbon balance disturbance is twofold: in the vegetation (plant) and in the soil1 which both constitute important carbon sinks. Changes in land use affect the oxidation and formation of carbon in both plants and soils, which results in changes in carbon-CO2 fluxes between the land and the atmosphere (Delucchi 2011). Many factors influence the magnitude of the disturbance: climate region, type of the land converted, nature of the new land, agricultural practices. Considering the two sinks together, forests accumulate and maintain carbon stocks better than grasslands and even more than croplands. Land-use intensity increases from forests to grasslands to croplands (Poeplau et al. 2011). Depending on the carbon fraction of both the land which undergoes the change and the new land, the impact can occur in both directions: either a release of carbon generating GHG emissions e.g. a forest is converted into a cropland, or a sequestration of carbon inducing GHG uptakes from the atmosphere e.g. afforestation. Therefore it is crucial to understand the impact of LUC on carbon balances in order to reduce GHG emissions since it can either be beneficial or harmful to the climate (Ben Aoun et al. 2015).
LUC can either result from the replacement of other types of lands i.e. direct LUC or from the displacement of existing crops i.e. indirect LUC (Broch et al. 2012). Direct LUC refers to the substitution of a given land for a cropland entirely dedicated to biofuel production. Indirect LUC occurs when the substitution for biofuel crops of a land dedicated to food crops reduces the availability of lands for food supply; this decrease is compensated by a switch of the de-mand for food in other locations where lands are converted into food crops (through market responses). For example, corn first produced for food supply purposes are turned into energy crops to be used for biofuel production. Many energy policies in different parts of the world2 foster a switch from fossil fuels to biofuels, resulting in an expansion of lands for energy crops hence an increase of LUC. As a consequence, considerable attention has been drawn towards the potential significant emissions due to LUC up to the possibility of switching a positive3 environmental balance into a negative one (Searchinger et al., 2008; Tyner et al., 2010). LUC from biofuel production can also help mitigate climate change. Lignocellulosic biofuels such as Miscanthus4 have indeed the potential to reduce emissions specifically when replacing a cropland (Stewart et al., 2007; Poeplau and Don, 2014; Qin et al., 2015; Harris et al., 2015). Based on these findings, the main biofuel policies namely the Renewable Energy Directive (RED) in the European Union and the Renewable Fuel Standard (RFS2) in the United States, progressively incorporate these impacts in the environmental requirements for the development of a biofuel production. A recent project for an indirect LUC Directive has been presented to the European Commission in October 2012. It aims at reporting estimated iLUC emissions in order to speed up the transition to advanced biofuels production (substantial GHG savings). Especially, it focuses on decreasing the GHG emissions from iLUC when growing the first gen-eration biofuels (agreement on a cap) as well as limiting the land and water use conflict with the food production sector. Therefore, the second generation biofuels may contribute considerably to the 10% objective of the RED. These biofuels are produced from lignocellulosic biomass, woody crops, agricultural residues or waste. They recently entered the market of biofuels with for example the French Futurol project. Their environmental balance seems promising com-pared with first generation biofuels.
Life cycle assessment is the common method used to quantify environmental impacts from projects (Finkbeiner et al., 2006). However, decisions are based on economic evaluation. What are the steps between environmental impacts quantities and environmental impacts values?
Evaluating a project that impacts global warming requires (i) giving a monetary value to GHG emissions (usually to CO2eq) at each point in time and (ii) aggregating costs and benefits of the project over time by discounting them to a chosen time period (De Gorter and Tsur, 2010). These two steps are fully part of the time dimension which characterizes a cost-benefit analysis, the common tool employed in economic assessments of projects.
Cost-benefit analysis: how projects are evaluated
Cost-benefit analysis first converts all costs and benefits to money equivalents based on willing-ness to pay, and second to a present value by applying the discount rate. The discount rate has a major role in determining whether a project should be accepted or not. And its role is even stronger when the project includes costs and benefits which occur over very long periods. LUC impacts are a good illustration of long-term impacts on the environment since this is directly linked to GHG emissions which remain in the atmosphere for hundreds of years.
In a cost-benefit analysis, a project which yields economic and environmental outcomes is usually evaluated by monetizing the environmental outcome at each date and applying the discount rate. A cost-benefit analysis is always carried out in comparison to a baseline or status quo (Pearce et al. 2006). Consider a project starting at time t = 0 with time horizon T , and resulting in a sequence of outcomes f(DCt ; DEt )gt=0;1;:::;T with DCt the economic variation relative to the status quo and Et the environmental-related (emissions or sequestrations of GHG) variation relative to the status quo. Put differently, DC and DE underlie variations between the status quo and the project assessed respectively in terms of economic (monetary) impacts and environmental impacts. The rate at which next period’s consumption is discounted at time t 1 is denoted by rtC (with the convention rC0 = 0).5 The relative price of the greenhouse gases emissions (measured in CO2eq) i.e. the price of carbon at time t, that is, the willingness-to-pay for mitigating one tonne of CO2 emitted at time t, is denoted by pt . The net present value of the project denoted by NPV is then: t=0 Ps=0 (1 + rs ) NPV t=0;1;:::;T = å t C f (DCt ; DEt ) g T DCt + pt DEt (2.1).
In the case of biofuels, the usual baseline is the fossil fuel (gasoline for bioethanol and diesel for biodiesel) which is a substitute for biomass-based fuels. In the context of LUC only, the baseline is the land which undergoes the change e.g. if Miscanthus is planted on a former cropland then the cropland constitutes the baseline for the analysis. The decision maker thus evaluates if the project improves (NPV > 0) or worsens (NPV < 0) social welfare.
Willingness to pay
In the context of global warming impacts such as those from LUC, projections of carbon prices over time are available, in general from international institutions such as the International En-ergy Agency (in the World Energy Outlook annual reports). However, biofuels for example affect other environmental goods or services such as water quality or biodiversity which are not traded. In the absence of markets, i.e. when no price is available for the environmental good, the WTP can be estimated through various methods.
Generalities
In the context of cost-benefit analysis, WTP has a major role. WTP allows to give a monetary value to nonmarket goods to be directly compared with market goods within cost-benefit anal-ysis. In the environmental context, this is the largest amount of money an individual is willing to pay for an environmental improvement or to avoid an environmental damage.
WTP can be estimated in various ways. Two main approaches have been developed namely revealed preferences and stated preferences. The method of benefit transfer complements these two approaches and will be described in subsection 2.2.1.
Dual-rate discounting
In the environmental context, the use of a single discount rate may not be appropriate. Weikard and Zhu (2005) point out two situations under which dual discount rates (one for consumption, the other one for environmental quality) should be used. First, when relative environmental prices are not available, dual-rate discounting is an equivalent way of evaluat-ing projects. Indeed, the authors show that the difference between the consumption and the environmental discount rates is equal to the environmental price growth rate. Second, if con-sumption and environmental quality are not substitutable, relative prices cannot exist, which results in a necessary evaluation of economic and environmental impacts separately with their own discount rate. They further argue that the lowest (limiting) discount rate in this context is the one that should be used in economic evaluation.
Table of contents :
1 General Introduction
1.1 Challenges
1.2 Research objectives
1.3 Structure of the thesis
2 Environment, time and the structure of preferences: the multidimensionality of the notion of value
2.1 From quantities to values
2.1.1 An illustration through biofuel production projects
2.1.2 Cost-benefit analysis: how projects are evaluated
2.1.3 Willingness to pay
2.1.3.1 Generalities
2.1.3.2 Formal description
2.1.4 Discounting
2.1.4.1 Generalities
2.1.4.2 Formal description
2.1.4.3 Dual-rate discounting
2.2 Heterogenous values across agents
2.2.1 Heterogenous willingness to pay
2.2.2 Heterogenous discount rates
2.3 The substitutability concept at the core of the values heterogeneity?
2.3.1 The different definitions of substitutability
2.3.2 Substitutability and WTP
2.3.3 Substitutability and discounting
2.4 Substitutability and heterogenous agents: from the individual to the collective value .
3 From quantities to values: the land use change time-accounting failure
3.1 Introduction
3.2 Background
3.2.1 Land use change time distribution: effective vs. accounted for in policies
3.2.2 Carbon prices and the Hotelling rule
3.3 Theoretical framework
3.3.1 The model
3.3.2 Discounting effect (rp = 0 and 0 < r 1)
3.3.3 Carbon price effect (r = 0 and 0 < rp 1)
3.3.4 Combined discounting and carbon price effects
3.4 Numerical illustration
3.4.1 Scope and assumptions
3.4.2 Data and computation
3.4.3 Results
3.5 Conclusions
Appendix A Proof of Proposition 1
Appendix B Proof of Proposition 2
Appendix C Proof of Propositions 3 and 4
Appendix D Land use change impacts time profile: formal description
Appendix E Data
Appendix F Land cover effects
4 Context-dependent substitutability: impacts on environmental preferences and discounting
4.1 Introduction
4.2 Conceptual background: the link between substitutability, WTP and discounting
4.2.1 Substitutability
4.2.2 Substitutability and WTP
4.2.3 Substitutability and discounting
4.2.4 Substitutability and preference representation
4.3 Theoretical model
4.3.1 General framework
4.3.2 An example
4.4 Contextual substitutability and income elasticity of WTP
4.4.1 General results
4.4.2 Numerical illustration with the CDS function
4.5 Contextual substitutability and discounting
4.5.1 General results
4.5.2 Some suggestions for income shocks testing with the CDS function
4.6 Conclusions
Appendix A Proof of Proposition 5
Appendix B Discount rates’ expressions
Appendix C An comparative analysis of the CES and the CDS functions
Appendix D Proof of Proposition 8
5 Beyond perfect substitutability in public good games: heterogenous structures of preferences
5.1 Introduction
5.2 Background
5.2.1 Why substitutability is an important concern in public economics
5.2.2 Substitutability and public good games
5.2.2.1 Linear public good games
5.2.2.2 Nonlinear (non-additive) public good games
5.3 Experimental environment
5.3.1 Theory
5.3.2 Experimental treatments
5.3.3 Experimental design and procedures
5.3.4 Predictions
5.3.4.1 Standard predictions
5.3.4.2 Behavioral conjectures
5.4 Preliminary results
5.4.1 Visual inspection
5.4.2 Structure of preferences and treatment effects
5.4.2.1 Structure of preferences and type
5.4.2.2 Structure of preferences and heterogeneity
5.4.2.3 Structure of preferences and order effects
5.4.3 An exploration of behavioral determinants
5.4.3.1 Free-riding
5.4.3.2 Inequality aversion
5.5 Conclusions, limits and perspectives
Appendix A Solving for Nash equilibrium
Appendix B Solving for Pareto optima
Appendix C Compressed payoff table format
Appendix D Instructions given to participants
Appendix E Final questionnaire
Appendix F Order effects: visual inspection
Appendix G Econometric tests
G.1 Panel Stata results: FE vs RE models
G.2 Panel Stata results: crossed-variables
6 General conclusion
6.1 Study objectives
6.2 Main results
6.3 Main contributions
6.4 Limitations
6.5 Future research
