Review of Economic Studies, Ltd.

A Note on Welfare Propositions in Economics

Author(s): T. De Scitovszky
Source: The Review of Economic Studies, Vol. 9, No. 1 (Nov., 1941), pp. 77-88
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/2967640
Accessed: 05-12-2015 10:01 UTC
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/2967640?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact [email protected].

Review of Economic Studies, Ltd. and Oxford University Press are collaborating with JSTOR to digitize, preserve and
extend access to The Review of Economic Studies.

http://www.jstor.org

This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

Note

on

Propositions

Welfare

in

Economlcs
Modern economic theory draws a sharp distinction between positive
economics, which explains the working of the economic system, and welfare
economics, which prescribes policy. In the domain of welfare economics the
impossibility of interpersonal utility comparisons has for a long time been
believed to impose strict limitations on the economist, which kept this branch
of economic theory in the background. Recently, however, there has been a
reawakening of interest in welfare problems, following assertions that these
limitations are less restrictive than they were hIithertosupposed to be.' The
present note attempts to analyse the problem in detail.
I
The aim of welfare economics is to test the efficiency of economic institutions in making use of the productive resourcesof a community. For analytical
and historical reasons it is useful to distinguish between welfare propositions
based on the assumption of a fixed quantity of employed resources and those
that regard that quantity as a variable.
The former are concerned with the allocating efficiency of the system ;2
i.e. with its ability of best allocating a given quantity of utilised resources
among their various uses in consumption and production. They can be conceived of as criteria for judging institutions and policy in a closed community
whose potential resources are fixed and can be trusted to be fully employed,
either because of the automatism of the system or because of the existence of a
governmental policy aiming at full employment.
The latter, which may be called welfare propositions in the wider sense,
are in addition to the above problems concerned also with the total quantity
of resources available to an open group and the degree of utilisation of those
resources. They are thereforerelevant, first of all, to problems of international
trade from the point of view of a single country; and secondly, to the general
problem of employment.
II
All the welfare propositions of the classical economists-viz., perfect
competition, free trade, and direct taxation-belong in the first category; a
fact which has not always been realised. They are all based on the principlethat
1 Cf. N. Kaldor: "Welfare Propositions of Economics and Interpersonal Comparisons of
Utility," Economic Journal, vol. 49 (I1939), p. 549; J. R. Hicks: " Foundations of Welfare
Economics," Economic Journal, vol. 49 (I939), p. 696. See also N. Kaldor " A Note on Tariffs
and the Terms of Trade," Economica (N.S.), vol. 7 (I1940), p. 377; and J. R. Hicks: " The
Rehabilitation of Consumers' Surplus," Review of Economic Studies, vol. 8 (I94I), p. Io8. The
present note is a criticism of the principle enunciated in Mr. Kaldor's first-quoted article and
underlying the argument of the others. It is not presented in polemic form, in order to enable
the reader not acquainted with the articles here quoted to follow its argument.
2 This expression was suggested to me by Mr. George Jaszi to whom I am also indebted for
reading the manuscript and making valuable suggestions.

77

This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

78

THE REVIEW OF ECONOMIC STUDIES

given the total quantity of utilised resources, they will be best distributed
among different uses if their rates of substitution are everywhere and for every
person equal; for only in such a situation will each person's satisfaction be
carriedto that maximum beyond which it cannot be increasedwithout diminishing someone else's. Perfect competition, free trade, and direct taxation are one
(probably the simplest) among the many ways of achieving this aim.
By limiting our universe of discourseto two commodities and two persons,
we can illustrate this principle on a simple diagram. Let us draw the indifference maps of the two individuals superposed on each other, one of them
reversed, with the axes parallel and in such a position that their intersection
gives the quantities of the two goods jointly possessed by the two people.
Every point of the rectangle enclosed by the axes corresponds to a given
distribution of the two goods between the two persons, and the two indifference
curves going through that point show their respective welfare positions. At
some points, indifference curves do not cut but are tangential one to another.
At these points the rate of substitution of the two goods is equal for the two
persons, and they represent optimum situations, because once such a point has
been reached no redistribution of the two goods can increase the welfare of
either person without diminishing that of the other. The locus of all optimum
points gives the contract curve.
We judge the allocating efficiency of economic institutions by the criterion
whether or not they enable people so to redistribute goods and services among
themselves (irrespective of their initial position) as to arrive on the contract
curve. That perfect competition or, from the point of view of the universe, freetrade are efficient in the above sense can be proved by showing that all pairs of
offer (reciprocaldemand) curves drawn from any point within the rectangle
intersect on the contract curve. Similarly, excise taxes and, from the point of
view of the universe, import and export duties are inefficient,because they can
be represented as distortions of offer curves that make them intersect outside
the contract curve. The arguments based on this diagram can be generalised
for any number of persons and commodities.' It implies only one limitation:
the quantities of goods available to the community as a whole must be fixed;
for they determine the points of intersection of the axes and the position of the
contract curve. This shows that the propositionsillustrated by the diagramare
allocative welfare propositions; and it also appears to limit their applicability
to the problem of the exchange of goods whose quantities coming onto the
market are given. It can be proved, however, that our arguments are equally
valid when instead of these quantities those of the factors utilised in their
production are consideredto be fixed. For the formal proof of the geometrical
arguments and their generalisations the reader is referred to the original
sources and to textbooks dealing with the subject.<
1 This also holds good for
2 Cf. F. Y. Edgeworth:

all arguments based on other diagrams in this note.

Mathematical Psychics, London, i88i, and " The Pure Theory of
International Trade," Economic Journal, vol. 4 (I894) ; Alfred Marshall: The Pure Theory of
and his Principles of Economics, Bk. V,
Foreign Trade (I879), London School reprint, 1930;
Chap. II. Note on Barter and Mathematical Note XII ; A. P. Lerner; " The Symmetry between
J. R. Hicks: Value and Capital,
Export and Import Taxes," Economica (N.S.), vol. 3 (1936);
Oxford, 1939, etc. For the best analysis of the nature of this kind of diagram see A. L. Bowley;
T4e MpothemqticalGroundwork of Economics, Oxford, 1924.

This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

A NOTE ON WELFARE PROPOSITIONS IN ECONOMICS

79

III
We have seen above that allocative welfare propositions are based on the
criterion of economic efficiency. They state that of alternative situations,
brought about by different institutions or courses of policy, one is superior to
the other in the sense that it would make everybody better off for every
distribution of welfare, if that were the same in the two situations. This is
different from saying that one situation is actually better than the other from
everybody's point of view, because a change in institutions or policy almost
always redistributes welfare sufficiently not to have a uniform effect on everybody but to favour some people and prejudiceothers. It follows from this that
economic welfare propositions cannot as a rule be made independently of
interpersonalcomparisonsof utility.
It would hardly be satisfactory, however, to confine the economist's value
judgments to cases where one situation is superior to the other from the point
of view of everybody affected. It is doubtful if in practice any choice comes
within this category; besides, there would not be much point in soliciting the
economist's expert opinion when everybody is unanimous, except in order to
enlighten people as to their true interest.
Favouring an improvementin the organisationof productionand exchange
only when it is accompaniedby a corrective redistributionof income fully compensating those prejudicedby it might seem to be a way out of the difficulty,
because such a change would make some people better off without making
anyone worse off. For instance, it might be argued that the abolition of the
Corn Laws should not have been advocated by economists in their capacity of
pure economists without advocating at the same time the full compensation of
landowners.out of taxes levied on those favoured by the cheapening of corn.
Yet, in a sense, and regardedfrom a long-run point of view, such propositions
are not independent of value judgments between alternative income distributions either. For, going out of their way to preserve the existing distribution
of income, they imply a preferencefor the status quo.
There seem to be two solutions of the problem. First of all, in addition to
admitting his inability to compare different people's satisfaction, the economist
may postulate that such comparisons are impossible, and that therefore there
is nothing to choose between one distribution of income and another. He may
then make value judgments on the sole criterion of efficiencywithout bothering
about concomitant shifts in the distribution of income, since he considers one
income distribution as good as any other.' In this cawse,however, he cannot
claim that his value judgments are independent of interpersonal utility comparisons,-becausethey depend on the assumption of their impossibility.
Secondly, the economist may put forwardhis welfare propositionswith due
1 This, I think, was the attitude of the classical economists;
at least of those who did not,
like Bastiat, impute ethical values to the distribution of income under perfect competition. It
seems to be the correct interpretation of that fairly representative statement of Cairnes': " . . .
standards of abstract justice . . . are inefficacious as means of solving the actual problems of .
distribution. . .
If our present system of industry (perfect competition) is to be justified, it
must . . . find its justification . . . in the fact that it secures for the mass of mankind a greater
amount of material and moral well-being, and provides more effectively for its progress in civilis4tion than any other plan,"

This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

THE REVIEW OF ECONOMIC STUDIES

8o

Yo

x~~~~~~~~~~~~~~~a

X0

Fig. i. Diagram i
emphasis on their limitations, as being based on the sole criterion of efficiency.
He may then point out the nature of eventual"redistributionsof income likely
to accompany a given change, and stress the necessity of basing economic
policy on considerationsboth of economic efficiencyand of social justice.' Such
an attitude, which I think is the only correct one, may diminish the force of the
1 Or, of course, he may also renounce his claim to purity and base his own recommendations
on both criteria,

This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

A NOTE ON WELFARE PROPOSITIONS IN ECONOMICS

8x

,I

Y*~

~~~~f

Fig. i, Diagram z
economist'swelfare propositionsbut does not makethem less useful. The above
considerationsqualify also the welfare propositions to be discussed below.
IV
When we come to the problem of welfare propositions in the wider sense,
we can no longer illustrate a change in economicinstitutions or policy on a single
diagram. For such a change will no longer mean a mere redistribution of
income and alteration of the rules of production and exchange; but may also
involve a change both in the total quantity of resources available to the comThis content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

82

THE REVIEW OF ECONOMIC STUDIES

munity, and in their degree of utilisation. The former may be due to the
imposition of a duty on international trade, which from the point of view of an
individual country alters the quantities of imports and retained' exports
available for home consumption; while the latter may be caused by this or any
other change, if it affects the propensity to save or the inducement to invest
and thereby changes employment. Analytically there is no difference between
the two cases. In both, the quantities of resources available'for consumption
are changed, hence the relative position of the indifference maps is altered;
whence it follows that welfare propositions in the wider sense must involve the
comparison of two diagrams. Since these are constructed from the identical
two indifference maps and differ only in the latter's relative position to each
other, such comparisonsare not the hopeless task they might seem at first sight.
For we can represent some (not all) welfare positions on both diagrams; and
it is possible to represent on one diagram the welfare positions corresponding
to all those points of the other diagram's contract curve that are inferior to its
" own " contract curve. This follows from the fact that our diagrams admit
the representation of all welfare situations that are inferior (worse from the
point of view of at least one of the two persons) to their contract curve, while
welfare positions superiorto the contract curve cannot be representedon them.
Our welfare propositions may necessitate the comparison of points on the
contract curves of the two diagrams, or of points suboptimal to them, or of a
point on one contract curve with a point suboptimal to the other contract
curve. The first case is that where the system's allocating efficiency is at an
optimum both before and after the given change; the second, where it is
suboptimal both before and after the change; the third, where the change
affects allocating efficiency. Taking an example from the theory of international trade, the first case may be illustrated by the imposition of an import
duty by a country in which taxation is direct and domestic markets are perfectly competitive ;1 the second case can be representedby a duty imposed in
a monopolistic world; and the third by a duty which favours the formation of
monopolies or is linked with an excise tax on the home production of import
substitutes.
V

Let us draw two diagrams (Fig' i), both consisting of the superposed
indifference maps of individuals A and B, but with the difference that in-the
second, B's map has been shifted by OBOB ; SO that the joint possessions of A
and B have increasedby xoxl of X and YoYiof Y comparedwith what they were
in the first. This shift will bring into a position of tangency indifferencecurves
that in the first diagram have neither touched nor intersected, and will thus
make the second diagram's contract curve superiorto that of the first diagram
throughout its range. This follows from that fundamentalpostulate of economic
I A tariff on foreign trade is not incompatible with the tariff imposing country's domestic
trade and production being of optimum allocating efficiency. The reader must not let himself be
confused by the fact that similar diagrams have been used for illustrating the waste caused by
tariffs from the point of view of the universe as a whole. We are here solely concerned with the
effects of a tariff on the welfare of a single country, consequently the indifference maps that
constitute our diagrams belong to inhabitants of the same country.

This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

A NOTE ON WELFARE PROPOSITIONS IN ECONOMICS

83

theory that indifference curves can never have a positive slope, and it will be
the case whenever the, shift in the relative position of the indifference maps
represents an increase in the quantity of at least one of the two commodities
without a diminution in that of the other. From the fact that the second
diagram's.contract curve is superior to that of the first, it follows that the
latter can be represented on the second diagram by tracing the locus of the
points of intersection of all the indifferencecurves that in the first diagram are
tangential to each other. This will give us a curve on each side of the second
diagram's contract curve, and the area between them represents welfare
positions that are superior to the first diagram's contract curve. Hence, a
change that brings the welfare of our groups from a point of the first diagram's
contract curve onto a point of the second diagram'scontract curve (or at least
within the area between the broken lines), can be said to be desirable with the
same generality and significance with which perfect competition or direct
taxation are said to be desirable on the ground of their allocating efficiency.
In other words, while it need not actually improve everybody's position, it
would do so for every possible distributionof welfare if the change were to leave
that distribution unaffected.
The above argument is an explicit formulation of the statement that
getting more of some (or all) commodities at no cost of foregoing others is a
good thing. This may be consideredas overpedantic,since that statement seems
to be obvious; on the other hand, it is subject to the same limitations that
qualify allocative welfare propositions (cf. section 3 above) ; and besides, it is
not even always true. Increased plenty is a good thing only if it is not linked
with a redistribution of welfare, too retrogressive from the point of view of
social justice; and if it does not lead to a serious deteriorationof the allocating
efficiency of the economic system. For the former there exists no objective
criterion, but there is a simple test for the latter. To test whether a diminution
in allocating efficiency has not obviated the advantages of increasedplenty, we
must see if after the change, it is possible fully to compensate people prejudiced
by it out of funds levied on those favoured by the change, without thereby
completely eliminating the latter's gain. From the geometricalargument above
it follows that if this test is fulfilled for one initial income distribution, it will
be fulfilled for all possible initial income distributions, and vice versa. Our test
is completely general also in the sense that it is applicable whether or not the
initial situation is of optimum allocating efficiency. (I.e. whether or not it lies
on the contract curve).
VI
of
The kind change contemplated above, where the quantity of some or
all goods is increasedwithout a diminutionin others, is likely to occur as a result
of increased employment, capital accumulation, technical progress, better
utilisation of strategic advantages in international trade (by putting a duty on
the export of goods for which foreigndemand is inelastic), and the like. Another
kind of change, especially important in international trade, is that where the
quantity of some resources is increased and that of others diminished.'LIn
1This is the effect of import and export duties whenever the foreigners'reciprocaldemand
for exports is not inelastic and employment is given.

F2

This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

THE REVIEW OF ECONOMIC STUDIES

84

4.L.

yo~~~~~

Fig. 2, Diagram

Fig. 2 this is representedby a parallel displacement of one of the two indifference maps in the negative direction; so that the quantity of X is diminished
by x0xl and that of Y increasedby yoyl. Nothing general can be said about the
relationship of the two contract curves in this case without detailed knowledge
of the shape of the indifferencemaps. It is possible that the change will result
in superiorwelfare positions throughout the whole range of the contract curve,
in the same way as was depicted in Fig. I. This is especially likely to happen
when the increase is large and the diminution small. When on the other hand,
This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

A NOTE ON WELFARE PROPOSITIONS IN ECONOMICS

85

-s

-o

--

04

Fig.

2,

Diagram

the diminution is large and the increase small, the change may result in inferior
positions throughout the contract curve; a situation which can be visualised
by thinking of diagram 2 (Fig. I) as showing the initial, and diagram i the new,
position. Between these two extremes lies-the more general case in which some
sectors of the new contract curve are superior to the old one, while others are
inferiorto it. Its simplest example is illustrated in Fig. 2, whereP0 is a common
point of the two contract curves, to the left of which the new contract curve,
ii7r, represents welfare positions superiorto the correspondingwelfare positions
of the old contract curve, PP; while to the right of P0, the old contract curve
is superior to the new one. In each diagram the broken lines show the welfare
This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

86

THE REVIEW OF ECONOMIC STUDIES

positions correspondingto the other diagram's contract curve wherever that is

inferior to the diagram's own contract curve.
The economic meaning of this is that the identical change in the composition of the national income would improve generalwelfarefor some hypothetical
welfare distributions and worsen it for others. Imagine members of a community divided into two groups accordingto their preferencefor goods Y and X
respectively.' Then assume a change that increases the quantity of Y and
diminishes that of X, but leaves the distribution of money income between
our two groups unaffected. From the point of view of individuals, the change
will appear as a shift in relative prices; which, given the distributionof income,
will be likely to make those with a special preferencefor Y better off, and those
with a liking for X worse off, than they were before. Assume next that the
membersof our first group are rich and those of the second poor. Then the gain
of the first group expressed in money (or in terms of any single commodity)
will be greater than the money equivalent of the loss suffered by the second
group. Therefore, if we so redistributed income as to restore approximately
the initial distributionof welfare, there would be a net gain, making membersof
both groupsbetter off than they were before. Conversely,if the people favoured
by the change were poor, and those prejudiced by it were rich, the money
equivalent of the former's gain would be insufficient fully to compensate the
latter's loss, so that a redistribution of income tending to restore the initial
distribution of welfare would result in a net loss of satisfaction for everybody.
What significance are we to attach to this case ? To refrain altogether, as
the classical economists did, from making welfare propositions relating to it,
seems unduly restrictive. It is true that as we have seen such a change would
improve general welfare for some welfare distributionsand worsen it for others;
on the other hand, we are not interested in all possible welfare distributions.
There are only two distributions of welfare that really matter. Those actually
obtaining immediately before and after the change contemplated.2 It seems
therefore sufficient to concentrate on these and to investigate how the change
would affect general welfare if it were to leave the distribution of welfare
unaffected and if that were both before and after it, first what it actually is
before, secondly what it actually is after, the change. Whenever these two
comparisonsyield identical results, we can make welfare propositions of almost
the same generality and significance as the allocative welfare propositions of
the classical economists; especially since the identical results for the two
welfare distributions imply a strong presumption in favour of the same result
holding for all intermediate welfare distributions as well.
We propose, therefore, to make welfare propositions on the following
principle. We must first see whether it is possible in the new situation so to
redistribute income as to make everybody better off than he was in the initial
situation; secondly, we must see whether starting from the initial situation it
1 The term " preference " is used in a loose sense. It denotes the whole shape of indifference
surfaces and not only their slope at the relevant point, which in equilibrium conditions is the
same for everybody.
2 The reader's attention is called to the fact that in reality the distribution of income is not
given as we have assumed in the argument above. As a rule, the change will affect the distribution
of welfare not only by shifting relative prices but also by boosting some industries and depressing
others, and thereby redistributing money income.

This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

A NOTE ON WELFARE PROPOSITIONS IN ECONOMICS

87

is not possible by a mere redistribution of income to reach a position superior

to the new situation, again from everybody's point of view. If the first is
possible and the second impossible, we shall say that the new situation is better
than the old was. -If the first is impossible but the second possible, we shall say
that the new situation is worse; whereas if both are possible or both are
impossible, we shall refrain from making a welfare proposition.'
We can illustrate this procedurein Fig. 2 for the special case when allocating efficiencyis at its optimum both before and after the change. Each situation
can then be represented by a point on its respective contract curve and compared with the correspondingpoint on the other contract curve. If both points
lie to the left of POon their respective contract curves, the change will increase
generalwelfare, because starting from the new situation on the second diagram's
contract curve it is always possible to travel along that curve by redistributing
income and arrive at a point which is superior to the initial situation from
everybody's point of view; whereas starting from the initial situation on the
first diagram's contract curve, it is impossible by travelling along that curve
to reach a position superior to the new situation. If on the other hand, both
points lie to the right of the commonpoint P , the change can be said to diminish
generalwelfare on the same reasoning; while if one point lies to the left and the
other to the right, we can make no welfare propositions relative to our group.
VII
Our two criteria for making welfare propositions bear a close resemblance
to Paasche's and Laspeyre's formulae in. the theory of cost of living index
numbers. There, just as here, the difficulty lies in comparing averages whose
weighting is different ;2 and the solution is sought in comparing the two real
situations not one with another, but each with a hypot-heticalsituation, which
resemblesit in weighting but is otherwise identical with the other real situation.
In the theory of index numbers, budgets of different dates or places are compared each with the cost of the identical bundle of commodities at the prices
of the other date or place; and these two comparisons, expressed as ratios
(Paasche's and Laspeyre's formulae), are the limits within which the true
differencein the cost of living must lie.3 In welfare problems,of course, we can
aim neither at a " true " answer nor at its quantitative expression without
measuring satisfaction and comparing different people's. But our two criteria
are exactly analogous to Paasche's and Laspeyre's formulae. For we compare
the first welfare situation with what general welfarewould be if the satisfaction,
yielded by the physical income of the second situation were distributed as it
was in the first; and contrast the second situation with the welfare that the
first situation's physical income would yield to each person if it were so distri1 It need hardly be recalled that in the situation discussed in section 5-that
is, when the
quantities of goods and services all change in the same direction-this last case can never occur,
and we can always make welfare propositions.
2 Because the general welfare can be conceived of as average welfare.
3 Cf. Henry Schultz: " A Misunderstanding in Index Number Theory," Econometiica, vol. 7
and A. A. Konuis: "The Problem of the True Index of the Cost of Living,"
(I939), p. i;
Econometrica, vol. 7 (x939), p. IO.

This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions

88

THE REVIEW OF ECONOMIC STUDIES

buted as to make the distribution of welfare similar to that of the second

situation.'

VIII
Mr. Kaldor and Professor Hicks have asserted that it is always possible:to
tell whether a given change improves general welfare, even if not all people
gain by it and some lose. The test suggested by them: to see whether it is
possible after the change fully to compensate the losers at a cost to those
favoured that falls short of their total gain, is fundamentally identical with
the first of our two criteria. The objection to using this criterionby itself is that
it is asymmetrical, because it attributes undue importance to the particular
distribution of welfare obtaining before the contemplated change. If the
government had a special attachment to the status quo before the change and
would actually undertake to reproducethat welfare distribution by differential
taxation after the change, then Mr. Kaldor's test would be sufficient. For then,
the economist could regard that particular welfare distribution as the only
relevant one and would be entitled to use it as his sole standard of reference.
But in the absence of such a governmental policy there can be no justification
in attaching greater importance to the welfare-distribution as it was before
than as it is after the change.
To illustrate the pitfalls of this one-sided criterion, imagine a change, say
the imposition of a duty on imports, that brings the welfare of A and B from
P1 (Fig. 2) on the contract curve of diagram I onto 7T2on the contract curve of
diagram 2. According to Mr. Kaldor's test this change is desirable, because by
redistributing income we could travel from f2 along the 7T7Tcurve to 7r,, which
is superior to P1. But once the tariff has been imposed and situation 7T2established, it will be free trade and the resulting (original) situation P1 that will
appear preferableby the same test, because starting from P1, income could be so
redistributed (travelling along the PP curve in the first diagram this time) as
to reach P2, which is superior to qf2. So the two situations can be shown each
to be preferableto the other by the identical criterion: an absurd result, which
can only be avoided by using our double criterion.
T. DE SCITOVSZKY.
Washington,D. C.

I We say that the distribution of welfare is similar in two situations if every member of the
community prefers the same situation. A more exact definition would be unnecessary for our
purposes; besides, it is also impossible, since welfare cannot be measured.

This content downloaded from 193.108.160.146 on Sat, 05 Dec 2015 10:01:56 UTC
All use subject to JSTOR Terms and Conditions