I

EFFICIENCY

IS t'rtIC1l'llCY, I;, it simply Olll' vahlt' ;JIllung othcrs---ollt' that t'umo

Illists but not the rest of LIS find vt'ry attra<:(ive-or it is sOl1lehow a
tl!1ldalllenral idt',1 that we callnot do without, \V hleh we tllll!ilt not to
thillgs~
sacrificc flJr otht'r

Efficiency and Rational Individual Choice

reflcct Oil wll,lt wc know ,Ibout ollr ratiolul "ecollolllic 111;111."
Lllion:ll ,'hooser UOIIIO I:(ollolllims lLIS a well-[(JrlllCd utility
fimctioll s,ltisfying the requirelllents oj" urility 1henr\' we c\alllillcd in
SectioJl'i 2,2 alld 2"t His nrdlTellccs ;11',' ,'haraetl'ri:n'd
dellland Cllr\,(.'S
sI11)';titution of goods ill Section
J 1"1110 1:'coIIlJlllio/S Ius ;1 prcrl'rl'lll'e ttll' pin;!.
How Illuch pilla is it r.ltiOlul fur /-101110 h'ollOlllims to COllSllllll'~
Suppose tint the cost oj" pin,l
comtlllt ,It II(): we Illmt rCIllclllhn
that
"cost" wc 111,':111 the toLl I "P/IOlll/llil), ({lsi.' of C()Jl';Ulllil1g
tl)]'golle opportlillitil'" to satist~, othl'r prctl-rcllces
,,), poillt ,)). It is 1I11porLmt to stress that "cost" does !lot
Ill'cl'ss,lrilv mcall a mOlletary p;IYllIellt, or sOllll,thing that VOli don 'I
ill '\1 emt oftaklllg this (,(Hirst' IS that I have to take tc,t,"). III
the l'CllI10111ist's sellse, the "nl')t" ofg"ttillg your rJ1',t choice of;1
is th:ll you h:ld to !()rgo your ,ccontl choice of ;1 bo:\ of ,'hickell
whcl1 you h,1Vl' to choose hetweell good things, the cost of
your (kcision i, th,!t
thillg you didn't cIH)osl'. This cm be
in knns of 1101110 1:'(oll(lllli(JIs\ [llrgolll' utility--call this
It i, the utility you would h,lve received ti'OI11 vour scCtmd
elwin:, No\v consider Air, decisloll to ('(mSUllll' the first slice of
It satisfies hi, prd(:rl'lHc [llr pina:
his utility ft!llctioll
We call represellt tim
SOIlIC Lltilitv bClldit
cdl it
(th,H is, if the utilitv bendits ar~' grl':Her thall, or
thell it \vill he ratioll:d tll]' hilll to purcha,~' the slice of
pizza. But ,hould ht' hill' oul\' Olll' slice) Well, we kllow that since the
cost of nizz;! is comLlIlt per LlIlit, the CO'it of the second pil'cc will he
But ht'CHISC of decreasing marginal utility, the bcnefit<; of the
s<:coml piece will be Ic" th.111 twill' 1/(/: c.111 this I/(il) + 1/1(/1) III
/I IS ;1
Illlillber less th;]]1 11(1))), The cruei,1i idc;l
hert' is that \vh,ltevcr the lItilitv benefits
the tlN pien' I/I(h) I. the
is hertel'
of the second
will be pO';itivl'
Let

Efficiency

OVERVIEW

II
!

e h:Lvc tlllLs !:ll' t,xllsed OIl individual Lltionalitv: wh,!t is it to

Ill' a ratiollal actor, and how call we flJrlllally ;llodcl such :11l
;Ictor' This eh;lptcr begills with that COllClTll, but thcll turm to all
how r;lliollal actors illter.lct-~Ihl' subject of th(' ITIIl,lindcr
of thl' book. Thl' lliaill COIlCl'f1l of this chapter is to explore the
relatioll betwce]] rational actio]] :lIld the idea of ctrlcicllcy: the
starts With eHlciellcy of the COIISUllIptioll dccisions of olle perSOIl, and
thell IlHlV~'S to the idea of all dh,'jcnt cxchange: all intl'l';lctioll
llt'twl'cll two ('collolllil';dly ratilln;ll ,Igt'nts. The illiporLllIt ide,1:> of
l)dYl'to slIj1criorily ;llld Pareto optill/alit)' an' introduced. Thl' secolld p.lrt
Df till' chapter briefly skl'tches \vl,\I-knllwll Elilurcs of
v:lriollS Ilotions of "c:\tlTl1;i1itics,"

3.1 RATIONALITY AND EFFICIENCY

Everyone' knows that cconomics is about diicicllcy, and most of us
have SOlllC pro Of COil ;lttitud" about IbM, SOIllC of 1I~ ,Ire all [(JI'
while others i1lsist that dlicie1lt')' i, a cold eUJI1011lic value
that mllst lIot ,'Ollle before t'quit)', COllcern !(H the needy, or prort'c
tiOll of tht' envirolllllent. Tim, it is ,;Iid wc Ileed to "trade ott'''
efEcicllCY against otht'r y;dues. I But while most of us kllovv' vl/hether
we art' "fe)!' it or Jgill' it," we art' oftcll Ilot surt' what "it" is. Just what
72

liS

As a

or

but slllalkr than thc utility orthc first picce (hence

is what IS me,lnt bv
73

\vill be
ulility.

r
CHAPTER 3

Total Costs

EFFICIENCY

Total Benefits

3[/i(C)]

/I(b)

Chicken wings for All

4

+ [JI(b)-m],

,,'10

I"

utility
tmnefits

I I (b) + [/I (b)

2[/1 (c)]

II(b)

Pi128 lor 2

Pizza
for A!I

Betty

Idc)

I)

Chicken wings for

Number
FIG U R E 3-1

FIG U R E 3-2

01

Efficient Consumption

'where !,(b) 'm n

It will be rational f()r Air to COllSUllle two slices if the utility benefits of
two slices is greater thall, or equal to, the utility costs
two slices~
+ 111(/1)-111
21f1(r)l. Because the costs are constant but the
benefits arc decreasillg. at S011le point it will be the Glse that
or as ecollomists say. "Illarginal" (utility) benefIts that
frolll sOllle slice of pizza will be less thall the marginal (utility)
costs he had to iucur ill order to get th;lt slice; 111 that else it would be
irration;ll for Homo h((lllolllims to COIlSUIlIl' that additional slice of
because the preferellces he then would be satisf)llng an' ranked
below the prdl'rellces he is fillgoing. Figure 3-1 is a graphic repre
sentation of a specific example of this simple choice problelll. 011 the
left vertiCll axis we measure HOlllo l;'rollollliws's total utility costs while
Oil the right vertical axis we llIeaSLIre his total utility benefits. In
Figure 3-1 it is rational ftlr } {Olll" /;'(01101111015 to purchase three slices
of pizza (but no

or

Homo Erollolllims will conslIme up to the point wht're 1I1Il~~illill

bCll~/its eqllal lIIil~~illl/l (os!.'. This is olle definitioll of efllciency. 1101110
chooses in this way not because ht' "values
more th;1I1 a ratiollal actor seeks to "III<lximize utility"
this decisi()ll is simply required
rationality. To' choose any other

74

A Simple Edgeworth Box

would be to choose a lower

over ;1
of course, is the essellce of'

, is sil1lply

There is, thcll, a clOSt' rebtioll betweell etlicicllcy and ratiollal

choice: to be ntion;ll is just to choose ill a WJY tll.lt hest satistll's 01lt''s
that II1cans that the lIIarl!:inal g,lillS ;lre at least as
great as the
sitllations in
Chapter 4 when' rationality and dli('il'llcy break ;Ipart; III many of
these cases we are 1IIKLTt,1II1)lIst what i, the truly rational
to do,

Efficiency and Rational Exchange

Let liS 1l10Ve to a two-person case. W l' will rL'prCSl'1lt the ell oice
problelll in terms of indiHl'rence curves, which. it will be
plot olle's preferences over bundles of goods
dClTe;lsing
rates of substitution between thelll (Sectioll 1.3, poillt
Alf has four slices of pizza ~\Ild Betty has eight BulEtlo chicken wings.
And suppose that they have pretty Illllch the sallie preferences over
alld chicken
so that their indifti.'rcIIlT curves are
the sallle. Figure :'1-2 gives their indifference llIaps; Betty's (solid)
indifference curves start in the lower left. Alfs
curves COllle down from the upper right. SllppOSC that ill this
worth Box ,':2 Alf :lI1d Betty arc at Doint PI: AII' has Illlir slices of plzza
and 110 wings while Betty has

75

,
EFFICI

CHAPTER 3

to remelllber that Air is inditterent bctvveell all

1t is
combinations of pizza and chicken wings on tht' indiffl'rence curve
: however he prefers ;111 hundles on curve Alf:! to ;111 bundles 011
. The dotted arrow cOllling do\vn frolll the top right comer
shows that as All' J1loves "southwest" he reccivcs incrc,)singly prc
fie'ned combin:1tions of pizza and chickcll wings. If hc moves all the
W;IY to the "southwest" comer, he receivcs all the pizza and all thc
chickcll wings. The solid arrow represcnts the direction that Betty
to 1110\'C: startillg at the lowc1' lett corner, .,he prcft'rs c01llbi
!lations of chickell wings a!ld pizza that are to the "l1orthclst." So
is indiflt'rCIlt lwtwcen all hundlcs of pizz;1 and chicken wings on
the indiffercnce curve Betty I, hut she prdi:rs ;111 the bUlldk, on lktty,
to any of the hundles Oil Betty I. (AmI, of course. sht' is indit1(.'rl'llt
between the hundll's on Iktty 2')
Now any point III the
f()]'lllCd
the Alfl-LkttYI inditt'er
l'nee curves improves thl' utility of both AIr ;1l1d Bettv over point PI:
allY pOInt ill the "eye" Illoves cach to a higher indiffercllce C\lrVl'.
Pick allY point ill the "cye": you will sec that Betty and Alfh;lve both
llloved li'OI11 PI ill their prefl'rred dlr~'d.JollS. What this show, is that
both em he made better orf hI' excilange, CI'1'II t//01I1;1r 11(1 IIddillOIl/1i
/UIl'C /ICC/! (/'('(/(ed, An exchange that moved both Alfand Betty to
poilll p, on Figure :\-2 would Ilnke both of thcm better ofl; both ;In..'
raised to a higher indifTlT~'l1cl' curve: !lote that j>, is Oil
Bctty>. We Clll say thcn that point
Oil 1~lguIT ..1-2 is
to point I> I: at ]e;lst olle p~'rson is better olLmd 110 Olll' is worse olf 111
this case, both pcople- arc better olr
But ;dthough the h;lrg;lill at p, is Pareto-superior to PI, thcre arc
still Pareto-superior bargaills that Alf and Betty can make stlrtJllg at
1>2' The gains fi'om trade have 110t becll exhamtl'd: ,0 long as Pareto
superior l110ves are :1V;1ihhlc, AILlIld Bettv can kccp (lll trading and at
least olle will bcnefit. Whcn :lrc the possihle gains ii-olll trade
l'xhaustecP We call easily scc ti'OI11 Figurl' 3-3 til;]t when tll",y reach
;1 h;lrgain at which their illdit1l'rclKl' curves :Ire Llllgcnt, Pareto
su perior IlIOVl'S arc ex h;lllstt'd.
Consider, Il)f exalllple, point l\. At point 1\ AII' em only lllOVl'
ineiittt>[('IlCt' curve if Betty 1l100'l'S to a lower indiflt'rcllcl'
that would make her worse
Jnd a 1110ve
IS on Iv i>;lrt'to-,lll)f'r]or if no one is made worst' otr So, too, starting at
\v;IY th:11 lkttv can rise to a highn inditlercllce
curve is if Alf moves to a lower Ollt' -that is, he IS worse off. Point 1',
is thus a P!1YCfO-Olllilll!1/ h:m':lin: any departure frol11 point P v would

76

NCY

Chicken wings for

Pizza for
Betty

Pizza
for Alf

P2 P

FIG U R E 3-3

4
Chicken wings

Betty

A Contract Curve

lJJ;]kc either Alfor Lktty worSt' otr Point I'" therdl)re, is etlicient: all
the gaillS 11'0111 possible exdLJIlgl" have 11lTIl Cxh;llisted. But
IS Ilot unique in this lcg:ml: ally b;lrg;lin that occupics a
which AU' and Bctty\' illllitTercllcc curvcs are tl11gcnt i, J>areto
Thc line A-B (which is called the (Ollimet n/rl'c) represcnts
,Ill slIch possihle etliciellt b;]rg;)il1S~(lIleS tiIat usc all thc possihle
11'0111 tLldl', ()bviously, st,)fting tJ'OI11 point 1'1> sOllle
the el11cient
contr,lCts Llvor Betty while others :llT better for AWl
rates of
Notice two thin~s, hl'sl, the
ration;ll
people
substitution is lTuci,d ill
varied
to
trade with c;]ch other. Because they eKh prellT

or

hundles

<T(lods. if AIr is nin;l-rich he will

Betty,

over

those last two slices 01

[(Hlr chich'll wings)

77

CHAPTER 3

EFFICIENCY

slices of pizzJ. Thus, as I have depicted the problem, it is irratiol1Jl for

Alf Jnd Betty to refrain fi-om trading: if they keep their bundles at PI
they Jre satisfying lower- over higher-ranked preferences.
We are now in a position to understand the concept of PJreto, or
allocative, efliciency.~ We can say that distribution )):, is Parcto
slIper/or to (more efficient than) I) I if and only if no person is on a
lower indifll'rence curve ill 1)2 than that person is in [) I, and at least
one person is on a higher indifference curve in I):, than she is in ]) I. If
despite the possihility of a move to a pJreto-superior distribution we
stay in ]) I therc is at least one person who could achieve J higher
level of preference satisf;lction without lowcring anyone else's. Thus
in Figure 3-3, the distrihution identified by P, is Pareto-superior to
the distribution of P I. As in all the cases we luve discussed thus tlr
(but see Section -1-.2) there is something irrational ahout maintaining
Pareto-inferior distributions. In addition to being Pareto-superior to
hoth P I and P 2 , point P, is also Pareto-opt/II/III just becallse there is no
alternative distribution which is Pareto-superior to it. That is, if Alf
and Betty have arrived at P ,. there is no way in which one of them
can be raised to a higher indifltTcnce curve without the other l110ving
to a lower curve.

Paretian Welfaristn The Pareto criterion is often IInderstood not

simply as J requirement of rat/olltllit)' qlla cftiCIency, but as a standard
by which wc cm judge the /1/01'111 desirability of a distribution or, in
gencraL a social stateS To many it seellls c1car that distribution 1)2 is
morally better than D I if (alld only it) somc person's wcltlre is
"highcr" in D:, and 110 one's is lower than it was in I) I. Especially
in politics, it is thought. what is good fllr pcople-their weltlrc
mnst be thc (sole) critcrion ofa good policy. This view has clear roots
in utilitarian Illoral theory, which identified promoting hUl11an hap
piness as the sole goal of morality and politics. Recall the rcmark
quoted in Section 1.1 from Nassau Willial11 Scnior, a leading political
economist of the ninetcenth century; economics, he said, could
aSSUl11e that everyone seeks wealth because "wealth and happiness
are vcry scldol11 opposed." The nltimate aim was clearly human
happiness even if the proximate aim of econol11ics was the growth
of wealth. Now contemporary wclfare (,(oIIOI1l/rs typically understands a
person's "weltlre" to bc mcasured bv her utilitv function. If the
utility of Betty is II in di,tribution ))1 :md /1+n in' U 2 , then it is ,aid

her welfare is higher in I):, than in I) I, and D2 is a better distribution

than 1)1. Here, however, things get complicated. The early utilitar
ians such as Jerel11Y Bentham and his followers believed that "utility"
(pleasure) was a cardinal measure (it could be measured along a
metric) and that, when contcmplating a move from I) I to U:, we
could scnsibly add the utility Alf received from the move fi-om I) I to
))2, to the utility Betty received, and then subtract the loss of utility to
Charlie (who, let us say, was better offin I) I)' Having done our sums.
we could then decide whether, ovcrall, the move from )) I to 1)2
increases overall aggregate utility. Bnt we have secn (Section 2.3) tlLlt
there is no particularly good reason to add von Neumann-Morgen
stern cardinal utilities of different people; unless there is a special else
for some additive function, it is simply arbitrJry to SUlll up cardinal
ntilities. The contcmporary welf;lrist seems to have a problem:
how to compare social states without ll1terpersonal comp,lrisons of
uti Ii ti es?
The Pareto criterion sce11lS to onl'r a way out of this problem: if
no one is worse off in ])2 than she was in ]) I, and at least one person
is better ofT in ]):, than he was in ]) I, then ]):, is P,lrcto-superior to
I) I. And since the welElre economist has identified a person's welLlre
with her utility, It looks as if we call say that ))2 does better tr011l the
perspective of human welElre. Now it is otten thought that this
cannot be a very usefi.il criterion of "moral betterness": it oilly yields
a judgl11cnt that 1)1 is better than I) I if 1/0 (J/ll' is worse off in I )2' But
how otten is it the casc tlLlt no one is ll1ade worse ofP On the Parcto
test, if in ])2 one l11illion people are made better otf th,m they were in
I) I, but one person is worse off; we canllot say th;lt 1)2 is Pareto
superior to I) I. Is there ever, we might well wonder, a Pareto
superior 1I10Ve to be made? We ,Ire now in ;1 position to see the
econol11ist's deep attractIOn to market transactions. Under certain
idealizing conditions (e.g., full information, no third-party ef1ects),
each market transaction moves us to a Pareto-superior distribution.
When people trade, they prefer what the other persoll has to what
they offer to give lip, and so we move to a Pareto-superior lhstribu
tion. As long as we have not exhausted the possibilities fllr
exchange-as long as there are trades that people W;ll1t to make
we have not exhausted the possibilities ftlr Pareto-superior moves.
Althongh market transactions are otten moves to Pareto-superior
outco111es, it is much harder to see how a collective public policy em
meet the Pareto tcst. It is hard to think of any uniform policy that
does not disadvantage someone.(> To avoid thi~ conclusioll (i.e .. that

78

79

Is the Pareto Criterion a Moral Ideal?

......

CHAPTER 3

the Pareto critenol! must he violated

wclbre economist, :llld "Parctiall"
adoptt'd what is known ;t, the

dislrilJIIli,lll

J) 110

OJ is

llllit(JrllL

han'

(,1'ell 11'111'1/ SOIl/C

do

lite

sOllle

lose

U)' IIIOI'llIg./I'OIiI

lJ I

or

80

Betty's

--_93
D:>

10

could 1111pCIISI/le IiiI' loSl'!'s

-; To grasp what it l1IellllS to say that a persoll wilid be
t()r a
cOllsider AIf~ who, wc ;m.'
is the
sole persoll who has het'LL 11Lilde worse
tht' 11I0ve tl'OI1l Dl to 1).2
make the casc SlllLp1<:, as,UlI](;' that l'verYOl1C else is hctter otr ill
D21, To say that All' ha, bccll madc worse 011' llle,IIlS that he is Oil a
lowcr iLlditllTl'!1ce curve ill I), thaLl hl' was ill I) I, No'x iLlLagilll'
after the move to I), the
trallSterred ellough of their
to
Alf [() raise hill! h;lck to thl' iLlditl('rcllce curve tint he
this would brillg Ilbout a Ilew distrihutioll 1)" which is iLldced
to I) I because everyolle' e:\t'el't All' is :It II higher
indifflTl'Llcl' curve ill I); th;l1l they were ill ))1, Illld Al(i, LlOW back.
OJl the ',lIlle inditll'i'l'LlCl' cUn'e (as he v\'IlS ill l) 1
W l' CllIl say.
th;lt 1)2 i, Kaldor-Hicks Paretosllperior to 1)1 iftlll'rc is a distribution
I) \ that (I) could be L1roduced bv rl'distribulilll! the
1110Vlllg Ii'om I) I to I), alld
I) \ is
the llorJlul
l'arl'lo
to I) I, Note the Kaldor-llicks test SllyS that
) and (2)
to I) I ('1'('1/
1"lS hcCII
I )jstrihutioll I) is that ill which
has b\:'l~n
lllade, bllt Kaldor-Ilicks docs lIot say silllpl)' th,lt I), is P.lrcto
to 1)1: it says th,lt I), is
to (more dllcicLlt
I) 1 evell thollgh SOIllC pcopk
illcurrl'd losses
the 1110VC ll'Olll 1)1 to 1)2' Bel'allsl' I (<'lIld give rise to I)\. alld
becausc I), 1I'(1111d be Pareto -sllperior to I) I, I), is Kaldor-I-!Jeks
Paleto-sllplTior to I) 1
To lIUllY tIllS S('l'IllS vcry odd:
which W;lS
b;lSCd Oil the dClllld th:!t
fill SOllle ClIIl
IS !lOW elllployell to .
that hellL'tit SOllll' at the expCIlSl' of
DtlLers, The lllove ,i'olll I) to 1):1 lIukes Stlllll' peop1<: worse ofl~ yet it
IS .Iust!tlell as a
Kaldor-Hicks looks like a
backdoor way of
U)lllplmSlJllS of utilit\' loss
alld gil ins within II Paretillll ti':ullcwork.
Even if we put aside the controvlTsilll I<aldor-Ilicks interpn'l:ltioll
the P:lrl'to critt'rioll, ll}JOll rdlectioll thl' Pan:'to test is l10t as U11COI1
troversi.d as is ofkll thought,;'; Much o!'thl' ,lppl'lll of the Parl?tu criterion
lies ill the question "Who could pos<;ibly object to all improvell1ent that
lllakes everYOlll' better olP" Figure :~.,.f suggests IIll answer
(IS IllllSC 11,110

EFFICIENCY

IlIlll'e

o
FIG U R E 3-4

utility
A Series of Paretian Moves

wc start at point O. and lliake the P,ln'to-superlOr l1l0Ve' to

A. which is OIlL' possible distributioll alollg l) I" the "P.lrl'to
Frolltier," the set of possibk P,ll'l'to improvemL'llts frolll O. Ollc't' WL'
are :It A. tht' Pareto impwvclllt'llts to I)::> Ilrl' liLllitl'd to thosl' lwtwt't'll
poillts W :llld X; other
thllli ill l) I. so are llot Pareto illlproVl'lllCllts, Assume thcll tInt at soml'
distributioll 1)2 becolllcs ;1 po'isibllity: Alf ,llld Bctty makc the
Pareto move to poillt B 011 1).2' Suppose now that ill the ttlture 1)1
becolllcs possibk; llO\\' the possible PIlrL'to improveLlll'ms are limited to
thOSl' bllillg bct\\'el'll Y and Z, We em sce that successive applicatiollS
of the Parcto critcrloll Illove distrihwiollS alollg a p.lth that is incH';!s,
illglv [wlldiei:ll to AlfaLld urless bCllefit to Bctty, Ifwc Iud bcell abk to
JUIllP to I), all at Ollce, cvcrythillgoll it would be;1 Pareto
over 0, hut ollce we have llladc the interlllediatc lllOVe'S to A IlIld B,
lllOSt of 1)\ is c:\duded by the P;lfeto crlterio]l, Perhaps.
would havc good rcaSOll to objcct to the illitial Pareto LllOVC to
The Pareto prillciple allows a wide r;Jllgc of movcs, ,llld it Illay matter a
lot which of those is ;1Cltllllly made, alld ill wh;lt ordeL

Welfare and Preferences

aside thest' problems with the
P;lrcto criterioll, it also Sl'elllS doubttttl that we should accept the
idc'iltiticatiuLl of "prelerl'IlCe SJtlsLlctioll" with "welf;lre, " We l!;lve
,cell that prekrl'llcl's Ill'l'd !lot be Jbnut olle's own good or self..
interest: anytilllc OllC rallks all outcome above another, Olle hilS ,1
2,1), l{,l'C;:lll our last-Llllmgo rdt!sl'r frolll Sectioll
is llot to take the List Lll:lllgO, though. as Sl'll llotl'S,
she would like th;Jt lllJllgO and would WdCOlllC SOllleOlle thrustillg
81

T
CHAPTER 3

EFFICIENCY

the mango on her. III that case it seel11S doubtful that we should say
that her welfare is enhanced by satistying her pn:ferellce not to take
the last mango, since her "civility" preference instructs her not to do
what would be good filr her. Those who identity "we!f:1re" with
"preference satisfaction" often simply seem driven to stipulating that
in this case one's wel6re IIIl1sf be advanced because one is
what one "prefers." Here, I think, is a perfect example of the way that
the ambiguity between the techniGlI and ordinary senses of
2.1) leads to serious confusions.
f(x P;lretian
Preterellees regarding others also posc
welf:llis111. Suppose Alf is a prude who
'.I LeweY. It seems tlut
~1 sOllwwhat racy book such as Lady
to Paretian wcltJris11I
book
cannot he approved of by the Pareto.
'\vorsc ofT" whell she fe;lds the book:
we might normally think she has ;\ right to
illditrerel1ce curve. The way in which the Pareto principle call COI1
flict with all individual's rights to decide what she is
to do has
been analyzed
All1arty~l Sen.') Sen conceives of a person having ;1
right as having :lllthority to dt'cicle the soci;Ji preference over at le~lst
one pair of alterIlJtivt's (x,y) such that if~l person chooses x
l' th:1t is
the social preference
us cl11 this social prdl'H'llce xP),); and if the
person chooses y>- x then yPx (i.e., the social prell'rence is y over x).
Sen shows that attributing sllch rights to two persons, and asslll11ing all
possible
of social states are permissible, the s(leiJI olltcOl11e
selected by the rights can conl1iet with a version of the Pareto
accordillg to which, if fOf evcryone x>- y, then xfly. That
is, if everyone prefers x to y, then the social preferellce nHlst he x ()vt'r
y. Sen nicely :;un1111arizes his ;lrgUll1ent:

it,

(l

is sou:dly better than

<l

sil1l'e A does not wallt to read

t1 is P:lreto-superior to

<,. But
If)

,md (f Ph
So we get hfl" (by Mr. B's right), oPa (by Mr. A's
Pareto, since in both Mr. A's and Mr. B's
<1>- b); so wt'
gct /l/JoJ>iI/)/)-;m illtDllsitivl' result. Sell saw this not as a case
rights. but as showing "the unal'cept:lbility of the 1l;IITto
universal ruk." 11 Sometimes it seems that a cOJ11mitment [0 Pareto
ClIl ICHlllS astr;IY. <- )I1C W;IY to respolld to this problem is to
exclude sOl11e sorts of prcfl'rCllCl'S tj'OIll cOllsider,ltion: thus we
restrict Ollr welt:ll'e Paretianislll to
ellces over diftl'rcllt St;ltcs of one's own litc) and so ignore
that other people do rather th:m not do ccrt;lill things
not read Llther th;lIl rcad ccruin hooks). But this
solve all the

>-

r.1l~e liS

over lllany expellSlve OllCS. Given thesc

structures, P;l1Tlian wclt~ms11l approves ofdistriblltiolls tint
both to hiuhcr indifil'rl'nCl' curves-- l!ivilH! llll' l'xpl'n,ih', ;Illd
12

you

There is a book (e.g. Lilly Chatterley's L(1/!('I') which may be

read hy Mr. A ("the prude") or Mr. B ("the bscivious") or
neither. Given other things, thest' three alternatives define
social st;Hes, <I, Ii and {) respectively. Consider now the
following possibility. The
A most
(1 (no one
then !l (''I'll take the hurt on
and lastly /)
that lasciviolls lapping it up"). The Iasciviol1s
that lilywhite baby a
prefers 1110St tl Cit will
lllce stwek"), then b ("it will be fun"), and last () ("wh,Jt ;1
waste of a
book"). 011 grounds of individual freedom,
si nee B wan ts to read the book rather than 110 one reading it,

Fair and Unfair Starting Points Perhaps the 1llOSt seriolls probkll!
P,ll'l'ti,l11 wl'IElI-isl1l is th,lt it is inst'llsitive to the distributions ti-oll!
which we
Suppose th,lt ;111 of Alfs :111d
over quantities ofpiZl';t ;1l](1 chickcll wings, hut
and ;111 the chickcli wings ami Bl'tty has nOllt'o
P;m'to principle, this is all dlicient distributioll. Since Bctty h;IS notil
ing to t]';J(Je, thne is no way to nuke her bettl'l' off without lllaklllg AII'
worse ()f1~ so wc have achieved Pareto optimality. 13m this hardly
set'lllS a moral reason to e111hr;1Cc the distribution. Thc hl'~lrt of the
Paretian project is to make pcople hetter ofT, and WhCIl no 011(' call he
made hettcr off without lowering S()llleOIH' else's utility. the Pareto
criterion Ius llothing lllort' to say---we h;lv(' achieved "optil1lalitv_"
But li'ol11 the perspective of adv;mcing 11l1111:m Welf~l1T it is hard to
conceive of situ atlOllS where Betty has IlothiIlg ;lS "optind." Plausihle
versiollS of Parl'ti;lll welf:ll'ism thus seelll COllllllitted to S0111(, idca of;1
flir starting point, :mel flim can hold that P;lfl'to-:lpproVl'd moves l11:1lk
from the illiti:l! t1ir startim: poillt arc mOLll

82

83

.....

EFFICIENCY

CHAPTER 3

3.2 EFFICIENCY, EXTERNALITIES,

AND PUBLIC GOODS
Externalities and Property Rights
to the cHIciellt
pizza: how l1lucb
ideJ \1;IS
bel'n
our allalysls of all et1iciellt COllSlllllptioll decision: you
up to tbe point where your marginal bendits
your 111;lrglllal costs. lfyoll stop producing pizza wilile yuur
benetlts are still greater thall your Illan.:rinal l"()sts. it looks
illsoClr as production of;1I1 extr;1
t~lCtion th;1Il it costs (rl'llIt'mber, our concern is opportunity costs). To
produce pizza above :md
the pOlIlt where your
bcnetlts equal 'y'OUI" lllarglllal costs 111l';IllS th:lt your LIst pin;! cost
you Illore (ill tLTlIlS of prdtTt'IHT s;ltis6ction) than you received ill
which again looks to be all irratio!l,11 decisioll. So ;I ration:11
lIl.lXilllizer will produce up to the point where her lllargilLIl
benet!ts equal her lIlargi 11;11 costs.
The nwbklll is that this lIeed not be the dticiellt level of
our f(1ClIS tI'0111 tile eltlcicllt

that you

('xlcrlulilic\.

All

prodllcn's or C(lIl\UlIlcr's cost,

extcrtl:liities
11IlPOSl'S
,I ratioll,li ccollomic :Igcnt Air
ClIg,lgC ill it up to the point were /iis Illarginal bCl1dits l'l]lIal /iis costs,
Alf willllot uke accollnt of the (mts to Betty. But if AIr
lip to tilt' Jloillt wltere his III;lrgilla] cmts :lIld 11l'netits are
there ;Irl' additiollal co,t, to Betty, it looks ;IS if the tot;11 ,()cial costs
(thl' cost, to All' ami
of All's last unit of productioll l'xceednl
tlte entire social bcndits. This would violate thl' P;lreto cnterion: AI!
has Illoved them both to a 11l'\' distributioll (with thl' extr;l
lInit of pizz,l) which bendits Alf at ;I cost to Betty. A sillli!;1r ;l11alysis
applies to external bCllefits: if Illy production 11;1\ bClIdits to you ;1S
wcll as to Ille, then if- I SlOp production whcn lily m;uginal bCl1dlts
Illy margiml costs, fWIlI a socia] Doint of view (which includes
the bcnetits and costs to everyone) the
social mamillal bcnefits still excecd soci:d 111argin:11 ,'osts, sincl:' lily
not taken into account the bellefits you receive.

this impact

1S

bellefit

84

Extnmlities ,1re a chlt-f source of "market f~lilure"-the i:lilurc of

the market to produce eHicicllt results. Oilly if Bl'tty)iilly illftTIIIlIi:cs all
the costs alld belld!ts ofber :lcti\"ity will slie stop at just the poillt wlll'rt'
social lIl:uginai bendils c<]u;II social lll,nginal costs. So too III a trade:
Ii"AII';1nd lkttv ftdlv mternalizt' the costs :lIld [wnd!ts oCtheir trade
C.lll Wl' S:IY lhe trade lll'ces'<lrilv moves lI) to .l Parl'to-sllplTior state. If

then' :l1T tlllrdpartv costs

extcl1l:diIlCS), AII' ;llId Betty

Iluke trades where the soci:1i marginal costs exceed the social

belldits because they do lint take ,ICCollnt of the costs to Ch,lrlic: if

[x'lIeiits exist, All' ;llId Belly Illay Ilot tLltlC CVCll thollgh the
bend!ts l'XCCl,d tlte sociallliarglllal costs.
the market would SCelll~()(IIICC ctticicllt out
,~:- ~-s~~';;-;:~l~'opert\: 'l~lgT;t~-\\Ill'I~'hv :~
.--.---------~--~.-~

--j.--,~-,"'"

.. _--_. -------"-- - "._

"r

,Igcllt ill/cllltlii:cs
((1.\1-' dlld /Jmc/irs
IllS tlllil'il)': he
obt.lillS the full bClldits, ;Illd P:IVS the fidl cost" of his :llllvlty.
Consider thl' \\l'II, kllO\\"1I os," ufthe "tr,lgcdy olthe COIIIIIIOIIS" such
;IS tisheri,'s.: 1 M:I1lY lisheries ,Irolllld till' \~orid ,liT overtished. result
illg ill ;] deplctioll of,tocks. No\\" it would Ilro\J,lblv be to the belletit
of fishl'r1l1,llI Alf to rnlucl' his each
llext year it" If(' ,,'lIld h1' ("II/i"l"IIl <'/" <)hlllillill,I; ,111 lite li//ilre
III's
r('dll(1'd rield Ihis l't'lir. But hl' call1lot: if lktty :lIId CiJ:lrlic lish ,IIIYW,I\,
the stocks will still 1)(' dl'pktn!. AII' will h:wc p:lid ,I cost but wlil llot
g:1111 lhl' full bl'lId!ts orbis rcstLlillt. COllvl'1'sciy, Betty ;Illd Charlie do
lIot pay the full costs of their overllshlllg, sillce the costs of
:Ire trallsi(:rn'd to other tlshermcll slich ,]S Alf Thlls the fisherics :11"('
;lIld the IIl.1rgill;1I soci,d costs l'xcecd till' lI];1rgill,11 bell efi t>,
A schellll'
tklt illll'rn:liI7l'd ,III costs ;lIld bendit,
-;~;-u-fJ~~-\T the probklli.
C<llme we h,IVl' this probll'1I1
t'COllOIlIIC

IWClllSl' p1"llperty rights OHT iish in the on'.I11 .Ire (lItticult to .

tioll,llizl' (tlsh tl'lId to swilll ;IWlllld ,I
I-!O\Vl'Ver, thl' ide:11 of;] Systl'llI
property rights th;]t illtl'r1];lhzcs
benciit;. .Ind co,ls is lIllre:1Iiz,lblc II II less we restnct \vhat Ullints :I.S :'11
l'~tl'rtl:i1ity. Think abollt SUI's case ofMr. Prllde's alld Mr. la,clvious's
abollt rl'adillg Lad)' L'11'lIIcrlC(S Ltll'lT. If Mr. LISl'iviom

ext'rL"i,l'S 1m ngln :l1ld rC:1ds till' hook. thnl' is all

Mr. Prude IS 11l,l(k \\'or,,' on~ SIIKl' Mr. Prude prl'il'l"' tllal 110 olle

rcads it to Mr. Lasclviolls rc,ldillg it. Mr. LlscivlOliS It:]'; lJl'g:ltivdy

impactcd the utility fUllctiol1 of Mr. Prude. If' people havc prCfl'rl'lH:es

over
thell extl'rI1alillcs will hl' t'Vt'l"Y\\'hl~

- -.. AlfprdlTs tlw people shop at farge't

W,;I-M,lrt; if

or

tT;;;,;'

so, every tLlIlS;lctlOI1 at Wal-Mart involves a Ileg:,tivl'

85

EFFICIENCY

CHAPTER 3

Jre apt to result ill this level. But

to
solution we must know what our property and other
we call identify externalities, and so etTiciellt out
cOllles. Say that Alf ,vallts to build a LlvenJ on his land and
objects. What is the eHlcient outcollle? If Alf has the
011 his
side, thell the efficiellt outcome is that he builds it; if Betty Ius the
rights Oil her side she h,1s :J right not to have her property v~lllles
lowered by living Ilext to a tavern-then the et11cicllt outcome is that
the tavern is Ilot built. Ifboth have ri!!hts then we must still sOInellOw
up the costs and benefits.

Ronald Coase proposes

view: on Coase's view ,lChievillg all etlitient outcome does

not depend Oil thl'
way that the illiti~]l
Suppose that we live in a world flTe of tr:msactioll and l)~lrgallllllg costs,
and in this world Alf has a flCtory that produces \vhatchacallits tint
generates smoke as a by-prodult; suppose that Betty has a l:ll1ndry, :llId
her costs :Ire increased because of the Slllokl' frolll Alfs tactory. All's
product's a ncg:]tiw externality. So if Alf produccs
whatch;lc'l11its up to the point whlTl' hi, lllarginal costs equal his nLlr
too many whatchacallit~ will bc produced. Suppose that
per year; Betty presently earns $24,000 fI'olH her
but she would make $3 L()O() if Alfs sllloke didll't illne~N'
her costs. AlL theil,
:Ill
against pollutioll, we call still achieve Pareto ctllClcncy: Iktty em pay
Alf$.\()(JJ to stop producing whatch;lcIllits; he will be hetter otralld so
will she, so thc 11IO\'C is Pareto-superior.
It is importallt that Coasc's theorelll applies rl'g.lrd1ess of how the
property rights arc divided betwel'll Alf and Betty: ;1 Pareto outCOllll'
Gill be reached whether All' Ius a right to pollutl' or Betty has ;1 right
that he llot pollute. SUPPOSl' that Alf is now lllaking $ ](I,O()() pro
whatchacallits while Bettv's llrotlts rC!l1;)l1l the S,lIl1e: ;f;2,t,()()()
has

One possible solution to this difficulty might be called the r(l;/lls

.1011111011,
to which Alfs actioll h~ls a negative externalBetty if and only if it violates a right of lktty's. Rights, we
i1llpinging Oil Ihose
individual.
The rights
constitutes cost
who
was
especially
based solution
to other, ttlr every
to others
cost they Impose upon
don't like. Mill
fi'OI11 performing ,lctions that their
should
~lrgued that slich "costs" should be ignored, md
set back
be said to illlpose recogniznl social costs 011 OtlH'TS whcn
as
rights."
1(,
"certain interests, which ... ought to be considered

liascd

Thus, says

or

thC1ll
,lilY loss
t:llsdlOOd or
unf:Jir or ungencrous usc of
over them; even ,eHish abstincncc frolll dd('lld
injury---these arc lit
of moral
lIlg them
ill
grave
cases,
of
ll10ral
retribution
and
reprobatioll,
.
1
17
PUIlIS 1ll1ellt.
lIltliction

OIl

The idea, then, is that we identity a crucial set of intel'l"t~ (or sets of
prefercnces over certain aspccts of ollr lite), and hold that if an action
or transaction imposes costs 011 other parties ill tenm of tliese I)/'c/i'r
CII(CS, the action or transaction has a sori:ll cost
., rights h,]ve heel]
That the :lCtiOl1 involves a s(lei:11 cost doe, not show that it
should be [)fohibited, since the social bcnefits Illay still OlltWl'lgh tht'
costs. I K
This results in ,I 1l1Oralistic
knuw which subscts of a person's prdcrcJlces are
rights bd(lrc we elll know what constitutes all eHlUl'nt level of
:lCtivity. It: :IS Mill el1lph:ltiCllly ,lrgucd, 110 one has :l right that others
don't read (rather thaIl reael) books onc tlnds offensive, Mr. Prude
incur, 110 cost when Mr. Lasciviolls reads Lady CIIil I Icr/C)' 's LOFer
there is no externality became IlO right was viobted. There arc,
real worries about this moralistic view. For Olle, it dot's not
make sense of a core argument of most liberal political econol1lists,
viz., that we should evaluate systellls of
ri~hts ill terllls of
their efficiency-promoting characteristics. If we are to say that
property rights promote eftlciellcy we must be able to first
what an efficiellt level of production would be, ami then show that

her $7,( J() 1 to rdi-ain from

According to
theu,
costs, parlies to

1m WliFit)'

'lr resollrces

ill tile IlhsCII(C

IPitil ex/mltllities

c:r tr,II!SII(fi,lll

11'111 (~l;ln' 10

!Inri

S,llll!'

or

til(' illilitll dislriiJlfliOIl

{!ropcrl)' rll;h/s.
Coase's theorem calls into qUl'stion olle of the traditional jllstificJtiolls
for government regu1:JtiolJ. In the abselKe of a peJil'l't schellll' of prop
erty rirrhts that internalizes both costs ;Illd 1)l'lletit'i, it has beell widely
al/Ndtloll

I'(:I;(JI'(I/css of

87

86

CHAPTER 3

EFFICIENCY

government is necessJry to re6rulate the "l1larket fililure" that

n:'sults from externalities. l3ut Coase shows that Illarket trallS;lctiollS
em solve the problel1l of externalities and get us
outcome (thOtl\lh the actual costs involved in
precludc

argument is

not

quite

as it seelllS. Three

,IS

must be

lIlind.

For ,lll

goods argulllent f()r sLlte action, it Illust

that evervone wants the .l!;ood, hut that in

Public Goods
Rebted to the question of extcrnalitie, is the special Cl,C of
Public L':oods ;Jr(:' defllled in tenllS of two char.lctnistils. [)irs/,
Consider clem air. I f it

everyolle \V,lIlts the

I,)r it to not h:\\'iIlg It.

;l'way ti'OIl1 Betty. Once tht" good is then:" consulllcrs do not compete
f,x it; everyolle can fi-ccly lise II withollt diminishing the amollllt left
ft.)r others. Suolld, we Cl1lllot control the flo\>,' ofbcnd"it'i ti'olll plIhhc
;lre I/ollc.w/lld,i/J/c. If a public goud is provided, it is
provided f()r all to use. If we clean the air, cveryolle 1l.IS c1c;m ;lir.
We canllot exclude those who h;lvc llot 11,lid their share. A pure
is one that
meets these two conditiolls. In most
,lI"e thus

ecolloillists have disputed the "pl1bliclll'~S"

this list. The classic textbook example of;\
warns all

It is not the case tlut 1ll;lrh,ts lll'VlT

so dl"icil'Iltlv. Suppose A1t\ go,lt wanden; illtu

l';lls hcrvcggics, ,lilt! Bl'tty's dog walldl'l's illto Ales property,

his go,]t ,0 th:lt it dOl" lint gIVe' 11l1lk.' I A [t'llce would he J

ASSUllll' th,lt clCh would bcnefit hy

would he hl'tln orfhllil,liJllr till'

fellCt'

:llolle thall

lIot

tlut till' other budd

1) the ()tlln
lleithl'r build,. III 'meh a case,
SillCl' l"leh would prdi:r to pay I()r the (,IHire good r;1Il1er thall do
\virhout it, the puhlic goud will be provided ('lIld, we sh;1I1 SCl' .
SCt,tioll -+.J, Ollt' Pl'!',OIl will pay t(lI' rhl' l'lItire good). I'wvi'U)11 uf
do lIotl'onstitlitc ,I lll,lrkct t;lilurc ulltil we add ttlltlwl"
conditIOns, such as rlLlt IJO Olle illdividu,d's utilitv ftlllclioll is sudl
th:lt It is r.ltiOIl:d tllr hllll to
leveL':: tilere ,In' a
Illllllhn of pl'0l'll' (SO caell IS temptcd to
or we dn llot kllow how IlI,IIlY Ill'Ople's l'OOPlTllioll is
lll'ccssary to s('("llre the good), ell'. As thl' Illllllber ofilldIVidu:11,
ill\"(lln'd Illcrl';ISl'S, the Ill'cd fill' St)lllC sort off(l1'lll,11
,lbollt ;dlocltioll of C( 1lI tri hu tiolls bl'(()Jlll" IH'Cl'"a rv.
per SC, but fluhlll'
that
lHlIllhl'r of'agellts thar arc IHost Ilkl'lv no!

evny itl'lll Oll

was typiCllly a
ship
,1Ild it
who lild not l1;lY fill'

the
next
tend to be
but the crux of the explanation appears (at least at first
Evell if everYOlle' prdtTS having the public good to lIot having it, each
of us will reccive it [()l' trct" if S()llleOlle else pays It.)r it. After all, the
belldirs are nonexcludahle: if ~lIlyone gets the
So we typically have an incentive to
pays, holds back from p~lying,
Thus t1w classic public good argumellt is fllr state action to fix the
market's failure to generate efficient outcomes, In the interests of
it is ottell said, goveml11l'llt Illust require everyone to
cOlltribute to the nroduction of sHch [!oods. While DowtTtill, the

hy the
.).

WI1l'thlT puhlic good, wdl he

Imrkct.'

volullt,lrily sllpplil'd, alld wht'thn

llllivcrs:d nmtrihul\()11 1S l'ttiul'llt, also depl'luls 011 the n'latioll

c'1l] l1htllll'1I1,h
:1.

thc goods alld individual cOlltributioll. We

three hasic types of rt'btiollS:

COlis/ali/ Re/llrt1s,

It' (; is til(' totll

allloullt

or ill('

;llld (;. is rhl' u)lltrihutioll of allY illdividual "

88

89

......

CHAPTER 3

EFFICIENCY

GI +
+ + ... Gil' Each individual
adds to the amount of the
here is picking IIp aHer your
in a public park: each person's cOlltributioll helps secure
the good of J clean park, where ;111 call walk \vithOl1t feaL

then

is that while Alf's p(() is the total costs of his share of

producing the good, his benefits tJ,(b) are just a slllJII part of the
total SOCIa! benefits, since everyone gallls trom his contribution
, the good is nOllexclu(bble and nonrival). As ;1

Dtionalecollolllic agent Alfstops contributing when his

costs equal the marginal benefits 10 ililll. But this will not

adequately take account of the over;]11 social

his contribution, since all others will benefit tl'om the

of the good be will provide. This is a c1;]ssic case where

will be underpnwided bv uncoordinated individual

b. 'J11Ycslwld Ilt Ihc Ttl]!. If C is the totll amount of the

and C, is the contribution of ,mv
individual i, then there exists some perSOJl k slIch
that C {G I +
+ (;1 + ... e /}
(C;I
C2 +
+ . . . + (;1: + . .. '}. The cOlltrihutioll of individual k
and those \\iho ttl110w her add nothing to the ,lmOllllt of
secured. Suppose that \ve all support a candidate
f'Jr office, ;lIld she needs 1O,()()O sign;ltlJreS to get 011 the
ballot After she has obtained 1O,()()O vil1id signatures,
additional signJtures do hef 110 good (;It least fi'olll a
do lIot help her get on the b'll1ot.

A threshold at Ihe

/J01l01II II'I'CH'

Ihe k

pcrsoll

is also the n, or

IdSf

[n tbe second and very speri;]! case, l'I'CY),olle's (Olitrihllfioll

ifill(' good is to he s('(/fred (/I <1/1, All example is ;1 crew of.l
unless everyone rows. the bO;lt will not make
the strong current. No public good is seclired
ulliess everyolll' colltributcs-tbe public good of reaching the
destinatioll will not be achieved lIllless everyone does her'
Here in In illtcrestin(!; case whnc the public good is apt to he
,Khieved
purely volulltary choices based silllply on individual
utility maximization, since no lIldividuill 11;Is an incentive to
over- or lI11dercolltrihute:''i
1)(Tsoli.

III the Bottoili. If C is the total illlJOUllt of the

good produced, and (;j is the contribution of any

individuill i, thell there exists sOlne person k such that

;2 + C l
... G/ }
0, but {(;I + C 2 + (;1 + ... C;

> O. Until the contlibution of indivldu;J! k is

no public good i, achieved, Again, think of our

candidate: until the 1O,OOOth sih'1li1ture is obt1inl'd, no good
i1t all is produced.

c. nm'sllOld

,"'I (,ISe (:f(0IlSt,1111 retllfllS lip to II t"re~/101d Ililhe top. The third case is
illtaestlllg because here it is illclliciclIl to relluire everyolle to
contribute all the tillle. SUDDme at Alf's ColleQc there is :1 IaWIl

We em identify both purl' and mixed CISl'S

that combine these types). We might have,
good that
constant retllrns up to a top tllresllolll, Nit tl1t'n 110
more returns thereafier, or which has a bottow threshold, :lIld COil
stant returns thereatter, :Illd so on. Consider three interesting rast's: (i)
a simple case of constant returns; (ii) ;1 threshold at the bottom whefe
the k person is also the II, or last person; ;1I1d (iii) a case of cOllStant
returns lip to a threshold at the top. Will the good be provided
each person maxi]l1izinl! her own lltilitv. or is sOllle sort of coordi
nation or

across from
would prefer a nice lawll between the
two buildings to a shoddy Olle, But evcryollc ;dso is
inconvenicllced by having to walk all the way around the 'lILld
the walkways are). Each persoll would preft.'r {having
nice lawn alld cuUinl' across diagonally I ~the shortest route
-to {having a nice lawn 1/111/
using the walkways}. If everyone Cllts ,leross, the lawn will be
1() out of a 1O() peoplc do so, there will be IlO
the threshold ~lt the top; after the ()(Jth person
across the \;]WI1, no rI.lrther public good IS
COl1sider three policies:
110 one crosses, so we
h;lve iI beJlItifllllilwn but everyone is
,1re
everyone crosses 1o'X, of the time. The

(i) .!I silllple ((ISC (!f WI/Sllllit

reflmls. [n the
case of COIl,t:l11t

returns each individual contribution

seClIfe a b'Te:lter level

of the t)Ublic l'ood. So AWs actioll

seeun's some of the

good: he will incur some costs

but since he \NillltS the good,

he also secures some benefit fi'0111 his contribution, u(b), The

90

91

JII..

EFFICIENCY

CHAPTER 3

Ilriil-'il)' IIlakes s('//.I(, , The problclll, we saw, is that

:lllothcr persoll neg,Hively impacts Illy utility, and this is not
taken Illto :ICCOtlllt ill her decision, she imposes ,1II n:ternality on
JIlL', But if Illy prdi'rl'llces ;1I'e about wlut she should ;lIld should
not do, ,I!..' wdl illlpose an extcrnality Oil l1Ie simply by living her
Jill' ,\0, she secs lil. This nroblt'1Il of adetlllatdv ddillillg all
of the Paretu

st'cond and tbird policies art' Pareto illlpruvel1ll'nts Oil the first,
Ten people can be 1Ilade better off (they (;111 cut across tilt'
without allY additional costs to othns. or t'veryollc can be madt'
better ofF 10%, of tbt' time: the st'cond and third policics acbicvc
. but at a lower
HOIIIO
if WC ,liT
for choosing between the
(but if wc aSSU1lle that t;limess is part of
the thml is to bc pret(''rred; see
public IlO1iCV bJst'd Oil the

cost. As t;lr ;lS

of dliciency doe, Iwt necessarily seck to Clllllin;lte

it
tht' hCIll'fits without p,p;ing the
level of
might even
policy h docs) seek to ,\(,(111'(' ;11l
frcc-riders (wIth llolicv b. 1

S/;:ct(/J('t/ II/('

( ;oll"i"('I('(1
<laio1l, <llId

(>'Od.\'(' liI('Of(,lII,

!,IIMi( yood" Imil

"1','(l'i/)('''

,,0111('

(llses 111"('1'1' I '0111 II tl1 1')' 11(liOIl lI'il/ s('(/Ire rliClII,

NOTES

to ullderlnine the basic ide;l that uSlully

None of this i,
The point is tlut
the st;ltc should t;IX cvnyonc to secure pllblic
of sUtl' :Ictioll
Wl' should lw ,Iware that the necessity and
iIlllllnii;lll'
intl-r
to secure universal contribution is hy no Illl"lIlS an
thl'
pursuit
ence ll'OI1l the Illere existence of a puhlic good

'1 he

Scc Anhur M, ()"llll, 1:'.1/1,l/il),

2, N,lIIlCd after h;lllels

(1 :->,1'1-1 ')2()) who depicted ,IIICfllall\'C
alJoclti()IlS of resource,. all" pmslhilnics f(lf COlli nets, ill this way,
,1, {)I COllI'''' If wc
ill ,'lIdmvlllclll dkcts (St,ctioll
alld Air and
a prl'tl-n'lIcl' to kl'C'!' wh,1I Ihcy alrt':ldy ILl'll', thcll they 111.11'
ILHlc, We Cll] IIO\\' l1l'ttl'r set' why l'lldowllll'1lt dt<:l'Is worry
he) In the very 1ll',lrt of

-L

allol'ative from

SUMMARY

P,lfer()

helH'lits

IS

.'\"U,t/
27,

I have tried to explain the Ilotion of ettlcil'IKY and

In this

its rcLltion to rationality, Tim eh,IDter h:ls:

WII)' II rrlliolldllVllSlfllICf lI'ill (O/lSllIlIC iI good "l' '0 Ihe

/ier 11I(J~~illdl
<"Ilidl Il('r 1II,1I;,(ill<l1 ((lsls, W c h,IVl' a).;o

/'\'1)/<11111'11

scen that :1 rational producer should produce lip to the point

whert' her marginal bCllefits l'qu,d hn

11'11)' 1',111011,11 (OIlSllIlIers Jl!illllldk1' ('.\'(lldllg1'.1 Illdl are

Ptn1'lo

illlprOIJ1'lIlCllts, The Edgl'worth Box \\,:IS expL1l1lcd.

considered the rcLltioll of the contract curve to

of Pareto superioriry ,wd

Explaillcd tile idell

COllsidered Il'ilellieY tlu' Pafet,) prillciplc is ,"lIililhir' <IS

tlH' 1I0lhlll

1111 extcrn,llif)',

pr"pcfl), 1'('(111.1

tlwr

PUJ'cto
,I IIl<lf'll

iilelll,

socd cosl-bl'lll'ilt

,111.1 ((}lIsitiered II'/lctl/(,/, 1111' id('al

illlcm,lli:c

,111

II/(, (0.11.1 dlld hCllef/ls "(

93

92

't

CHAPTER 3

in

1\. Rmsell Hardin makes much of the

illld S"ci!'I)!. Pl'. 10-11

Y.

thi~

See 1m

of;1 P;jn:rian LIberal." For an

extcnded, and ,](ccssiblc, disclIssion, sec Sell's,

amI

and Rights," p. 211i.

10. Sell,

Ibid., p. 2.1S.
12. See here H;llI~llI;1Il and McPhersoll,
Pl'. 71i-7') .

(I/Id

FCOI/Ollli(

iv/ora! I'hilos-

Game Theory

.1. See DCllnis Mucller, I'IiMic Choicc 1lI, p. 25.

of the (:Olll 11101 IS."

14. Sec (;;lIT<:t Hardin, "The

S. Sec \):lvid Schl11idtz, Thc LiIIIlIS ('f (;OI'I'I'IIIIICIII:

Coods
4,

1(,. John Stuart Mill, OJi

P;I[;1.

/111

0/1

the I'Jlhlic

.1.

7. Ibid .. Chapter 4, para. 14.

OVERVIEW

11\. Ibid., Chapter 5, para. J.

II our l'Xamillatioll of utility theory ill Chaptef 2 we f(,KlIscd Oil ~]
fatiollal agellt as olle who has prc(erellces over outcollles and a sd
of fixed adioll optiOIlS, alld who elll correiatl' outcollles with :lctioll
her nrl'it'n'lllTs O\,l"r outCOllll'S determined hCf
J wt' begall to consider how rational
also arc ch;lracterized by [he :lddition,ll
interact, alld eSpl'Ci;llly how ration;11
economic agellts wi\l l'nga!;!:l' in efficiellt trallsactions, 13m the
of rational intL'lactioll ill Chaptcr :'1 tlKtiSed
moves -roul!:hlv. cases in which ;1 !;!:l'!lts, interests
of cOI1t1ict. We
galll"s -a gl'IltTll tl1l'Ory of what
other ratioml agl,'nts. :lI1d especially wht'n wh;1t is best for
not be best "Jr AIr In this Ch:lptl'r I L'mploy ollly the gellnal idea of
individuals as utility lllaxillJizers: the specific, additional features th;lt
are n:'quirl'd fiJr HOIII" /;'wllollliOlS are not cClltral to this
(Again we Sl't' ""hy it is Sll illlportlllt to distmguish } [01110 f:'t'ollolllims
from r;ltiOllal utility maximizers ill ,
The ch;lpter begins with thc sllllpiest sort of g:II11C, a "zefO-<;lI III "
gamc ill which wh;ltevl'r one person \villS the other lusl's. This is a
game
pllfe ((lllt7ia. We then IllOVC on to the f~l1llOUS "I'risolll'r's
Dilemma" and other "variable-sulll" games' ,-soll1e will bc f()Cuscd
on cOIlt1ict, but we will also look at g:lI11l'S in which
will cooperate. It i'i illlportallt to realize that

I'). Scc ROIL]ld Coase, "Th .. ProbleIll of Social Cost."

tllliows Muclkr, Public Choice Ill. pp. 27-.10.

20. The Coasl' theorclll ,'Jll he applied to more' realistic cases, wlll're
activitl"s have variable costs-the Illore
pllrcha~c of
the extcl'Ilalilics. SCe' MIIl'IIl'!', PJlbI,C Choicc III,
ibid., pp. 1()Il:

:2

22.

Will

has a
not

2.1. But thIS i, !lot to say that govcl'Ill11clltal. COC1'civl' actlOlI is necessary. Sec
J)~tvld ScilIllidtz, '/1U' Lillli!s of (;'WCrJllIU'111.
~4.

nice OVLTvicw of tlIe cxtClll III wlm'h lIllhvidual \'01

Muclkr
t1mary dlOin's Will seClIrl' public goods, alld how this result:, III 111111er
1'1I/J1i( Ch"icc Ill, pp. IlilT. TIll' disClISSlOIl that t(Jllows draws ()Il

~5.

Sce ibid., p. 22. But see the

this part of Mudh'r.

of elK

;tSSULlllCl'

III

S"l'liOll

4 ..1.

2(,. Sec David Lyolls, Till'

'!( I Jrilildl'idlli,\lII, PI'. ()I)n~ 162f1: I

d"t;1I1 ill Illy SonoIIl/li/')sop/IY, PI'. lli2f}:

FOri/IS 1IIId Liltll'rs

havc discussed this case

ill mOlT

or

95

94

.....