‘Arrow-Debreu model of general equilbrum. 1 1xrnoDUCTON;
TUTE MODEL EQUILIBRIUM IV PARETO OPTIMALITY, V WHAT
Ik isnot easy to separate the significance and influence of the
‘Arrow Debreu model of general equlibrum from that of
futhmtiea ccomonies Hass In an extaordinary series of
papers (Arrow, 1981; Debreu, 1951; Arrow-Debreu, 1954),
fw ofthe oldest and most important questions of neociassical
cconomics, the Vablily and elcency of the market ster,
were shown to be susceptible to analysis in a model completely
Taithul to the neoclassical” methodological premises of
individual rationality, market clearing, and rational expects
fons, through arguments at least as_clegant as any in
‘economic theory, using the two techniques (convexity and
fixed point theory) that are sil, aller thirty years, the most
important mathematical devices in mathematical cconomics.
Fileen years after its birth (eg. Arrow, 1969), the mode! was
still being reinterpreted to yield fresh economic insights, and
‘sven years later the same mode! was sill capable of vielding
fnew and fundamental mathematical properties (eg Debreu,
1970, 1974) When we consider thatthe same tWo men who
derived the most fundamental properties of the model (along
‘with McKenzie, 1984) also provided the ost significant eco-
fnomic interpretations i ho wonder that its invention has
helped earn foreach ofits creators, i diferent years, the Nobel
Prize for economics.
Tn the next fe pages I shall try to summarize the primitive
mathematical concepts, and theit economic interpretations,
that define the model, i give hint of the arguments used to
xablshthe model's conclusions Fall, on the theory that
‘mode is equally well desenbed by what it cannot expla, Uist
Several phenontena that the model isnot equipped to handle
Commodities and Arrow-Debreu Conomodiies (A.1) Let there
be L commodities, [=1,..,L. The amount of & commodity Is
described by real suber. A lst of quantities of all
commodities is given by a vector in Rt
‘The notion of commodity is the fundamental primitive
concept in economic theory, Each commodity is assumed to
Ihave an objective, quantifable, and universally agreed upon
(le. measurable)" destription. OT course, in Teality” this
eseiption is somewhat ambiguous (should two apples. of
iferent sizes be consdered two units ofthe same commodity,
‘or two diferent commodities?) but the essential quantitative
Sipect of commodity cannot be doubled. Production and
consumption are defined in terms. of transformations of
commodities that they cause. Conversely, the set_ of
‘commodities isthe minimum coletion of objects necessary to
‘describe production and consumption. Other objects, such a8
Financial asels, may be traded, but they are not commodities.
U6
‘General equilibsium theory is concerned withthe allocation of
commodities (between nations of individuals, across ime, oF
Cinder "uncertainty ete). The’ Arsow-Detwrew model etiiee
thos allocations Which can be achieved through the exchange
of commodities at one moment in time
Wis easy tose Ut is often naportant to Ue agent at
economy to have precise physical desriptions of eommodities,
5 for example when placing an order for a particular grade of
Steel or ofl The Jess erude the categonzation of commodities
‘becomes, the more scope there is for agen to Wade, and the
‘greater isthe set of imaginable allocations. Two agents may
tach have apples and oranges, There is no point in
‘tchanging one man's frit for the other man's fut, but both
‘ight be made beter off if one could exchange his apples for
the other's oranges. Of course there need not be any end fo the
‘stinctions which in principle could ‘be drawn between
‘commodities, but presumably finer details become les and less
important. When the descriptions are so precise that further
refinement cannot yield imaginable allocations which increase
‘he satisfaction of the ages in the economy, then the
commodities are called Artow-Debreu commodities.
‘A field is better allocated to one productive use than another
depending upon how much fin has fallen omit bu ts also
better allocated depending on how much rain has fallen on
other fields, Thit illustrate the apparenty paradoxical
Usefulness of including inthe deserption of an Arrow-Debres
commodity characteristics of the world, for example the
‘commodity’s geographic location, is temporal locaton (Hicks,
‘ate of nature (Arrow, 198%; Debre, 1989, Radner,
\ and. perhaps even thc mame of ils final consumer
(Arrow, 1969), which at fst glance do not seem intrinsically
‘connected with the object ill (but which are in principle
observable).
Ticks, perhaps anticipated by Fisher and Hayek, was the
first to Ruggest an elaborate notion of commodity this idea
has been developed by others, especially Arrow in connection
with uncertainty. Hicks was also the fist to understand
“parently complicated transactions, pechaps involving the
exchange of paper assets oF other noncommodities, over many
lime periods, in terms of commodity tade at ope moment in
time. Thus saving. or the lending of money, might be thought
fof as the purchase today of pariclar future dated
commodity. The second welfare theorem, which we shall
Shortly discuss, shows that an ‘optimal’ series of transactions
‘an always be so regarded. By making the distinction between
the same physical object depending. for example, on the rate
fof nature, the general equim theory of the supply and
‘demand of commodities at one moment in time can
incorporate the analysis of the optimal allocation of risk (@
concept which ar removed from the mundane
‘qualitesoffresh fut) with exactly the same apparatus weed to
Snalyse' the exchange of apples and oranges. Clasilying
physical objects according to ther location likewise allows
transportation costs to be handled in the same framework,
Distinguishing commodities by who ultmately consumes ther
could allow general equilibrium analyss to. systematically
‘nluae esternsities and pubic goods at special Css, hough
this has not been much pursed.
Tn realty, it i very rare to find a market for a
Arrow Debreu commodity. ‘The more finely the commodities
fare described, the less likely ae the commodity markets to
have many buyers and sellers (ce, to be competitive). More
commonly, many groupe of Arrow Debreu commodities are
traded together, in unbreakable bundles, at many moments in
time, in second bes’ transactions. Nevertheless, this under
Handing of the limitations of teal world markets, based on the‘Arrow-Debreu model of general equim
concept of the Arrow-Debreu commodity, is one ofthe mort
ppowerfut anya! tools of systematic accounting svaiabe To
the general equilibrium theorist. Similars, the. model of
‘Artow-Debret, with it idesiation of a separate market for
tach Arow-Debreu commodity, all simultaneously meting is
the benchmark ‘against which the eal economy can be
measured
Consumers. (Act) Let there be H consumers, = I, #.Q
‘Each consume Aven imagine consumption plans xe Re
lying in some consumption set 2°, (A.2) Xe a closed subet
in BE which is bounded ftom below. a
Each consumer also has well defined preferences > hovet
very pa (e y)EX"2X% where x py means x i a least as
Sesrable ay: Typiealy it ie asumed that (A.3) > ie a
‘complete, transitive, continous ordering, a
‘Notice that 1m geeralequiltrium consumers make choices
between entire consumption plans, not betweeen Individual
commodiis. A single commodity has sigaieance to the co
Sumer only in relation to the other commodities he has
consumed, or plans to consume. Together with transitivity and
‘completeness, this hypothesis about consumer preferences em
bodies the neolasial ideal of rational choice
‘Rationality has not alvays been a primitive hypothesis in
neoclassical economics, It was customary (eg. for Bentham,
Jevons, Menger, Walas) to regard satisfaction, or uty, asa
‘measurable primitive; rational choee, when it was thought tO
‘eeur at al, was the consequence ofthe maximization of uti.
‘Rad’ since ‘uty was often thought to. be instantaneously
‘produced, sequential consumer choke on the basis of sequential
Inctantancane ality maximiration wae sometimes expity
discussed a rational Gee eg Botum-Baverk on saving and the
reasons why the fae of interest is always positive)
‘Once utility is taken to be a funewon not of instantaneous
consumption, but ofthe elie consumption plan, thea rational
choice equivalent to utlty maximization. Debreu (1951)
proved that any preference ordering > defined on X¥ xX
Saisfes (A.)-(A3) if and only if there is 2 utlity function
8: Xt-oR soch that x Peay exactly when uh) Bu)
Under the influence of Pareto (1909), Hicks (1989) and
Samuelson (1947), neoclassical economics has come to take
‘aUonalty as prtilive, sal wilt animation ay x logical
consequence, This has had a profound effet on welfare eco
omic, and pechaps on the scope of economic theory as wel
Inthe ist pace, itutity is not dicey measurable, then it can
only be deduced from observable chores, asin the proof of
‘Debreu. But at best this wll give an ordinal” wilt, sine if
//R—R is any strc increasing function, then represents
24 fad only if vbw? represents, Hence there ean be
fo! meaning to inerpersonal wtlity comparisons the Ben-
thamite sum EE. us very different from the Bentham sum
Eff" Inthe second place, the ideal of rational choice
for preterence, feed from the need for measurement, is much
tore easly extended 1o domains not directly connected 10
platforms, of ‘social states. The elaboration ofthe nature of
{he primitive concepts of commodity and rational choice,
veloped av the basis of the theory Of masket equibiun
prepared the say for the methodological principles of neo.
lassiel economics (ational choice and equlibrium) 10 be
applied to questions far beyond those of the market.
"Although the rationality principles Is in some respects a
weakening of the hypothes® of measurable uility and instan:
‘ancous utility maximization, when coupled with the notion of
consumption pln it is also a strengthening ofthis hypothesis,
fand avery strong assumption indeed. For example there isnot
room in ths theory for the Freudian spt psyche (or sel.
Geception), of for Udyscus-ke changes of raft Pethaps mote
Importantly, a consumer's preferences fr example how thi
hts) donot change according tothe ol he play ia the proce
‘of production (eg on whether hei capitalist or landowner),
not do they change depending on other consumere preferences
forthe supply of commoditis. As an instance of this last case
ote that i follows from the rationality hypothesis that the
Surge inthe microcomputer industry influenced consumer
choice between typewriters and word processors only through
availability (va the price, and not through any learning ee.
(Consumers can ‘lear’ in the Arzow-Debrev mode, eg. thet
‘marginal rates of substitution can depend onthe sae of nature,
but the rate at which they etn is independent of production
for consumption it depende on the exogenous relation of
the state. We shall come back to this when we consider
information) It for no other reson, the burden of calculation
ad attention which rauona! choles: over consumption plane
imposes onthe individual sso large that one expects rationality
to give way to some Kind of bounded rationality in some future
‘general equilbrium models
“Two more assumptions on preferences made inthe model of
‘Acrow-Dsbreu are nonsatiation and convety
(For cach x0 tre ina ye wh > eh
that y pax and not 2 >a
(AS) Fe a eonven set td ys conten, Leg fy ax and
Dee et then ly +0 —Oxlmx a
‘The nonsatition hypothesis seems entiely in accordance
with human nature, ‘The convexity hypothe implies that
‘cammadiiee are infitely diiibe, and that mintuee are at
least a good as extremes. When commodities ae distinguished
‘very finely aceording to date, so that they must be thought of
438 flows, then the convexity hypothesis is untenable. In a
Standard example, a man may be indieret between drinking
4 glass of gin or of seotch at a particular moment, but he would
‘bemuch Worse offi he had to drink a pane of half gin-halt
‘otch On the other hand if the commadties were not so finely
ated. then they would be more analogous to stocks, and @
consumer ght well be better off with aie of gin and a lite
‘of scotch, than two lites of ether one. In any case, s we shal
ema lteycvesy agent ual sia tte het (FE
if there are many agens) then the onconvenitis in preferences
ace relatively unimportant
‘each agent Mis also characterized by a vector of tial
‘endowments
(AQ MeXMERE forall het. A o
‘The endowment vector e* represents the claims thatthe com-
sumer has on all commodities, not necessarily commodtie in
his physical possession. The fat that ee X* means that the
‘consumer eat ensure his own survival even i hei deprived of
Allopportunty to ade. This 4 somewhat strange hypothesis
for the modern world, in which individuals often have labour
buat few other endearment =e tae Dubin the hypthese
could be relaxed; in any cafe survival isnot amiss that i
Addressed in the Arrow-Debreu model,
{Each individual slao endowed wih an owaciship sha
of each of the firms j= os J
(A) For all he tesco Hey besocedh dy 20, and for al
Fe deced Bvdy =i o
Firms. (A.8) Let there be J fms, j= by. a
‘The firm in Arrow-Debrew is characterized by ss initial
seibution of owners, and by ils technological capacityArrow Debreu model of general eulibium
¥,cR! to transform commodities. Any production plan
yew, where negative components of y tter to inputs and
positive components denote outputs, is Teasible for firm j if
Ye, A-customary asumplion made inthe Arrow-Drbert
‘model is ree disposal: f= 1,» L any commodity, and r,
i the unt vector in RE, with one in the [th coordinate and 20
chew, then
(A9) Fer all Land k>0, —kye¥,, for some
Poked o
Ahough sts strange, when thinking of nuclear waste ete,
to think that any commesdy ca be disposed without cot (fe
‘without the use of any other inputs), a5 we shall remark lates
this assumption can be faxed, if megative prices are introduced
{ort weak monotonicy t aseimed).
"The empirically most vulnerable assumption to the Arrow
Debrew model, and one erucsl to ts lop, I
(A.10) For each j,¥, i closed, convex set conaining 0.
‘This convexty assumption rules out indvsiblsis in pro
duction (half a tunnel) increasing returns to scale, guns
from specialization, el. AS with consumption, ifthe indivis:
Toes of production are small elativ to the ize of the whole
‘economy, then the conclusions we shal shortly pesent are not
‘uch affected. But when they are large. or when there are
Significant increasing return to sae, the model of competitive
‘equilibrium that we are about to examine simply mot appic=
thie. Neuerheets, convent consistent withthe traitionally
important eases of decreasing and constant returns to scale in
production
We eoncuue by procting Unee Gal assumptions wed in
the Arrow-Debreu mode
(AND Leto = Ets
BEES (pel zp ay ye Pye hod
ti Falperyresiyand 7 i
Beer (pe ppetece sm Mm ye, yeF
Then FAB. #9 and Ks compact
Assumption (Al) requires thatthe leel of productive activity
‘hats possible even if the productive sector appropriates all the
resources of the consuming sectors bounded (ts well ascosed)
‘Notice that these asumpuons are conssent with ems
‘owning initia resources, as well as indwiduals In the orginal
[Arrow Piven mde (1084), the firme wnte pried fre
‘owning initial resources (they were assigned tothe firm owners,
With complete markets there wit ference, but with incom
plete markets the eater assumption is restrictive)
(A.12) The economy is reducible a
We sall not elaborate thi assumption hee. It means that for
any twoagents hand the endowment eof agent his postive
in some commodity . which (Caking into account the poss
Dities of production) agent ’ could use to make hinself
Strely better of I certainly seems reasonable that each agent's
Inbour poser could be asd to wake abuthes agent lite Uf
Tasty, we assume thi
(4.13) The commodities are not dutinguished ssoording to
Which firm produces them, or who consumes them C)
Assumption (A.13) is made simply for the purposes of
interpretation. When put together withthe defition of com-
petitive equilibrium, it plies that there are no externalities fo
Production or consumption, no public goods, ete. Matbe-
‘atcally, however, (A.13) has no content. tm other words if we
‘dropped assumption (A.13), the Arrow-Debreu notion of com
petitive equiloum would sil make sense (even im the presence
‘of extraltes and public goods) and it would still have the
‘optimality properties we Shall elaborate in Section ML, but it
would require an entirely diferent interpetation. Consumers,
Tar example, would be charged diforem price forthe tame
physical commodities Game, that, according, to dae, ocation
fand state of nature). In more technical language, a Lindahl
quilbria i special cae of an (A 1)-(A.12) Artow-Debrew
quit, with the commodity space suitably expanded and
interpreted, Thus each physial unt of «publi good is replaced
by Hf goods, one unit forthe public good indexed by which
agent consume it Also the physical technology set describing
the production of the public good is replaced by a diferent set,
in the Arrow-Debreu model, lying in higher dimensional
syace, where the output of the sime amount of ff goods. In a8
‘Aviom Debreu cqulltriam, consumers wl kely pay aierent
Drees for these H goods, Le for what in reality represents the
same physical public good. Hence the differential pay principle
for the optimal provision of public goods ‘elvtdated by
Samuelson, which appeared to point toa qualitative difference
between the analytical apparatus mended to describe optimality
in public goods and private goods economies, is thus shown (0
beveaplicable by exactly the same apparatus wed for private
goods economies, simply by multiplying the number of conn
‘dies. The same device ca also be used for analysing the
‘optimal provision of goods when there are externalities, pro:
‘ded that nopaive prices ae allowed Assumption (A 13) tus
Seriously limts the normative conclusions that ea be drawn
From the met From a dresnptive unt af oem, however
rationality and the price taking behatiout which eqoilibriam
implies, make (A.13)nevessry
ice ithe Final primitive concept m the Arrow-Debreu mode!
Like commodity it is quanifabie and directly measurable. AS
‘Bebreu has remarked, the fundamestal role which mathematics
plays in economists party owing tothe quansiable nature of
these two primitive concepts, and to the rch mathematical
‘eationship of dual vector spuces, sto Which 1 satura (0
lassi the collections of price values and commodily quan
tives Properly speaking, pice only sensible (and measurable)
ltsarelaionhipbetmeen two commodiin sc. a0Flatie pie
Hence there should be LL teatve pres an the
‘Arrow-Debrev model. But the definivon of Arrow-Debreu
‘equuibcium immediately mptes tat suthees to ge L~ Tot
these ratios, and all the ees are determined
For mathematical convenience (namely to teat prices and
‘quantities as dual vectors), one pre specified for each unt
‘uaniy of each commodity The relauve prie of two com
fhodities can be obiained by taking the rato of the Arrow
Debreu prices of these commodities.
T shal proceed by specifying the defininon of Arrow-Debreu
‘equilibrium, and thes I make a umber of remarks emphasizing
Some of the salient characterises ofthe dfition. The longest
remark concerns the dillerences between the historical devel.
‘pment of general equilibrium, up unt the time of Mieks and
Simueson and the particular Arrow Debreu model of general
"An Arfow-Debreu economy E isan array E = (L, H,J(X,
a OD) kale Hyg J). sting
assumptions (A.1)-(A.18), An Arrow-Debrew equilibrium san
aay (A). (81), (Ft Leh elec HJ leone)
stieying
For all j= 1... S540
ween ponArrow Debre model of general equilibrium
For all a1... H.S€B%B), where BY)
sfreniacedaad
ohn
Jo
and if x €84(9), then not x>43%,
Fora f= tanks Eat
‘The most sting feature of general equilbrium is the
justapositon of the great diversity im goals and resources it
flows. tozether with the supreme coordination it requires.
Every desire of each consumer, no fatter how whistal, 5
met precisely by the voluntary supply of some producer. And
thio tr forall markats and consumers tmulteneoucly.
“There isa symmetry 1 the general equilibrium mode, im the
way that all agents enter the model ndvidually motwvated by
Seltintrest (aot as members of cstinet clases motivated by
class interest), and simultaneously, 0 that no agent acts por
to any other on a given markel (eg. by seine price). IT
workers” subsistence were not assumed, for example, that
‘would break the symmetry: workers income could have to be
fuaranteed fist, otherwise demand would (discontinuous)
Sollase. As 5, atthe aggregate level, supply and demand
Squall and simuttancousy determine. pices in equilibrium,
botn the consumers mana! rates subsulucon and the
producers’ marginal rates of transformation are equal t0
Felative pees (assuming. diferentiabilty and. inteority).
‘There are gains fo trade both through exchange and through
production This point of view represent 4 sgnifiant break
Sith the clascal tradition of Ricardo and Marx. We shall
Some (the main diffrence between the classical and
ncoeasical approaches shortly. Another diference is that
here need not be fixed coefcenls of production in the
Mrow-Debreu model = the sets Y are much more general
Aso in an Arrow-Debreu equlibrium, there is no reson for
here 10 be a uniform rate of prof. Tere is none the less one
spect of the model which these authors would have greatly
"Pproved, namely the shar d¥ which allow the owners of
‘rss to collet profs even though they have contributed
vothng to production.
TNouke tat Im generat equilbrum each agent need only
concer himsalf with his ova goal (preferences opis) and
hie pees. The implicit assumpion that every agent "knows!
Ihe peices s highly non-trivial. Ie means that at cach date
scl agent capable of forecasting perfectly all Ture prices
‘0 the end. of time It yim thi sense that the Arow
Debreu model depends on “rational expectations. Each agent
ust also be informed of the “price g of eath firm J, where
‘fa fs}. Fis that produce under constant returns (o
le must also discover the level of production, whick cannot
< deduced from the pries alone.) Assuming that the “man
1s the spot’ (Hayek's expesson) Knows much better than
ryone ise What he wants, of best how his changing
vironment is suited to prodorng his prodvet,decenteaied
Seison making would seem to be highly desirable, si not
‘eompatible with coordination. Indeed, harmony through
versity one ofthe sacred doctrines ofthe liberal tradition.
“The greatest triumph ofthe Aerow-Debreu model was to lay
ut explicitly the conditions (roughly (A.I)-(A.18) under
hich if is possible to claim that a properly chosen price
stem mst always exist that, lke the invisible hand can
lade diverse and independent agents to make mutually
compauble choices. The idea of general equilibrium had
bradually developed since the time of Adare Smithy mowly
throug the proneering work of Walras (1874), Von Neumann
(0937), Wald (1982), Hicks (1939) and Samuelson (1917) BY
hele 19905 he definion "of equllbeum, including
‘overs shares n the rms, was wellesablshed, Buti was
‘Arrow-Debreu (1954) that spelled ou precise mcroeconomic
sssumplions a the level of the individual agents tht could be
‘sed 10 show the model was consistent,
"The axiomatic and rigorous approach that characterized the
formulation of general equilrivm by Arrow Debreu has been
enormously influential. It is now taken for granted that a
‘models not propery defined unless thas been proved 0 be
Jogiclly consistent. Much ofthe clamour for microsconomic
foundations to macroeconomics’. for example. i a desire to
see an axiomatic clanty sila to that of the Atrow-Debcea
model applied to other areas of sconomies. OF course there
‘Riomane and ngorous; one thinks especially of Von
‘Neumana-Morgenste's Theory of Gamer (1944) But game
theory was, at the time, on the periphery of economics,
Competitive oquiivium is at its heart
‘The central mathematical technigues. convexity. theory
(separating hyperplane theorem) and Brouver’s (Kakutan's)
fixed point theorem. used in Arcow-Debreu are, thirty years
later ‘sll the most importa: tools used sn mathematical
economies. Both elements had played a (hidden) role in Von
‘Neumans’s work. Convesity had Been prominent in the work
or Koopmans (1951) of acti analyse inthe work of Kuhn
fand Tucker (1951) 08 optimieation, and in the papers. of
‘Arrow (1951) and Debrew (1981) on optimally Fed pont
‘Georems fad been used by Von Neumann (1937), by Nash
(1950) and especially by MeKenzie (1954), who one month
tarler than Atrow-Debrey had published a proof of general
fquilbrium using Kakutan' theorem, albeit ina model where
the primitive assumptions were made on demand Tunctons,
rather than preferences McKenzie (1989) also made a early
ontnbution to the notion of an stedieble economy
(assumption (4.9)
“The frst fruit ofthe more presse formulation of equilibrium
that began to emerge in the early 1930s was the transparent
Semonsiration af the Fret snd second welfare theorems that
‘Arrow and Debreu simultaneously gave wp 1981. Particularly
noteworlby is the proof that every equilbrium is Pareto
‘pal $0 snple ai lating ds demons tat
i's no exaggeration to cll jt the most frequently stated
Argument inal f neccasical economic theory.
‘Among the confusions Ui were eared away by the careful
suiomave treatment of equim was the fekance of the
Fest 2, omcing ay.
‘Re poo atte second well here, on the ter band
des sje convey of ie pretence td positon st
ttioughnct tee foetal norte sry o he
‘Gd allcnion 8) Eventi pnd on Minkowshs
theorem, which asserts that between any two disjoint convex
see in RY there must be 4 separating hyperplane
Tn this connection let us mention one more cemarkable
mathemati property of the ArrowPohren model. Let us
fppos tht all production takes place under constant returns
forsale: if ye Y, then 50 i 4), for 10. We say that a
{asile allocation & for the economy E sin the core if there
i no coalition of consumers S'= (Ty... Af} such that using
‘only their iil endowments of reources, a8 well as access
to all the production technologies, they cannot achieve an
llloeation for themselves whch they al prefer to &, The core
is meant to reflect those allocations which could be rsitained
‘when Bargaining (he formation of coalitions) is costes. Tn
4 satus quo core allocation, any labour union or cartel of
‘Somers dt tnewtens to miield a goods from the market
Knows that another coalition could form and by withholding
its goods, prevent some members of the original coalition from
being better off than they wore under the satus quo, Is 35)
to see that any competiine equilibrium in the core
Debreu-Seart (1963) building on eatlier work of Scart
showed 6y using the separating hyperplane theorem, that it
‘gents are small relative tothe market n the sense they made
precise trough the notion of replication, then the core consists
‘only of competitive alleations. Such @ theorem can also be
Dproved even af there are small sonconvexiies tn preferences
(oe Aumann (1964) fora diferent formulatuon of tbe stall
agen’.
[EMSTENCE OF EQUILIANIOM. Suppose that agents’ preferences
and fms production ses are srily conven, and that agents
fice peter mace of any commadiy to lest (eet mono:
tonicity) and that they all have sritly positive endowments,
Leta be the st of Lepr vectors all nonnegative, summing
tone, Let /'(p) be the commodiy bude mos prefered by
grat, given the sctly pose pies pe By +. Similarly
(2) be'the profit maximizing choice of fm. give prices
PEA, .- Fil, et /(9)= 20. 4p) ~¥, 5p) ~ 6. is
{any do show that fis 3 continous function’at all pe A,
‘price Fed, «is an Artow-Debreu equilibria prige if and
only 78) 3.
Th pera there is no reason o expec a continuous fanction
toltaye evo. Tas Weld vould grove ely wir great diielty
ina special case that an equllbium necessary exists Now
observe thatthe funetion must sais Wala Law, p'f(p) = 0,
foe al p80 fs na abivary.
‘Consider the convex, compact se A, of prees ped with
1 3e>0, forall Consider also the continuous enction
408, mapping p to the elses point fi & to fp) +p. By
Brouwer's fxed point theorem, there must be some p with
(9) =. From strict monotony 1 follows that feanmot be
fn the Boundary of 4, if is chosen suficently small. From
‘Walras Law it follows tha iff iin the interior of 4, then
{(P)=0. The demonstration of the existence of equilibrium
by Arrow and Debrey, as modified later by Debreu (1959),
fallow a sila ok
"Note the esenta ole of convent in two parts of the above
poo. It was used with respect to agents characteristics (0
uarante that thir optozing behaviour is continuous. And
{was abo used to ensue thatthe space has the fixed point
property. Smale (1976) has given a pathfollowing proof (ce-
{ated to Scars (1973) algorthn) tat on closet mspestion does
not require convexity ofthe pce space. (Derker (1974) and
Bulasto (1986) have given homotopy poof) This not only
of computational fmportanse. appears that there may be
‘conomie. problems, dealing with general equilibrium with
incomplete markets, im which the price space is intnsclly‘Arrow-Debreu model of general equlibriam
ronconvex, and in which the existence of equibium can only
beiproved using pathfalowing methods (ee Dull Shaler,
1385).
‘To weaken the assumption of strict convexity, in the above
proof, one can replace Brouwer’ fixed point theorem with
Kalulant’s. An important conceptual point aes in connection
with set montoniety. Ithat is dropped, and the production
sts do not have fee disposal, then in order to gurantee the
existence of equlbrium, the’ definition must be revised to
require either (9) =0, oP) <0 and f= 0. There may be
free goods, like air, in excess supply. One cannot drop meno
tonicity and fee disposal without allowing for negative prices.
‘Final, t can be shown that if ter are small nonconvexities
in sither preference or production, and ial the agets are small
ftlative tothe. market tether in the replication sense of
Debreu-Seart, or the measure fo sense of Aumann), then
‘thre wil be prices at which the markets nearly clear. On the
other hand, inerensing reluray to sale over a broad range
“efntely incompatible with equilbrum
LOCAL UNIQUENESS AND COMPARATIVE STATICS. Another prop-
centy of the excess demand function f(p) is that iis homog-
neous of degree zero. 50 instead of taking ped, lets ft
‘i= L Sinilaey. fet Pip) be the LI vector af exces demands
for goods (=2.....£ If Fip)=0, then by Walras’ Law,
So)
Suppose furthermore that agent characteristics are smooth
‘Then F(P) is diflerentable funtion ID, F(A) bas fll rank
av'an equilbriam then ps localy Ungue- Moreover te
‘equilibrium p will ove continuous, given continuous, small
changes inthe agents characteristics, such as their endowments
TED, F(p) has ful rank at all equilibria then thee are on)
4 Ftd number of equiibria. Debreu (1970) called an economy,
EE regular i, Fp) bas fll ank at all oquiiboium P of
‘The problem of trying to give sulicient conditions on prefer
‘noes ete. fo guarantee tit D, Phas ful rank in equilibrium
hhas proved intractable (except for restrictive, special cases). But
‘Debreu (1970) solved the problem in classic tye, appealing to
the transversality theorem of differential topology (ar Sara's
theorem), to thow that Hf one were content with epulriy for
most aif economies, then the problem is simple ‘He proved
that for almoet all economies Dj hat fall rank at ery
‘quitrium Hence, in almost all eednomies comparative statics
(Ghe ckange in equilibrium, given exogenous changes to the
economy) i Well dened
‘Observe that excess demand F depends on the agents
characterises, ineluding their endowments, 0 we could write
Flesp). Now the transversality theorem says that (given some
technical conditions) i D, Ft, f) has full ank a all equilibria
FB forthe economy Ele) with endowments efor al then for
‘almost all ¢, D,F(e,p) has full rank at all equltsium p of
Ee) But its eaby to show that D, Fe, p) always has fll rank
‘Along siniar lines, Debreu proved the "generic regularity" of
ui,
“There is one unfortunate side fo this comparative sates
stor) One wou ke to show not only tha comparative statis
are well defined, but also that they have a definite form. In a
oncave progeamiing problem, for stamps, a small increase
finan input results in'a decrease in that inpat's shadow pace,
and an reat n output approximately equa tothe siz of he
‘put inrease multiphed By its orginal shadow price. Given
the strong rationality hypothesis ofthe Arrow-Debreu model
fone would hope for some sot of analogous result. Following
‘conjecture of Sonnenschein, Debreu proved in 1974 tha gives
any function f(p) on 4, satisfying Wares” Law, he could find
fn’ Artow-Debreu economy such that /ip) its aggregate
‘excess demand on 4. Thus assumption (A.1)-(A.13) do not
petit any a prior predictions about the changes tat must
‘Sccurin equilibrium given exogenous changes tothe economy.
‘An increase inthe aggregate endowment of a pariulat good,
for example, might cause ts equiltium price to rie, The
Possibility of such pathologies is disappointing. It means that
fo make even qualitative predictions, the economist needs
‘denied data on the excess demande F
. WAT TIE MODEL DOESNT EXPLA
We have already discussed the implications of the notion of,
Arron Debreu commodities and the second welfare theorem
Tor insurance, namely tht since every Pareto optimal allocation
is supportable as an Arrow-Debyeu equilibrium, every optimal
allocation. of sk. beating canbe accomplished) by the
production and trade of Arrow-Debreu commodiis, Le
without recourse 10 additional kinds of insurance markets
Specializing in sks. Every Arrow-Debreu commodity is as
‘much a diets in location, oF tims, or physical quality a it
[s for risk. This leads toa great simplification and economy of
analysis, Buti also means, that from the positive point of
view, the Atrow-Debreu economy cannot diel provide an
danaljis of insurance markets (except asa benchmark case) In
this section T shall uy to point out a few of the ober
phenomena which recede into. the background inthe
Kreow'Debrev model but "whieh would emerge if the
assumption of a nite, but complete set of Arrow-Debres
tind ennsimers wae dened
four cursenly active lines of research which
attempe £9 come to grips in general equilbrium framework
With some of thee phenomena, while preserigg the
Fundamental neoclassical Arrow-Debrew principles of agent
‘optimization, market clearing, and rational expectations, that |
think are paticularly worthy of attention. They are the theory
‘of general equilibrium with incomplete asset markets which
fan be traced back to Arrow’s (1953) seminal paper on
‘Scconties: overlapping generations economies, whose study
inated by Samuelson (1988) sn his clasic consumption
Iban model, the Cournot theory of market eathange mith fem
leaders, first. adapted "to general equilibrium by
Shapley-Shubik (1977), and the model of rational expectations
‘equlibram, proneeres by Lieas (1972)
Let us note Bist of al that in Arrow-Debreu equilbcum
ther ismo trade in shares of firms. A stock erties s not an
‘Avrow-Debreu commodity, for its possession entiles the
‘owner to additional commodities which he need not obtain
through exchange. Note alo that ia Arrow-Debres equilib:
sium, the hypothesis that all prices will remain the same, m0
‘matter how an individual frm changes it production pan,
fzuarantecs that firm overs unanimously agree on the! frm
bjetive, to maximize profit. If there were a market for frm
es there would not be any trade anyway. Since ownership
Of the frm and the income necessary to purchase it would be
perfect substitutes. Tn an incomplete markets equiium,
Sitorentsourees of revenve. are not nesesarly potest
fubsittes, There could be ative trade of the stock market,
Of course, such a model would have to specify the firm
objectives, since one would aot expet unanimuty. The theory
ff stock market equilibrium is stl in its infancy, although
Some important work has already ben done. (See Deze, 1974,
nd Grossmas-Hart, 1999)
Banksuptcy isnot allowed in an Artow-Debrew equilibrium,
‘That follows from the fact thit all agents must meet theit
2Arrow Debrea model of general equlibium
budget constraints. Ina game theoretic formulation of
Seofbvium Goch as 1a Satu shony) ie ached
frposing an infinite bankrupty pena. Since every
Kivu Ewes equsiom i Prete optima thee would be
0 bene in reducing the bankeupc) penalty t0 the point
Where someone might choose to" g9 bankrupt But with
Tncomplete mast sich a poicy might be Pareto improving.
ten allowing for the deadweight lom of imposing, te
enti
Money docs ot appear in the Arrow-Debres model. Of
cours, al ofthe reasons for ts ea fe eine: transactions
and, precautionary demand, store of value, unt. of
Sounty ete are already taken ear of the Arrow Deb
‘model One could imagine money inthe model at data zero
Sexy agent could bouton tnney fun the cea bake At
Sry date afterwards he would be required 10 france is
purchases out of his stock of money adding to that sack fom
ais: atthe lant date ie would berequed fortum othe
tunk exactly what be. borrowed (or le face an ininite
Tuankraptey pent). In such a model the Arow-Debrea
prices would appeat as money prices, The absolute lve of
‘money prices and the aggregate Amount of borrowing would
tot be determined, but the alecations of conmodites would
‘the sme atin Arrow Debres. Tere sno point in aking
the tole of money exp in the Arrow Debret mode since
‘is no eft on te real allocations, However, one cond
the same mode! wth incomplete ase! market, the presence of
tanlcly financial secures can be of eeatsnicance tothe
‘al allocasons
Ta the Arcow-Debrew model, all ade takes place at the
tegoning of tins market more opened st ater dates for
{he mame Arrow-Debros commodiicn, thm no. additonal
trade would take place anyway. Al the other exteme, one
‘ght consider 8 model in which at every date and ats of
‘ature ony those Avrow-Debreucommoditis could be traded
tic were indeed bythe corresponding (dt, stat) pir An
Tntemediate case woud so permit the trade of ome (but aot
Al) difereniy indexed Arrow-Dbreu commodities. Now the
‘Arto Detrew pools of the existence and Pareto opty
ff equilibrium do not apply to such an incomplete markets
tconoms, as Hart (1973) fs ponte xt. We have already
Soted the exitenoe problem. ‘ar for efowncy. tne Pare
plialiy of Atrow-Debreu equa might suggest the
fresmpton that, though there might be los 0 hminating
Irarkets, rade om the remaining markets would be as een
35 possible. In fact, it can be shown (geserialy) that
sgulibeium tages do aot make effeent use of the existing
‘market
"The Arrow-Debreu. model of general equilnium is
relentlessly neoclassical n fact thas Become the paradigm of
the neoclassical approach, This stems in part from its
individualistic hypothesis, and its elebrated conclusions about
the potential efbeacy of unencumbered markets. (Although
‘Avro, for example, has always mainained that a proper
Showing how ‘neffcint is the limited teal world. marke,
‘stem,) But stil more telling is the fact thatthe assumption of
2 lte number of commodities (and hence of dates) forces
‘pon the mode! the interpretation of the economic process as
one-way activity of converting given primary resources into
final consumption goods IF thee is universal agreement about
when the world wil end, there cam be no question about the
Teproduction ofthe captl stock. In equity i wil be ru8
own to zo. Similarly when the world has 2 definite
‘beginning, so thatthe fst market transaction takes place after
the owneship ofall esoutees and techniques of production,
iD
and the preferences of all individuals have been determined,
fone cannot study’ the evaluon of the social norms of
Consumption in tems of the historical development of the
‘lations of prediction One certainly cannot speak about the
production ofall commodities by commodities Grafs, 1960)
{since at date zero there must be commodities which have not
‘been produced by commodities, i. by piyial objects which
are trade).
‘Te seems natural to suppose that as L becomes very large, $0
thatthe end ofthe world's put of unt the distant ature, that
‘this event cannot be of much signifeance to behaviour now.
But let us not forget the rationality imposed onthe agents. Fs
‘off as the end of the world might be. te perfcly taken into
account. Thus, for example, socal scusty (unde as ts in
the US by tases vo the yousy) could wot exit if satonal
agents agreed on «nal stopping tine to transactions.
Consider a model satsying all the assumptions
(ADMALTS), except that Land H ae allowed 10 be init,
such as the overlapping generations model. Tecan be shown
{hat there is robust collection of economies hich have 4
continuum of equilibria, most of which are Pareto sub
optimal, which difer enormously in ime O behaviour. Thus in
mode! where time dacs act have a definite end, the
‘optimality and comparative sates properties of equilibria are
radically diferent. (For example, there may be a continuum of
guile, imdeted by the lve! of penod 0 real wages
(versely elated tothe rate of profit) ofthe eve! of output oF
employment. The interested reader can eonult the entry on
OVERLAPPING GENERATIONS MODELS "A systematic. study of
‘conomies where ony Lis allowed to be infinite was begun by
Bowley (1972). Such economist tnd to ve propertia svat
1 thove of Arrow-Debreu)
‘There is no place in. the Arrow-Debreu model for
asymmetric information. The second welfare theorem, for
‘example, eles on limp sum redistribution, is edstibutons
that occur in advance ofthe market interactions, But if agents
fanaot be" distinguished. except through their market,
behaviour, then the redistribution must be a function of
market beaviour Rational agents, antiipating this wl
stort their behaviour and the optimality of the redtbution
wl be fost
Similay nthe definition of equiirium no agent takes into
account what other agents know. for example about the state
ff nature. Thus it quite possible in an Arcow-Debvea
‘ulibram Tor some ignorant agents to exchange valuable
commodities for commodites indexed by sates that other
Sgeats know will not occur. Ths problem received enormous
tention in the finance erature, and “some cit (ee
Grossman 1981) that i has been solved by extending dl
Arrow-Debreu.defiition of equilium to 2 rational
‘expectations equim” (Lucas, 1972; se also Radner, 1979)
Bur tis definition i self suspect; in particular, it may not be
smplementable
‘Eten if ational expectations equilibrium (REE) were
{o ghips with the most fundamental problems of asymmetne
intormation. For ike Arrow-Debreu equiltium, in REE all
trade is conducted anonymously through the market at given
Prices. Implicit in this definition isthe assumption of large
hhumbers of traders on both sides of every market. But what
has come to be called the incentive problem in economies
revolves around individual or frm spec uncertainty, ie
trade in commodities indexed by the names of the traders,
which by defniuon involves few traders.
"This Brings ve to another major riddle: how are agents
supposed to get to equilibrium i the Arrow-Debreu mode?Arrow-Debreu model of general equa
“The pioneers of general equibrium never imagined thatthe
sconomy ses neveear in equilibrium: Walrar, for example,
Proposed "an "expliitttomnement procedure which he
onjectured converged to equibrum. But that idea is awed
In two respets: in general, It can be shown not to converEe,
find more importaaly, ii an imaginary process which no
fachange is permited_ until equilibrium is reached. This,
itustrates a grave shortcoming of any equilibrium theory,
ramely that it cannot bepin to specify outtomes out of
quilbvium. The major ers of labour market clearing inthe
19805, and again recently, argues strongly that thre are limits
to the applicabilty of equilibrium analysis.
‘One is led naturally 10 consider market games, in which the
‘outcomes are well-specifed even when apents do not make
their equilibrium moves The most fam market game f
‘Courno’sduopaly ode, which has been extended to general
equilium by Shapley-Shubik (1977). When there are large
fnumber of agents of each type, the Nash equilbna of the
Shapley-Shubik game give ney igetical allocations to the
competitive allocations of Artow-Debreu. This jusiies (0
fist. approximation) the price taking behaviour of the
Acrow-Debreu agents. But note thatthe informational
equiements of Nash equilibrium are at last twice that of
‘Arrow-Debreu competiive equiltrium (each agent must
‘know the aggregates of bids and ofers on each marke). ts
abo extremely interesting. that trade takes place inthe
‘Shapley-Shubik game even if there is only one trader on each
side of the market. Hence many problems in asymmetic
‘nformaton wi have wo pave i ie Atsow-Debreu del,
frecause they involve 100. fine a specifeation of the
commodities to be consistent with price taking, might be
Sone in market game context. Finally, can be shown
that REE isnot consistent with the Shapley-Shubsk game, oF
sndeed wth any continuous game.
‘We have indicated some ofthe ways in which itis possible vo
extend general equiltrium analysis to phenomena outside the
Seope of the Arrow-Debrew model, while at the same time
preserving the neoclassical methodological premises of agent
Optimzation, rational expectations, and equilibrium. Tt is
Important t note that thee variations have extended the