download (590kb) - munich personal repec archive
TRANSCRIPT
Munich Personal RePEc Archive
Landscape in the Economy of
Conspicuous Consumptions
Situngkir, Hokky
Bandung Fe Institute
7 May 2010
Online at https://mpra.ub.uni-muenchen.de/22948/
MPRA Paper No. 22948, posted 30 May 2010 06:37 UTC
1
Landscape in the Economy of
Conspicuous Consumpt ions
Hokky Situngkir
Dept . Computat ional Sociology
Bandung Fe Inst itute
Abstract
Psychological states side by side with the bounded rat ional expectat ions among social agents
contributes to the pat tern of consumpt ions in economic system. One of the psychological states are
the envy – a tendency to emulate any gaps with other agents’ propert ies. The evolut ionary game
theoret ic works on conspicuous consumpt ion are explored by growing the micro-view of economic
agency in lat t ice-based populat ions, the landscape of consumpt ions. The emerged macro-view of
mult iple equilibria is shown in computat ional simulat ive demonstrat ions altogether with the spat ial
clustered agents based upon the emerged agents’ economic profiles.
Keywords: conspicuous consumpt ion, behavioral economics, agent-based simulat ions.
2
Envy is ever joined with the comparing of a man's self;
and where there is no comparison, no envy!
F. Bacon, Sr.
Thou shalt not covet thy neighbor's wife;
thou shalt not covet thy neighbor's house, nor his field ...
nor anything that is thy neighbor's.
the 10th
rule in ten commandments
1. Introduction
There have been broad recent understandings on human economic behaviors outside economic
discourses. Economic decisions are always becoming one of the most complex things in economic
discussions. One of them is the way we choose our pat tern of consumpt ion. It has been a common
understanding that the pat tern of consumpt ion is st rongly related to the social and economic status.
A lot of things are being marketed not only for the funct ionality or degree of necessity solely, but
also related to the profile beneficiaries. M ost luxurious stuffs are placed in the market as a sort of
conspicuous consumpt ion. They are being sold and bought for the need of good reputat ion,
pecuniary st rength and good name of buyers, or social leisure of having more than others. This is
direct ly reflected in our daily urban and highly organized industrial society [19], from the type of
cellular phones, dressings, style and brand, even place of lunch and recreat ion to the luxurious home
living and daily vehicles.
While income is an important measure for social class, the pat tern of consumpt ion is the way to
show other people conspicuously the represented social class. A good and unleashed descript ion of
the relatedness of between consumpt ion is described in [18]. A lot of things are thus, bought in
order to be shown to others. Furthermore, research surveys have also confirmed how social
st rat ificat ion correlated to the cultural consumpt ion [2].
Nonetheless, those are economic phenomena since it has been direct ly related to the concavity of
the supply and demand curves for part icular products in the market , but yet , it is also related to a
deep emot ional and psychological t rait of human species, envy. This is related an interest ing field
related to behavior economics. Inequality has been understood to be one of source of unhappiness
among people (cf. [7]). Furthermore, the interest ing relat ions between inequality with the state of
well-being or happiness are related to the emot ional states of human being. As it has been noted in
[13], emot ions serves an adapt ive role in helping organisms deal with key survival issues posed by
the environment . Emot ional based decision making among economic agents might have been one of
explanat ion to the deviat ion of rat ionality in the sense of convent ional understanding (cf. [10]),
beside the realizat ions on the social boundedness on which rat ional choices must be taken ([16] &
[9]).
Recent economic discussions have int roduced a lot of interest ing discourses related to this. The
evolut ionary game theory as harmonious mixtures between biological studies, economic behavior,
and mathemat ical theories on games [17] has out lined some applicat ions into the recent problems
in economics [5]. One of interest ing points are related to the formalizat ions of the social pat tern
regarding to the game of the conspicuous consumpt ion [4]. The discourse of the evolut ionary game
theoret ic analysis on economic issues, the emot ional based economic decision making as well as the
boundedly rat ional agent based model employment in computat ional analysis are those becoming
main issues mot ivated the paper.
3
The paper is begun with the discussions related to the overview of the mathemat ical models related
to the conspicuous consumpt ions. M ost development of the model has been explored in the fashion
of analyt ically evolut ionary game theoret ic models. Discussions are cont inued to the implementat ion
of the models for wider theoret ical explorat ions by incorporat ing the acquisit ions of computat ional
simulat ions. By the end of the paper, some demonstrat ions of the toy model are presented with
some out lines to some conjectures in further development .
2. Overview of the M odel
The presentat ion of the paper shows lat t ice-based populat ions as landscapes reflect ing the
allocat ions of ordinary and conspicuous expenses by economic agents. As described in [8], the lat t ice
based populat ions, or more generally the neighborhood st ructure usually “ tends to favor the long-
term co-existence of st rategies which would not co-exist in well-mixed populat ions” . This is the
dynamics as once int roduced in [1, 6] and thus implemented in cultural disseminat ion model [12],
and also used as model to discover some dynamical characterist ics in corrupt ion [14]. The plat form
is part icular kinds of agent based model [11] in which we ut ilize as computat ional experiments
media. Another similar previous work related to advert ising could also worth for ment ion [15].
Imagine a landscape where people are represented on lat t ices and grids. Each agent is given the
same amount of money, and it is on their decision to allocate an (0,1]x = amount of the money for
ordinary necessit ies. While the savings and investments are neglected, the 1 x− fract ion of the
money is thus allocated for luxurious expenditures. The later expense is thus becoming the source of
envy among economic agents, things that influence their apprehension on their surroundings and
thus give impact to their decision making. Thus, the expected pay off on each round of the game
thus depends on each agent ’s ordinary consumpt ion ( u ) and her allocat ion of for the conspicuous
consumpt ion (U ),
U cuϕ = + (1)
where 0c ≥ denotes the constant marginal rate subst itut ion. The ordinary consumpt ion can be
stated as a concave ut ility funct ion,
lncu c x= , (2)
and as 0x → we have ϕ → −∞ reflect ing the importance of the ordinary consumpt ion, while the
ut ility funct ion due to ordinary expenses would be ( ,0]cu ∈ −∞ - a fact that no one will completely
neglects the ordinary consumpt ion expect for the limit ing case of 0c = .
The main of the focus in the game is thus the conspicuous consumpt ion: how much is the fract ion
needed to sat isfy the social effect of envy when other surrounding people have more (or less)
fract ion for the conspicuous expenses. An evolut ionary modeling of this has been analyt ically
analyzed by the calculat ing the gradient dynamics in the fashion of cumulat ive dist ribut ion of
funct ion [4].
Our approach in the paper is related to the implementat ion of the models showing the economic
behavior as pointed out in [19] as consumpt ions of the excellent goods that is socially related to the
evidence of wealth. In this game, if an economic agent compare her allocated expenses for excellent
goods is bigger than other, she would be in the psychological state of envy – a thing that mot ivates
her to allocate bigger consumpt ion if they interact again in the future. This envy however must be
constrained due to her allocat ion of ordinary consumpt ion which is a necessity and cannot be
4
nullif ied. Here, when an agent allocates a fract ion of x to the ordinary consumpt ion (thus allocate a
fract ion of (1 )x− for conspicuous one) and find out that other agent that interacts with her pay the
fract ion of y x< for the same respect ive expenditures (thus (1 )y− for excellent goods), then she
will have disut ility in the proport ion of
( , ) min{0, ( )}r x y y x= − (3)
In the analyt ical model as proposed by [4], the payoff can be writ ten,
1
0 0 0
( , ) ( , ) ( ) ( ) ( ) ( )
x x
U x D r x y dD y y x dD y D y dy= = − = −∫ ∫ ∫ (4)
where ( )D y is the cumulat ive dist ribut ion funct ion of other’s decision on the fract ion to the
ordinary consumpt ion and give us a consequence to the total payoff of an agent ,
0
( , ) ln ( )
x
x D c x D y dyϕ = − ∫ . (5)
It is easy to see that that in the limit ing case
0
( ) 0
x
D y dy =∫ , a corresponding agent that allocates
sufficient ly small amount of x for ordinary consumpt ion gets the maximum payoff related to her
envy to others. The gradient dynamics, on which agents adjust the value of x is.
( )x
cD x
x x
ϕϕ ∂ = −∂
@ (6)
Figure 1. The landscape of allocat ion for ordinary consumpt ion
after the rule of envy is implemented for T=250 rounds
3. Experiments with spatially-bounded agents
A numerical simulat ions is conducted in [5] in order to see the possible stable equilibria of the
problems by a discret izat ion of equat ion (5) in such a way,
5
1 1
1ln ln
N N
i i
i i
c x F x c x FN
ϕ= =
= − ∆ = −∑ ∑ (7)
With the gradient as analyt ically shown in eq. (6),
n n
n
cF
xϕ = − (8)
where i
F denotes the cumulat ive dist ribut ion funct ion obtained from the numerical integrat ion of
the probability measure i
f in the i -th interval among
1
N
i i
k
F f x=
= ∆∑ (9)
as i
f is the average density of M number consumers with st rategies divided in x∆ equal N
number of intervals, as to 1
xN
∆ = . Consequent ly, we have 0
( ) 0F x = and ( ) 1N
F x = , a sort of
rank among whole agents – from which the name rank dependent consumpt ion came from.
When the focus of our at tent ion is the mult iple equilibria, this approach has given interest ing
results. Yet , another perspect ive can also be offered related to the economic agent ’s boundedness
related to the conspicuous consumpt ion. In reality, the people are “ t rapped” in social surroundings
in which envy and vanity is sourced upon cognit ively. People do not have to compare his
expenditures with those he would only “ meet” on television, but people that are socially related to
them.
Figure 2. Different average equilibra at different values of c.
Thus,
cells or
view is mor
the ri
neighb
total n
the se
as,
( ,i jϕ
where
neighb
On an
( ,x i j t
where
consu
payoff
her nei
s, we t ransform
or square
is more lik
right grid
hborhood
l neighbors,
set of st rat
, , ) lni j t c=
re N
xυ
is
hbors, inde
any
, , 1)i j t + =
re (N N
x iυ υ
' ‹“ ›« fi
sumpt ion.
ayoff is larger
r neighbors’
nsform the
are lat t ices
like a torus
grids with t
d (with conn
ors, [ ]Nυ =t rategies x
[ln ( , ,c x i j t=
is the decis
ndexed as N
+ =
( ,x i
(x i' ‹“ ›« fi
( , , )N N
i j tυ υ
' ‹“ ›« fi
n. Thus, an
er than the
ors’ decision
the discret iz
s located at
torus: the low
the left
connect ions
8= ). In e
(0, ]x ∈ ∞
] 1, )
N
j tλ
= + −∑
cision of n
{1,2Nυ =
), , , (i j t iϕ ϕ
( , , )N N
x i j tυ υ
' ‹“ ›« fi
, )N
j tυ
' ‹“ ›« fi is th
an agent wo
the average
ons for ordi
Fig
t izat ion as
d at a two
lowest two
ft one. T
ns betwee
each t ime
, ]∈ ∞ and as an
) ( , ,N
t x i jυ
+ − ∑
neighbors
1,2,...,8} , t
( , , )i j tϕ ϕ< ' ‹“ ›« fi
, ) , N
t othυ
' ‹“ ›« fi
the averag
t would sta
age payoff of
rdinary con
Figure 3. Ag
as suggeste
two-dimensio
st two-dimensio
The myop
een nearest
me step t T
an adaptat
, , ) ( N Nj t x i + −
ors for ordi
, that has ma
) ( ,N N
t i j tυ υ
ϕ ϕ< ' ‹“ ›« fi
otherwise
rage of the
stay with t
ayoff of her nei
ary consumpt ion
Agents’ equili
sted in eq.
sional virtu
sional grids ar
myopic agen
arest and ne
1, 2,...,t T=tat ion of eq.
( , )N Ni j tυ υ
rdinary con
as made the
, , )N N
Nj t
υ υυ
' ‹“ ›« fi
se
the averag
the same
eighbors a
ion otherw
ents’ equilibrium po
q. (7) into r
rtual world
ids are past
ents evaluat
next -nearest
...,t T , every
eq. (1), (3)
, )
consumpt ion
he respect i
' ‹“ ›« fi
verage deci
same fract ion
ors and chang
rwise.
rium points for c=
representat
world of ,i j L=asted toget
luate their
rest neighb
ry agent ev
(3) and (4),
(10)
ion and λct ive agent
(11)
ecision of
ion of ordi
nge her de
c=0.1
ntat ion of
, 1,...,j L=
ether with t
eir payoffs
hbors, thus
evaluates
), we can wr
10)
λ denotes
nt envious,
11)
of neighbors
ordinary con
decision into
of agents p
while the
th the highe
ayoffs in the
hus each ag
s her payoff
n write the
tes the num
s, ( )x t y t<
bors for or
onsumpt ion
into the ave
6
placed in
the global
ghest , and
e M oore
agent has
ayoff from
he pay off
number of
) ( )N
y tυ
< .
ordinary
ion if her
average of
n
al
nd
e
s
m
off
of
y
er
of
4. Computational Experiments
From
among
condu
simulat
consu
strate
averag
subst i
consu
One of
social
interv
for cequilib
higher a
Intere
end of
higher
agents),
becom
. Computational Experiments
From the mod
ng agents
nducted in
ulat ive proc
sumpt ion a
st rategies for mor
rage allocat
st itut ion ( csumpt ion s
of interest
ial classes.
rvals in wh
0.1c = is s
ilibrium po
er and low
rest ingly, t
of rounds
er allocat io
nts), while
come a uniqu
. Computational Experiments
model, we ca
ts due to th
n square lat t
rocess is s
and thus aft
for more. The
ocat ion of or
c ). It is cle
should be
Figure 4. T
est ing thing
s. It is wort
hich agents
is shown in
point near
ower profile
y, this can a
nds of simulat
cat ion for or
ile most of
unique prop
. Computational Experiments
can do co
their alloca
lat t ices o
s shown in
s after rou
he not -cha
ordinary c
clear that
be allocated.
The emergen
ing that we
orth to note
ents stop ch
in figure 3.
ar the value
rofile agents. T
also be vis
ulat ion, it i
ordinary co
of agents ar
operty that
computat i
llocat ions of
of 30L = in figure
rounds of sim
hanging stat
ary consumpt
at the high
ated.
rgence of so
shows
we can lear
ote that th
changing t
3. However,
lue of the a
ts. Thus, the
visualized a
it is obvious
consumpt i
ts are in the
at we coul
tat ional exp
of ordinary
30 for Te 1: the in
simulat ion
states refle
pt ion as sh
gher the va
f social class
s the high, l
arn from o
the averag
g their allocat
ever, it is inte
he average a
the equilibr
d as the clu
us that the
pt ion (i.e.:
the state of
uld have g
xperiments
ary and con
250T = ro
init ial rand
on stay at e
flect the equ
shown in f
value of c
cial class: the allocat i
gh, low profile agents.
our simulat
rage value
locat ion for
nterest ing to
e as some l
ilibria is show
clustered ag
there are som
.e.: lower prof
of medium
gained from
nts that wo
conspicuous
rounds of
random cond
at equilibria
equilibrium
n figure 2 for
c , the mor
the allocat ion for o
file agents.
ulat ion is t
e as shown
for ordinary
g to see that
e less agen
own to be
agents spat
some clust
rofile agen
ium class of
from the com
would reve
us consump
of games
condit ion of
ria when age
um and can
2 for variou
more fract io
r ordinary c
s the emerge
wn in figur
ary consum
that most of
ents are kee
e in a kind
spat ially as
lustered age
ents), lowe
of ordinary
computat
eveal the u
umpt ions. O
s with var
of allocat
agents do n
can be seen
ious value
fract ion of incom
dinary consumpt i
ergence of
ure 2 cam
umpt ion. O
most of agents
keeping th
nd of interv
as shown in
agents to b
ower one (i.e.
ary consum
tat ional age
Cluster
Clusters
unique equ
. Our simul
ariable c =ocat ion to or
o not chan
een on the co
ue of margi
ncome to or
nsumpt ion
of spat ially
came from a
One of the
ts are findi
their statu
rval.
in figure 4
be in the
i.e.: higher
sumpt ion. T
agent-based
usters of lower
usters of highe
7
equilibria
ulat ion is
.1c = . The
ordinary
ange their
e constant
marginal rate
to ordinary
lly located
a sort of
he results
nding their
status as the
4. At the
e state of
er profile
n. This has
ed model
ower profile
gher profile
ia
is
e
ary
eir
t
e
y
ed
of
eir
e
e
of
le
as
del
ile
from t
of eco
consu
and low
From
lucky
are se
accum
consu
locat io
in suc
our sim
accum
throug
the model
economic c
sumpt ion in
lower prof
From our simu
cky enough t
set or at le
umulated p
sumpt ions.
ocat ions filled
uch ways e
simulat ion
umulated p
ughout the
among agents
del as int ro
c classes b
in the stat
rofile agents amo
Figure 5.
mulat ion, w
h that they
at least they
payoff rel
s. Obvious
lled with suc
ys emerging
ons. Those
payoffs t
the spat ial l
Fig
ng agents thro
high pay off
low pay off
t roduced in
based du
state of mu
ents among
. The landscape
are filled with agents wh
, we could
ey have alr
ey have ne
relat ively
usly, the e
uch lucky ag
ing the spat
se agents ar
there wo
al landscape
Figure 6. Th
throughout the c
ay off
pay off
in eq. (5) a
due to th
mult iple equ
g the sea of a
e landscape of pay
led with agents wh
ld also obs
already clos
neighbors w
y higher th
emergent
cky agents. Ro
at ially clust
ts are cluste
would be
pe related
The equilibri
ut the computat i
) and (6). Fr
their profile
equilibria:
a of averag
payoff after
ed with agents whose higher accumula
observe som
lose enoug
ors with such
than the
nt spat ial p
Rounds by
lustered ag
stered one
e such gra
ed to the op
equilibrium points as
mputat ional simulat i
). From the
rofile in thei
a: there are
rage profile
ter T=250 ro
se higher accumula
some inter
ugh to the
uch allocat i
e others. F
l pat terns
by rounds i
agents whos
e another
gradat ion.
opt imum a
ints as the at t ract
nal simulat ion.
he big pictu
their allocat
are “ islands”
le persons.
rounds: the darker the
se higher accumulated pay
erest ing as
e equilibri
ocat ions of or
. Figure 5
s are show
s in our sim
hose relat ivel
er and to t
n. This refl
m and best fract
the at t ractors based
on. The inter
cture, there
ocat ion to
nds” repres
s.
the darker the
se higher accumulated payoffs
aspects sp
ria when t
ordinary c
5 shows t
own here.
simulat ions
at ively high
the cluste
eflects the
st fract ion of or
based on en
he interval of the equili
re has bee
to ordinary
esent ing th
the darker the red cells
s.
spat ially. Som
n the random
ary consumpt
ows this in o
e. In the fig
ns they acc
igher payoffs
stered agen
he diffusion
n of ordinary
n envy
f the equilibria is sh
been an eme
ary and lu
the higher
cells
. Some age
ndom init ial
pt ion. Thos
our landsca
figure, dark
accumulate
ayoffs at the
ents whos
ions of st rat
nary consum
ria is shown.
8
mergence
luxurious
er profile
agents are
ializat ions
hose have
ndscape of
arker red
ate payoffs
he end of
ose lower
st rategies
sumpt ion.
wn.
e
s
ile
e
s
ve
f
ed
ayoffs
f
r
s
9
However, the process of the diffusions along the rounds of our computat ional simulat ions can be
seen vaguely in the phase map corresponding to the mapping of ( ) ( 1)x t x t→ + . This is shown in
the figure 6.The random init ializat ion has been at t racted to the intervals of equilibria. This is a sort of
at t ractor of envy. Whatever the init ial st rategies, they will always be at t racted to the intervals as
shown in figure 3, and the equilibrium st rategy decided by agents are thus related direct ly to their
respect ive spat ial posit ions i.e.: in what kind of neighborhood she is placed upon. Thus we have
demonstrated that the evolut ionary processes do not only depend upon the st rategies that are
being used in the games but also the respect ive posit ions of the strategic occupants.
5. Concluding Remarks
We have shown the discussions incorporat ing the computat ional agent based models of the game
on allocat ing ordinary and conspicuous consumpt ions among economic agents. While the task came
from one of the proxy of the evolut ionary (economics) game theory, the presentat ion is based on an
implemented fashion of computat ional simulat ions. The lat ter has been enriched the works on the
boundedness of agents due to the myopic evaluat ions of agents on their neighborhood and some
spat ially emerged facts while macro-view equilibria have been reached over rounds of simulat ions.
An interest ing result is shown related to the varying, yet clustered populat ions and an at t ractor of
envy on the landscape of allocat ions between ordinary and conspicuous consumpt ion. While the size
of the allocat ion for conspicuous consumpt ion is related to the economic profile of agents, it is also
visually shown in 2-dimensional lat t ices the emergence of higher, lower, as well as medium classes
of economic profiles.
While in general the model depicts how emot ional state – in this case, envy – could become a source
of the concavity of the demand funct ion, the promising future works could be conjectured to see
how evolut ionary game theory can be explored for other factors and variables used by agents, e.g.
the development of economic discourse based on psychological behaviors, evolut ionary paradigms
on mult iple equilibria, as well as computer simulat ions as toy models. The further works on this
research inquiries would also be conjectured to the recognit ion of psychological aspects e.g.
happiness or well-being-ness, to the emerging macro-pat tern of economic decisions. This is left for
future work.
Acknowledgement
Author thanks Surya Research Internat ional for the provision of financial support in which period the
paper is writ ten. All fault remains author’s.
Works Cited
[1] Axelrod, R. (1997). "The Disseminat ion of Culture: A M odel with Local Convergence and Global
Polarizat ion" . The Journal of Conflict Resolut ion 41 (2): 203-26. Sage.
[2] Chan, T. W. & Goldthorpe, J. H. (2007). "The Social St rat ificat ion of Cultural Consumpt ion: Some Policy
Implicat ions of a Research Project" . Cultural Trends 16 (4): 373-84. Rout ledge.
[3] Friedman, D. (1998). “ On Economic Applicat ions of Evolut ionary Game Theory” . Journal of
Evolut ionary Economics 8: 15-43. Springer-Verlag.
[4] Friedman, D. & Yellin, J. (2000). “ Cast les in Tuscany: The Dynamics of Rank Dependent Consumpt ion” .
UC Santa Cruz Economics Working Paper No. 455.
[5] Friedman, D. & Abraham, R. (2004). “ Landscape Dynamics and Conspicuous Consumpt ion” . Paper
presented at the 2004 Proceedings of the Society for Dynamic Games. URL:
http://www.vismath.org/research/landscapedyn/articles/Tucson4.pdf
[6] Gilbert , N. & Terna, P. (2000). “ How to build and use agent-based models in social science” . M ind &
Society 1 (1): 57-72. Springer.
10
[7] Graham, C. & Felton, A. (2006). " Inequality and Happiness: Insights from Lat in America" . Journal of
Economic Inequality 4: 107-22. Springer.
[8] Hofbauer, J. & Sigmund, K. "Evolut ionary Game Dynamics" . Bullet in of the American M athemat ical
Society 40 (4): 479-519. AM S.
[9] Kahneman, D. (2003). M aps of bounded rat ionality: psychology for behavioral economics. The
American Economic Review . 93(5). pp. 1449–1475. American Economic Associat ion.
[10] Kirman, A., Livet , P., Teschl, M . (2010). "Rat ionality and Emot ions" . Philosophical Transact ions of The
Royal Society B 365: 215-9. The Royal Society.
[11] M acy, M .W., & Willer, R. (2002). “ From Factors to Actors: Computat ional Sociology and Agent Based
M odeling. Annual Reviews Sociology 28: 143-66. Annual Reviews.
[12] Parravano, A., Rivera-Ramirez, H., & Cosenza, M . G. (2006). " Int racultural Diversity in a M odel of
Social Dynamics" . Physica A 379 (1): 241-9. Elsevier.
[13] Plutchik, R. (1980). Emot ion: A psycho-evolut ionary synthesis. Harper & Row.
[14] Situngkir, H. (2003). “ M oneyscape: a generic agent-based model of corrupt ion” . BFI Working Paper
Series WPD2003. Bandung Fe Inst itute.
[15] Situngkir, H. (2006). “ Advert ising in Duopoly M arket” . BFI Working Paper Series WPF2006. Bandung Fe
Inst itute.
[16] Simon, H. (1990). “ A mechanism for social select ion and successful alt ruism” . Science 250 (4988):
1665-8. AAAS.
[17] Smith, J. M . (1982) Evolut ion and the Theory of Games. Cambridge University Press
[18] Solomon, M ., Bamossy, G., Askegaard, S., & Hogg, M . K. (1999). Consumer Behavior: A European
Perspect ive 3rd ed. Prent ice Hall.
[19] Veblen, T. (1898). The Theory of the Leisure Class. M acmillan.