"Exit, Voice, and Loyalty": Further Reflections and a Survey of Recent Contributions

Author(s): Albert O. Hirschman

Source: The Milbank Memorial Fund Quarterly. Health and Society , Summer, 1980, Vol.
58, No. 3 (Summer, 1980), pp. 430-453
Published by: Wiley on behalf of Milbank Memorial Fund

Stable URL: https://www.jstor.org/stable/3349733

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].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Milbank Memorial Fund and Wiley are collaborating with JSTOR to digitize, preserve and
extend access to The Milbank Memorial Fund Quarterly. Health and Society

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty:
Further Reflections and a Survey
of Recent Contributions*

ALBERT 0. HIRSCHMAN

School of Social Science,

Institute for Advanced Study, Princeton

Y BOOK EXIT, VOICE, AND LOYALTY: RESPONSES TO

Decline in Firms, Organizations, and States was

1970.1 Reactions to it and applications of its co
been fairly numerous and I have myself had quite a few
It will therefore be difficult to bring these matters
passably structured paper. In the following, I shall limit
broad areas of inquiry which have been so arranged
further reflections figure rather prominently thoug
exclusively in the first two sections while the latter
heavily weighted with reports and comments on th
contributions of others.

New Economic Arguments in

Favor of Voice

As most economists who have made contributions to political science

in recent decades, I have occasionally used economic models and
modes of reasoning to dissect political phenomena. While such exer-

* This paper first appeared in 1974, in Social Science Information (1):7-26, and is
reprinted with the permission of the publisher and the author.
1 By Harvard University Press, which published a slightly revised paperback
edition in 1972.

Milbank Memorial Fund Quarterly/Health and Society, Vol. 58, No. 3, 1980

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 43I

cises in interdisciplinary imperialism can be genuinely enlightening,

only a small part of my work has been of this particular kind. In fact, in
much of Exit, Voice, and Loyalty I have been guilty, not of imperialist
ambition or designs, but rather of the opposite: namely, of the desire
to convince economists of the importance and usefulness, for the
analysis of economic phenomena, of an essentially political concept
such as voice. Perhaps it is in recognition of these somewhat treason-
able services on behalf of political science that political scientists rather
than economists have honored me by calling together a seminar with
my book as basic document for discussion.
In the large portion of my book which was an essay in persuasion on
behalf of voice I argued that voice can and should complement and
occasionally supersede exit as a recuperation mechanism when busi-
ness firms, public services, and other organizations deteriorate. My
approach was both positive and normative. I explained the conditions
under which voice comes into existence and can be expected to be
powerful, but I also argued that, in some situations, the proper balance
of institutional incentives ought to be adjusted so as to strengthen
voice in relation to exit. I now find that my advocacy of voice was not
exaggerated, but, on the contrary, too timid. This is not surprising.
Since voice is an entirely new category for economists, our thought
processes are not properly attuned to it and it will take some time to
uncover all the situations in which the importance of voice has been
underrated. In this section I propose to discuss three such situations.

When the Cost of Voice Turns

into a Benefit2
In discussing customers' or members' choice between exit and voice I
naturally gave some attention to the cost of exit as compared to the
cost of voice. This comparison tipped the scales against voice, for I
considered exit to be generally costless, except when loyalty is pres-
ent, while resort to voice is typically costly as buyers of a product or
members of an organization spend time, effort and perhaps even
money in the attempt to exert influence on the firm or organization
with whose products or policies they are dissatisfied.3

2 Some of the arguments of this section have been previously put forward in
Hirschman, 1971, Introduction, pp. 4-7.
3 See Exit, Voice, and Loyalty. See below, pp. 439-440, for situations in which
exit is costly for reasons other than loyalty.

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
432 Albert 0. Hirschman

This was good economic reasoning, appropriate to "normalcy." It

took the explosion of protest activities after the Cambodia invasion
and the Kent State shootings to remind me that, in certain situations,
the use of voice can suddenly become a most sought-after, fulfilling
activity, in fact, the ultimate justification of human existence.
In other words, while normally felt as a chore and a cost which one
tries to minimize or shirk, the activities connected with voice can on
occasion become a highly desired end in itself. How is it possible to
account for this strange mutation?
In addition to choosing and allocating their time and income be-
tween various consumer goods, individuals also decide how to appor-
tion their activities between all private pursuits, on the one hand, and
such contributions as they choose to make to the "public happiness"
on the other. Decisions to make such contributions appear to be
subject to a number of characteristic properties in comparison to
private consumption decisions. For one thing, simple observation
reveals that the preference for participation in public affairs over the
"idiocy" of private life is much more unstable, and subject to much
wider fluctuations, than the preference for, say, apples over pears or
for present over future consumption. Events such as demonstrations,
marches, riots, and revolutions are "participation explosions," that is,
they result from a sudden enormous intensification of the preference
for public actions for which there are no parallels, with the possible
exception of the world of fashion, in the realm of private consumption
choices.
The reason for this instability of the taste for participation in public
affairs lies in the peculiar dual character of this activity. On the one
hand, such participation is equivalent to expressing a demand for
certain public policies, and since such public policies, once estab-
lished, can be enjoyed or "consumed" by everyone in the community,
the demand for public policies has the earmarks of the demand for
public goods. It follows that actual participation on the part of those
who favor a given policy is undermined by the well-known tendency
to lie low and to hide or understate the true intensity of one's demand
for a public good in the hope of getting a "free ride" through the
exertion of others. This is a major reason for the much lamented
"apathy" in relation to public issues. What looks like apathy is often
not absence of interest in a public policy, but considerable interest

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 433

combined with the expectation that someone else will exert himself on
one's own behalf.
But there is another side to the story which has not been analyzed
by the public goods theorists and which works in exactly the opposite
direction. In the case of the acquisition of normal public goods-say,
public parks or police protection-the usual distinction between the
value of the services rendered by these amenities and their cost is
sufficiently clear cut. But ambiguities arise when one transfers these
categories to public policies. The cost of obtaining or pushing through
these policies is the cost, in time spent, risk shouldered, and perhaps
money expended in the course of their advocacy. In other words,
striving for these policies through various acts of participation and
voice is their cost which, in accordance with the theory of public
goods, tends to be shirked by the individual. However, it is in the
nature of the "public good" or the "public happiness" that striving for
it cannot be neatly separated from possessing it. This is so because
striving for the public happiness will often be felt not so much as a
cost, but as the closest available substitute for it. We all know that
participation in a movement to bring about a desirable policy is (and,
unfortunately, may be for a long time) the next best thing to having
that policy.4
Uncertainty is an important element in this strange transformation
of means into ends, and of costs into benefits. Success in the advocacy
of a public policy is always uncertain: nobody knows the size of
citizens' advocacy or protest that is needed to impose, change, or stop
a given public policy. If a citizen feels strongly, he may therefore
experience the need to negate the uncertainty about the desired

4 Elsewhere I have shown that the distinction between private and public
goods goes back to Pascal who contrasted "particular things which can only be
possessed by a single individual" with "the true good [which] must be such
that all can possess it at the same time" (Pensees, 425). Pascal refers here to
God who is indeed the quintessential public good since His possession, unlike
that of public parks, can never become "rival" or "exclusive." But Pascal's
analysis went one step farther: God is also the archetype of that category of
public goods in whose case striving cannot be separated from possessing and
this important property of some public goods was again contrasted by Pascal
with private goods: "The hope Christians have to possess an infinite good is
mixed with actual enjoyment [. . .] for they are not like those who would hope
for a kingdom of which they have nothing, being subjects; rather, they hope
for holiness, for freedom from injustice, and they partake of it" (Pensees, 540).

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
434 Albert 0. Hirschman

outcome by the certainty of participation in the movement to bring

about that outcome. In a more rational vein, uncertainty may act at
times as a discriminating monopolist as it extracts from each person
with a "taste" for a certain policy the full amount he would be willing
to pay to have that policy; this would happen if each individual
becomes convinced that his contribution makes the difference be-
tween success and failure of the movement.
No matter what the precise explanation-one could simply take
refuge in the definition of man as an animal with the ability and
propensity to transform means into ends-the sudden, historically so
decisive outbursts of popular energies must be explained by precisely
this change in sign, by the turning of what is normally sensed as a cost
that is to be shirked into a benefit, a rewarding experience, and a
"happiness of pursuit" in which one simply must share.
The possibility of this mutation is fundamental for the understand-
ing of political change: achieving change often requires such a muta-
tion. It is also helpful in reconciling the conflicting views on political
participation in a democracy which are perhaps best epitomized by
Rousseau's Contrat social, on the one hand, and Benjamin Constant's
De la liberte des Anciens comparee a celle des Modernes, on the other. The
total participation considered as essential for the preservation of lib-
erty by Rousseau and the strictly limited participation advocated by
Constant can both characterize, in turn, the same polity, whose good
health may actually be served by some alternation or oscillation be-
tween the Rousseau and the Constant model.
The moral of this excursion into political theory for the exit-voice
alternative is clear: if active concern with the public happiness can on
occasion be felt as a benefit and as an important contribution to the
private happiness rather than as a subtraction from it and as a cost,
then voice will have an occasional edge over exit in those situations
that clearly impinge on the public happiness. This means that voice
can be expected to play a role in relation to those goods and in
particular to those dimensions of goods and services that have a strong
public interest component. Thus, deterioration in the taste of a firm's
food product will give rise to exit; but the presence of a health hazard
will lead to voice. Similarly, in the case of automobiles, unattractive
design will lead to exit, while safety problems will bring out voice.
Wherever the public interest is involved, voice will not be felt as a cost
but as a benefit by some people at some time, and, in this way, one of

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 435

the primary handicaps of voice in relation to exit will be reduced and,

on occasion, eliminated.
A recent illustration of these matters is the changing role of the
shareholder in the corporation. As I had mentioned in my book, exit
had long reigned supreme in this area, in obedience to the Wall Street
rule "if you don't like the management, you should sell your stock"
and in spite of remonstrances against this practice on the part of some
financial writers (see Exit. . . , p. 46). In connection with the ordinary,
private-regarding, return-maximizing activities of investors there has
been no overwhelming change in this respect even though, according
to some indications, institutional investors, such as trust departments
of banks, have tended to vote against management proposals some-
what more frequently (Eisenberg, 1969). But a considerable shift
occurred when institutional investors took an interest in, and became
concerned over, corporate policies and practices in such matters of
public concern as pollution or racial discrimination. In these situa-
tions, the concerned investors generally decided not to exit by selling
their stock, but to use what influence they could marshal in order to
modify corporate policies.
The institutional investors that were most active or, perhaps, reac-
tive in this field were the large private universities. They had to
respond to campaigns, such as "Campaign G.M." in 1970 and again in
1971, which had been launched by citizens' groups outside of the
universities, but soon received support from important student and
faculty groups. Committees were appointed and new policies de-
veloped. As a result, universities generally decided to take a more
activist role in shareholder meetings and, in general, in relation to the
management of corporations in which they are important stock-
holders. At the same time, a consensus developed on rejecting exit
(i.e. sale) as the only or even as the proper response to discontent in
matters where the public interest was at stake, as can be seen from a
very cautious committee report issued at Harvard University (1971).
The most emphatic and elaborate statement on the problem appears in a
book, The Ethical Investor by John G. Simon et al. (1971) that grew out
of the discussion of these issues at Yale University. Here exit is
advocated only as a last resort after voice has failed:

We have expressed dissatisfaction with attempts to cleanse a

portfolio through the sale of morally or socially objectionable hold-

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
436 Albert 0. Hirschman

ings. Such efforts [.. .] tend to involve one in illusions about moral
purity [. . .] we advocate such action when other attempts to correct
or avert a serious wrong have failed. (p. 53)5

The corporation thus stands as an example of a private organization

in which the relation between management and stockholders was
dominated by exit until such time as some activities of the corporation
were shown to affect the public interest; and as soon as stockholders
had complaints on the ground of these activities, the use of voice
seemed the more natural choice. It is of course possible that the use of
voice, once well established in connection with public-interest issues,
will contaminate the private, hitherto exit-dominated areas, and will
come to play a greater role in stockholder-management relations in
general, for better or worse.

Ignorance of Consumers and Producers

A second, not completely unrelated way of identifying goods, services
and organizations that are or should be voice- rather than exit-
intensive was suggested to me through recent papers of Richard R.
Nelson and Michael Krashinsky (1972) and Kenneth Arrow (1973).
In discussing institutional alternatives for the delivery of day-care
services, Nelson and Krashinsky make a distinction between goods in
which "the consumer can be assumed to be an expert in knowing what
he likes, e.g. sweet juicy oranges" and such services as day care for
small children whose quality is difficult to fathom for the parents.
Besides, clear quality standards for day care are simply not available.
Arrow addresses himself somewhat similarly to situations in which
there is a disproportion of knowledge between seller and buyer-as in
the case of medical services or in that of complex technological prod-
ucts such as drugs and automobiles-and he underlines the impor-
tance of ethical codes (about disclosure of information, for example)
which sellers ought to observe in such situations as a restraint on
socially undesirable profit-maximization. Both Nelson-Krashinsky
and Arrow are concerned with institutional implications of market

5 That voice is in fact more effective than exit in changing corporate policies
that are objectionable to some of its stockholders on public-interest grounds
is shown in Malkiel and Quandt (1971).

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 437

situations in which the buyer lacks knowledge about product quality

or is far inferior in this respect to the seller. As in the case of
exploitation of the consumer by a monopolist, some form of public
intervention or self-policing on the part of the producers or sellers
seems to be the answer to these situations since the consumers are
assumed to be in an inferior and impotent position in which neither
exit nor voice on their part is likely to perform as an adequate
safeguard of their interests.
The case of day care goes beyond these important, but still tra-
ditional concerns about "market failure." As Nelson-Krashinsky
almost intimate at one point, we have here a situation in which ignor-
ance about quality, about what one is really after, is by no means
limited to the buyer or consumer or member. Day care is typical of a
whole class of services for which, for a number of reasons, a strong
demand arises at some point; some people are willing to pay for it or
feel that it should be offered as a public service and some other people
step forward claiming to be able to accommodate the new demand or
are willing to do so to the best of their ability. The reality of the
situation is that demand for a service has arisen in advance of real
knowledge of how to satisfy it; society then delegates some of its mem-
bers to search for the best method of filling the new demand and of
supplying the newly arisen need; and the institutional question is here
not one of protecting the consumer, but of educating the producer, of
providing him with as much information as possible about his perfor-
mance. In such situations, the contribution of voice can clearly be of
the greatest importance, simply because the information it supplies is
rich and detailed as compared to the bareness and blankness of silent
exit. Moreover, exit may fail to supply even the bare information
about the existence of discontent, if dissatisfied consumers switch
back and forth between various equally unsatisfactory suppliers so that
each individual supplier gains new customers as fast as he loses them.
This phenomenon has been described in my book under the heading
"Competition as collusive behavior."
Producers' ignorance, or a substantial degree of such ignorance,
about ways and means of satisfying certain demands is probably more
widespread than is generally realized. It characterizes a large portion
of sectors that show a high rate of growth in modern economies,
namely education, health, and leisure activities. When the delivery of

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
438 Albert 0. Hirschman

health services can proceed along standard lines within well-charted

territory as, say, in the case of minor dentistry, consumer dissatisfac-
tion with one dentist is likely to take the form of exit. But the
individual who has some as yet poorly articulated complaint with
respect to his general physical or mental health is probably well
advised not to abandon his family doctor or psychiatrist at the slightest
disappointment, but to help them grope on his behalf and to collabo-
rate intensively with them through active use of voice.
To repeat, the second new criterion for discriminating between
exit-prone and voice-prone situations can be defined as ignorance and
uncertainty, shared by consumers and producers, about the manner of
procuring a desired good or service and, in fact, about their precise
nature. It is clear that there is a strong affinity between this criterion
and the first one which centered on the presence of a "public happi-
ness" component. Generalized ignorance and uncertainty about what
one is after exist typically when motivation to solve a problem is
outrunning understanding,6 and this situation arises in turn when
there are pressing public demands to "do something" about a poorly
understood problem. In such situations, then, the use of voice rather
than exit is to be expected and recommended on both counts.
The ignorance criterion is also helpful in accounting for swings from
the predominance of exit to that of voice in relation to the same goods
or services. Ignorance and uncertainty with respect to the desired
nature of a good or service are not always something that is con-
quered once and for all. For a number of such goods and services,
doubts are periodically reborn in the light of new experience. In fact, in
many cases doubts about the desirability of the product in its present
form arise for the first time after a more or less prolonged period of
unquestioning acceptance. Recent examples are automobiles, DDT,
and some drugs. Products which are subject to cycles of acceptance
and questioning are typically such complex and ignorance-intensive
services as psychiatric help and higher education. It is then quite
proper that exit should be the principal reaction of dissatisfied stu-
dents when no fundamental questions are widely asked about the
value and current methods of the university, while voice will pre-
dominate during periods of generalized loss of confidence in the
traditional system.

6 This topic is discussed in Hirschman, 1963, pp. 235-238.

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 439

Vertical Integration and Marriage

as Institutionalized Voice

I now come to a third criterion which throws light on the difference

between exit-prone and voice-prone situations and which had been
neglected in my book. It is somewhat symmetrical to the first criterion
which dealt with situations in which voice becomes less costly than it
would be under "normal" circumstances, or is even sensed as a benefit
by the customer-member. Just as I failed to question adequately
whether voice is costly under all circumstances, so I took the costless-
ness of exit too much for granted. I did allow for the existence of a
cost of exit to the extent that loyalty was present; but exit can imply
considerable cost in purely economic market situations even in the
absence of loyalty. Such costs are least in evidence in the case of
consumer purchases on which I had principally focused; but they
come to the fore in interindustry transactions when a firm buys a
specialized input from one among a limited number of potential
suppliers. In such situations, the firm will often spend considerable
time and money in apprenticing the supplier and these costs would
have to be incurred over again if the firm were to switch to another
supplier because, for some reason, it becomes dissatisfied with the
current relationship. The same applies to the supplier firm to the
extent that it has shouldered some of the costs of apprenticeship-
they would be largely lost to it if it had to look for a new customer.
This situation will therefore cause voice to become relatively more
attractive than exit for both firms in case of friction, but it also is one
of the basic motivations for vertical integration of firms as Professor
Oliver Williamson (1971; 1973) has pointed out. Integration can
indeed be considered as an arrangement, not for suppressing voice
through hierarchy, but rather for institutionalizing and routinizing it;
it is voice from one unit to another within a unified organization with
a common goal, supplemented by an established mechanism for ad-
judicating any unresolvable disputes that may arise between the
various producing units. The logic that makes for this sort of in-
stitutionalized voice has asserted itself also in the Soviet Union where
"direct ties" between input-using and input-producing firms, and be-
tween producers and retail outlets, have made their appearance. This
is regarded by Spechler (1970) as a "major innovation without over-
whelming support from the highest places."

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
440 Albert 0. Hirschman

It is not easy to think of analogues for this phenomenon outside of

the economy. But marriage could perhaps be interpreted as a similar
institutional arrangement: when a man and a woman have reached an
advanced degree of mutual understanding and adjustment, the costs of
exit from this relationship are high-one has to start from scratch with
the next partner. An attempt will therefore be made to take care of
remaining and recurring frictions through voice, and marriage can be
considered, just like the merger of two firms, as a way of routinizing
this voice-with unresolvable disputes being referred to the psycho-
therapist in lieu of the Executive Vice-President. There are probably
other situations in which the decision to enter a formal organiza-
tion is not prompted so much by fundamental agreement on values
and goals or by the desire to eliminate conflict, but, on the contrary, by
the need to bring necessarily recurrent conflict more frankly and more
routinely into the open without risking, each time, the survival of the
relationship. It is precisely because voice hides here behind the facade
of organization, hierarchy, and harmonious unity that I failed to be-
come aware of these situations.

Exit and Voice: The View from the Top

In 1970 a reader of Exit, Voice, and Loyalty remarked to me that the

book was written almost entirely "from below," that is, from the point
of view of customers or members as victims of deteriorating quality,
and that he would have liked to see the topic of exit versus voice
treated from the point of view of top management of various organiza-
tions. No wonder he expressed such a preference, for this was right
after the Cambodia events and the poor fellow had just been ap-
pointed president of one of our more turbulent colleges! I could of
course point out to him that throughout the book and especially in my
concluding chapter I had made an attempt to look at those institu-
tional combinations of exit and voice that might be optimal from the
point of view of the survival and strength of the organization. But I
admit that I had not addressed myself directly and systematically to
the possible manipulation of exit and voice as "management tools," to
use the language of our business schools. I shall try to do a little better
now, although only in connection with one particular organization: the
State.

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 44I

Fortunately, Professor Rokkan (1974) has since made a consider-

able contribution to this area through his paper. As I wrote him after he
sent it to me, I was first rather stunned to see the sweep of European
political development since the Middle Ages reinterpreted through
my concepts, but while I am still a bit puzzled about the marriage he
arranged in the process between Talcott Parsons and myself, I do find
the historical scheme he traces in the second part of his paper remark-
ably illuminating.
Let me briefly paraphrase this portion of Rokkan's argument in
order to bring it into contact with the general notion of exit and voice
as "management tools." Every state-and indeed every orga-
nization-requires for its establishment and existence some limi-
tations or ceilings on the extent of exit or of voice or of both. In other
words, there are levels of exit (disintegration) and voice (disruption)
beyond which it is impossible for an organization to exist as an
organization. At the same time, an organization needs minimal orfloor
levels of exit and voice in order to receive the necessary feedback
about its performance. Every organization thus navigates between the
Scylla of disintegration-disruption and the Charybdis of deterioration
due to lack of feedback. A territorial organization such as a national
state must by its very nature suppress exit in the form of secession
(though not necessarily the emigration of individual citizens); hence,
feedback is here bound to take primarily the form of voice. But, as
Rokkan shows, in the center of Europe the attempt to suppress territo-
rial exit-and to assert the right to control the movement of men and
commodities across borders-required so great a concentration of
effort and authority that the attempt to achieve manageably low levels
of exit led also to the crushing of voice, which was reduced in the
process to levels far below those required for long-run stability and
health. The countries of the European periphery (and a few others)
found it easier to control their borders and therefore "managed to
keep a better balance between exit controls and voice channeling
during the crucial phases of state-building" (Rokkan, 1974).
Rokkan is probably right in asserting that, particularly during some
initial phase of organization, attempts to restrict exit and to choke off
voice tend to go hand in hand and to feed on each other. I had looked
primarily at selective manipulations of one of these two levers and had
suggested that they would be engaged in by management, not in order
to receive feedback about its performance, but on the contrary to

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
442 Albert 0. Hirschman

encourage that particular reaction mode that is least unsettling to it

and least dangerous to its perpetuation in power (Exit .. ., p. 124).
Rokkan's paper provides a stimulus for going over this terrain with
greater care and perhaps somewhat greater charity toward the political
managers. For one thing, just as the process of state-building required
restricting both exit and voice, so liberalization and widening of par-
ticipation may not be possible, or may be extraordinarily difficult to
handle, unless exit and voice controls can be eased jointly. The reason
is simple: the forces of criticism and dissent that have been dammed
up by stringent voice and exit controls may be so powerful, especially
during a period of economic transformation, that, if they are released
into one channel (usually voice) only, they will exceed tolerable levels
or, at any rate, such levels as are thought to be tolerable by the rulers.
Illustrations from recent and current history come to mind im-
mediately. The history of Europe in the 19th century would probably
have been either far more turbulent or far more repressive and the
trend toward representative government much more halting, had it
not been possible for millions of people to emigrate toward the
United States and elsewhere.
This proposition represents an application to Europe of the "labor
safety valve" theory which was originally put forward in the United
States to explain, by the availability of the "frontier," the lack of
militancy of the American working class during the 19th century in
comparison to its European counterpart. While the theory has been
strongly controverted for the United States, a European safety valve
theory might well be proposed: for all the class-consciousness of the
European workers, it is well known that their revolutionary accom-
plishments did not quite come up to the expectations of, say, Marx
and Engels and one reason may be the availability of overseas emigra-
tion. Some empirical backing for such a European safety valve theory
has recently come to my attention: according to a study by J.S.
MacDonald (1963-1964) of emigration from rural Italy for the de-
cade preceding World War I, the socialist vote and labor militancy were
high in those Italian provinces that showed low rates of emigration,
and vice versa. MacDonald argues quite convincingly that the differ-
ential response to rural poverty-labor militancy in central Italy and in
Apulia and emigration in the rest of the South-can be explained by
differences in land tenure and other aspects of agricultural organiza-
tion. But at the same time, his data suggest strongly that the

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 443

availability of mass migration reduced the total amount of social

conflict with which the Italian State might otherwise have had to cope.7
It is interesting to look at contemporary non-European politics in
the light of this European experience. Today the safety valve, or outlet
for excess voice which emigration represents, is largely nonexistent,
except for the Mediterranean countries in relation to Western Europe.
This may well be a factor rendering more difficult the introduction and
maintenance of a measure of voice in the newly independent and
industrializing countries of the twentieth century.8 Similarly, the dif-
ficulties of liberalization in the countries of Eastern Europe (including
the USSR) are intensified by the insistence of their political managers
on maintaining strict exit controls.
The manipulation of exit and voice controls on the part of "man-
agement" can be constrained in yet another way. In my book, I
considered lack of feedback as the principal cost an organization
incurs as a result of suppressing voice and exit. But, at the level of the
state, a very important immediate cost, in contrast to lack of feedback
and information which is primarily a cost in the longer run, can be the
need for repression. Many countries find it entirely impossible (that is,
unacceptably expensive) to control their frontiers and some rulers
may not be willing to go beyond a certain degree of repression in
limiting or choking off voice, possibly for humane reasons, but also
because they know by now that unlimited repression brings rule by
the secret police. The immediate cost of repressing exit and voice-
admittedly, in the case of some sadistic rulers, some of this cost also
turns into a benefit-is then an important element in explaining the
behavior of states in relation to the limitation of exit and voice. I
believe, for example, that the puzzling and unique permissiveness of
the Cuban socialist regime with respect to emigration can be inter-
preted in this fashion: Fidel Castro was determined to establish an
authoritarian political order with a strictly limited amount of voice,
but at the same time he did not want Cuba to become a state with an
all-powerful secret police and, given the size of the internal opposi-
tion, he preferred to let as many disaffected Cubans as possible depart

7 I am grateful to Samuel P. Huntington for reminding me of this early study

in exit and voice.
8 In line with this reasoning, the poor countries might well demand the
opening up of immigration into the rich countries on political as well as on
purely economic grounds.

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
444 Albert 0. Hirschman

rather than having to subject them to permanent police surveillance

and worse.
The interaction of these three variables-suppression of exit, sup-
pression of voice, and repression-can also be observed in other
settings. One might even propose a theorem: a state can control only
two out of these three variables. In Cuba, Fidel Castro chose to
suppress voice and to limit the amount of repression: so he had to put
up with an unexpectedly large loss of skilled manpower as hundreds of
thousands of Cubans chose to emigrate. In Stalin's Russia, complete
suppression of exit and voice yielded repression of a size and kind that
surely had not been fully intended at the outset, while in post-Stalinist
Russia, the decision to set limits to repression, combined with the
continued strict controls on exit, has led to the voicing of considerably
more dissent than the authorities had planned for.
I do not wish to make too much of this theorem. Its merit is to
create a richer field of forces than the usual two-way alternative
between participation and repression. The trouble with the theorem is
that the freedom to exit will not always act as a brake on voice: Fidel
Castro may have been particularly lucky in that so many Cubans
thought of Miami as a potentially satisfactory second home. As we
know from Ronald Dore, in countries such as Japan the permission to
exit is likely to be as feeble a restraint on voice as the permission to
commit suicide.

Exit and Voice in Political

Parties and Politics

Political parties in polyarchies are among the rare organizations in

which both voice and exit have well-recognized, important roles to
play. They should be therefore privileged topics for the testing and
refinement of the hypotheses I developed in my book.
Before reviewing the work of others in this area, I cannot resist
pointing out that the critique of the Hotelling-Downs model which I
put forward in chapter 6 of my book was confirmed by the decision of
the Democratic National Convention to nominate George McGovern
for President in 1972. Once again, as eight years earlier in the case of
Barry Goldwater's nomination by the Republicans, it has been shown
that those members who are farthest from the center can wield con-

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 445

siderable power in the party even though, according to Hotelling-

Downs, they have "nowhere to go" and should therefore be powerless
while the party was expected to cater to the middle vote which can
wield the power of exit. My point was of course that power grows not
only out of the ability to exit, but also out of voice and that voice will
be wielded with special energy and dedication by those who have
nowhere to exit to.
One proposition I put forward about party politics dealt with the
probable amount of internal democracy within parties as opposed to
bureaucratic or machine control. I suggested that in a two-party sys-
tem articulation of opinion on the part of party members and there-
fore a degree of internal democracy are more likely to be forthcoming
than in multiparty systems because in the former ideological distance
between parties can be assumed to be greater and loyalty stronger so
that dissatisfied members will ordinarily voice rather than exit, the
opposite being true for multiparty systems. This was of course a very
general deduction from an admittedly primitive political model: in it
there is just one spectrum of opinion (left to right) and the ideological
distance from extreme left to extreme right is everywhere the same.
Clearly the world isn't that simple, as Professor Val Lorwin soon
pointed out to me in correspondence. In particular, so he stressed,
there are democracies such as Belgium and the Netherlands, whose
religious and cultural cleavages have made for a multiparty system in
which parties may occupy some position along the left-right con-
tinuum but are also, and sometimes principally, identified with a
religious or language group. In this situation, a country can obviously
have more than two parties and yet the distance between any two of
them need not be any shorter, and may in fact be larger, than in a
country with a two-party system, but without overriding cleavages. It
is therefore quite in line with exit-voice theory when Lorwin (1971)
writes in an article on the smaller European democracies (Belgium,
Netherlands, Austria, Switzerland): "The pluralism due to segmenta-
tion (= cleavage) has, on the whole, made for more, rather than for
less, participation in voluntary organization" (p. 157).
The foregoing does not mean, of course, that cleavages, and what
the Dutch call verzuiling, that is, the organization of parties, interest
groups, etc., along strictly confessional or language lines, are guaran-
tees of democracy. While verzuiling may strengthen feeling of iden-
tification and participation within parties and other organizations, it is

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
446 Albert 0. Hirschman

only too obvious that the cleavages which give rise to verzuiling, also
tear countries and communities apart: Nigeria, Pakistan, and North-
ern Ireland are among the more recent examples. Looking at the range
of these outcomes an economist cannot suppress the mechanical and
perfectly unhelpful thought that there may be some optimal degree of
verzuiling which would assure internal democracy and participation
within organizations while permitting peaceful and democratic coexis-
tence of the various segmented groups in the wider society.
My model of political parties was excessively simple and general
from a number of other points of view. For example, I did not
distinguish between the voters, the party members, and the party
leaders. Clearly the propensity to voice rather than exit can be ex-
pected to increase along this dimension. It follows that in the more
traditional European-type parties, where members are supposed to be
permanently active and represent a sizable fraction of the total vote,
one can expect voice to be more in evidence than in "electoral" or
"catchall" parties of the American type. To the extent that "catchall"
parties predominate in two-party systems, this structural factor may
then detract from the propensity toward voice that parties in a two-
party system were expected to exhibit, according to my analysis, in
comparison with multiparty systems.9
An important further complication is dealt with in an article on
party organization and strategy by Wellhofer and Hennessey (1973).
In my scheme, dissatisfaction with one's party arises exclusively on
ideological grounds as the party pursues policies that are not to the
liking of some of the membership. Another potent reason for dissatis-
faction is quite simply the failure of a party to grow and to score at
election time. In this view, a party must supply its voters with the
satisfaction to be on the winning side and its active members with the
more tangible benefits of a widening supply of party jobs and, eventu-
ally, of public offices. If members are dissatisfied with the party's
performance in these respects, their possible reactions are once again
exit or voice, and Hennessey and Wellhofer show that, contrary to
what one might expect, exit is not necessarily the dominant reaction of
those who are primarily success- and office-oriented.
In any event, the voice or exit pattern of those who are primarily
policy-oriented is likely to be quite different from those who are

9 I am indebted to Aristide Zolberg for discussions on this point.

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 447

primarily success- and office-oriented-it is easy to imagine party

moves that will arouse the former while delighting the latter and vice
versa. Interesting remarks along such lines are made in a paper by
Schlesinger (1972).
In my opinion, it may be even more realistic to assume that every
"political animal" is part ideologue and part reward-oriented, and is
therefore willing to trade off a certain amount of opportunism on the
part of the party for its power and success at the polls. Voice and exit
behavior would then be understood as a function of such trade-offs.
Similar mixtures of motivations lie behind resignations from public
office, a subject to which I addressed myself toward the end of my
book. I showed that exit from the United States Government had
fallen into excessive disuse and speculated about possible psy-
chological and institutional reasons. The record of the last few years
has, on the whole, confirmed my analysis, although there have been a
few interesting exits. An important research project in this area is now
in progress: Professors Thomas Franck of New York University and
Edward Weisband of the State University of New York at Stony
Brook have compiled and are now analyzing all resignations from the
Cabinet and from certain other top administrative positions that have
occurred since 1900 in the United States and Great Britain. A princi-
pal question they are interested in is whether resignations were ac-
companied by reasoned declarations of dissent or were silent, in
deference to some loyalty code or simply on opportunistic grounds,
and whether resignation behavior of one or the other kind had notice-
ably different effects on reentry into public office. It is my hope that
the study will lead to a better understanding of how the present
scarcity of exit behavior in the United States has come about.

Exit and Voice in the Urban Context and

in the Organization of Public Services

In the field of urban studies, the exit-voice dichotomy was obviously

one of those ideas whose time had come. Without having seen my
book, Professor Oliver P. Williams (1971) writes in Metropolitan
Political Analysis: "There are essentially two options for those who
wish to employ a location strategy to change their access within the
urban complex. They can move or they can change the characteristics
of the place they presently occupy" (p. 29).

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
448 Albert 0. Hirschman

To appreciate the change this approach means in comparison to

earlier analyses, it is useful to recall the well-known paper by Tiebout
(1956), which had celebrated mobility as making possible an efficient
allocation of public services to the consumer-citizens in a metropolitan
area. Each municipality was viewed as a firm offering a differentiated
set of services to customers with different preferences; and, in the
model, the only way in which a customer could express his prefer-
ences was by moving-there was no room in it for voting or for other
political action tending to make his own community more to his liking.
Together with the advocacy of competition in education by Milton
Friedman, which is mentioned in my book, this article can stand as the
perfect expression of the economist's bias against voice and in favor of
exit.
In the fifties it was perhaps forgivable to search for the hidden
rationality of the drive to the suburbs. In the sixties, of course, the
overt irrationalities of the phenomenon exploded. The merit of exit-
voice theory is to call attention to hitherto neglected, alternative
courses of action. Very much in this spirit, Williams calls the current
preference for exit "mobility as a substitute for formal politics"
(p. 110) and raises critical questions about the institutional framework
that has led to this abdication.
A striking case of convergence with my work is an article of Profes-
sors Orbell and Uno (1972). In 1966, they had conducted a sample
survey in Columbus, Ohio, to elicit information from residents on the
kind of neighborhood problems they were concerned about and on
how they were planning to react to whatever problems they perceived.
It turned out that people's intentions could be arranged into two
categories: either they planned to move or they intended to amelio-
rate the problems they experienced through political action. In the
process of writing up the results of their survey as an article for the
American Political Science Review, Orbell and Uno learned about my
book, came to correspond with me, and subsequently decided to use
my terminology in presenting and analyzing their data. As the survey
and my book were asking very similar questions, Orbell and Uno were
able to test a number of my hypotheses. One of those confidently
expected findings which it is nevertheless nice to see empirically
confirmed was that "blacks are more likely to voice in response to
problems than are whites of similar status who live in similar urban
areas" (p. 484). The reason for this greater propensity to voice on the

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 449

part of blacks is of course their lower mobility because of de facto

segregation of housing in numerous urban and suburban areas.
On the whole, the study confirms what we know about proneness of
whites to exit rather than to voice from urban areas as a response to
neighborhood problems; but Orbell and Uno uncovered interesting
differential behavior patterns not only for whites and blacks, but also
for higher-status and lower-status whites, with the latter particularly
prone to exit while higher-status whites are also often inclined to take
political action. Once people reach the suburbs, the pattern changes of
course radically and the first reaction to newly arising neighborhood
problems in the suburbs is voice rather than exit. This leads Orbell
and Uno to an interesting analysis of "exit-fatigue," the reality of
which anyone who has recently moved can readily confirm.
The concept of exit-fatigue leads me to open a brief parenthesis
about the general topic of exit-voice sequences. One might ask: does
exit (and subsequent entry elsewhere) lead to more exit in faster and
faster succession? Or do exit and voice typically alternate? Obviously
such questions cannot be given a uniform answer but they serve to
lead us to diverse situations and to the identification of contrasting
sequences and critical variables. For example, when highly structured
and hierarchical organizations lose their hold on some members, as
the Catholic Church did during the Reformation, or Communist par-
ties in the West during various twists of the party line, then such first
exits tend to be followed by many others as is shown by the proneness
to splintering among both Protestant sects and groups belonging to
the ex-Communist Left. On the other hand, there are cases where
exit-fatigue after a first exit leads to the determination to be "loyal," or
to use voice within the new community, as in the case of those who
have moved to the suburbs or who have emigrated to a new country
(see Exit ..., pp. 112-114). Finally there must be many sequences
about which it is difficult to have a strong intuitive a priori feeling: for
example, is a person who remarries after one divorce more or less
likely to divorce than when he or she married for the first time? I am
told that the second time around one tends to try harder to make the
marriage work; on the other hand, those who divorce once contain a
good proportion of men and women who really cannot abide married
life, but insist on trying again and again.10

10 For some data for the United States, see Bell, 1967, pp. 509 ff.

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
450 Albert 0. Hirschman

To return to the city. Both Williams and Orbell-Uno lament the

victory of exit over voice, that is, of mobility over politics which is
responsible for the deterioration of the American city over recent
decades, and both end up speculating about remedial policies. One of
the more interesting findings of Orbell-Uno is that many people seem
to be considering both exit and voice in response to neighborhood
problems so that the actual exit decision may often win out only by a
narrow margin. Hence small improvements in the attractiveness and
efficiency of voice could make a great deal of difference and the
further deterioration of the central city is not an irresistible wave of the
future after all. It could be arrested and reversed by improvements in
the political process, supplemented perhaps by economic measures
that would tax exit and subsidize nonexit. As to the political process in
the urban context, both the movement toward decentralization and
"community control," and the proposals for metropolitan integration
are relevant to the strengthening of voice, but these huge topics
obviously fall outside the scope of the present paper, as does the
closely related issue of ghetto-improvement versus ghetto-dispersal.
The exit-voice framework has also been found useful in connection
with the search for optimal ways of organizing urban public services.
In this field Dennis Young proposed in 1971, once again without prior
knowledge of my book, a systematic survey of three possible ways of
improving efficiency: one was systematic performance evaluation, the
second decentralization which can be considered a form of intensify-
ing voice, and the third competition-exit. In an introduction to a
projected reader in this field (1972) he uses the exit-voice framework
explicitly in order to look at a wide range of public services, from
taxis, garbage collection, cable television, fire protection and police, to
school systems, day care for children, medical care and criminal cor-
rection. A systematic examination of organizational alternatives yields
some surprising conclusions: for example, the introduction of a mea-
sure of competition is recommended for garbage collection,l1 whereas
decentralization and other ways of strengthening voice are advocated
for police departments. A particularly interesting combination of exit
and voice is proposed for the organization of day care for children, on

" This is consistent with the notion that competition is at its best when the
consumer knows exactly what sort of good and service he is after (see above,
p. 436).

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 451

the basis of the already noted contribution of Nelson and Krashinsky

(1972), and even more ingenious proposals for introducing some
measure of both exit and voice are made with respect to criminal
correction and prison reform.
Finally, the exit-voice framework appears to be particularly applica-
ble to certain debates around the British National Health Service. The
partisans of the NHS have taken a stand against extending some of its
benefits (cheaper medicines and laboratory tests) to those who would
avail themselves of private rather than public medicine for precisely
the reason why I advocated "locking in" of the dissatisfied customer
in certain situations. The defenders of the NHS, well aware of its
possible failings, feel that NHS needs precisely the potential exiters
-educated, vocal middle-class people-as critics within the ser-
vice; hence exit should not be made too easy or cheap for them. A
detailed review of the issue and the debates around it as well as a
thorough examination of the analytical problems involved will be
supplied in the forthcoming doctoral dissertation of Hugh B. Davies
(1973), a British graduate student in economics at the University of
Pennsylvania. Mr. Davies, whose thesis advisor is Professor
Williamson, has also extensively reformulated some of my own analy-
sis of exit and voice and, in the process, has made it more amenable to
mathematical treatment. While I naturally welcome this effort, I have
nevertheless somewhat mixed feelings about it. If Mr. Davies is suc-
cessful he may well spawn a large and increasingly complex mathemat-
ical literature with the result that it will become ever harder for me to
read the papers that will be written on exit and voice, let alone to
comment on them in the easygoing fashion which I have been able to
use this time around.

References

Arrow, KJ. 1973. Social Responsibility and Economic Efficiency.

Public Policy 21 (Summer):303-318.
Bell, R.R. 1967. Marriage and Family Interaction. Homewood, Ill.:
Dorsey Press.
Davies, H.B. 1973. Exit, Voice, and the British National Health
Service. Department of Economics, University of Pennsylvania.
(Dissertation in progress.)
Eisenberg, M.A. 1969. The Legal Roles of Shareholders and Man-

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
452 Albert O. Hirschman

agement in Modern Corporate Decision-Making. California Law

Review 57:50-52.
Harvard University. 1971. Report of the Committee on University
Relations with Corporate Enterprise. January. (Mimeo.)
Hirschman, A.O. 1963. Journeys toward Progress: Studies of Economic
Policy Making in Latin America. New York: Twentieth Century
Fund.
. 1970. Exit, Voice, and Loyalty: Response to Decline in Firms,
Organizations, and States. Cambridge, Mass.: Harvard University
Press.
. 1971. A Bias for Hope: Essays on Development and Latin
America. New Haven: Yale University Press.
Lorwin, V. 1971. Segmented Pluralism: Ideological Cleavages and
Political Cohesion in the Smaller European Democracies. Com-
parative Politics 3 (January):141-175.
MacDonald, J.S. 1963-1964. Agricultural Organization, Migration
and Labour Militancy in Rural Italy. Economic History Review,
second series, 16:61-75.
Malkiel, B.G., and Quandt, R.E. 1971. Moral Issues in Investment
Policy. Harvard Business Review (March/April):37-47.
Nelson, R.R., and Krashinsky, M. 1972. Some Questions of Optimal
Economic Organization: The Case of Day-Care for Children,
March. (Mimeo.)
Orbell, J.M., and Uno, T. 1972. A Theory of Neighborhood Problem
Solving: Political Action vs. Residential Mobility. American Politi-
cal Science Review 66 (June):471-489.
Rokkan, S. 1974. Dimensions of State Formation and Nation-
Building: A Possible Paradigm for Research on Variations within
Europe. In Tilly, C., ed., The Formation of National States in
Western Europe. Princeton: Princeton University Press.
Schlesinger, J.A. 1972. Riker versus Downs and the Definition of
Political Parties. (Mimeo.)
Simon, J.G., Powers, C.W., and Bunneman, J.P. 1972. The Ethical
Investor. New Haven: Yale University Press.
Spechler, M. 1971. The Economics on Product Quality in Soviet
Industry. Department of Economics, Harvard University. (Un-
published Dissertation.)
Tiebout, C.M. 1956. A Pure Theory of Local Expenditure.Journal of
Political Economy 64 (October):416-424.
Wellhofer, E.S., and Hennessey, T.M. 1973. Models of Political Party
Organization and Strategy: Some New Analytic Approaches. In
Crewe, I., ed., Political Elites. London: Croom, Helm.

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms
Exit, Voice, and Loyalty: Further Reflections 453

Williams, O.P. 1971. Metropolitan Political Analysis. New York: F

Press of Glencoe.
Williamson, O.E. 1971. The Vertical Integration of Production: Mar-
ket Failure Considerations. American Economic Review 61
(May): 112-123.
. 1973. Markets and Hierarchies: Some Elementary Considera-
tions. American Economic Review 63 (May):316-325.
Young, D.R. 1971. Institutional Change and the Delivery of Urban
Public Services. Policy Sciences 2:425-438.
. 1973. Introduction, in Innovation and Reform in the Eco-
nomic Organization of Urban Public Services: A Reader.
(Mimeo.)

Address correspondence to: Prof. A.O. Hirschman, School of Social Science,

Institute for Advanced Study, Princeton, NJ 08540.

This content downloaded from

31.205.8.52 on Mon, 21 Aug 2023 17:59:30 +00:00
All use subject to https://about.jstor.org/terms