Group Title: Working paper - International Agricultural Trade and Policy Center. University of Florida ; WPTC 06-02
Title: Global warming : to believe or not to believe?
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00089809/00001
 Material Information
Title: Global warming : to believe or not to believe?
Series Title: Working paper - International Agricultural Trade and Policy Center. University of Florida ; WPTC 06-02
Physical Description: Book
Language: English
Creator: Ranjan, Ram
Publisher: Working paper - International Agricultural Trade and Policy Center. University of Florida ; WPTC 06-02
Place of Publication: Gainesville, Fla.
Publication Date: March 2006
 Record Information
Bibliographic ID: UF00089809
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.

Downloads

This item has the following downloads:

WPTC_06-02 ( PDF )


Full Text

WPTC 06-02


I '-ional Agricultural Trade and Policy Center



GLOBAL WARMING: TO BELIEVE OR NOT TO BELIEVE?
By
Ram Ranjan


WPTC 06-02 March 2006


WORKING PAPER SERIES


'V


i~fr


UNIVERSITY OF
FLORIDA


Institute of Food and Agricultural Sciences









INTERNATIONAL AGRICULTURAL TRADE AND POLICY CENTER


THE INTERNATIONAL AGRICULTURAL TRADE AND POLICY CENTER
(IATPC)

The International Agricultural Trade and Policy Center (IATPC) was established in 1990
in the Institute of Food and Agriculture Sciences (IFAS) at the University of Florida
(UF). The mission of the Center is to conduct a multi-disciplinary research, education and
outreach program with a major focus on issues that influence competitiveness of specialty
crop agriculture in support of consumers, industry, resource owners and policy makers.
The Center facilitates collaborative research, education and outreach programs across
colleges of the university, with other universities and with state, national and
international organizations. The Center's objectives are to:

* Serve as the University-wide focal point for research on international trade,
domestic and foreign legal and policy issues influencing specialty crop agriculture.
* Support initiatives that enable a better understanding of state, U.S. and international
policy issues impacting the competitiveness of specialty crops locally, nationally,
and internationally.
* Serve as a nation-wide resource for research on public policy issues concerning
specialty crops.
* Disseminate research results to, and interact with, policymakers; research, business,
industry, and resource groups; and state, federal, and international agencies to
facilitate the policy debate on specialty crop issues.









Global Warming: To Believe or Not To Believe?


Ram Ranjan
Postdoctoral Associate
International Agricultural and Trade Policy Center
Department of Food and Resource Economics, University of Florida
Email: rranian(@ifas.ufl.edu, Ph: (352) 392 188-326; Fax: (352) 392 9898



Selected Paper to be presented at the 3rd World Congress of Environmental and
Resource Economics, Kyoto 2006








Abstract




This paper looks at the issue of formation of collective beliefs over risk of natural
catastrophes such as global warming. The dynamics of belief formation is analyzed in an
adaptive learning framework where individuals are exposed to a new set of information
periodically. The impact of this information in setting of beliefs is influenced by the
adherence of individuals to a particular cohort with certain values and interests which may
be helped or undermined by the formation of these beliefs. In this paper we demonstrate
three crucial points. One, the level of learning from random climatic signals could be
easily adulterated by the individual's reference frame which is defined by his group
adherence. Two, there is interaction between different groups with different ideologies
that shape the long run convergence of beliefs. Three, the process of belief formation
also influences public policies where various groups compete to get favorable policy
decisions made by the government. Contrary to the common perception that public beliefs
influence public policy over uncertain events such as global warming, this paper argues that
the link between beliefs and public policies is much complex than that of simple
unidirectional causality.









Introduction

Due to the uncertainties that are argued to persist over the causes and consequences of

global warming, it is becoming exceedingly hard to come to a common consensus over

the adequate course of action, both at national and international levels. In light of these

uncertainties, the steps required to curb carbon emissions and take other mitigative

measures become even more challenging to implement as they come at irreversible costs

to society in terms of forgone consumption or capital accumulation. Further, the

enormous challenges associated with modeling the geo-physical aspects of global

warming in its entirety, make it impossible to accurately predict the extent of causality

and damages associated with anthropogenic carbon emissions. While one may argue

that the risks of global warming are fairly known, the risk perception of global warming

is not a static phenomenon and varies with nationality, race, education, gender, etc.

Further, it is subject to revisions from influx of new information. Consequently, the

ability of the governments to reach a common consensus is heavily influenced by public

perception, as it may be extremely hard for governments to take cutbacks in consumption

and growth against public wishes even if that might seem to be a prudent thing to do.

Therefore, formation of public opinion over the risks of global warming is an issue of

great concern and needs to be studied in greater details.

Public opinion could differ on the basis of gender, race, education, political

affiliation, etc. Bleda and Shackley (2005) argue that businesses would not change their

perceptions towards climate change until affirmative signals are received consistently for

a long period of time. They further argue that reality is perceived by businesses after

being filtered through a reference frame and is not perceived objectively. Consequently,









experienced reality may differ from actual reality due to perceptions which are based

upon their interests, etc. It is further argued that direct signals of climate change may be

subject to misinterpretation as isolated weather related signals and thus could be

discarded if re-interpretation of these signals requires significant organizational changes

(Berhout et al 2004).

Another example of skewed risk perceptions is of a recent case in Britain of

refusal by parents to vaccinate their kids for mumps, measles and rubella under the

unsubstantiated fears that such vaccination could lead to autism. Parents not only refused

to avail of this free vaccination program, they also vehemently refuted and propagated

their own beliefs against scientific evidence demonstrating no links, whatsoever, between

the vaccine and autism (Nature 2006). Interpretations of signals or experiences have

been found to be governed by the frame of reference of the receiver and could be resilient

to objective revisions (Daft and Weick, 1984).

This paper looks at the issue of formation of collective beliefs over risk of natural

catastrophes such as global warming. The role of group adherence, interests and other

factors in influencing an objective interpretation of climatic signals is explored in an adaptive

learning framework. In absence of conclusive scientific evidence, public policy formation

may be subject to other influential forces such as popular opinion, lobby groups, political

interests, etc. For instance, in case of global warming, individuals who see the cut-back

on carbon emissions as a threat to their individualistic life styles are more likely to

discount the available information related to global warming. Whereas, individuals who

are egalitarian in nature would be more willing to accept the possibility of global

warming as they would be more concerned about the intergenerational equity. These

cohort loyalties can be easily seen in the current societies where people identify









themselves as liberals or conservatives and are likely to weigh the same information

differentially to come at different conclusions. Such differences in public opinion due to

group loyalties are likely to lead to an impasse in formation of efficacious public policies.

Formation of beliefs, especially over risky events, is influenced by individual's

adherence to certain cohorts in the society, their level of education, race, age, sex, etc.

Women have been argued to be more risk averse than men. Also, Kahan et al. (2005)

mention that whites are more likely to be less risk averse as compared to minorities to

certain types of risks involving abortion, guns, global warming, etc. This phenomenon

often termed as the 'white male effect' is explained by their more individualistic and anti-

egalitarian lifestyles. Gusfiled (1986) argues that individual's perceive their status in a

society by their adherence to a particular group. Consequently it is possible for risk to be

perceived by the impact it would have on their status within a particular group and

society.

However, it is unlikely that all issues could be seen as black and white by

different groups, with the possibility of most issues falling under a grey area comprising

some believers and non-believers in each cohort. This leads to the question of whether

belief formation is a dynamic process and is it possible for society to come to an

agreement on several key collective risks despite adherences to cohorts pulling in

different directions.

Therefore, it is crucial to understand the formation and evolution of risk

perception of society in order to recommend policy measures. This is especially

important as a low perception of even higher levels of real risks would make such









mitigative measures hard to implement as governments would seldom take steps to go

against the majority opinion.

In this paper, first an adaptive learning model is designed, based upon an earlier

work by Ellison and Feudenberg (1993). Learning from climatic signals is incorporated

to analyze long run propensity of the system to dwell into various possible states of the

system. The dynamics of belief formation is analyzed in an adaptive learning framework

where individuals are exposed to a new set of information periodically. The impact of this

information in setting of beliefs is influenced by the adherence of individuals to a particular

cohort with certain values and interests, which may be helped or undermined by the

formation of these beliefs. The impact of group adherence on influencing the

interpretation of signals is explained and the interaction between competing groups is

taken up to explore their influence on each other's opinions. Finally, the role of belief

dynamics in public policy formation is analyzed within a political economy framework. In

this paper we demonstrate three crucial points. One, the level of learning from random

climatic signals could be easily adulterated by the individual's reference frame which is

defined by his group adherence. Two, there is interaction between different groups with

different ideologies that shape the long run convergence of beliefs. This interaction is

chiefly ruled by the level of benefits (both tangible and intangible) that individuals derive

from either adhering to or rejecting a particular notion about global warming. Three, the

process of belief formation also influences public policies where various groups compete

to get favorable policy decisions made by the government. Failure or success of these

efforts also influences the long term benefits to the groups from sticking to a particular

view point, thereby making the process of belief formation endogenous. Contrary to the









common perception that public beliefs influence public policy over uncertain events such as

global warming, this paper argues that the link between beliefs and public policies is much

complex than that of simple unidirectional causality. In fact, when public policies have

differential impact on different cohorts in society, formation of beliefs is endogenous to the

impact of public policies on the cohorts' common interests. Thereby, belief formation may

be subject to certain inertia from cohort adherence which may inhibit or alter optimal

selection of public policies.



Model

We build upon the adaptive learning model for technology adoption proposed by Ellison

and Feudenberg (1993) to incorporate a collective risk formation framework. Consider a

population that is exposed to periodic information related to the possibility of a natural

hazard. This information could be in the form of observation of phenomenon supposedly

correlated with global warming1. Individual beliefs are influenced by several factors.

One of them is the knowledge over what every one else in society believes. Other factors

are a person's cultural background, his level of education, income, risk attitudes, etc.

Information received through signals is parameterized by c which is uniformly

distributed between { c, a }. The individual belief of global warming phenomenon is

influenced by his observation of the actual value of c in each period. This parameter

could stand for an increased frequency of weather related events such as hurricanes or an

increased variability in events such as mean summer or winter temperatures. We only

1 For instance, hurricanes are considered to be caused by oceanic warming which is directly related to
global warming. An increased frequency of hurricanes could then be taken as a proof of the global
warming process. Recent evidence does find a 'smoking gun between global warming and hurricane
intensity' (New Scientist 2005). The individual's decision to believe it or not as such is based upon his
observation of these related events that occur periodically.









consider a specific signal perceived to be most significant in belief formation related to

global warming here. Assume that if the value of c is positive, he is going to become a

'believer' in global warming. If it is negative he is going to remain a 'non-believer'. It is

also possible for the believers to turn into non-believers after subsequent observations of

the contrary evidence. Other factors such as culture, race, education and risk aversion

also influence an individual belief patterns but are not taken up here.

In order to incorporate dependence upon rest of the population for belief

formation, we borrow from the literature on adaptive learning and allow for what has

been termed as 'popularity weighing'. Popularity weighing implies that an individual is

more likely to fall in line with the majority belief pattern as the number of believers or

non-believers increases. Ellison and Feudenberg (1993) use the function m(1-2x)to

incorporate popularity weighing, which implies that once the fraction of population

(given by x) is more than half, the sign of the term in the function reverses. Suppose that

the individual is going to turn into a believer when: c > m(1 2x). Here, a positive value

of c is not enough for him to change his beliefs. He is going to wait until a majority of

the population believes positively about global warming, after which even a slight

negative value of c would be enough for him to change his beliefs. It is possible that this

fraction of population beyond which the sign of the term m(1 -2x) changes is less than

half, based upon several factors. For instance, when there is heterogeneity in the

population in terms of their risk perception, education, race, etc., the impact of 'popular

opinion' may start showing as early as when only a third or a fourth of the population has

turns into believers.









In order to see how popularity weighing influences risk perception, figure 1 below

demonstrates the weighted probability of a person (or a fraction of population) turning

into believer even as the actual probability of a positive signal remains constant. If one

assumes that the random event is drawn from a uniform distribution, the probability that

c > 0is given by 1/2. However, when there is popularity weighing involved; the

a m(1 2x)
probability that E > m(1 2x) is given by- Notice that this term will always
20

be less than half as long as x is positive. Also, this value will be increasing in magnitude

as x increases as shown below.

INSERT FIGURE 1 HERE

In order to explore this issue in more detail we construct a simple example. Let

0 < x(t) <1 be the fraction of people who have turned into believers at time t. There is

inertia in the system as a result of which only a fraction of the population can believe or

reject its current stand per unit of time. More specifically, the fraction of people who

have turned believers can take the following possible steps:

113
(1) x(t)= 0, ,1
4'2 4

The probabilities of forward and backward steps are given by:


p(x() x()) = p( > m(- 2x))
(2) 4 2
p(x( ) x( ))= p( 2 4

For M=2;o=4, the probabilities of transition between states are given as: {pl, p2, p3, p4,

p5}= {0.25, 0.375, 0.5, 0.625, and 0.75}. Long run fraction of time spent in each of the

states is given by: {P1,P2,P3,P4,P5} = {.323,.129,.097,.129,.323}, where pi is the










probability of transmission from x(0) -> x( ) and so on and P1 is the long term fraction


of time spent in state x(0). Notice that both the sates when the entire population is either

a believer or a non-believer are equally attractive in the long run. Due to a popularity

weight of 2 (on m), the fraction of times spent in states 0 and 1 is equal, thus implying

that in the long run belief patterns are going to fluctuate wildly between the extreme

states. When the variance in the random signals is reduced,(m=2;o=3;) the long run

distribution is given as: {P1,P2,P3,P4,P5} = {.375,.094,.062,.094,.375}, which leads to

an even higher fluctuation between the extremes. When the popularity weight is

increased to {M=3,o=4} the distribution is given

as: {P1,P2,P3,P4,P5} = {.4,.07,.04,.07,.4}. It is obvious that the fraction of times spent

in each of the states is a function of the key parameters such as the extent of popularity

weighing, m, and the range of the random evento-. In order to explore the possibility of

full convergence to extreme states where every one is either a believer or non-believer all

the time, we follow the analysis in Ellison and Feudenberg (1993) and present the

conditions they have derived in Appendix A.

While full convergence of beliefs is definitely of interest for policy purposes, it is

yet to be determined how beliefs may converge in a society that is heterogeneous.

Society is characterized by various interest groups with differing stakes in the outcomes

of global warming and polices aimed at containing it. These interest groups or cohorts

may be prone to interpreting the same signal in a different way, thus accepting or

discarding the notion of global warming. A possibility also exists that these groups might

try to influence each others beliefs either through direct communication or through









indirect ways such as influencing public polices that punish or reward behavior that

augments or mitigates the risk from global warming.

Next, we consider the evolution of beliefs when they may be influenced by

individual's adherence to certain cohorts in society with common interests and well

settled belief patterns that are characterized by race, education, ethical and ideological

beliefs.



Cohorts and Interaction

People belonging to a particular group are likely to treat such threats as challenge to their

own ways of life and position in the society (Kahan et al. 2005). The risk of global

warming requires serious cutbacks in emissions which come at a cost of the current

energy-consuming lifestyles of the developing and developed nations. Even within a

country, there are various groups that would react to such threats differentially.

Assume that there are two cohorts now, L (liberals) and C (conservatives). The groups,

liberals and conservatives, are defined here in terms of their location on the scale

representing individualism and solidarism as the two extremes (see Kahan et al. 2005 for

more discussion on individualism and solidarism). Liberals are assumed to be solidarists;

more receptive to the risk of global warming and are more likely to revise their beliefs

based upon positive observations over the global warming phenomenon. Conservatives

are assumed to be more individualists and therefore would tend to care less about the

collective impact of global warming. However, the population dynamics of believers

within that cohort is now given by: accept the risk of global warming if: E > 1 + n(1 2x),

where / is the amount of discounting done on any random periodic event and is a function









of the allegiance to the liberal group. In this framework, / is the cohort adherence

weighting and n(1-2x) is the popularity weighting within the cohort. Similarly, the

population dynamics within the conservative cohort is given by: e > c + m(1 2x), with

m >n and c > That is, the conservatives are prone to discount the periodic

information at a higher rate than the liberals due to their stronger group allegiance and are

also more likely to pay attention to what others believe within their cohort due to their

hierarchical pattern. Liberals are more likely to be egalitarian and thus independently

assess the information. Assuming o=4;n=2;1=1; we can derive the steady state

distribution of beliefs within the liberal cohort as: {0.74,0.12,0.04,0.03,0.04}. Notice that

the fraction of time spent in the states when risk of global warming is highly discounted

is very high. This is due to the extra group adherence weighing added to the model. For

the case of conservative cohort, we actually see convergence to the state when everybody

completely discounts the risk of global warming. For the parameters, o=4; m=2; c=2; we

get the steady state distribution as: {1,0,0,0,0}. Consequently, due to their higher group

adherence weighing and popularity weighing tendencies, conservatives are more likely to

converge to a state in the long run where every one is a non-believer. In order to

precisely derive the condition for full convergence, as given by equation (13):

o +m+c (c+m-) o
Xs > 1, x > 0 => x(t)- 0, where X >- and x Notice that
2m 2m

equation (13) would be satisfied when: j+c>m and m+c>c7, which would be

satisfied when a = m + c. That is for any given level of o-, since c > 1 and m > n, the

conservatives are more likely to converge to the state 0.









Next we allow for communication between the two groups. Besides believing the

believers within one's own cohort, an individual now also considers the number of

believers inside the opposite cohort. The problem for the individual within the liberal

group is then to accept the prospect of global warming if: c > / +n(- 2(x+ y)).

Similarly, the problem for the individual within the conservative group is to become a

believer if:c > c + m(l-2(x+ y)). Now, the dynamics within the two groups is

dependent upon the believer population in each of them. For instance, a high number of

believers within the conservative group would expedite convergence towards full belief

over global warming in the entire population, as it would provide positive feedbacks to

the signals received in each period. On the other hand, a high number of non-believers

amongst the liberal group would lead to the entire population moving towards a state of

non-belief. This is akin to providing resiliency in the system towards sticking to extreme

ends. Two caveats apply; first, starting level of believer (or non-believer) population in

the two cohorts would play a crucial role in determining which set of beliefs is converged

to by the entire population. Besides, the speed of convergence itself may increase.

Second, we assumed that individuals are influenced in a positive fashion by the believing

(or non-believing) population in the opposite cohort. It may happen that individuals

within one cohort react adversely to the beliefs held by majority population in the other

cohort. Consequently, a higher number of non-believers in the conservative cohort may

actually force the liberals to assign an even higher positive weightage to the periodic

signal. Strong group adherence may play an important role in this kind of behavior, as

individuals may perceive an increasing number of people holding contrary opinion in the

opposing group as a threat to their own set of values that their group stands for. In this









way, weightage applied to the signal would include all those other values that might get

threatened by adoption (or dis-adoption) of belief over global warming. We explore

these issues in the following example.

Consider the groups I and c. For simplicity we do away with the group adherence

weights / and c and consequently, the decision to believe for the liberals is now given

as: c > (n(1 2x) ql( 2y), where y is the population of believers in the conservative

group. When parameter ql is positive, it would imply positive feedback from beliefs

within the opposing group and vice versa. Decision to believe within the conservative

group is now made if: E > m(1 2y) q2(1 2x), where x is the population of believers

within the liberal group. Probability of a forward step for liberals is given by:

2 + ql(1 2y)- m(1 2x)
(16) P[E > (m(1 2x) ql( 2y)] =
20c

Probability of a forward step for conservatives is given by:

a m(1 2y) + q2(1- 2x)
(17) P[ > (m(1 2y) q2(1- 2x)] = 2y 2
20

Now, in order to look at the steady state distribution of the system, we divide the state

space into nine parts as follows:

(18) { xOyO, xOy.5, xOy, x. 5y0, x. 5y. 5, x. 5y, xyO, xy. 5, xy}

The transition matrix representing the probability of transition between these nine states

is shown in the Appendix. When there is no cross cohort learning, good possibilities

exist for full convergence of both groups towards turning into full believers or non-

believers. This can be demonstrated for a given set of parameters: (-=5; m=2; n=l;

ql=0; q2=0). The steady state distribution (say for the base case) is now defined as:









(19) {x0yO, xOy.5, xOy, x. 5yO, x.5y.5, x.5y, xyO, xy.5, xy}

{0.166,0.086,0.13,0.11,0.08,0.08,0.13,0.07,0.12}

Next we allow for learning between cohorts. That, is one group positively reacts to the

information set available within the other group. This is shown by making the learning

parameters ql and q2 negative, which implies that one group is more likely to discount

positive information related to global warming if the other group does that too and vice

versa. For the given parameters:(o-=5;m=2;n=l;ql=-2;q2=-2), the steady state is solved

as:

(20) {xOyO, xOy.5, xOy, x. 5yO, x.5y.5, x.5y, xyO, xy.5, xy}

{.34,.06,.04,.09,.05,.06,.04,.04,.23}

Now, notice the sharp increase in tendency towards the states xy and xOyO. However,

notice a counter intuitive result in the below examples which arises from reversing the

learning behavior from opposite groups. Now, liberals react adversely to an increase in

the number of believers in the conservative group by negatively weighing the information

based upon the population of believers amongst the conservatives. Notice that, now the

system spends significant and equal amounts of time in the states xOy and xyO. This is

due to the aversion of the liberals cohort towards conservatives' belief systems. For the

given parameters:(o-=5;m=2;n=l;ql=2;q2=-1), the steady state is solved as:

(21) {xOyO, xOy.5, xOy, x.5yO, x.5y.5, x.5y, xyO, xy.5, xy}

{0.12,0.09,0.16,0.11,0.06,0.09,0.16,0.08,0.09}

Next, consider the possibility that even conservatives react adversely to the beliefs of the

liberals and are likely to discount a positive signal related to global warming if the









majority of the population amongst the liberal group is a believer. For parameter values

(o-=5;m=2;n=l;ql=2;q2=1), the steady state distribution in these nine states is given as:

(22) {x0yO, xOy.5, xOy, x. 5y0, x.5y.5, x.5y, xyO, xy.5, xy}

{. 084,.076,.22,.097,.07,.07,.24,.07,.066}

Notice that highest possibilities exist for states xOy andxyO. This says that the system

has a propensity to move towards states where either liberals have turned fully into

believers and the conservatives into non-believers or vice versa as predicted before.

When the aversion of conservatives towards liberals is increased, this behavior is even

more pronounced. This is given by parameters: (o-=5;m=2;n=l;ql=2;q2=2), the steady

state distribution is now given as:

(23) {x0yO, xOy.5, xOy, x. 5yO, x.5y.5, x. 5y, xyO, xy.5, xy}

{0.06,0.058,0.26,0.09,0.06,0.064,0.28,0.055,0.05}

The extent of role played by group aversion in influencing risk perception could be a

significant issue, but can only be empirically or experimentally ascertained. So far we

have focused upon belief formation without considering the impact of these beliefs on

public policy. Next section takes up this issue.



Formation of Behavioral Thresholds

The above section assumed static thresholds over cohort weights / and c.

However, it is possible that each cohort's threshold may get adjusted in a dynamic

fashion due to collective reassessment of the situation or hardening of evidences. But,

this threshold so far has been assumed to be exogenously given. It is likely that this

threshold would signify what the individual may have to sacrifice in order to switch his









belief pattern from a non-believer to a believer. For instance, producers of carbon

emitting devices may have a lot to loose from a confirmation of the risk of global

warming. Their own forgone profits therefore would become a factor in influencing

their belief patterns. However, the collective threshold should be representative of

forgone profits of the entire cohort. In order to derive this threshold, we take recourse to

the political economy framework that assumes that different cohorts in the society

bargain with the government to further their own interests. The government in turn

selects policies that maximize its chance of survival which are a function of the revenues

received and the impact of such policies on its future votes.

In this section, there is interaction allowed between cohorts in terms of their

standings over risk perception within a political economy framework. Let the fraction of

believer population within the liberal and conservative cohorts be given as x and y. Let

S(x+y) be the political gain to the incumbent government from supporting the population

that believes in global warming and acting according to their demands. Then S(2-x-y)

would be the gain from supporting the non-believers. Let C, be the contribution made

by the conservatives and L, the contribution made by the liberals to influence public

decision making related to global warming. Let Cb be the benefit (perceived and

tangible) to the conservatives from the incumbent government not taking any policy

measures such as carbon abatement, etc. Similarly, the (perceived and tangible) benefit

to the liberals from the incumbent taking carbon abatement measures isLb. Negotiations

between government and the two groups happen in the form of a bargaining game where

the two maximize the product of their surpluses. The product of the surpluses of

conservatives and the government is given by:









(24) [s(2-x- y)+C, S(x+ y)-L,]{C -C,}

Maximization of this surplus with respect to contributions by conservatives leads to:


(25) C, = Cb + L, + S(x + y)- S(2 x- y)}
2

The above gives a best response function for the contribution of the conservatives as a

function of the contribution of the liberals. Therefore, we need to solve the best response

functions of the two groups simultaneously in order to derive their respective

contributions. The product of surpluses for the bargaining game between the liberals and

the government is given by:

(26) [s(x + y)+ L S(2 -- y) C ]{Lb- L

Maximization of which with respect to contributions from the liberals leads to:

(27) L, = Lb +C + S(2- x -y)- S(x + y)}
2

For any given level of population of believers within the two cohorts, there is going to be

a unique solution where either the liberals or the conservatives make the higher

contribution, thus influencing public policy. In order to see this, consider the scenario

where: S=l, Cb =1 = Lb. Then, the response function of the two cohorts is given as:

1 1
Cc = 1+Lc-2} andLc =-l+Cc+2}, solving which we get the level of
2 2

contributions where both the cohorts have the same response to each others contributions.

When the population of believers is low (say zero), the liberals are forced to make very

high contributions in order to be able to influence public policy. In fact, when x=0=y, the

contribution required from liberals in order to match the response of the conservatives is

1.6, which is more than their benefit of 1. Consequently, liberals will not be able to









influence public policy under this situation. Table 1 below shows the benefits to the two

groups from bargaining for given levels of population divisions.

INSERT TABLE 1 HERE

The figures below show the intersection of response functions for the two groups when

there are no believers in both groups and when there is equal number of believers in both

the groups.

INSERT FIGURE 2 and 3 HERE

A Nash equilibrium outcome like the one derived above does not say anything about how

it would be reached. It is more likely that one of the more dominant groups would act as

a Stackelberg leader and the other the follower. In here, we assume that the

conservatives act as leaders and the liberals as the followers. In this game the

conservatives incorporate the best response function of the liberals in their own surplus

maximization problem with the government as:

(29) [s(2- x y) + Cc S(x + y)- Lc]{Cb Cc

which leads to the below outcomes as shown in table 2:

INSERT TABLE 2 HERE

Notice that for the first two cases when the fraction of believers among the two groups

combined is 0 and 0.5, the conservatives are able to influence public policy. Their

surplus after contributions in the two cases is 1 and 0.5 respectively. That is, the surplus

keeps declining as the number of supporters of the global warming hypothesis falls. This

would imply that as the surplus from bargaining falls, the benefits from group adherence

also decline and therefore the group adherence threshold would fall too. This would

make it easier for the remaining population to adjust their beliefs faster, thus hastening









convergence. Similar logic would imply that as the population of believers declines,

convergence towards non-believing states is hastened.



Conclusion

In this paper we demonstrated three crucial points. One, the level of learning from

random climatic signals could be easily adulterated by the individual's reference frame,

which is defined by his group adherence. There is interaction between different groups

with different ideologies that shape the long run convergence of beliefs. This interaction

is chiefly ruled by the level of benefits (both tangible and intangible) that individuals

derive from either adhering to or rejecting a particular notion about global warming.

Finally, the process of belief formation also influences public policies in which various

groups compete to get favorable policy decisions made by the government. Failure or

success of these efforts also influences the long term benefits to the groups from sticking

to a particular view point, thereby making the process of belief formation endogenous.

Further, when the process of belief formation becomes endogenous to groups' interests

and the impact of public policies on them, the process of belief formation may suffer

from an inertia making objective judgment of climatic signals difficult.

While individuals tend to weigh risks of losses and gains, it is still not clear how

individual weighing of risks transforms into a collective weighing at the societal level. In

order to understand a possible link, a probable hypothesis could be that in absence of

individual's own lack of faith in the signal received from the periodic event, he or she is

more likely to weigh the observation by what others perceive the signal to be. When the

loss is personal, an individual would treat that risk differently as compared to the case









when the loss is at a much aggregated level, even though the amount of expected loss is

the same in both cases. This discrepancy may arise due to the interdependence of

individuals on society for learning the nature of risks. An individual is more likely to

discount the risk of an event that affects the entire society if every one else in the society

discounts it too. Whereas, when the risk is personal, he or she is more prone to rely on his

own judgments, thus showing the observed inverted s-shaped weighing of risks. Based

on accumulated evidence in economics and psychology literature (see summary in Hurley

and Shogren, 2005), one could assume that the individual assigns higher weights to low

probabilities of a catastrophe and lower weights to high probabilities of the same event

(also see Starmer, 2000). The weighing function usually follows an inverse S-shape as

has been assumed in the literature. Following Prelec (1998), a two-parameter weighting

function could be used to define the weight as:w(p)=e- (lnp) where 0 is the

parameter that primarily determines elevation, and y is the parameter that primarily

determines curvature. Elevation reflects the inflection (reference) point at which a person

switches from overestimating low probability events to underestimating high probability

events, i.e., the degree of over- and underestimation; curvature captures the idea that

people become less sensitive to changes in probability the further they are from the

inflection point (Tversky and Kahneman, 1992; Gonzales and Wu, 1999). By adding a

subjective weighing that discounts high probabilities, an individual cares lesser about the

catastrophe; the higher is its probability. The inflexion point of the inverted s-shaped

curve is critical and can only be determined empirically. This weighing scheme is shown

in the figure below.









Some of the future challenges along this line of thought could be to incorporate

individual weighing of risks in the models presented in this paper and figure out the role

of risk weighing in influencing belief formation from the other factors mentioned in this

paper. There are other factors not discussed in this paper that too could play a crucial

role in belief formation. Studies have found that people often learn of the quality of sea

water from beach closings instead of forming their own judgments based upon direct

observation. Media information could play a much more crucial and dominant role in

influencing belief formation instead of popularity weighing, which could be a slower

process. The influence of media information and government actions are crucial in

influencing belief formation. A lot of work is remaining to be done in deciphering this

simultaneity possible between belief formation and public decision process.












References:


1. Ellison, G., D. Feudenberg "Word of Mouth Communication and Social
Learning", The Quarterly Journal ofEconomics, 100(1), 1995: 93-125
2. Ellison, G., D. Feudenberg. "Rules of Thumb for Social Learning", The journal of
PoliticalEconomy, Vol. 101, No. 4, 612-643 (1993).
3. Kahan, Dan M., Braman, Donald, Gastil, John, Slovic, Paul and Mertz, C. K.,
"Gender, Race, and Risk Perception: The Influence of Cultural Status Anxiety"
(April 7, 2005). Yale Law School, Public Law Working Paper No. 86.
http://ssrn.com/abstract 723762
4. Pidgeon, N., R. E. Kasperson, and P. Slovic, Edited "The Social Amplification of
Risk, Cambridge University Press 2003.
5. Berhout, F., Hertin, J., D.M. Gann, Learning to Adapt: Organizational Adaptation
to Climate Change Impacts, Tyndal Centre Working Paper no. 47 (2004).
6. Daft, R. L. and K.E. Wieck, Toward a Model of Organizations as Interpretation
Systems, Academy ofManagement Review, 9, 284-295 (1984).
7. Bleda, M. and Shackely, S., The Formation of Belief in Climate Change in
Business Organizations: A Dynamic Simulation Model, Tyndall Center Working
paper no. 68 (2005).
8. Gusfiled, J. R. Symbolic Crusade: Status Politics and the American Temperance
Movement. University of Illinois Press.
9. Young, T. "The Evolution of Conventions", Econometrica, Vol. 61, 57-84, 1993
10. Developing Resistance, Nature 439: 1-2 (2006).
11. Gonzalez, R. and G. Wu. "On the Shape of Probability Weighting Function",
Cognitive Psychology 38, 129-166 (1999).
12. New Scientist, "Global Warming may Pump up Hurricane Power", 2005,
http://www.newscientist.com/article.ns?id dn7769
13. Prelec, D. "The Probability Weighting Function", Econometrica, 66(3) 497-527
(1998).
14. Starmer, C. "Developments in the Non-Expected Utility Theory: The Hunt for a
Descriptive Theory of Choice under Risk", Journal of Economic Literature, 331-
382 (2000).
15. Tversky, A., and D. Kahneman. "Advances in Prospect Theory: Cumulative
Representation of Uncertainty", Journal of Risk and Uncertainty, 5, 297-323,
(1992).











Appendix A: Derivation of Conditions for convergence based upon Ellison and
Feudenberg (1993)

Following their approach, let xg be the level beyond which the belief over global

warming is certain to be adopted. This can be derived noting that x is certain to move

forward if the minimum value of weighted random event is positive. This is possible

when c = -o :

(4) x(t) > xg -a- > m(l 2x)

This gives:


(5) xg >--
2m

Similarly, the value of x, say xf below which a backward step (towards non-belief) takes

place with certainty is given as:

(6) x(t) < xf < m(1 2x), which gives:


(7) xf < -)
2m

Also, realizing that the minimum probability of an upward step is positive when x=0, we

get this probability as:

(8) P(E > m), or,

S m -x'(2m)
(9) P[ > m] =2
20c 20c

Similarly, the minimum probability of a downward step is realized when x=l:

(10) P(E < -m), or,

( -m (xg -1)(2m)
(11) P[E < -m] =
20 20









From above we can derive the conditions for convergence of the beliefs as:

(12) xg < 1, xf < 0 => x(t) 1

(13) xg > 1, xf > 0 => x(t) 0

(14) xg > 1, xf < 0 => no convergence

(15) xg <, xf > 0,if x, > xg => x(t) = 1, however if x, < xf => x(t) = 0

Condition (12) implies that the belief over global warming will eventually get adopted by

everyone if xg < 1, xf < 0. Also note that when > ,: xg < 1, but xf > 0 Similarly,

when m < a, xg > 1, but xf <0 Therefore, the conditions for full convergence as given

by equations (14) and (15) do not hold. Further discussions and extensions over these

conditions (though in a different setting) are nicely detailed in Ellison and Feudenberg

(1993).










Table 1 : Benefits to the two Groups from Bargaining for a given Level of Believing

Population







C, C, x y

1.66 .33 0 0

1.33 .66 .25 .25

1 1 .5 .5

.66 1.3 .75 .75

.33 1.6 1 1












Table 2: Benefits to the two Groups from Bargaining for a given Level of Believing

Population when Conservatives Act as Stackelberg Leaders


X Y Cc Ic

0 0 0 1.5

.25 .25 .5 1.25

.5 .5 1 1

.75 .75 1.5 .75











Figure 1: Influence of Popularity Weighting and Risk Perception


0.2 0.4 0.6 0.8 1


Fraction of Believers


sigma 5;m 2;


Adjusted Probability












Figure 2: Intersection of Response Functions for the two Groups when there is no


Believer in Either Group


Liberals' best
response funct


Conservatives'
best response
function
-1


x=y=O












Figure 3: Intersection of Response Functions for the two Groups when there are

Equal Number of Believers in Both the Groups.






c Liberals' best
5 response function

4

3
Conservatives'
2 -best response
function


1 1.5


2 2.5


x=y=.5









Figure 4: Weighing of Risks by Individuals


1


0.8


Weighted
Probability 0.6


0.4


1
Probability


a=.5;b=.5;Plot[{p,EA(-a(-Log[p])Ab)}, {p,0,1 }];











Table 4: Transition Matrix between States
{x0y0, xOy.5, xOy, x.5y0, x.5y.5, x.5y, xyO, xy.5, xy}


Note: Elements in each of the nine parentheses denote transition from the state
corresponding to that parenthesis to all other states.




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs