Chapter Eleven - Twelve - Ok

CHAPTER
ELEVEN
LABOUR SUPPLY AND WAGE STRUCTURE
To have a complete theory of distribution, we need to explain not only the
demand for productive factors but also their supply. For the inputs to the economy
are not given. The land area of the world may be roughly unalterable but the
powers of the soil can be changed, depending on the rewards for doing so. And,
even if population is considered exogenous, the supply of labour is not- people
decide how much to work and what skills to acquire. In fact , both land and
labour are largely man-made, as, of course, is physical capital.
So we need to know on what terms the factors are supplied. This provides
us with the right framework for understanding the current distribution of income,
and for predicting how it will respond to exogenous changes or to new policies.
For example, suppose there were a technical change, like the discovery of the
computer, which raised the relative demand for graduates. How much would
existing graduates benefit ? If the supply of graduates was infinitely elastic, they
would not benefit at all. The more inelastic the supply, the more the benefit to
graduates. Or suppose that undergraduate education were more highly subsidised.
Who would gain? Not students, if the supply of students was infinitely elastic. So
a key question in this chapter (on labour) and the next (on capital) is: What
rorces determine the elasticities of factor supply?
There are many dimensions toa person's labour : his effort per hour, his hours
)l:r weck, hi s wccks per year, his length of working life, his skill, the area where
he works, 1 he j oh he chooscs, and so on. We shall concentrate first on the supply
or hours, showin ll ow simple utility thcory illuminates a whole range of behav-
pl1 "llllll\1' 11 :1 w,- tli L11 move to thc supply toan occupation, which provides
lil e spri11 glw:11d ln1 :111 :dysill tll ...: dclerminanl s or wagc slruclurc and carnings
i11 lpl :il ily l'1111111 y lVI" IIIIIH" to tli normali v tll"sli on: ll ow pro rcss iv...: should
1111' ll l'llllll' '" ' l11 '' \ , ,,,. SIT, tllis d p nds ' IIII' all y 1111 li tl\V ,l: 1slil is tll c
MICROECONOMIA 2 - DOCENTE: MARCO ARROYO - UNCP - HUANCAYO - 2014-1
304 FACTOR SUPPLY
labour supply response. The more inelastic the supply, the more consumer surplus
is available for taxation and redistribution.
11-1 THE SUPPLY OF HOURS
We begin by considering how an individual chooses to spend his time in a given
period. We shall ignore any long-term considerations that may affect his current
behaviour, and ask, first, how a change in the current real wage rate per hour
(w) will affect his choice of current working hours. Most people can vary their
working hours to sorne extent. They can either decide whether to work at all,
or for how long to work (either by their choice of job or within a given job ).
The Simple Theory of Time
In the simplest analysis we suppose a person values only two things: goods y
and leisure time T. But goods, let us assume, can only be bought out of wage
income, all of which is spent. If the individual faces a given constant real wage
per hour (w) and he works H hours a day, his income and expenditure per day are
y= wH
Hi s hours of work per day are 24 hours minus leisure per day:
H = 24----, T
So the problem
max u(y, T)
becomes
max u(wH, 24- H}
Differentiating by H and setting the result equal to zero, we obtain
U y W + U y(- J) =
At the optimum the individual equates the marginal benefit of work (uY w) to the
marginal cost due to the leisure foregone (ur ). If we express both benefit and cost
in terms of goods, we have
ur(wH, 24- H}
w= - ----
uy(wH, 24- H)
He equates the rate (w) at which goods can be transformcd inlo lcisun: to hi s
<>ubjective rate of substitution between goods and lcisurc.
The preceding equation givcs a n:lati on hctwccn thc w:1gc rat and IHlllrs
of work . ln ol hcr words, it rcprcscnts th suppl mvc. What sh:q> will thi s ll :vc'l
Suppose that in I,. igm 11 - 1 III L ri svs l'rotn 11 '
11
lo 11
1
, dii L' s: 1 IP :1 L' lll i11 a
,. ' ,. ,, , ( itWil llll' I:IX 'J'1i l' WP tk (' l IS Sid iJI'I' l (( IWP P!'1iilSIIIJ'. lllillll' IHTS :
Goods
(v)
S'
S
Real wage
(w)
\.-V2
LAfiOUR SUPPLY ANO WAGE STRUCTURE 305
Work (H) _ _ _ __ __ __L ___ _j
Figure ll-1 Supply of hours.
l. A effect ("it's worth working more now"), measured by RQ.
2. An mcome effe.ct ("you don't need to work so hard now"), measured by QP.
(We are assummg that leisure is a normal good.)
In our example, the substitution effect outweighs the income effect: the person
works
2
more. But, as we drawn the supply curve, there comes a time (for
w > w ) when the. reverse true and the supply curve is backward-bending.
Let us formahse all th1s. Suppose that we allow the individual to have sorne
unearned mcome Y as well as his wages. Then
y= wH +Y
The supply curve obtained from utility maximisation is now
H = H(w, Y)
? -
whcrc lhc signs rcfcr to thc partial derivatives. Now, if w rises, the total effect on
llours c: 1n. hy thc. 1ut sky dccomposition, be written as
(
l//)
rl 1v ,
'1
() J-I
11
rl Y
306 fACTOR SUPPL Y
If w rises by 1 (small) unit, the worker's " income" rises by H and the income
effect on bis labour supply is a fall, since oHjoY is negative (assuming leisure
normal).
The size of the substitution effect in labour supply is one of the oldest issues
in poltica! economy. Those who worry about incentives tend to imply that the
effect is large, while people on the left tend to imply it is small. But evidence is
not altogether lacking. As every child knows, male working hours have fallen
for the last hundred years. This suggests a backward-bending labour supply curve
for "prime-age males," and this is confirmed by cross-sectional and experimental
evidence.t In other words, for men the substitution effect is smaller than the
absolute size of the income effect. But at the same time as men's hours have
fallen, married women's working hours have been rising. This suggests for them
a rising supply curve, which is again confirmed by cross-sectional work. This can
be rationalised by saying that, given our social institutions, men have two chief
uses of time (market-work and leisure ), while women ha ve three (market-work,
home-work, and leisure). So when the wage rate for women's market-work changes,
there is more scope for substitution. And market goods (Iike dishwashers), which
can be paid for out of market earnings, may be quite good substitutes for
home-work.t
What does al! this imply for the "problem of incentives?" One might think
that if the supply curve is backward-bending, there is no problem: after al!,
heavier taxes make people work more. But this inference is a fallacy. For suppose
that in Figure 11-2a a tax reduces wages from w
0
to w
0
(1- t). The worker' s
choice point moves from A to B: he works harder. There will be no welfare
loss provided the gainers (the taxpayers) can compensate the losers. But can
they? To compensate the worker [while keeping his net wage at w
0
(1 - t )] the
taxpayers would need to give him an amount PQ. But if they did this, he would
move his choice point to Q and the taxes he paid would be PR. So the
taxpayers would only be receiving taxes of PR from which to effect compensation
PQ. So the excess burden of the tax is RQ. This does not mean the tax is
wrong. It may be needed for equity reasons or to finance public goods.
The only case where there is no excess burden occurs when there is no
substitution between leisure and goods (see Figure ll-2b). Then, if the worker
were compensated, he would consume at the original point A and the
compensation needed (P A) would equal the taxes paid. Thus in this special case
t See, for example, O. Ashenfelter and J. Heckman, "Estimating Labour Supply Functi ons," in
G . L. Cain and H. W. Watts (eds.), lncome Maintenance and Lahour Supply, Rand- McNally, Chicago,
1973; and C. V. Brown, E. Levin, and D. T. Ulph, "Estimates of Labour Hours Suppli cd by
Married Male Workers in Great Britain," Scottish Journal of Political Economv, vol. 23, 1111 261 277.
Novembcr 1976.
tOn all this sec J. Mincer, "Labour Force Participation ofWomcn." in H. G. Lcwis (cd.). !l xwcrx
of Lahour Economics, National Bureau of F.cono mi c Ncw York, \ 9(> .
Thc cxccss burdcn of a tax is thc additi on;d \u m11 su m inctliii L' whi h thc f(: iii L' rs frnn tl 1v t a x
( thc [:OVCI'I111 1CIIl) W() ll\d II L'Cd in () \'(\c l l ll h v l O C<l ll'[ ll' II S;I I L' lhl' \ (),SL' I S ( w \1 0 :1\C' pa y ll l): 1\ 11 [ ;\\ )
Thi s is ( )1\\ y <"!11ll i I P li1 l' S\1111 ni li ll' L' ll ll'[ l\' II S: \1111 }', Vi ll1 11 111ll1', 11' ' ' '\tll < : \ioll\1 1111' [ !I X w i ll' ll 1:< \
LABOUR SUPPLY ANO WAGE STRUCTURE 307
24w0 (1 - 1)
y

24
(a) s > O
Figure 11-2 The excess burden of an income tax.
)'
(b ) S= 0
an income tax is Iike a lump-su t . h .
take no evasive action. Looking
(?Y trtue of his tastes)

the elasticity of substitution (i.e., the less curved _Is that the gr)eater
greater the excess burden of the tax t C 1 m I erence curve ' the
d. . . onsequent y, the less the scope D
re Istnbutwn. If the elasticity of substitution i 1 . . . or
have markedly different values and th . s ow, different umts of leisure
ere IS more surplus t b d
final section we shall find that th 1 t. . f o e extracte , In our
. e e as ICity o substitu" b
affectmg the optimal rate of tax. IOn ecomes crucial m
;;h: curve of is backward-bending if the negat.ive income eff'ect
.f . wage outwezghs the positive substitution e_.rect . BL-tl
backward b d h !.!' even t_ it is
- en zng, t ere may reman a problem of incenves.
The fact that different peopl k d fli .
for the measurement of income hours serious problems
works 6 hours and B earns $20 and works o uppose $18 a day and
dis we must compare the

en owe With 24 hours a day but A t f .
hour, while B can only get $2 S A'can rans into goods at $3 per
have interpersonally comparable otTts set IS larger, and, unless we
A better off. u I I y unctwns or A and B, we must consider
So far we have assumed that leisure is alwa s a
so, as is shown by those people who o k f y It may not always be
(
. ti w r or ess m come than they co Id t
. wt 10 ul work in _' ) fr om thc social sccurity authorities. But utirt f t. u ge
.1 s th 1rs an :
1
stl h , lnndlc<_l Wt.tl . f 1 Y une wns such
' 1 1n our ramcwork.
1 11114 l ' t 11111 '' '1111 illld l l l p 111 ( 'lllt>l t' l 1
308 FACTOR SUPPL Y
Q 11-1 Draw lhe indifference map and opporlunily sel of someone who prefers working for a living
lo receiving lhe same income as unemploymenl rehef.
Qll-2 Suppose a Negative Income Tax wcre inlroduced by which each person was given $30 a week
and lhen laxcd al a rale of 25 percent on al! of his earning:;. Would the followmg people necessanly
reduce their work effort:
(i} Someone carning $100 a week when thcre were no taxes
(ii) Someone earning $160 a weck when lhere were no laxes
(Assume Jeisure a normal good, bolh here and henceforth Befare the N.l.T. there were no laxes and
no nonlabour income.)
Qll-3 Suppose a person has a utility function
u= y+ 2TL
2
where y is goods and T leisure.
(i} Derive his labour supply function.. . . . . ?
(
.. ) Whal is the mm1mum wage al wh1ch he IS wdlmg to work at al!.
11 . . 1 . ?
(iii} How will his supply of hours respond lo a small cut m a propor110na mcome tax
(iv) How will his supply of hours respond toan equiproporlional cut m a general value added
tax?
Qll-4 (i} Suppose a person has a job al $n per hour, where maximum hours (hul nol minimum
hours) are fixed by his employer. Would he be willing to take secondary work 1f 1t IS avaiiable at
less than $n per hour? . .
(ii) In many countries 1he old-age pension is reduced dallar for dallar 1f lhe p n s ~ o n r has
earnings. Suppose this rule were aholished. Would lhe Treasury neccssanly lose money' Assume
pensioners' only income is from the pension orfrom earnmgs. Assume that all earnmgs,_ olher lhan
those for which the pension is reduced, are subject lo a proporlwnal tax t. The pensiOn 1s tax-free.
(iii) Suppose an employer has been paying his workers $5 an hour for the firsl 40 hours per
week and $7.50 per hour for voluntary overtime. Average overtime IS 10 hours per week. The employer
also has lo pay $10 a week social security contribulion per employee (1rrespec11ve of hours worked}.
The workers now suggest aholishing overtime rates and replacmg them by a ftat rate system paymg
$5.50 per hour for each hour from zero upwards. Will the employer agree?
Qll-5 (i) Compare lhe same family al two points in time. In period 1 lhe husband can earn f4 per
hour and lhe wife f1 per hour. Now 1he husband's wage rises lo f5. WIII lhe w1fe work more
or less? . ,
(ii) Supposc now that in periods 1 and 2 the husband can earn f4 per hour but lhe w1fe s
wage rises from f 1 lo f2. Will she work more or less?
*Qil-6 If a person's Iabour supply curve is vertical and he has no unearned income, whal does
lhis lell us aboul his elasticity of substitution between goods and leisure?
Qll-7 (i} "If the supply curve of Iabour is backward-bending, leisure is a luxury." True or false? .
(ii) "If when his hourly wage rale rises, a man chooses so much more le1sure lhat h1 s
earnings fall, earnings are an inferior good." True or false?
Qll-8 "Turning to the markel for man-hours, suppose a tax is imposed on labour, the pricc to
the employers musl rise and so man-hours must fall." True or false?
A More General Theory of Time
The theory, as presented so far, seems to imply that pcoplc havc no l'cdin s
whatever about work as such. This can casily he rcmccli cd by wrtlmg
11 u( 1', : 11)
LABOUR SUPPLY AND WAGE STRUCTURE 309
The equilibrium conditions then include
The wage has to compensate for the (marginal) value of leisure lost minus any
value attached to working. This is obviously a more satisfactory approach, since
people are never concerned with the value of time as such (unless considering
the issues of birth and death); instead they are concerned with the clifferences
in value attaching to using time in different ways. From peoples' behaviour when
facing specifiecl alternatives, one should be able to infer the relative marginal
valuations which people put on different uses of time. For example, suppose that
it is possible to travel to work by two equally comfortable modes of transport,
one taking 10 minutes longer than the other but costing $0.10 less. For the
marginal person (the one who is currently indifferent between the two modes of
transport) the value oftime spent commuting must be $0.60 per hour less than the
value of time spent in sorne alternative way. All statements about the val u e of time
are of this form. Our first formulation of the work-leisure choice can be regarded
as a shorthand for the present approach, u(y, T, 24 - T) having been rewritten
as sorne other function u(y, T).
However, there is a weakness even in our present formulation. lt does not
enable us to consider the effect on labour supply of changes in the relative price
of goods. A more realistic and fruitful theory may be as follows. Utility ftows not
from goods or leisure, but from activities, each of which involves as inputs the
use of market goods and time. For example, going into the countryside involves
incurring the cost of transport (a market good) but also involves spending time.
Suppose that utility clepends only on two activities Z 1 ancl Z 2 . The first ( 1 ike going
into the country) is time-intensive in that the (fixed) time input per unit of output
(t) is high relative to the money value of the goods input per unit of output
(p 1a). The other activity (Iike eating) is less time-intensive. So
The individual's budgct constraint is
1 n labour supply analysis, it is often convenient to rewrite the budget equation
with thc va luc of thc incliviclual's endowment on the right-hand side- this is called
hi s .fitl/ inco11w. Doin g thi s, wc find
1( 1\ : ( >11 1 11 '11) is lhl I'IIII pr icc or / 1. S11ppnsc ll ()W lhal lhcrc is ; LIII in 1'1
l ' O IIlj )(' 11, ' : 1 1\' d h y :1 r: Ii 1111 r ,,() 1 h; 11 wl'i ; 1 I'l' is 1111 . 11 ; 111 n i. T ll t' 1 l' w i 11 lw :1 S 11 hs 1 i 1111 in ll
310 FACTOR SUPPLY
. . h n the amount of time devoted
from_ z2 to the time-.intenstve z , asked to prove this later.)
to relative price of goods inputs to
explanation of the secular decline of working hours m the
extensions of the analysis are referr_ing tzo
. . b th f e and goods mputs mto 1 an
if substitutwn ts allowed etweenb e dtm d A fall in p would lead to less time
z th bove effect could e re uce . 1 .
tOr 2' e a . d " d , Z Hence labour supply mtght not
and more goods bemg use to pro uce 1 . . . " d . , nd
f 11 depending on the relative elasticities of subshtutwn m pro uctwn ha
a ' . d not assume a constant wage rate per our
Second, hours and we can even allow for a person's
e:;ning His optimising solution remains pe:fectly
consumptwn o a e . omic theory becomes more reahsttc and
determnate and at the same ttme econ .
human.t for many problems simple theory ts adequate.
QU-9 Suppose that data were collected for a number of north of England towns showing .
x, trips to the Lake District National Park per head of population
M. money cost of getting to the edge of the Lake Dtstnct (f)
T, time spent getting to the edge of the Lake Dtstnct (mmutes)
A regression then showed that
x = a+ bM + cT
(i) What significancewould yo; in the National Park from
you expect cj b to be bigger or
, h 1 in poor areas than m nch areas. Y
Qll-10 (i) Queues in barbers s ops are onger .. l . their use of goods than North
(ii) Jndians (in India) are on average more econo,;ntca m
Americans and less economical in thetr use of ttme. Why. d f 11 . wi\1 induce
d
'b d the text an income-compensate a m Pt
*Qll-11 Prove that in the case escn e m h b d t l"nes in the space of Z , 2 2 in such a
. .
1
(t z + t z ) Hmt Draw t e u ge t . ( h
an m crease m etsure 1 1 2 2 . h h the original consumption pomt w ere
way that the new compensated budget \me goes t roug
consumption occurred before Pt fe\1).
11-2 THE SUPPLY TOAN OCCUPATION
ed that the individual will earn the same wage whatcver
We so far This is not generally the case. Suppose t.hal thcrc are two
occupatwn he prac tses. . . th e sk ill Hours ofwork are prescnhlxl hy 1 h.
occupations A and B requlflng e sam . . .
LAROUR SUPPLY ANO WAGE STRLiCT U RE 311
employer and are the same in both jobs. The annual real wages are wA and w8 ,
respectively. Which job will a person choose? Clearly the one which he finds the
more satisfactory. In other words he will weigh up the monetary characteristics
w and nonmonetary characteristics x of each job, and will choose A if
u(wA , xA) > u(w8, Xs)
This is simply the principie of maximising " net advantage."
Another way of describing his choice is to say that he will choose A provided
the monetary advantages of A over B more than outweigh any nonmonetary
advantages of B over A. Suppose A is, money apart, a lousy job, like being a
garbage collector, while B is an office job. Then for any given wg there will be
sorne value which is just sufficient to make a particular individual indifferent
between the two jobs:
xA) = u(wg, x8)
The difference is then a compensating or "equalising" differential, just
sufficient to compensate for the nastiness of job A. This differentialmight or might
not vary with w8 . It seems likely that it would. If the real wage in offiees doubled,
one would think that the wage needed to induce a person to eollect garbage
would at least double. For simplicity we shall assume that the absolute
differential which a person needs is always a fraction k of w8 . So k is the
proportional compensating differential, or what we shall henceforth call, quite
simply, the compensating differential.t Thus the individual chooses A, provided
or provided
WA- Ws k
- >
Ws
wA
-- > 1 +k
WB
But individuals differ in their attitudes to jobs. Sorne may even prefer garbage
collecting to office work. So we now have a separate compensating differential
k for each individual i. This measures his supply price to the occupation :
(
WA) = 1 + k
WB i
To obtain the supply curve to occupation A we take people in ascending order
of this compensal ing differential. The array of the ascending supply prices is the
suppl y curve. t\s we have drawn it in Figure 11 -3, there is one person who so
likes garb;1gc col lcctin ' thal he is willing lo do it provided wAf wB > 0.8. In other
l lll tll dii i!I I Y jllllll l !H I' 11 11 \ \ 111d " dllrf l l'll ll:ll '' !l l\l l l ' tl lll' tl ll'l'd IP 1\'1\'1 tn JHIIpiH!h'lli: d tll:111
,oh,o>lll i<' dl l lo'll ' lll io d,
312 FACTOR SUPPLY
Figure 11-3 The supply curve to occupation A.
A lo vers
words his k = - 0.2. As its wage rises he will be the first to en ter the
occupation. There are not many others who prefer A to B, and, if more than
four people are to be attracted into garbage collecting, it will have to have a
positive differential in its favour.
The more people differ in their attitudes, the more steeply the supply curve
will rise. So supply is less elastic, the greater the differences between people. This
is a perfectly general principie, and is analogous to the point that the supply of
hours is less elastic the more differently an individual values different units of
leisure.
The supply curve to an occupation is the array o.f the supply prices o.f the
different individuals, taken in ascending arder ofprice. The more individuals differ,
the more inelastic the supply curve.
The first principie has an important corollary. Market differentials are deter-
mined by supply and demand (see Figure 11-3 ).
The market wage differential is ther4ore the compensating differential oj' the
marginal man.
lf the supply curve is completely elastic, his differential will be the samc as t
compensating differential of everybody else. But if the supply curve is fairly
inelastic, dueto big differences between people, many of thosc in an
will be earning substantial rents, which have no allocativc functi on. could
be taxed away with littlc loss o f cfftcicncy, and possihly with gain in qttity.
That is why it is so import;tnl to know thc cl;tsti of stq pl y.
"' ..... , ." ' ' ' " " !":11 (ltti V ( lil l." (!ittll"II Si\lll aiong whk it ohs dilkt
(their intrinsic pleasantness) and one d. . .
evaluation ofthe pleasantness of . b ) along whtch people differ (their
other factors affecting occupatij:als h _e candnow extend _the analysis to include
e OJee an thus affectmg wage structure.
Qll-12 everybody agreed about the desirabilit .
dJfferentJal would necessarily be greater th . h h y of occupallon A relative lo B, the pay
an Wlt l e present dJversity of views." True or false ?
11-3 WAGE STRUCTURE AND HUMAN CAPITAL
The differences in income between eo 1 . . .
differences jobs, and partly fror!: dtfferent JObs arise partly from
ll_lany ways m which jobs differ in their at r people. There are
nsk, prospects, and training cost t t act!Ons m pleasantness of work,
o:iginal abilities, and in but a few. differ in tastes, in
dtffer, but jobs did then the lo (b other opportumties. If people did not
' ng-run ut not short run)
occupation would be infinitel 1 f S - y curves to each
"supply-determined" in the se y e ts IC: o long-run dtfferentials would be
price. Figure ll-4a ansmg from (uniform) differences in supply
. e so urce of wage diffe t . 1 b
occupattons A and B in such a world T . . ren Ja s etween two
allocates itself between the t .. here IS a gtven labour force No which
of A o ver B (k o) is that . the_ desired wage differential
and the poor, though unequal i. . nce IS I erenttal IS compensating, the rich
n mcome, are equally well off.
S
D
D
N;..
_ ____ L._ _____ N A
NB N8

(h) Rigid dirkrcnccs hc l wcc n pcopk :
NA N11 N::
\1' 11
/1' ' v,, N"
314 FACTOR SUPPLY
Real inequality arises from differences between people, especially in abilities
and opportunities. To illustrate this, Figure 11-4b gives a possible different account
of this same wage differential. In this case people are born so that they are either
able todo occupation A of them) or occupation B (Ng of them). Wages are
then "demand-determined" in the sen se that they reflect the demand prices for the
given inelastic supplies.
Economists vary widely in the importance they attach to "differences between
people" as determinants of wage inequality. In America such differences tend to
be given less weight than in Europe. If "all men are born equal," then wage
inequality can only be due to differences between jobs, and need cause no loss
of sleep. But if men are not born equal, then wage inequality may arise in part
from interpersonal differences and this, especially if the differences are in abilities
or opportunities, may be worrying. lnterestingly, those who believe that people
are not very dissimilar in abilities or opportunities often stress their differences
in tastes. But then differences in tastes are difficult to evaluate in terms of equity,
unless we have evidence on interpersonally comparable utilities.
However, despite differences of emphasis, all economists would agree that
differences between people do have sorne influence on wage structure. To see
how this works we shall take the various differences betweenjobs as our framework.
For each of these we shall ask what compensation is needed for the difference,
and how this compensation vares between people. In this context differences be-
tween people come in as factors affecting the elasticity of supply: the more the
differences between people, the less elastic the supply. We assume throughout that
everyone in a given occupation is paid the same. This is not the case in occupations
as conventionally defined. But we can make it so, by defining an occupation so
narrowly as to include only those doing similar work with similar skill m a
similar place, and consequently paid the same under perfect competition.
Pleasantness of Work, and Risk
Occupations differ, as we have noted, in pleasantness of work. If sorne people
would not do job A at any price and the rest would not do job B, then the
supply to each occupation is complete! y inelastic. By contrast, if tastes are similar,
supply curves are highly elastic.
Jobs also differ in riskiness. If job A adds to the income variance of each
individual more than B does, then, if the expected wage w A is lower than the
expected wage W
8
, only risk lovers will en ter occupation A. lf most people are
risk-averse, the average wage in A will be higher than in B, unless thc dcmand
for A is very low (see Figure 11-S).t The more people differ in their atliludcs lo
risk, the less elastic the supply curve.
t We continuc to keep N
1
.JN
11
on thc ct xis sincc this is t hc relevan! IP (krn:11rd. lf 11 all(l
H ctr thc onl y occupations, ;r s NA riscs. N11 fa lis hy lil e samc anr ollll l. Tir e wnrds wr II L' ll ;rl n11 g
thc :txis rl'ia te tu tl ll' cha racll' t pf thosc ind ivtd t1!il \ wlu1 Vlll t' t l)l'l' t qmt roll !\ 11 1 ti H: cn tll'."!hllld ing
!.O
S
D
NA
\:::==::;:======:::::====,.---+- NB
Risk Iovers Risk ha ters
Figure 11-5 Risk-prefcrence and occupational
differentials.
Training Costs and Prospective Returns
Sorne jobs require more training than others. Consider first the case of full-time
education. job A needs one more year of education than job B, and
that m JOb A earns the same, irrespective of age, and will continue
to_ do so m Likewise for job B. There are no tuition fees. Then people
wlll be wilhng _to supply themselves in job A if they are compensated for
the earn_mgs (ws) wh1ch they lose during their last year of education. lf the capital
perfect, the compensation must be such that the present value of the
hfetJme mcome a person can obtain in each occupation is the same. If he enters
occupation B, he i_s paid a (virtual) perpetuity of w8 ; if he enters occupation A,
he a perpetmty of wA but starting one year later (see Figure 11-6). So
1f the Jobs are equally pleasant and risky, he will only en ter A if
PV (occupation A) PV (occupation B)
Thus > Wu
i l + i - i
\ V A , WB

H r--r------------------18
316 FACTOR SUPPLY
. in that everyone is equally capable of
where i is the mterest rate.t g '1 h resent values are equahsed. So
for A, people will move from B mto A untl t e p
in equilibrium
WA- Wn = i Wn
. . t enough to pay the interest on the cost
The differential differentials
(wa) of gammg access to 1 . n
S
o i is the compensating differential. t f . . g 'or J. ob A capital markets
. 11 apable o tramm 1' '
Thus if everybody IS equa y e 'd d qually pleasant and risky, the wage
b A d B are cons1 ere e . F.
are perfe<;t and JO s an . . coro letely supply-determined, as m .gure
differential between the _two JObs IS it i:a compensating differential, but yew_ed
ll-4a. Viewed from the S!de of supply, ductivity of the worker Wlth
from the side of demand it reflects the greater pro
more " human capital." . . eds in exactly the same way. Suppose
The analysis of on-the-job trammg f 11 wed by a year's apprenticeship
th same educatwn, o o ) h
two jobs A and B reqUire e . h. d In J. ob A (say accountancy t e
1 mpetence 1s ac 1eve b B
before full professwna co ntributes little to current output. But m JO .
apprentlce, though learnmg a co .t roductively employed. Assummg
(say brick laying), the apprentlces are p ducts (w* and wt respectively,
that the apprentices are paid_ thelr_ be required the adult
'th * < w*) a compensatmg dfferentm Wl
Wl WA B'
wages w A and Ws su eh that
wA - w8 = w!)
. t value of lifetime incomes
The compensating differential agam equates the presen
in the occupations. . com are the supply prices to the two
The same analysls can_ be used to prospects (wA and wa) Someone
apprenticeships, each of whlch open up . 1 A o long as the current earnings he
is only willing to enter apprentce_shlph_m . are not greater than the present
foregoes by not doing an apprentlces lp m
S
s; : ow has thc samc va luc as SI pcr
. s Si per year for ever. o
1 11
t s now has the same va u e a
1
e as S>w per year for e ver.
Y
ear for ever, and $11's/ i now has the same lvda u . s
. . r F we shou req Utre
+ lf there were tut!lon ,ees
+
11_, - "'n = i(lln + F)
11 ' \
H n
value of the earnings differential (wA- w8). So in equilibrium,t we have
* * - (WA- Ws)
Ws- WA- .

So long as a person's training affects his marginal product equally whatever
employer he works for, his wage will always equal his marginal product
(assuming perfect capital and insurance markets). The employer has no incentive
to finance the training. For once trained, the worker will leave if he is paid less
than his marginal product, since he could earn that elsewhere. There is therefore
no inducement for the employer to pay more than the marginal product during
the period of training. This type of training has been called "general," as opposed
to "specific" training, which raises a person's productivity more if he works
for his present employer than for another.!
Specific training may or may not be financed by employers. The employer
may be willing to finance it, since, once the training is done, he will only have
to pay the worker his marginal product as it would have been without the
training. He can thus trap the returns to training by paying the trained worker
less than his actual marginal product. There is, however, the danger that the
trained roan may leave, which could lead his employer to make the worker
finance the training (as in general training). On the other hand, the worker
faces the danger that the employer may sack him, in which case he could not
reap the returns to his training, since these returns are specific to the work
done in the firm. Exactly who bears the costs of specific training remains an
unsettled issue, but, insofar as employees do so, workers will not be paid their
current marginal product-an important exception to the general presumption
that they are, and one which we shall ignore from nr-w on.
t We might appear to ha ve reached a circular result he re: adult wages are determined by apprenti cc-
ship wages and vice versa. But we are not saying this. For we ha ve not so fa r determined the absolutc
leve\ of the wages- only their ratio. To determine the absolute wages we would have to bring in
demand. Suppose for simplicity a production function
where * indicates apprentice, and also that

a nd
Thcn il'
r,
,, 1\'/ , 11.11'1'' " ,. d !'lt' lllllll ! ' d
1 1; " 111,,, // illll</111 111/1 11/ Noi t i!Hi i dll ""' """l l ,,.l illlllll !( , .... ., , ,, Nn1 \''"" 1'111 ,1,q1 '
318 FACTOR SUPPLY
\VA
wu
t
l.O L-- ------- ----..,. N A
Children of poor Na
Children of rich
Opportunity and Ability
Figure 11-7 " Opportunities" and occupational
differentials.
So far we have made the artificial assumption that everybody has the same
opportunities and is equally able. A key source of unequal opportunity is capital-
market imperfections. Not everybody can borrow at the same rate and sorne
cannot borrow at al!. (Those who cannot borrow at al! will be using their
subjective rate of discount to derive their equalising differences, while those
sufficiently affluent to be lenders will finance their human capital by lending less
and so will use their lending rate, see Chapter 12.) So the children of the rich,
who can borrow cheaply or finance themselves from diverted family investments,
will use a low discount rate to compute their equalising differences, while the
children of the poor, other things equal, will use a higher discount rate. So,
taking the case where job A requires one more year of education, the supply price
of individual i to occupation A is given by
where r is his marginal discount rate. The higher r, the higher the supply
price. If workers are arrayed in order of their supply prices, this gives us a
rising supply price (see Figure 11-7). The more the variation in r, the less elastic
the supply curve. Capital-market imperfections clearly help to explain why, despite
the supposed nonmonetary returns to education, real prvate financia! rates of
return to education have been higher than the real rates of interest at which most
people can lend.
People also differ in their abilities. Those who are less able will have to
undertake more training to be able to perform a given job. Suppose for simpl icit y
that once people have qualified for occupation B they need to spend diffcrcnt
lengths of time in training to fit themselves for occupation A. The ith individual
needs to spend n years training. Provided n is small, the cost to him of his lraini ng
is n w
13
So his suppl y price is givcn by
\\ '"
1 i '. , ,
l. O

Geniuses ldiots
NB
Figure 11-8 Ability and occupat iona l differen-
tials.
The_ is again l_ess the more ability vares (see Figure 11-8). In
eqmhbnum the ab_le w!ll be earnmg rents. If ability could be directly measured it
be wtthout misallocation of resources. But any tax
earmngs wtll have sorne allocat!Ve effect, as well as extracting rents.
In an older book the limited supply of persons for many skilled occupations
ha ve been treated under the heading " noncompeting groups." However this
Is not altogether satisfactory, since there is some price at which a person from
a v_ery home may be attracted to, say, medicine, and some price at which a
qmte be able to acquire skills. However, the old distinction
_dtfferences" and "noncompeting groups " is in a sen se pre-
served m our distmctwn between differences between jobs and differences between
people m the terms on which they are willing to do the jobs. The whole complex
of wage structure cannot be explained without attending to both of these aspects
as well as to demand. '
a?dition, of course, there are many labour markets which are not in Iong-run
eqmhbnum, and many where rates are determined by nonmarket forces. Sorne
are up_ by entry to these professions. Others are raised by
actlVlty. _It IS for example, that in the United Sta tes unionisat ion
the wages_ m the unionised sector by, on average, about 15 percent.
IS a fatrly small differential compared to the full spread of
?Ifferentials m _the labour force. To explain this latter, the preceding framework
IS the most frmtful we have.
Q_l_l - 13 '' As only have to pay unskilled people low wages and pay no more t ha n they
nc"d lo pay, t hrs s hows that low pay has nothing to d o with low product ivit y. " True, false, or
unccrtain ? G ivc rcasons.
Q ll- 14 S11pposc lhcrc w:rs :r pcrfcct ca pi tal market with a 10 percent interest rate a ll 1
e 11 1 1 . . 1 . . . . peop e were
" Ud Y .1) " l 11 " 1111n w:rs re" n i 111 111011 cha rgcs 1nd had no psychc 1 d 11
. , . . .. . . . . . ' 1 va ue, a n a peop e wllh
.1 W< I L" p111d l l1e of agc and ca lendar time
(1) Sll ill'"' '' Wl' " " '''1 vr fro1n ;r two-y"1r corr rsc clrn 11g 1ft 1 11x $? 10() . . , , , . - pcr annun1
ll 11w t i 11H' Ii lllllSI hl c: 11 ni ng ( : ll't cr ta x)?
( 11) 1 III'P" 1 1111 11 P lll't' IIII IIU' llll!ll' lt,' l\"'1,' i 11 t!J v fn 1 p,1:HI II :tf c" Wdl 1111s i 11c n:: 1sc
1 1 " ' 1 1 - 1' 11 1 t 1 11 1 1 1 ., 1 1111' , ltl ! 11 '1
320 FACTOR SUPPLY
(iii) Now allow for differences in opportunity or ability. Will increased demand for graduates
in crease graduate earnings in the short run? In the long run ?
Qll-15 " Since university education generales socia lly productive human capital and student time is an
important input to university education, students should be paid a wage." True, false, or uncertain ?
Qll-16 Suppose a ll public subsidies to education beyond age 18 were abolished ; what would be
the effect on the inequality of personal incomes ?
The Interpersonal Distribution of Earnings
We have so far concentrated on how the pattern of occupational wages is
determined. But the interpersonal distribution of earnings depend also on how
many people there are in each occupation, i.e., earning each wage. To think
about this we move to a slightly different, though related, framework.
In equilibrium each individual will have a set of productive characteristics
which have been priced by the market (by the operation of supply and demand).
But if two individuals with identical productive characteristics are working in
different occupations, the one in the occupation that is less desirable (to the
marginal person) will be more highly paid. So job characteristics also get priced
in the market. Thus the wage of the ith individual will depend on his own
productive characteristics (q) and on the nonmonetary attributes of his job (x;).
w; = f(q, x;)
This relation is known as a wage function. It is fundamentally a descriptive
relation, sin ce it is the reduced forro of the system of supply and demand functions
and does not enable us to identify either of them.
However, the theory of occupational supply does give us sorne clues to the
appropriate forro for the wage function. For suppose we take the simple
schooling model developed above. This indicates that in a perfect capital market,
where relative wages are determined entirely on the supply side,
w1 = w0 (1 + r)
where w; is the wage of someone with i years of education and r is the interest
rate. So, for someone with s years of schooling,
W
5
= Wo(l + r) ="= Wo ers
Thus
log w = log w0 + rs
Taking variances on both sides and assuming w0 and r constant, we find that
var (log w) = r
2
var(s)
The proportional inequality of wages (measured by the vari ancc of thc logs ) is r
2
times the variance of years of schooling. This simple modcl has bccn found to
explain a part of the interstate and intercountry vari ati on in incqualit y. But
within any onc country it cx plains onl y a small proporti on of in qualit .
Obv iously som of oth -r fa tors w ha v 1nntion d 11 d ((\ h hrPuglit in:
:. .. - .... .. .. .. ,. .. .. , ! . ., ., ni', .,,.,,., ,. wnrk 'X Il ' l i 11 . M11 l' li 1\'Sl ': ll h is 11 nw
in progress. on the.se. but clearly if we knew the function f (q, , x,)
together With the JOIOt d1stnbution of q, and x,, we should have a complete
description of the sources of wage inequality. Theory provides important
clues in this work. t
Qll-17 " lflog w, bs, (for all i) and there is a perfect capital market with interest rate r = b,
there 1s no mequahty between people m the present value of their lifetime earnings." True or false ?
11-4 THE OPTIMAL INCOME TAX
Peopl.e have been in labour supply, because they have always
been mterested m rehef. Those who were against liberal relief argued that
handouts would d1scourage the poor from working, while the taxes needed to
finance the. handouts would discourage the better-off from working. Those in
favour of relief believed these fears were exaggerated. The arguments,
though voc1ferous, were not always very precisely formulated. But much effort
has been recently to developing a systematic approach to the question
of the optimal degree of redistribution, taking into account labour supply
responses. t
. In . thi.nking about this we shall confine ourselves to the effects of
red1stnbut10n on the supply of hours, though similar lines of argument apply to
the effects u pon skill acquisition, and so on. We take each individual's gross
hourly wage W; as This implies that his marginal product is independent
of the lab.our by others: there is perfect substitution in production.
So, assummg the mdividual has no private unearned income, his gross income is
WH .
Theequity that people differ in their productivities w : productivity
frequency f(w). !o deal with this problem and to raise money
for I.ts own expend1ture on pubhc goods, the government introduces a tax-
subsidy :-vhich specifies the net income y for each level of gross income
wH. For Simphcity we shall assume that this function is
Yi = WH(l- t) + y (1)
The individual pays taxes at a rate t on all his income, but receives a tax-free
handout Y. This scheme is equivalent to an income tax with a constant marginal
. '1' For cxample, it points to problems that may arise if ability and family background are
111 cludcd Sllllllll a11 cously 111 f see G. S Becker Human Cap1'tal 2d ed Nat'o 1 B fE . , , , . 1 na ureau o conom1c
lkscarch, Ncw York . 1975. pp. 11 7- 146.
t Thc semina l. t hough d ifTicult . art icle is tha t of 1 A Mirrlees " An Ex 1 t. h Th . . __ . . . . p ora IOn m t e eory
ol O rtlllllllll I IICOIIH.: 1 l<cllicw ol Economic Studies, vol. 38. pp. 175- 208, April 197 1.
l:or :111 :1 11 :!1 sis on dkcts on cduca ti ona l choiccs scc A B At k. " H
1
. . . . 111 5011 , OW
1 11>>\1 \',,slv< Sli 11 11l il lm'l""'' '1'11 x lk 'l " i11 M P1rk i11 ( d) :"""' M . L
l t i i Hi t nl . Jl ) f \ ,, , . , "' , 11 0(1' 111 ongman,
, , Mlll il'l' ' 11111111 1 111 111 hl 'l Pl'lllll lli " 'i>l' il iik lVI I ' IH I\ '1 11
11
322 FACTOR SUPPLY
tax rate levied on all in come above an exemption limit (Y /t) and a corresponding
subsidy (Negative Income Tax) paid to those with incomes below the exemption
limit.
The government's problem is to find the optimal t and Y. Obviously the
citizens would like as low a t and as high a Y as possible. But the government
is constrained by the need to balance the books. So, if R is expenditure on
public goods, and there are Q members of society, then
Q
L tw H = Q Y + R
(2)
1
The government is also constrained by the fact that the tax-subsidy scheme will
affect hours of work. If all individuals have the same uti lity function
u= u(y, H;)
they will have the same labour supply function
H = H[w(l- t), Y]
(3)
(4)
Given these four constraints (1) to (4), the government then selects t and Y to
maximise its welfare function:
W = W (u , .. . , U , ... , UQ)
(5)
The solution to this problem clearly depends on the structure of the utility
and welfare functions, as well as on the degree of underlying inequality in
abilities and on the government revenue needed. Calculations by Sternt confirm
one's intuitions that the marginal tax rate should generally be higher :
l. The less elastic the labour supply response (i.e., the lower the elasticity of
substitution in the utility function)
2. The stronger one's inequality aversion (i.e., the lower the elasticity of sub-
stitution in the welfare function)
3. The wider the dispersion of abilities, and
4. The higher the government's revenue requirements
Let us look at these features more closely. We shall suppose that the (cardinal)
utility function is of the constant-elasticity-of-substitution variety :
v > O p > - 1 p =f. O
Here v measures returns to scale in producing utility. The elasticity of substitut ion
is 1/( 1 + p ). The lower the elasticity of substitution, the less elastic thc supply of
labour. United States estimates of male labour supply imply an clast ct y or
substitution of about 0.4.
i" N. H. Stern, "On thc Spccifi cat ion of Modc ls nf Opt i"' ' "" .1 ""m u/ "f/'"1>/ic
vo L 6. pp. 12:1 l 2. 1'>76 ; and N. 11. S! clll , " WI.' If:11 c Wciglil s a11d il ll l ' l:i sli c ily o f lil l' M:tl',i ll :t l
V:d un 1io1l o!' 111 ' 0 111 1,:, " 111 M /\ 111 ...: !111d 1( Noh:1y (rd .... ) S t/1(// '' 111 f\ f ut!t ' lll l : t 'lllltll ll lt ' l t tul \'\ 1.\,
l!l n,kw. ll lhltlld 11) // 11 1> 11111
Next we need to make a value judgment about the social welfare function.
Suppose it has the formt
1
W = - + + a ::::; 1 a =f. O
a
Clearly a is a measure of inequality aversion: the smaller it is, the more the
aversion to inequality. lt is commonly assumed that the product av = - l. Using
these parameter values (for p and av) together with reasonable estimates ofj(w)
and R for the U.K., Stern calculates, for illustration, an optimal marginal tax rate
of just over 50 percent and a mnimum income guarantee Y equal to one-third
of average income. These figures are not offered as prescriptions, but as
contributions to the developing attempt to incorporate equity considerations in
a consistent way into economic policy discussion.
The figures have an important corollary. Even when the optimal income tax
is levied, a great deal of inequality is left. Suppose we calcula te the marginal social
value of a lump-sum $1 given to people at different levels of net income; i.e., we
compute for a person i at each income leve!
awau
-a -;;- (y, H;)
u oy
If we take a person at the 15th percentile of ability (i .e., a fairly poor person),
his dollar is worth about 21 times the dollar of a person at the 85th percentile.
So difficult is it to achieve equality, when taxes induce unfavourable labour
supply responses (even with an elasticity of substitution as low a 0.4).
Even if our welfare function were of the Rawlsian maximin variety, it would
not be in the interests of the poorest man that complete equality were brought
about: higher taxes would so reduce the total available for distribution. Only in one
case would complete equality be desirable: when the elasticity of substitution in
the {CES) utility function is zero.
This can be shown by taking the case of two workers with different
productivities (w A < w8 ) . If we assume v = 1 and zero elasticity of substitution,
the CES function becomes, in the limit,
. (y T)
u= mm
So consumpt ion will always occur along a ray from the ongm. Suppose that
initially there is a marginal tax rate less than 100 percent. So A is less well off
than B; their utilities are marked as uA
0
and u
80
at the corresponding consumption
points in Figure 11 -9. Now suppose the marginal tax rate was raised and so was
thc handout, in such a way that uA was raised to
1 = 1-(uAO + UBO)
As bcl'orc, A ami B log lh r consume the whole national product. So where must
B hc , ., 11 loo was consumi ng al 11'"
1
. Thcn hi s consumpton
1
1
324 FACTOR SUPPLY
\ll BO
\
\
\
\
\
\
a
Slopc =;
\
\
\
\
\
\
\
24
Figure 11-9 Redistribution with zero elasticity
of substitution.
of y wou!d have fallen by the same amount that A's consumption had risen.
So national output must be the same as before. But this cannot be so. For B is
more productive than A and has increased his hours by as much as A has
reduced his. So total output must have gone up. Therefore B cannot consume
at Lt'" 1. He must consume more, e.g., at u
81
as we ha ve marked it. Does (uA
1
, u
81
)
give more welfare than (uA0, u
80
)? Yes, using any symmetrical quasi-concave
welfare function. For if both were consuming at uAI, this must be at least as
good as (uA0, uB
0
) . And (uA
1
, u
81
) is better.
So, by raising the tax rate and the handout once, we have improved things.
It follows that we can do so again and again, until eventually we reach a
marginal tax rate that approaches 100 percent in the limit. And at that limit we
reach a state of complete equality. The argument clearly does not depend on
having a linear tax schedule, and generalises to an economy of more than two
people.t
However, unfortunately there is a substitution response. Sorne of this problem
can of course be overcome by partially relying on commodity taxes, and taxing
more heavily those goods which are complements to leisure (see Chapter 6). But
a system in which all taxes were levied on commodities at a rate independent
of the customer's means would be unlikely to satisfy one's equity objectives. So
there remains a crucial role for income taxation, and in principie the rates of
al! taxes should be simultaneously determined.
Within any framework, emprica! estimates of labour supply are of central
importance to the makers of tax policy. They are needed not only for men, who
t For a more rigorous proof, sce N. H. Stern, "On the Specification of Modcls ol O pi imum Tax-
ation," Journal o{Public Economics, vol. 6, pp. 123 162, 1976. Note a !so thal il 111ilil y is al
decreasing returns to sea le (v < 1 ). lhc rcsult rcfcrrcd lO is slrcll glh.;ncd.
LABOUR SUPPLY AND WAGE STRUCTL!RE 325
as receive most attention, but for women. And we want them not only in
to the supply of hours but in relation to effort, occupational choice,
mtgratwn, and all the other dimensions of labour supply. Where other policies
(as on education) have labour supply effects, they too should be analysed in the
light of these. lf this is not done, statements about gains and losses from
government policy can be grossly misleading.
CHAPTER
TWELVE
CAPITAL AND INTEREST
The capital assets of any economy, including its on. its
past pattern of savings and investment. In each penod every md1v1dual dec1des
how much to save and how much productive investment to make. !f .the
capital market (loans market) is perfect, everyo?e and lend u?hmited
amounts at a given rate of interest. So the mdJvidual s plan savmg and
investment will depend on the rate of interest, bis product1ve mvestment
opportunities, and his preferences. Once he .has how much to save and
invest, he must (assuming no change m h1s holdmg of money) len.d out any
excess of savings over investment or borrow to finance any excess of
over savings. The market supply and demand for loans are thus a funct1on of the
rate of interest, and the rate is determined so as to equate supply and demand.
Thus we need first to look at individual behaviour (in Secs. 12-1 to 12-3) before
coming to the market (in Sec. 12-4). . .
This in essence is the neoclassical theory of the rate of mterest, p10neered by
Irving Fisher, which is the centrepiece of this is in fact, extremely
similar to the single-period model of general equ1hbnum, w1th only two
modifications: (1} all commodities are dated, and (2} there is a markct for
temporal claims. For convenience, we shall general! y assumc lhc1 e 1s _o n
consumption commodity in each period, but the theory can casi! y be gcnc1 a hscd
to more than one. . . . .
The assumption of perfect capital markets is obviously artd icial. In pra\.: (1 \.:C
people cannot borrow unlimited amounts al a g1vcn ral c of 111tcrcst. hccausc th
CAPITAL AND INTEREST 327
more they borrow the greater the risk of default. So in Secs. 12-1 to 12-3, which
concern individual choice, we allow for the possibility of capital-market im-
perfections. But in Sec. 12-4, where we move to the market, we confine ourselves
to the determinants of "the" rate of interest, rather than of a schedule of oppor-
tunities facing each individual.
Neoclassical capital theory has been much criticised, especially from
Cambridge, England. In the final section we examine these criticisms and attempt
to show which points in the critique are valid.
12-1 THE INDIVIDUAL: TWO-PERIOD CASE,
CONVEX PRODUCTION SET
We begin by considering a person who lives only two periods (periods O and 1)
and has to choose from a number of mutually "independent" investment projects.
(If two projects are independent, this means that one of them can be done,
whether or not the other one is.) There are three elements involved in any person's
choices over time:
l. His productive investment opportunities
2. His borrowing-lending opportunities
3. His preferences
To simplify matters we shall first considera hermit (Robinson Crusoe?) for
whom there are no borrowing-lending opportunities (2), and then a ne'er-do-well
for whom there are no production opportunities (1}, before bringing all three
elements together.
No Borrowing-Lending Opportunities
The individual's investment opportunities consist of a set of projects, each of
which involves a sacrifice of current consumption (c0 ) and an increase in future
consumption (e). So a project can be represented by a vector (Ac0 , Ac) which
when charted in consumption space corresponds to a move northwest from, say,
E to E' (Figure 12-1 ).
The rate of return (p) on the project corresponds to the ratio of the return
lo the cost and is given by
Ac 1 return
- Ac0 cost
return - cost
- - --- +l=p+l
cost
lla vin g fl l'or c;1ch pro jcct, and provided projects are not only indepen-
dcn t hu t sma 11 , ( ' n1 soc rcsolvcs t o considcr projects in order of their rate of return.
St1!)Jh1SL' tli :tt ir II L' d1d I W pro jccts (just gathcrcd bcrri cs ), hi s output of
l' llll S11111p11 011 :<HH I-. 1\1<11 ild 1 1' :11 cndow111 Cllt po int in Fi g11r 12-2. lf'hc did
tlt r Jl1 on t \V IIIi tl1 1' 111 1'.111'. 1 1 ( pl :1111 i11 1: 111' \' Illlid Jll lll hl \ '\' :11 / ;" . t\ 11d S1l
328 FACTOR Sl:PPLY
C
L__ _ _ ___ ____ c
0
Figure 12-1 An invcstment projcct.
on. By adding in more and more projects in order of their rates of return, he
traces out a convex production frontier rcaching northwest from E. The slope of
the curve at any point is 1 + p, where p is now the rate of return on the marginal
project.
Since Crusoe is isolated from the capital market, his production frontier is
also his consumption frontier. His problem is thus :
max u(c0 , e)
such that
'( E E) 0
j Co-C, C - C =
(production constraint)
where (cg, cD is the consumption vector at E and f represents his investment
opportunities. At his optimum, using our usual notation for derivatives,
.t uo
ft U
In geometrical terms, his optimum is at a tangency between his production
frontier and an indifferencc curve. The absolute slope of the production frontier
is .f
0
/f
1
= p + 1, as we have already seen. The absolute slope of the indifference
C
Production frontier (MRT = 1 + p)
'"' 'u
CAPITAL ANO INTEREST 329
curve is
dc1 u0 u0 -u1
- - = = -- - +1=r + 1
dc0 u1 u1
where r is the " rate of time preference," sin ce it meas u res the proportion by which
the value of consumption today exceeds that of consumption tomorrow. In
equilibrium therefore r =p. At less investment than the optimum, we have p > r,
but as more investment is done p falls and r rises till equality is reached. (The
reader could represent this by charting p and r against accumulated investment.)
At the optimal production and consumption point e, Crusoe has determined
not only his rate of saving and investment (cg - c, as a fraction ofhis total income
of cg), but also the rate of growth ofhis economy [(cj/cg) - 1] and his equilibrium
rate of time preference (r* = the absolute slope of the indifference curve at e,
minus 1). Once we know the (ex-post) discount rate (r*) we could equally well
describe Crusoe's problem as being to undertake all projects which have positive
present val u es when fu tu re flows are discounted at r*. The "present value (PV)
rule" is then: do all projects for which

PV = + --
1
- > O
l + r*
A little manipulation will show that this "present val u e rule" is equivalent to the
"rate of return rule": do all projects for which
p > r*
where

p =-- - 1

However, we shall shortly show the inadequacy of the rate of return rule when
production frontiers are nonconvex or when consumption in more than two
periods is at stake.
Except when there are capital-rnarket irnper.fections, the present value rule is
invariahly correct; though the rate of return rule gives the sarne answer in the
two-period convex-production possibilities case, it has no general validity.
Q12-l lf Crusoe does no projcct, he produces 110 units in each year. His utility function is u= c0 c 1.
How much will he invest/save in the first year, and how fast wi ll his economy grow
(i) lf he has a very large number of projects open to him, all with 10 percent rates of return''
(i i) lf !he available projects havc the following costs and ret urns?
i\co l'. c ,
11
11 1()
( ' 1. 11
330 FACTOR SUPPLY
C ,Y
e*
1
1
1
1
1
1
1
1
1
1
1
----j------
1
1
1
1
1
1
1
1
1
1
1
O '------'c;j ____ ----""v--eo, Yo
Figure 12-3 Perfect capital markets: consump-
ti on loans only.
No Productive lnvestment Opportunities
Now consider a person with no opportunities for productive investment, but
with opportunities for borrowing and lending at a constant interest rate (i). (Note
that the acquisition of financia! assets is not an investment.) The individual earns
his income (yg, if) at E in Figure 12-3, but he may wish to exp!oit his market
opportunities to choose a different consumption point. Any !oans he makes in
periodO must be repaid with interest in period 1 (borrowing being merely negative
lending, and repayment to others being negative repayment by others ). So bis
capital-market constraint is
Lending x (1 + i) = repayment by others
Hence
(yg- c0 )(1 + i) = C -A
So
E
F Yt C
Yo + --. = Co + -- -.
1 + 1 1 + 1
Thus
PV of income = PV of consumption = distance OV
This is the equation of his market opportunity !in e. lts absolute slope is 1 + i ami
it passes through both E and V. The crucial point for what follows, howcvcr, is
that a person's consumption possibility frontier can, in a perfcct capital mark ct,
be defined simply by its slope i and its intercept on thc hori zontal axis; in
other words, by the interest rate and the person 's prcscnt valuc (scc Fi gun.: 1 ).
Thc individual now maximiscs utility subj cct to hi s capital -m:1rkd conslr:,inl ,
hy again cquatin g hi s rat c o l' suhi c ti v lim prl'i'l'fL' II T l!l tli ohj l ti vL r:11l' pi'
,. . .,, , . Ilion :1v: 1iahk l<l i1n1 , in ll1i .' l ' : l. \ ' tll \' IIII L' I l ' 1 l!ll l' 1 knn r 1. 111
the he to consu_me at point e, and !ends (yg - c) (a pure
consumptwn loan) m penod O and 1s repaid (ro- c)(l + i) in period 1.
Q12-2 Suppose the individual has an income 110 in each period and can borrowj lend at 10 ercent
and u= c0 c1 . P
(i) What will he consume in the two periods?
(1i) How much will he borrow/ lend in the first period and repay/ be repaid in the second?
(m) What IS the PY of 111S mcome? Of his consumption'l
Productive lnvestment Opportunities and Borrowing-Lending Opportunities
We now ?ring together oppo_rtunities for productive investment and for borrowing
and lendmg. In a perfect capital market, such as we ha ve assumed, the individual's
problem can be solved in two sequential steps (in Figure 12-4).t
C , )'
Figure 12-4 Perfect capital markets, with variable
production.
This can also be seen from a glance at the mathematical structure of the problem:
max u(c0 , c1 )
s uch th:It
C )i l
. = Yo +
1 1 + i
capital-market constraint Co 1
a mi proclucti on constraint
11
ll
l.
[1
1
332 FACTOR SUPPLY
. ( ) 1 . h . . ses the present val u e of the stream
l. Choose the production pomt y w 11C . . O V) This guarantees that
of in come (net of investment outlays )(Le., maxlmJ_seh . ' ble It is done by
the consumption frontier is pushed _as far to the ng t as possJ . at the
undertaking all projects wi_t)h for which
market rate of mterest (t . e ru e ts .
PV = + O
o 1 + 1
Equivalently, we could do al! projects for which
1
p= --- - >1

So in equilibrium, i = P . h r red
' t. t to its maxtmum choose t e preJer
2
H ing expanded the consump 10n se ' h
. av . . t ( ) At this point if r* is the ex-post discount rate, we ave
consumptwn pom e '
r* = i =p.
Th . dividua! borrows c- Y in period O and repays (c - +Ji) in
. e m t how much of the loan lS a consumptwn . oan,
penod l, but _we say ,E- * (which equals the investment) lS an
unless we arbJtranly assert that }o Yo . h . h the person
N h t the collectwn of mvestments w lC
investment . ote t a . 1 f m the present value of the vector at E to
undertakes raJses hts present va ue ro 1
t e by an amount equa to
the present value of the vector a y, t. .,
* ___li_t_ - E + - _1_
(
* ) ( l )
Yo + 1 + i Yo 1 + i
Y* -
* ,E + _ 1 _ ___ = + - - -.
Yo - }o 1 +
0
1 +
or
. . . nd lend at 10 percent .His utility is u = Co C and, if he
Ql2-3 Suppose an mdiVJdua1 can borrowh_ a . 110 I n each perod. Ca1cu1ale Co' C as well as
. . b g/1endmg IS mcome IS .
does no mvcstmg or orrowm ' . . d
1
Assume the avai1ab1e projects are
hs borrowing/1ending in pcnod O and repayment m peno .
those listed in Question 12-l(i).
lmperfect Capital Markets .
. fect ca ita! markets, the recursive proccdurc JUSI
Once_ we allow for ose borrowing rate (iu) toan individua l cxcccds
breatks both a re constan t. lt is thcn no good tcllmg
his len mg rae e lxescnt valuc, until wc know whcthcr he is " lcnd r ot. _a
hun to max11111S .
1
T 11 . pr lcr n cs lo
horrowcr. Thi s dcp nds on hi s produ t ion posst .l t 11 '.'s :tnl lts . .: . ' .
. , . ,' 1 , , 1 t ()11 llllltll' t' :111d thL11 , II SIII 1
C, Y
( ISlopel = l+i1)
CAPITAL li'TE){EST 333
( ISiopel = l +iR)
Figure 12-5 Unequal borrowing and lending
rates for the same person.
and iL, construct his consumption frontier (see Figure 12-5). He may then choose
to
l. Produce at y and consume on the borrowing line (he does thi s if at y, r > i 11 )
2. Produce at y' and consume on the lending line (he does this if at y', r < ir.), or
3. Produce and consume somewhere between y and y' (a direct tangency solution
like Robinson Crusoe's)
When we know which ofthese cases maximises his utility, we can thcn charactcrise
his production decision as maximising the present value of his incorne at the
ex-post discount rate. If two people ha ve the same i8 , ir., the same tastes, and
the same endowment E, but one has good production opportunities and the
other bad ones, the former will tend to be a borrower and the latter a lender.
Sorne borrowers do not face constant borrowing rates but rates that rise with
the amount being borrowed, the reason being that the marginal risk of default
rises with the amount borrowed and the creditor therefore requires a premium
for lending more. In this case we trace out the investor's overall consumption
frontier by tak ingeach possible production point, drawing from it the corresponding
conditio nal consurnption frontier, and then finally taking the envelope of all such
fronticrs. For cxarn ple, if up to x can be borrowed at an interest rate i and
thcrcaft cr nothing, thc cnvclopc reproduces the shape of the production frontier
(scc Figure 12-6). At ; regular o ptirnurn, such as that shown at e, the rate of
tim ' pn.' i' ' !'l.' ll e ' ljl l:il s III L r:II C or rcturn 011 thc marginal projcct. In fa ct the
s:lll iL' L'VVI I il' , ,. lll :JrgII:il horrowi tt g r:1t c riscs contilll ltll isly, t hcr
ly 11 11111 1 (ll 11ll i111 (Y :ts tl 1: 'lllt ' 11 ( >
11
334 FACTOR SUPPLY
C , Y
Figure 12-6 Limited borrowing opportunities.
However, where projects are not small or divisible, the optimum may well
involve a comer solution.
With imperfect capital markets the only safe le is the general of maximising
u( Co, e
1
) subject to the production and capt.tal-market constrmnts.
This requires the investor to estima te bis utility for all feasible points (co, C)
that are not obviously dominated. b
For example suppose a person can borrow and lend at a constant rate
b
'more than x He has to decide whether to do a certam
cannot orrow Id eh period
indivisible project. He first determines what he wou consume m ea t . t
if he did the project, assuming he could borrow and lend wtthout cons ram .
He then checks whether this point is feasible, given the borrowmg constramt.
If it is not, then, if he did the project, he would have to consume at
Co = yg + + X
c
1
= yf + - x(l + i)
The utility given by this point must now be compared with that he could obtain
if he did not do the project. (For the case where he does not do the proJCCt,
he must also determine the regular optimum an? then check whether tt ts fcas1blc
and if not determine bis constrained consumptwn . .
1
.
1
B
ates may of course vary between mdtvtduals, from somct 1111
orrowmg r 11 '( lht cnl
fi t F people who havc no secur1ty to o cr as e>
near zero up to m 111 y. or . .. . . .
b
. tes ,
1
re oftcn very hi gh. lf thcy canno t ho rrow, thc
1
1.11 1s
0 1 rowmg ra < . . . ti . " ts
inflnit c and thcir invcstmcnt d cision oc urs al a dll . t tancn , d_s 1111 ll l L .'
l
. l' J . ( ' l'll ''llL' J:X 'L' J)I j 11 ; SllCiv t y wl ' l l' : l:lVl'l y l. ' JWlllllll n , \lllll.l ll
<l \01ll S01 '
capital cannot often be used as collateral, so that in poor families the private cost
of human capital investment is often met by an equal reduction in consumption.
Q12-4 (i) Suppose there is a project (A):

-600 1111
If he does not do the project, the investor receives 110 per period. He can borrow and lend at
10 percent but cannot borrow more than 500. u= c0 c1. Will he do the project ?
(ii) Would he do the project (B) below if the alternat ive was again no project ?

- 500 1111
12-5 Suppose two individuals ha ve the same tastes, endowments, and abilities to profit from investment
in human capital, but A has a higher borrowing rate than B. 8 oth borrow to finance their education.
(i) Which will invest most ?
(ii) Suppose the present value of the lifetime income of each is evaluated using A's discount
rate, which individual has the higher present value? '
(iii) Which is better off?
(iv) What conclusion do you draw?
12-2 THE INDIVIDUAL: TWO-PERIOD CASE,
NONCONVEX PRODUCTION SET
Perfect Capital Market
We now allow for interdependence between projects. If projects are mutually
exclusive or if one project can only be done if another intrinsically less productive
project has been done first, then the production frontier cannot be constructed
by simply arranging all projects in order of their rates of return. Instead we seek
that frontier that maximises y 1 for each y0 , after allowing for which combinations
of projects are permissible.
For example, suppose all projects are small and independent, except for
two large mutually exclusive projects A and B, each of which is more productive
than any ofthe small independent projects. Then ifproject A is done rather than B,
we have a permissible production frontier EAA' ; if project Bis done rather than
i\ , wc havc a pcrmissiblc procluction frontier, EBB' (see F igure 12-7}. Since each
or thcsc fronti crs is pcrmiss iblc, thc overall frontier must be the envelope of the
two. whi ch is This is nonconvcx.
11' IHlll C<l ll V xit is Jll'l'Sl'lll , it may be ri ght to prdc r a pro jcct with a lower
ratl ni' l l' llllll l< l :1 H PJnt wi lh a hi hcr Oll l' . Th 0111 <..:O IT " l r11l c is to
tl ll 1'1\'\1' 111 v. li111' <11' ll L'IIIl ll' . 1:o 1 n:l lll Jlh-. 11 1 J.' ll', lll t' 1/ 7 th\' <lptirnal
336 FACTOR SUPPLY
1
1
1
Figure 12-7 Nonconvex produc-
0 1 1 V ti on frontier.
production point is y assuming a perfect capital market with interest rate equal
to the slope of y V But selecting y means selecting project B and rejecting project
A, which has a higher rate of return. The reason is simply that project .B has a
higher present value. In order to maximise his tota l present value the mvestor
chooses among mutually exclusive projects those that have the highest present
values. For OV is nothing other than the present value of his original endowrnent
plus that of all his projects.
In the two-period case a project with a lower rate of return can only have
a higher present value than sorne other project if it is larger, as in the following
example :
Project t. yo t. y p PV (approx.)t
i = 0.01 i = oox
A - 1 1.1 0. 10 0.09 0.02
B - 2 2. 16 0.08 0.14 000
t Note that for small a, i, (1 + a)/ (1 + i) is approximately equal to 1 + a - i .
If the interest rate is 1 percent, B has thc la rgcr prcscnt va luc. ll owcv r, :ts
intcrest rates ri SC, thc cliflcrCil CC in prcs ' JI ( valtt c f:tlfs, :tll d th 'I'C lllii SI h ' S(llllL'
critica ! rat c or int crcst lcss !han X pl'l'tTJll :11 wlt il' h /1. h 111 \li'L' prnlil :lhk
ls e : 'tlrL 12 X). /1. 1 a11 y i11ll'rv. l r:JI , : I IHlVl' lili:-. kwl. /1. s li Pt il d h1 do1t v r:11l w1
PV
B
Figure 12-8 Present values at dif-
fcrent interest rates.
than B. A glance at Figure 12-7 confirms that, as interest rates rise, a switch
to project A is necessary if the consumption frontier is to be kept as far to
the right as possible.
. also occurs if a more productive project R can only be done
If an mtrmsJcally less productive project S has been done already. In order to
maximise his present value the investor must evaluate the present value of each
project that can be done on its own (such as S) and the present value of
joint (such as R + S). There is no point in his evaluating a project like
R on Its own (unless he has already decided to do S).
Imperfect Capital Markets
If interdependent projects ha ve to be evaluated in imperfect capital markets the
of maximising present values does not help until we know whether a
IS a or neither. The only safe approach is the general one of
maximismg utJhty subject to the procluction and capital-market constraints.
However, we shall henceforth revert to perfect capital markets.
Q12-6 An investor has the following projects available:
l?roject
t. yo t. y ,
A
-2 2.02
B (conditional on A)
-2 2.28
e
- 1
1.14
D (concl itional on C)
- 1
t.Ot
i\ 11 ;re indcpClll lclll , tlwl B can only be done if A has been cl one, and D can only
he done d ( hccn dPll L' . Wl11ch pmccts shoulcl he do if
(i); (),() '!
(11) 1 () 1() '1
Wlll t l iJII IIJI' tl . ,ll .. l tiol l ll ' tl !i d
(111) 1 II W """ 111 """"""' ltt llw ' """'' lt 'Sii l( lll \11 '., 11 1' lll ll '" ''ll>l t lt> " " ''"''' 1 ' 1111 ( .. ,
1
1

1
338 FACTOR SUPPL Y
12-3 THE INDIVIDUAL: n-PERIOD CASE
Permanent lncome Streams
Most investments last more than two periods, but we can_ essentia_ls
of the two-period analysis if we confine ourselves to proJects mvolvmg costs m
the first ear and a permanent stream of constant returns The reason
is that ,Ze thereby continue to rule out any possibilities of substitutwn
future periods. We simply modify the two-period model by relabellmg the
axis as permanent consumption (cP) and permanent income (Yv) o t e
curves become p, r, and i rather than (1 + p), (1 + r), and (1 + z), smce

p= - -

.
= --

and the utility function is defined as u(co, cP). The present value of a project is
given by

PV=
l
. . .
Ql2-7 Suppose there are the following two mutually:exclusive proJects for supplymg a gtven quan Y
of electricity (the real-life dilemma ts roughly as mdtcated).
Project
Thermal
Nuclear
L'1yo
-50
- lOO
Which method should be adopted if
(i) i = 0.10?
(ii) i = 0.20?
The General Case
15
22
Most choices are not between permanent in come streams, nor the . inter est
rate be the same between one pair of adjacent years and anothet pall. Gt vcn
perfect capital markets, the only generally correct investment rule I S
Yt Y2 1 ... -1 y,
max PV = Yo + + ( 1 )( 1 1 )
1 + 1 1 ' 2
11 ( l
1 1
PV
Figure 12-9 Present value of - l. 5, - 6.
where i, is the interest rate between periods l - 1 and t- subject always to the
production constraint.
From this it is at once obvious why the rate of return rule cannot be
used in the many-period case. The rate of return on a project is got by solving
for p in
Now, suppose we obtained a single solution for p. Which interest rate, out of
i 1 to i, , should we compare it with? On top of this, the polynomial in which
we solve for p may either have no real roots (if L < O) or multiple roots.
The maximum number of roots having economic significance (i.e., greater than
- 1) equals the number of sign changes in the stream of net returns. Thus a
mining project that had to be filled in when it was finished could easily have
two rates of return. An example of multiple roots would be the project ( - 1, 5,
- 6). This has roots of 1 and 2, but as Figure 12-9 shows, should only be done
if the interest rate is between 100 and 200 percent.
The deficiency of the rate of return approach is that it does not ask the
key question: If the project is done, can the investor maintain his planned pattern
of consumption (as it would be without the project) and add something to it?t
However, the rate of return concept can be saved (after a fashion) if we confine
it to the two-period case and treat n-period projects as aggregations of two-period
projccts, each of which is judged by the rate of return rule. If we can decompose
-- T hi s pc: rkcl c<tpil ;il m;rkcl. lf lhcrc wcrc no capital market, we make our
prod uc t it\ 11 tkc i :-> iPil :-1 l l) m : l '\ lllii ' V 11(( '0 , . , e, ). T hi s i s equivalen! t o maximi sing
I 'V
wl w1t II H' 11 1 11 11 1111 ' n ll li lll ild 111 111 1 11 >lt11111111
1
!l '
1
11
1
340 FACTOR SUPPLY
the n-period project into (n- 1) two-period projects each of which pass the rule,
then the total project should be passed. Thus, suppose i = 0.05. If we take the
project (- 1, 1, 1) this can be decomposed into two projects:
A: periods O and 1
B: periods 1 and 2
-1, 1.05
-0.05, 1
p = 0.05 = i
p = 19 > i
The first project just passes, and the second passes with flying colours, so the
overall project passes. Project A has no adverse effect on the individual's ability
to pay for his original consumption plan, while if he sticks to his plan he will
be left at the end of period 2 with a surplus of 1 - 1.05 (0.05) to spend on
consumption then, or when suitably discounted, in sorne earlier period.
J n the case of projects ending with negative returns, the rate of return
language has to be modified somewhat. Reverting to the project ( - 1, 5, - 6), we
can decompose this into
A: periods O and 1
B: periods 1 and 2
-1,1.05
3.95, -6
p = 0.05 = i
The first project passes the rate of return test. Of the second we ask: lf the
investor lent out the 3.95 he gets in period 1, could he earn enough to repay
6 in period 2. The answer is a resounding no. But had the interest rate been 100,
the answer to the comparable question would have been a bare yes, for the
subprojects would have been
A: periods O and 1
B: periods 1 and 2
-1,2
3, -6
p=1=i
p=l=i
However, this face-saving operation has little point, other than to remind us of the
underlying rationale of the present value rule.
From a substantive point of view, the main change which enters when we
move to the many-period case is that the length of life of different projects
affects the way their relative present values change when interest rates change.
The lower the interest rate, the better the relative performance of long-lived
projects, as can be seen by comparing projects A and B below. Thus more durable
machines are likely to be built in periods when real interest rates are low. By
the same token, if two projects are the same length, lower interest rates favour
projects whose returns are concentrated later in the period (see example below).
Project
A
B
Stream of nct returns
- 1, 1.1, O, O, O (forever)
- 1, 0.1 , 0.1,0.1 , 0.1 (forcvcr)
PV (approx.)
i = 0.01
0.09
9
i = 0.10
o
o
" .,, ,. ,., .,, rnli' s { II ' III II ' lltrt/'1' /un lit tr/ untl , 11111 ' 11 ,ul'!i, r) lurur t
Ql2-8 (i) At what interest rate do the following projects have the same present v 1 ? a u e.
A: -1, 1, 1.1
B: -1, 1.1, 0.99
(ii) Which has higher PV at interest rates lower than that?
12-4 CAPITAL-MARKET EQUILIBRIUM
We so far seen _how the individual's demand for borrowing will de end
on preferences, h_1s production possibilities, and assuming perfect
mar ets, on the real mterest rate. Given this it is easy to see how th t
rate is d t d . ' e m erest
e_ermme m a two-penod model (see Figure 12-10).
Constder first the world of Figure 12-3 where there is n d
investment Wh th . ' o pro ucttve
1
en e rate rises, borrowers will naturally want to borrow
.ess, as _present consumptt?n IS more expensive in terms of future consumption.
and, bemg they wtll tend to consume less in both periods.
of lenders less clear. The htgher interest rate in creases the rewards to
en mg, but this it raises lenders' real income, making them tend to
consume more m both periods. Thus the supply curve of funds may be rising
or There may thus be multiple equilibria, of which alternating
eqUI 1 na w1 e stable.
?ow introduce productive investment, the probability of multiple
eqUI I r_ta ts reduced. When the interest rate rises, borrowers' demands for
borrowmg are further cut by the fact that fewer projects are worth doin so
the demand curve for funds becomes flatter At the same t1'me Jend g,
r 1 d . ers are more
WI mg to en , as fewer of thetr own projects are worth doing so the su
curve _bends back less sharply. However multiple equilibria may exist pp y
Ftgures 12-3 and 12-4 ha ve been drawn to illustrate an interest rate . which
produces equilibrium in the loans market, assuming there are the same large
D
-
/)
: )( ll rt' 1 10 l l ll' <' '1 111 1<111111111 ni lh, l' ! lj lll :il
ll! l ll k l' l
342 FACTOR SUPPL Y
number of people characterised by Figures 12-3 and 12-4. Total le_ndi_n? (in
Figure 12-3) equals total borrowing (in ). Moreover, for each mdiVldual
his total inftow of funds must equal h1s total d1sposal of funds. So
Income + borrowing = consumption + investment +lending
+ net hoarding of cash
The same is true when summed over individuals. Thus if in the _whole community
total borrowing equals total lending and there is no net hoardmg, 1t follows that
for the community
Investment = income - consumption = saving
How would a change in the nation's investment affe_ct the
interest rate? This clearly depends on the of . . m the
intertemporal utility function. If the elasticity of substitutJOn IS the mterest
rate is determined entirely on the supply side- by the supply of funds. But,
more realistically, we must think of the interest rate as determmed by both the
supply and demand for funds. . . . .
We have so far dealt with a world m wh1ch there IS only consumptwn
good in each period. Suppose there are many. The in?ividual then face a
capital-market constraint in which each dated co11_1modity has a _to
one of the dated commodities, taken as a numeraire. Thus, ?Y up a claJm
over a specified form of consumption toda y, can _acqmre. to som_e
specified form of consumption He wJll utih_ty to
production opportunities and th1s capJtal-market constramt. Th1s Will
his excess demand in the market for each commodity, and the market wJll
determine the relative prices, such that the excess demands for good, summed
over all individuals, add to zero. There will be no such as th_e n:oney
rate of interest unless we introduce money as a commod1ty. There 1s_ s1mply
a set of relative prices between all dated commodities; and rate at say,
butter, in year t, can be substituted for butter in year t + 1 1s_ not necessanly the
same as the rate at which butter in year t + 1 can be for butter m
year t + 2. The strength of the whole theory is that 1t IS not hnked to any
notional steady state.
12-5 AGGREGATE CAPITAL AND THE
CAPITAL CONTROVERSY
The reader may have noticed that we have not so far mentioned thc conccpt
of capital. This is because there be dragons here. Joan Robmson and h r
colleagues have made their attack on the concept of aggrcgatc captt al th '
centrepiece oftheir attack on neoclassicalthcory."l" But thc conccpt ts untH.:ccss;tr
"1" For ; usd ul C<l l k cti on or arti ck s " " ho th Sl\l cs " ' llll' l 't llllii )V\""Y , ,.,. ( : . ( .
N. 1 . (Lt b ), ('ul'lfll/1111, ( ; 1'1111' 1/r , l \ ' 11' 11111, ll ,1111111111hW<IIIl t , l ' l /1
11 :11 1.'111 111 111 1\ 1
for theoretical purposes. At no point in this chapter have we needed a measure
of the aggregate capital stock at a point in time. All we have assumed is that
at any moment, producers believe they can add to future output by the use of
current resources, according to sorne determnate relationship. However, for
empirical work it is often conveni ent to use m.easures of aggregate capital, and we
must therefore be aware of the dangers of doing this.
To understand Joan Robinson's objection, we can refer to J. B. Clark's one-
sector model of the distribution of income. Here there is only one kind of good
which serves both for consumption and as a capital good. Then if output is
produced by capital and homogeneous labour, we have
y = f(K, L)
The rate of profit is determined by the stocks of labour and capital, at a leve!
equal to the marginal product of capital. Since output and capital are measured
in the same units, the marginal product of capital is a pure number. lf there
are constant returns to scale, it also follows that as capital per man (K/ L)
in creases, the rate of profit (i) falls, and the real wage ( w) rises.
This approach to the rate of profit is obviously wrong, since consumption
and capital goods are not the same. Supposing there were only one capital good,
and one additional unit of investment (now) yielded a permanent increase in the
output of consumption goods of LlcP. If the value of the capital good in units of
current consumption were LlK, one might still be tempted to say that the
marginal product of capital was LlcP/LlK and that this explained the interest
rate. This unidirectional analysis would however be completely misleading, since,
viewed from another angle, the value of the machine is determined such that it
equals the present value of its output. In that case
LlK = LlcP
i
In fact, of course, the interest rate and the quantity of machines are simultaneously
determined by intertemporal preferences and the possible combinations of present
and future consumption implied in the physical productivities of machines and
labour. In this sense Figure 12-4 is a partial analysis for the individual, since
the amount of future consumption output he can produce by giving up different
amounts of current income to bu y capital goods depends on the final equilibrium
set of prices. However, on reasonable assumptions there would be such an
cquilibri um, whether there is one or many capital goods.
Evcn so, as the Cambridge (England) school points out, it does not follow
t hat, if dif'fcrcnt steady-state equili bria are compared, the rate of profit necessari ly
l'a ll s as thc val uc or capit(ll pcr man (measured in consumption goods) rises. To
gel a 1\;cl f'or thc argumcnt, considcr thc case where the consumption good can
h prod11 cd h two p()ss ihl 1 chnologics.
\ Js itt g , ( IIIIf of Ollfpttf ;n he p rod11 'L'd ll l':tr 1 h :I J1J1 lying
f IIIII (S ll( l : tlllllll 111 VI' 11 1 . \ Js ill fL'cl iii OJ ogy 11, 1 111111 Pi' PI II Jlll( l':lll ()l'
344 FACTOR SUPPLY
produced by applying 2 units of labour in year t - 3 and 6 units in year t - 1 .t
Technique
A
B
Labour requirement per unit of output in year t
t- 3 ( - 2 t- 1
7
2 6
lt is simplest to think of all the labour being used to improve the quality of the
raw materials, so that all capital is working capital. But the example is chosen
for simplicity only, and the result which follows could equally occur when all
capital is in the form of durable machines.
We now examine the different possible steady states in which such an economy
could operate if it had a constant labour force. (A steady state is one in which
output per unit of capital is constan t.) To do this it is convenient to start by
positing the interest rate i and then find what real wage w and value of capital per
man (K/ L) would be consistent with it. Now, producers will always choose
that technique which minimises the cost of producing a unit of output, cost
being measured by its present value measured in the year of sale. Unit cost
using technology A is
CA = w[7(1 + i)
2
]
and unit cost using technology B is
C
8
= w[2(1 + i)
3
+ 6(1 + i)]
So technology A is used whenever
2(1 + i)
2
- 7(1 + i) + 6 >o
and when that expression equals zero, producers are indifferent between A and
B. Since the expression is a polynomial, it may have more than one root (as we
saw in the case of the rates of return on n-period projects ). If this happens, there
may be more than one interest rate at which producers switch between techniques
A and B. In fact in the example given the position is as follows:
i > 1
1 > i > 0.5
0.5 > i
A cheaper
B cheaper
A cheaper
This is the so-called phenomenon of " reswitching" or " double-switching." As the
ir.terest rate falls, a given technique can be found profitabl c, thcn unprofitahle
again. This, if it happens in the real world, has dcvastating implicati ons for th '
conclusions usually drawn from thc onc-scctor modcl or d ist ri hu t ion. To scc t h is,
J T hi s cx ampk: is l : t k l' ll r r ul ll P. S: 11 l llli.: l ".t l ll , .. , , Si ir llll lll l}' 1 Jp," ( )1111/f /1 .lullf'/1 (// uf J, ' IJ//UII/1( ' \ ,
v' , XII, PP x \, N"'''' '' ' ""' l' lt 'c'
CAPITAL AN D INTEREST 345
we first compute the levels at the "switching" rates of interest. Iftechnique A is
used, the real wage IS gtven by equating the sale value of output to the present
value of Its costs at the moment of sale. So
w[7(1 + i)
2
] = 1
Similarly for technique B,
w[2(1 + i)
3
+ 6(1 + i)] = 1
At the switching of interest i = 1, is the same irrespective of
techmque and IS m umts of output. Stmtlarly at = 0.5, the wage using
both techmques ts 1; 15.75. Thus far, the one-sector model's results survive: given
constant. returns, wages rise if interest rates fall . But unfortunately it is not
true that value of capital per man also rises (capital being valued
m umts of consumptwn).
If technique A is u sed, the value of capital in year t is the capital value of
work-m-progress m year t - 2 plus that done in year t - 1. If there are
L workers and K 1s the value of capital in units of output then
()[Lw(1 + i? + Lw(1 + i)] = w[(l + i)
2
+ 1 + i]
Since, if technique A is used
1
W= - --
7( 1 + i)
2
(
K) A = j (l + _1_)
L 7 1 + i
As_ the falls, Kj L rises, as in the one-sector model. The same is true
usmg techntque B. The problem arises when we switch from one technique to
the other.
. Consider the switch at i = l. The value of capital per worker using technique
A IS
But, technique B, the of capital is the valuc of work-in-progress done
m ye,u t - 3 plus that done m year t - 2 plus that done in year t - 1. If there
a re L workers of whom 2Lj8 work at the first stage of product ion and 6Lj 8 at
thc sccond stage, then
(
/\.!. )
11
1 j2L11' 6L l
t. l X 1(1 f- i).\+ (1 + i) 2+ (1 + i)] + Xw(l + i)
11 .: ,; 1 ( 1 1 i)
1
1 ( 1 1 i )" 1 1 1 1 1:
346 FACTOR SUPPLY
A
0.5
A
Figure 12-11 Thc reswitching of techniques.
At i = 1, using our earlier calculation that w = 2
1
8, we find
( =
12 + 2) = is < 2
6
s
Thus when i falls to 1, the value of capital per worker falls, as shown in Figure
12-11. It then rises steadily until i = 0.5, when it rises discontinuously as technique
A comes back into use.
Looking at the matter from the more traditional angle, a given value of capital
per man is consistent with more than one rate of profit, depending on what
technique is being used. This contradicts the story of the Austrian economists
like Bohm-Bawerk, according to which as production beco mes more " roundabout"
and takes place over a longer period of time, interest rates must falland capital
per man increase. The reason is, of course, that when one compares techniques
A and B, it is quite unclear which of them is the more roundabout. If, instead,
we hada point-input point-output case, there would be no ambiguity, nor would
there be if, for a given technique, labour was always applied uniformly over time.
The importance of reswitching as an empirical phenomenon has been littlc
investigated by those who stress its importance, but it must be more importan!
the less easy it is to tell unambiguously whether a process is roundabout and
mechanised, or unroundabout and unmechanised.
What the reswitching debate has done is to counsel caution in thc use of
simplified models. It in no way undermines the generali scd ncoclass ica l thcory
oEtlined in Section 12-4, which is not concerned with any artili cial stcad y-s tat
situation. However, in emprica! work therc rcmains thc prohl cm ora grc :1ti on.
This is an acute problem whcthcr onc is ITI asurin g c;1p ital scrviccs, bl our scrv i 'L s.
or output.ln gc ncral,aggrc ati on is o 1d mss ihl wli nL lhc it L' Ili S I n he :1 grl ;11 d
;1rc cili1L' r p rkct c<> ll>p k ll> u il s 01 w1 kc t lillll t'S < >tl wrwisv. li ll ' ll' ill> llll :!ll y
a danger of estimation bias. The danger is reduced if the relative weights which
are u sed m any two items correspond to their "shadow prices" (i .e., to
_relat!ve effec_ts whtch an additional unit of each would have on output or
utthty). The spectal problem with capital equipment is that, because once installed
It IS less readdy marketable than labour or current output, the prices available
for aggregatmg capital are less relevant than those for aggregating labour or
current output. In a world of economic change, the annual value of the historie
costs of a (even after adjusting for inflation) may bear little relation to
the value of Its services. And due to the costs of resale, second-hand
markets often provtde no further guidance. This poses real problems for
econome_tnc1ans. But the neoclassical theorists emerge unscathed, provided they
state thetr theory m sufficiently general terms.t
Ql2-9 Suppose output can be produced automatically by two machines (there are no others). Each
machme 1s bllllt m one year by one worker. They yield the following ftows of output:
Machine
A
B
t + 1
2
t + 2
o
2
lf the value of capital per man incrcases, must the interest rate fall?
. t See, :.or example, K. J. Arrow and F. _H. Hahn, General Competir/ve Analysis, l-loldcn- Day.
San Francisco, 1971. For a sunphfied exposJtJOn see C. R. S. Doughcrty. " On thc Ratc of Rcturn
and the Rate of Profit," Economic Journa/, vol. 82, pp. 1324- 1350, December 1972.

Chapter Eleven - Twelve - Ok

Transféré par

Informations du document

Description originale:

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Chapter Eleven - Twelve - Ok

Transféré par

Droits d'auteur :

Formats disponibles

CHAPTER

(?Y trtue of his tastes)

Vous aimerez peut-être aussi