⇤
Stanford University
Abstract
This paper considers the normative assessment of bonded labor from the perspectives of lib-
ertarianism and Paretian welfare economics. I argue that neither theory can account for our objec-
tions to bonded labor arrangements; moreover, they fail in interesting ways. Reflecting on their
normative failures focuses us on other considerations besides individual choice and efficiency.
Such considerations include: the effects of labor markets on workers’ preferences and capacities;
the exploitation of the vulnerabilities of the poor; and the permanent binding of one person to
another.
⇤
Marta Sutton Weeks Professor of Ethics in Society and Political Science, Stanford University.
This paper was presented at the “Equality and Markets” Conference at Stanford University, at
BAFFLE, at the University of Toronto Law School, at the NYU School of Law Colloquium on
Law, Economics and Politics and at a conference in Ramat Gan on Labor Rights in the Era of
Globalization, organized by Yossi Dahan and Faina Milman-Sivan. I have benefited from written
and verbal comments from Barbara Fried, Josh Cohen, Yossi Dahan, Rob Reich, Seana Shiffrin,
Elizabeth Anderson, Elizabeth Hansot, Marc Fleurbaey, Meir Dan-Cohen, Eric Rakowski, Andrew
Levine, Lewis Kornhauser, Jonathan Wolff, and many other members of the audiences. Thanks
also to the editors of this journal, and to an anonymous referee.
Most anyone ought to know that a man is better off free than as a slave, even if he
did not have anything. I would rather be free and have my liberty. I fared just as
well as any white child could have fared when I was a slave, and yet I would not
give up my freedom.1
INTRODUCTION
One of the most momentous achievements of modern capitalism was the transition
from a system of bonded labor, indentured servitude and forced work to a system
of formally free contractual labor. Fifty years ago, it might have been assumed that
just as indenture and bondage declined in the industrial world, it would eventually
disappear everywhere with capitalism’s globalization. But bondage has not faded
away. In parts of the world today, labor bondage and similar practices persist under
other names (e.g., debt peonage, attached labor, serfdom, and debt slavery). In a
bonded labor arrangement, “a person is tied to a particular creditor as a laborer
for an indefinite period until some loan in the past is repaid.” 2 In practice, this
indefinite period can last a lifetime. Bonded workers are often completely servile,
forced to exhibit deference and subordination to their employers both on and off the
job.3 Even those who have defended the economic rationality of such relationships
have noted the “ugly power relations” involved in the phenomena.4 The ILO has
estimated that some 12.3 million people, many of them children, are held as bonded
laborers around the globe.5
Although bonded labor is considered by many to be a paradigm of un-free
labor and is often analogized to slavery, it is sometimes chosen voluntarily. Indeed,
while slavery itself is usually rooted in an initial act of coercion, it is not even
necessary for slavery to originate in violence and force. Slavery has been reported
to be voluntary in a number of important historical instances. 6
1
Reverend E.P Holmes, a black clergyman, testifying before a congressional committee in
1883. Quoted in ERIC FONER, NOTHING BUT FREEDOM: EMANCIPATION AND ITS LEGACY 7 (1983).
2
Pranab Bardhan, Labor Tying in a Poor Agrarian Economy: A Theoretical and Empirical
Analysis, 98(3) Q. J. ECON. 502 (1983).
3
Julie A. Schaffner, Attached Farm Labor, Limited Horizons and Servility, 47 J. DEV. ECON.
241-70 (1995).
4
PRANAB BARDHAN, THE ECONOMIC THEORY OF AGRARIAN INSTITUTIONS (1991) see ch. 12 “A
note on interlinked rural economic arrangements.”
5
A Global Alliance against Forced Labour, 54 WORLD OF WORK MAGAZINE (Aug. 2005),
available at http://www.ilo.org/wow/Articles/lang--en/WCMS_081360/index.htm.
6
Cf. ORLANDO PATTERSON, SLAVERY AND SOCIAL DEATH ch. 4 (1982).
87
How might labor bondage arise voluntarily? Since poor peasants have
no assets, they have no formal collateral. Furthermore, the wages they receive
tend to vary with the agricultural seasons: wages are lower during the lean seasons
when unemployment is high and higher during the peak growing seasons when
unemployment is low. In many cases, their survival from one harvest season to
the next will depend on borrowing for consumption during the lean seasons—they
simply do not earn enough during peak seasons to save.7
Consider the case of a landlord who offers credit in exchange for a laborer’s
agreement to pledge his future services as collateral for the loan. In so doing, the
landlord increases his power to enforce the agreement since he can directly deduct
the amount due from the worker’s wages during peak season. Furthermore, the
laborer now has access to credit that might not otherwise have been available to him.
From the perspective of contract theory, it is not evident that there should be any
legally imposed limits on competent adult landlords and laborers who seek to enter
such contracts.8 For, if agents are rational and can foresee the future consequences
of their contractual provisions, then there is no reason not to allow borrowers the
freedom to commit to providing indentured services to lenders should they lack the
resources to repay their loans.
In this Article, I examine the ways that two different frameworks with
different underlying normative assumptions view the phenomenon of bonded labor:
libertarian laissez faire theory and Paretian welfare economics. Libertarians believe
that agreements between competent adults should be respected.9 Paretians endorse
exchanges that leave both parties better off in terms of their preferences. Each
theory gives us reasons for endorsing many if not most instances of the practice of
bonded labor, reasons that are intuitively plausible. But each theory also ignores or
dismisses other considerations that might lead us to view these very same instances
of labor bondage more critically.
I argue that neither theory fully accounts for our objections to bonded labor,
objections that are codified in the laws of most developed capitalist countries.10
In the United States, for example, there are important restrictions on labor
contracts: The state does not enforce voluntary slavery contracts, debt slavery,
specific performance as a remedy for breach, or contracts that are considered
7
There are, of course, other forms of bonded labor: in factories, in prisons, and among
prostitutes. My discussion focuses on the phenomena of labor tying in rural areas.
8
See the discussion infra on children.
9
Some libertarians will be accepting of regulating transactions that have external third
party effects. See the discussion infra.
10
Indeed, many developing countries have signed the international labor conventions
against slavery, child labor and bondage, although these are often not enforced.
88
10.2202/1938-2545.1032
11
Some theorists think that we have paternalistic reasons to prevent people from entering
into voluntary, but self-destructive or degrading activities. See, e.g., the discussion of dwarf tossing
in AMY GUTMANN & DENNIS THOMPSON, DEMOCRACY AND DISAGREEMENT 252-54 (1996).
12
Admittedly, much of the phenomenon does occur precisely under the circumstances of
dire poverty and market failure.
13
JOHN STUART MILL, UTILITARIANISM 14 (1957). Mill himself considers any defense of
voluntary slavery to be self-defeating: it is self-contradictory to say we should be free not to be
free. See his discussion in ON LIBERTY (1865). This argument faces some serious problems, since
Mill defends liberty as non-interference and there is nothing paradoxical about giving someone the
ability to enslave themselves without interference.
89
I. LIBERTARIANISM
Libertarians believe in the principle of freedom of contract within the bounds of
justice.15 If two or more rational adults agree to an exchange, provided that they
are actually entitled to the goods they are exchanging and that no one else’s rights
are thereby violated, the government (and other agents) ought not to interfere. In
Robert Nozick’s pithy formulation, libertarians do not forbid “capitalist acts between
consenting adults.”16 While economists typically value free market exchange as an
instrument of efficiency, the ability to freely exchange one’s own property is seen by
libertarians as closely connected to the freedom and inviolability—the separateness
and sanctity—of the individual person.17
In evaluating the permissibility of a particular bonded labor contract, a
libertarian will consider whether the goods and services to be exchanged were
acquired by legitimate means, and whether or not the exchange was voluntary. If
these conditions are met, then according to the libertarian the exchange should be
allowed. In Anarchy State and Utopia, Nozick claims that respect for the principle
of freedom of contract entails that individuals even have the right to sell themselves
into slavery.18
Since the libertarian justification of the freedom to participate in bonded
labor or slavery contracts refers to the idea of voluntary choice, its application
seems to depend on an understanding of what makes an act of exchange voluntary
as opposed to coerced.19 What coercion consists of, however, is an elusive matter:
The distinction between offers that are coercive and those that are not is notoriously
14
See ROHINTON MISTRY, A FINE BALANCE (2001), for a novelist’s moving depiction of the
lives of India’s poor in which lifetime bondage to a creditor appears as a rational response to an
irrational world.
15
Nozick has three principles of justice: justice in acquisition, justice in transfer, and a
principle of rectification. See ROBERT NOZICK, ANARCHY STATE AND UTOPIA (1974).
16
Id. at 163.
17
NOZICK, supra note 15, at 32.
18
Id. at 331.
19
Indeed, it is curious that given the centrality of the idea of “consenting parties” to the
justification of market exchange, there has been so little attention paid to it in economics.
90
10.2202/1938-2545.1032
difficult to draw. Indeed, coercion rarely takes the form of direct compulsion that
deprives individuals of all choice. For example, when a gunman threatens “your
money or your life,” what makes the offer coercive is clearly not that you have no
power to choose. “Coerced” cannot be the opposite of “chosen,” since even in our
gunman case the victim will intentionally make the necessary physical movements
to surrender (or not) his money.
On Nozick’s account, what makes an offer coercive is that it decreases an
agent’s position with reference to their legitimate baseline situation. 20 Even though
an individual confronted with a gunman’s threat might well decide “freely” to hand
over her money, she is coerced given that the gunman has no right to her money
and by taking it he impermissibly worsens her situation, violating her baseline
entitlements.21
Nozick’s point is that coercion is essentially a normative concept.22 Two
people can agree on all the facts about an exchange and still reasonably disagree as
to whether one party to the exchange was coerced by the other. Whether or not they
each find an offer coercive is parasitic on their prior determination that the coerced
party had a baseline entitlement that was violated. In other words, in Nozick’s
view, calling an offer “coercive” or not, simply follows from one’s prior view of
what is morally justified. Thus, if I believe that a trade made under circumstances
of poverty, lack of education, and so forth, is coerced it follows that I believe that
people have a right—of some kind—not to be in such circumstances.
Libertarians, of course, do not generally think that the state has any affirmative
duty to improve the background circumstances of an individual, no matter how
bad these circumstances are. In thinking about the legitimacy of bonded labor
contracts, then, a main issue between egalitarians and libertarians is the nature of
agents’ underlying entitlements—the morally acceptable baseline for agreeing to
(or threatening not to) contract.
What kind of underlying entitlements—systems of property rights—do
libertarians think that people have? Libertarians tend to think that property rights
rest on something like the rights of first claimants.23 As long as a landlord was
20
Robert Nozick, Coercion, in PHILOSOPHY, SCIENCE AND METHOD: ESSAYS IN HONOR OF
ERNEST NAGEL 440 (Sidney Morgenbesser, Patrick Suppes, & Morton G. White eds., 1969).
21
For most libertarians, this baseline involves few, if any, positive rights. In particular, it
does not involve rights to sustenance or to aid.
22
One worry is that Nozick’s approach makes it impossible to ask whether a form of
coercion—say by the state—might be justified. For an attempt to distinguish between coercion and
compulsion see Joel Feinberg, Coercion, in ROUTLEDGE ENCYCLOPEDIA OF PHILOSOPHY 387 (Edward
Craig ed., 1988).
23
Of course, the justification for the initial appropriation of natural resources is notoriously
vexed.
91
the first to produce goods on the land or acquired property in land via a voluntary
transfer from the first claimant, libertarians will grant the landlord an exclusive right
to determine how this land is used. On the libertarian view, the state acts unjustly
if it prevents the landlord from using his land and the surplus that it generates as he
wishes. If an individual landlord wishes to loan some of his surplus to others, then
he should be free to decide the terms under which he is willing to forgo the private
use of some of his own resources. It is the landlord’s piece of good luck if he finds
others willing to accept his highly unequal terms but, on the libertarian view, this
good luck generates no injustice.24
While libertarianism is often seen as a theory that embraces and justifies
a pure form of capitalism, it is important to notice that it can also embrace and
justify a system of voluntary feudalism.25 Indeed, if a feudal lord acquired his land
by first title, and offered employment on his land to all those who wished to live
under his protection, then a libertarian would have to condemn a state that acted to
break up his monopoly power. This means that, at least in theory, libertarianism is
compatible with the permanent direct subjection of one individual to another. This
puts libertarians in a difficult position, for there seems something self-defeating about
a theory that invokes the values of individual freedom as a basis for permanently
stripping individual freedom away.26
At this point, the libertarian may protest that there are obvious rejoinders to
the charge that libertarianism endorses feudalism and feudal type relations between
people. He might point out that, in fact, much of the lords’ initial acquisition during
feudalism was based on plunder, fraud, and violence. Libertarians do not accept
as legitimate, agreements based on force or fraud. If the agreements that are today
reached between a landlord and a bonded laborer had “dirty” origins—if they are
based in employer malfeasance or maintained only by physical violence—then this
is a reason for not respecting these agreements. 27
Not all libertarians worry about the origins of property rights—rather,
some libertarians emphasize the importance of respecting individual rights in
property;; however these rights have been erected.28 But even these libertarians
recognize some limits on private property rights.29 Nozick argues that a person
24
See NOZICK, supra note 15, at 263-64.
25
Samuel R. Freeman, Illiberal Libertarians, 30(2) PHIL. & PUB. AFFAIRS 105 (2001).
26
I’ll return to the suggestion that libertarianism is not a stable theory below.
27
Some instances of bonded labor do originate in force and are maintained by violence.
See KEVIN BALES, DISPOSABLE PEOPLE: NEW SLAVERY IN THE GLOBAL ECONOMY (1999).
28
See JAMES BUCHANAN, THE LIMITS OF LIBERTY: BETWEEN ANARCHY AND LEVIATHAN 59-60
(1975) for a discussion of slavery as the result of bargaining from an anarchistic equilibrium.
29
Randy Barnett may be unique among libertarians in advocating debt slavery for those
who impose costs on others through wrongdoing. See RANDY E. BARNETT THE STRUCTURE OF LIBERTY
35 (1998). Thanks to Arthur Ripstein for the citation.
92
10.2202/1938-2545.1032
cannot legitimately acquire all of the water in the world. And some libertarians
have argued that individuals have non-alienable rights to self-ownership, so that
they cannot contract themselves into slavery.30
Let’s consider in more detail the example of a very poor laborer who has
agreed to bond himself to a landlord in order to attain a loan. In theory, a state can
set limits on the private property rights involved in this transaction in a number of
very different ways:31
(i) The state can give the laborer subsistence income, or other employment
choices, to enlarge her background alternatives;; if she still enters into the
bonded labor contract, it can refuse to enforce the contract as written.
(ii) The same as in the case above, except that if the laborer under improved
circumstances still enters into the bonded labor contract, then the state will
enforce the contract as written.
(iii) Although it has no affirmative duty to improve the background options of
the laborer, the state can refuse to enforce the bonded labor contract in any
fashion, if the laborer breaches.
(iv) The state can enforce the contract through specific performance, but alter
the substantive terms of the exchange to make it less lopsidedly favorable
to the landlord. For example, the state can set legal limits on the amount of
interest that a lender can charge for a loan.
(v) The state can refuse to enforce the contract through a specific performance
decree, but give the employer some other sort of monetary or equitable
remedy (e.g., allowing the employer to garnish the wages of the employee
when she goes to work for someone else). 32
(vi) The state can enforce the contract as written through a specific performance
decree, and in the event the employee fails to comply with it, enjoin her
from working for anyone else.
(vii) Same as above, but the state can enforce the decree by jailing the employee
if necessary.
Which of these scenarios is a libertarian committed to accept? Libertarians do
not generally hold that the state has any affirmative obligation to improve the
background circumstances of a laborer and hence they tend to reject (i), (ii), and
(iv). Libertarians believe that treating people with respect precludes the state from
Thanks to Barbara Fried for suggesting that I separate out the different ways that an
31
individual might have a right not to be in the circumstances which give rise to bonded labor.
32
In fact, specific performance is an exceptional remedy for breach of contract in employment
contracts Anglo-American law. The normal remedy is money damages.
93
compelling its citizens to transfer any of their resources to others, even if those
others are in dire need. But in practice, almost no libertarian would go all the way
to either (vi) or (vii). Nozick, for example, doesn’t;; when faced with some de facto
monopolistic markets, such as a monopoly over water in a desert, even go as far as
(v).33
In Anarchy State and Utopia, Nozick defends a version of the “Lockean
proviso” which argues that an initial act of appropriation must not leave anyone
worse off than they would have been had there been no appropriation at all.34 This
appeal to the Lockean proviso—and to welfare based restrictions on the principle of
freedom of contract—sits uncomfortably within the confines of libertarian theory.35
After all, the Nozickean-style libertarian’s commitment to liberty is in principle
meant to be independent of its consequences for human welfare. Libertarians
generally assume that individuals have no welfare rights that give them claim over
the labor or property of others. For, once we admit that the welfare consequences for
individuals can form a basis for evaluating other people’s entitlements, why should
we not contrast the private property ownership regime embraced by libertarians
with other alternatives such as a mixed economy that limits background property
rights and redistributes income? Perhaps a poor landless peasant would be better
off under some alternative form of ownership than he would be under a libertarian
regime of private property. At the very least, the peasant might be better off if his
background assets were more equal to his employer’s.
It is striking how strongly Nozick’s version of the Lockean proviso resembles
the Pareto efficiency criterion appealed to by welfare economists. But although
libertarians sometimes cite the efficiency benefits of a laissez faire economy,36 the
core of the libertarian case for private ownership and laissez faire is that these are
necessary for individual freedom, not that they are more efficient than alternatives.
If we shift the ground to consider the effects of private property and unrestrained
freedom of contract on wellbeing, we have opened the door to the consideration
of other property arrangements and market regulations that might be superior to
laissez faire on welfare grounds.
Even if we admit some version of the Lockean proviso, however, it won’t
necessarily kick in with respect to bonded labor arrangements. What if the employer’s
33
Not all de facto monopolistic markets violate Nozick’s Lockean proviso. See infra.
34
NOZICK, supra note 15, at 178.
35
Recall that Locke’s founding principle was a positive duty to protect and preserve human
life. It was this duty that was the basis of his restriction on individual property rights. See Second
Treatise of Government, in TWO TREATISES OF GOVERNMENT ch. v (Peter Laslett ed., 1988).
36
For a libertarian argument that stresses the negative efficiency consequences of government
intervention in people’s lives see MILTON FRIEDMAN, CAPITALISM AND FREEDOM (1962) and RICHARD
EPSTEIN, PRINCIPLES FOR A FREE SOCIETY: RECONCILING INDIVIDUAL LIBERTY WITH THE COMMON GOOD
(1998).
94
10.2202/1938-2545.1032
monopolizing power is based not on appropriated natural resources, like land, but
instead simply on social power and capital? Nozickean libertarians would deny the
application of the proviso at all in that situation, unless one could trace such power
all the way back to first appropriation of scarce natural resources.
Of course, the fact that libertarians think that people can legitimately enter
into bonded labor agreements does not itself entail that they think that the state
is required to enforce such agreements in any particular way (or even to enforce
them at all. Nozick, for example, says surprisingly little about the enforcement
of rights.) My point, however, is that there is nothing in libertarian theory which
rules out the state demanding that a laborer comply with the terms of his contract
by specific performance or from imprisoning him if he fails to perform his part of
the agreement.
A libertarian might yet have another objection to press against many
bonded labor arrangements. Jane Humphries has pointed out that there is no infans
economicus responding to market signals: Most children are put to work by their
parents.37 Libertarians might claim that parents are not entitled to bond the labor of
their children;; at least not into adulthood. To the extent that the libertarian defense of
the principle of freedom of contract rests on the idea that individuals are voluntarily
contracting on their own behalf, or for others who have (voluntarily) designated
them to act on their behalf, then there is room for libertarian criticism of practices
that involve parents using their children’s labor as collateral for their own loans.38
Nozick, like most libertarians, says little about children’s positive rights. In
fact, some libertarians argue that while it would be a good thing if parents cultivate
their children’s abilities, they have no obligation to do so. Their only obligation
is not to harm their children.39 However, to identify “harm” to children we have
to know the appropriate baseline for comparison. There are tricky issues here for
libertarians since young children cannot provide for their own needs but require
others to do so for them.40 Is a child harmed if she becomes malnourished through
neglect of others? Or does she have rights over the time, nurturing, and labor of
her parents?
37
Jane Humphries, Cliometrics, Child Labor, and the Industrial Revolution, 13 (3-4)
CRITICAL REV. 269 (1999).
38
Parents transact with respect to their children’s labor;; except in exceptional circumstances
children do not make the decision to work. See Debra Satz, Child Labor: A Normative Approach,
17(2) WORLD BANK ECON. REV. 297-309 (2003).
39
See NOZICK, supra note 15, at 38, arguing that children have a right not to be eaten by
their parents (!)
40
Libertarians who base their doctrine on a principle of self-ownership have struggled with
the question of why children are not owned by their begetters. Susan Okin advanced this objection
to Robert Nozick’s views in JUSTICE, GENDER AND THE FAMILY ch. 4 (1989).
95
41
Rawls considered long term stability to be critical to the justification of his two principles
of justice and devoted a large part of Theory of Justice to this concern. See A THEORY OF JUSTICE
398-99 (1999).
42
See Elizabeth Anderson, Ethical Assumptions in Economic Theory, 7 ETHICAL THEORY &
MORAL PRACTICE 345-60 (2004).
96
10.2202/1938-2545.1032
43
Cf. HAL VARIAN, MICROECONOMIC ANALYSIS 94-97 (3rd ed. 1992).
44
As its critics have noted, the concept of Pareto optimality does not deliver very much: a
state can be Pareto optimal with some people in extreme poverty and others living in extreme luxury,
as long as making the poor better off requires redistributing from the rich.
45
GERARD DEBREU, THEORY OF VALUE: AN AXIOMATIC ANALYSIS OF ECONOMIC EQUILIBRIUM
94-95 (1959).
46
See John Broome, Should Social Preferences be Consistent?, 5 ECON. & PHIL. 11 (1989).
Although the principle endorsing Pareto improvements seems attractive, there are actually some
powerful objections to it. See ANNE PHILLIPS, WHICH EQUALITIES MATTER? (1999).
47
See Kaushik Basu, The Economics and Law of Sexual Harassment in the Workplace, 17
J. ECON. PERSPECTIVES 141 (2003) for an argument that we should accept all Pareto improvements
while rejecting Pareto optimality as a normative criterion.
97
so as to avoid costly recruitment of workers in the peak season.48 He also argues
at a later date that such contracts can provide risk-averse laborers with insurance
against income fluctuations through varying seasons and risk-neutral landlords with
assured cheap labor during peak season. 49 Braverman and Stiglitz analyze how
labor-tying contracts can give incentives for agricultural laborers to work more
productively in the slack season.50 Srinivasan argues that attempts to reduce the
landlord’s power by restricting his credit activities will lower agrarian output and
make tenants worse off.51 Some economists have even argued that under non-ideal
conditions, labor bondage is Pareto optimal.52
Of course, a Paretian justification of bondage as optimal depends on
the assumption that the parties to an agreement are rational and have adequate
information—but a peasant who agrees to borrow from a lender at the interest
rate of 20% per month may not understand what he is actually agreeing to. It
also depends on the absence of transaction costs. Thus, Paretians might argue that
the circumstances of bonded labor—circumstances characterized by transaction
costs and asymmetric and imperfect information—give us reasons to endorse state
intervention in such markets, where such intervention is possible and likely to be
effective.53
48
See Pranab K. Bardhan, Wages and Employment in a Poor Agrarian Economy: A
Theoretical and Empirical Analysis, 87 J. POLITICAL ECON. 479 (1979).
49
See Pranab K. Bardhan, Labor Tying in a Poor Agrarian Economy: A Theoretical and
Empirical Analysis, 98 Q. J. ECON. 14 (1983).
50
See Avishay Braverman & Joseph E. Stiglitz, Sharecropping and the Interlinking of
Agrarian Markets, 72 AM. ECON. REV. 695 (1982).
51
On the Choice among Creditors and Bonded Labor Contracts, in THE ECONOMIC THEORY
OF AGRARIAN INSTITUTIONS 203 (P.K. Bardhan ed., 1989).
52
One reason that a Paretian might think bonded labor arrangements are Pareto optimal
under certain circumstances is that if the parties to a bonded labor agreement are free to bargain over
their property rights and have information about the effects of different labor and credit policies,
then following Coase, we can assume that an alternative allocation would have emerged if it were
more efficient. The Coase theorem states that property rights make no difference to efficiency as
long as the parties are free to bargain their way to an alternative allocation of property rights. If a
bonded laborer could be alternatively employed in a more productive way, then he would have been
able to persuade his landlord to accept a contract that did not include bonded labor.
Coase was, of course, very careful to say that his theorem held only in a world without
transaction costs. We do not live in such a world. To the extent that transactions costs prevent
parties from reaching an optimal bargain, all bets are off: there is no reason to think that the existing
system of distribution is optimal.
53
For literature on flaws in the formation of people’s preferences, see AMOS TVERSKY AND
DANIEL KAHNEMAN, AVAILABILITY: A HEURISTIC FOR JUDGING FREQUENCY AND PROBABILITY 185, 207
(1974).
98
10.2202/1938-2545.1032
What about under more ideal conditions, where there is adequate knowledge
and no transaction costs? In such a world wouldn’t Paretians have to reject
restrictions on the freedom to contract as inefficient? Doesn’t prohibiting a landlord
and peasant from contracting as they wish stand in the way of obvious gains from
trade?
One possible argument for banning such exchanges goes like this. When
an individual acts so to maximize his welfare, we should not assume that he does
so within a set of constraints that is exogenously defined. The set of choices that
an agent faces is often endogenous, affected by the choices of others. Allowing
laborers the option to bond themselves to their creditors may make alternative
options (that the laborers would actually prefer) unavailable to them.54
Consider the practice of child labor. Poor parents in developing societies
often send their children to work because they can see no other way of supporting
their families. For each individual family, child labor looks to be the family’s best
option for survival: Child labor can generate the income that keeps family members
from starving. At the same time however, the widespread availability of child labor
can serve to drive down the wages of the adult laborers, thus making child labor
necessary for every family’s survival. The institution of child labor restrains the set
of alternatives available to poor families, so that they now have no better choice than
to send their children to work. While an individual act (token) of child labor may
be Pareto improving when we consider its consequences for a single poor family,
the practice (type) of child labor may make other families worse off by changing
the range of options that are open to them.55 Interestingly, permitting child labor is
54
See Garance Genicot, Bonded Labor and Serfdom: A Paradox of Voluntary Choice, 67 J.
DEV. ECON. 101 (2002). Genicot argues that the existence of bonded labor hinders the development
of welfare enhancing credit opportunities for laborers. Given poor peasants’ lack of collateral,
asymmetries of information, and ineffective enforcement institutions, credit contracts tend to be
implicit self-enforcing agreements. The loss of future credit opportunities from lenders provides
peasant borrowers with an incentive not to default on their loans. In the absence of such incentives,
enforcement would be extremely costly, and creditors would have less reason to make such loans.
Bonded labor, by providing an alternative opportunity for obtaining credit besides formal credit
institutions, decreases the costs of reneging on the agreement. Bonded labor thereby renders the
implicit promise to repay a loan from a formal credit institution unenforceable. Genicot argues
that the existence of bonded labor can lead local credit institutions to deny loans to poor laborers;;
it might also prevent such institutions from forming or thriving. A ban on bonded labor, if this
contributes to the creation of formal credit institutions, might make some poor laborers better off by
enabling them access to credit on terms they would prefer.
55
As Derek Parfit has noted, our evaluation of a set of acts is not the same as our evaluation
of each of the acts contained in the set. Even if each single act of labor bondage is a Pareto
improvement, it does not follow that the practice of labor bondage is. See DEREK PARFIT, REASONS
AND PERSONS (1984).
99
in the interests of some employers since it provides them with a cheaper source of
labor than adults. Such employers may seek to manipulate the choice environment.56
And, once child labor is the norm, the incentive for the state to invest in education
may be diminished, thus further restricting parent’s options.
In cases of multiple equilibria, Paretianism does not yield a determinative
result. Let’s consider the limits and implications of a Paretian argument in support
of banning labor bondage, given the possibility of multiple equilibria:
First, recognizing the ways that the set of choices that an agent faces are
endogenously determined, this line of argument undercuts one classic defense of
the market. It is circular to endorse a set of market relationships on the grounds
that they (optimally) promote the satisfaction of preferences if those preferences are
themselves substantially the result of the very market arrangements under question.
More crucially, the recognition of the endogenous nature of choice sets leaves open
the question of institutional evaluation. If two different systems create and satisfy
different preferences, on what grounds should we choose between them?57
Second, the endogeneity of choice sets shows us that the distinction between
a voluntary choice and an imposed form of servitude is not so clear.58 Powerful
agents often act to restrict the set of choices open to less powerful agents, who then
choose the best option that is now available. (Indeed, the choices of poor peasants
are inevitably made against a background of property rights and market institutions
that they are not able to choose.)
Third, because the argument is made in terms of Pareto non comparable
multiple equilibria, they do not entail that it is better than not to have a legal ban on
debt bondage. Prohibiting bondage makes some agents better off and some agents
worse off than they would be otherwise;; allowing bondage makes some agents
better off and some agents worse off than they would be otherwise. Parents, for
example, who have no objection to child labor, might prefer an equilibrium outcome
with low adult wages and child labor to an equilibrium outcome without child labor
but with only somewhat higher adult wages. Indeed, many types of transactions,
including types of transactions we might well want to endorse, are likely to make
some agents worse off.59
To draw the conclusion that bans on labor bondage are justified (or bans on
child labor), we have to leave the ground of Paretian reasoning and appeal to other
56
See Debra Satz, Child Labor: A Normative Approach, 17 WORLD BANK ECON. REV. 297
(2003).
I do not mean to suggest that this problem arises only for the Paretian welfare
57
economist.
58
See Genicot, supra note 54.
59
Many transactions generate pecuniary externalities, for example.
100
10.2202/1938-2545.1032
normative considerations or standards. Broadening our conception of welfare, for
example, might lead us to consideration of questions of interest and need, as well as
to attempt to make interpersonal comparisons.
Of course, there is indeed, no reason for economists to limit themselves
to the Paretian idea of efficiency. Indeed, some economists endorse the idea of
potential Pareto improvements, in effect, cost-benefit analysis. In theory, if banning
child labor raised the productivity of labor so that winners could compensate the
losers, then the ban would be justified. One problem for this approach, of course, is
that compensatory mechanisms are rarely in place and in their absence it is unclear
why economists would accept a policy that makes some people worse off.
Alternatively, one could move away from the assumption that preference
satisfaction is the correct measure of welfare. On Amartya Sen’s view, for example,
there are certain basic functionings—“beings and doings”—that an individual needs
to achieve a certain quality of life.60 These basic functionings include nourishment,
literacy, life expectancy, satisfying work, and the ability to appear in public without
shame. Well- being or quality of life or welfare does not consist exclusively in
a person’s level of subjective preference satisfaction but is primarily a matter of
the objectively defined functionings that she actually achieves. Sen designates a
person’s “capability set” as the set of functionings that are really, that is, effectively
open to her. If we accept Sen’s view, we might try to rank alternative equilibria
in terms of the important functionings that they actually make open to people,
including such considerations as whether the contracting parties live at the mercy
of their creditors.61
Fourth, and finally, the endogenity of both preferences and choice sets in
bonded labor suggests that in some exchanges, the parties themselves –their culture,
their values and their preferences—might be partly constituted by the exchange
itself. This insight—that certain kinds of exchanges not only distribute things but
also distribute power and shape the kind of people we become—is missing from the
conventional welfare economist’s approach to markets.62 If we incorporate the idea
that our preferences and capacities are not fixed into our economic models, then we
cannot ignore the ways that a set of social arrangements is likely to change us.
60
See JEAN DRÈZE & AMARTYA SEN, HUNGER AND PUBLIC ACTION (1989);; AMARTYA SEN,
COMMODITIES AND CAPABILITIES (1985, republished OUP 1999).
61
See AMARTYA SEN, INEQUALITY RE-EXAMINED (1992).
62
Interestingly, this recognition was an important part of classical political economy’s
heterogeneous approach to markets. See Debra Satz, Nineteenth Century Political Economy, in
NINETEENTH CENTURY PHILOSOPHY (Allen Wood ed., forthcoming 2007).
101
Let’s consider bonded labor more closely. Although bonded labor can be the product
of an agreement and although it can improve a worker’s welfare, it can have other
features that make such labor arrangements (and especially state enforcement of
these arrangements) especially problematic. I have already alluded to some of
these problematic features in the case of children.
Labor bondage arises in circumstances in which some people lack the resources
to protect themselves against life threatening risks due to crop failures or seasonal
unemployment. Lenders take advantage of the vulnerability of workers in these
situations, offering credit on terms that are well below what people who were
less desperate would accept.63 The background inequality and desperation behind
this agreement arouses our suspicion that the exchange is unfair and exploitative.
Perhaps in more ideal circumstances, we could imagine bonded labor that did not
arise on the basis of a desperate exchange and which was not exploitative. But this
consideration certainly should give us pause in accepting libertarian and welfarist
arguments in favor of existing systems of labor bondage.64
Although most bondage arrangements allow for the termination of the contract
once the debt is repaid, in practice, these arrangements often last a lifetime.
Bonded laborers are not free to leave their employment, even when they can obtain
employment on better terms (and thus accelerate repayment of his debt). Agricultural
workers in India and Pakistan, where labor bondage is a frequent occurrence, rarely
pay off their debts in their own lifetimes. Indeed, when they die before their debts
are repaid, their children and grandchildren are sent to work in their place.
Why shouldn’t people be able to contract themselves to an employer for a
lifetime? It might be argued that no one should be able to bind her future self in
such a manner. Not only does an individual lack complete knowledge about her
future self, but also her conditions are likely to change in ways that she cannot now
predict.
In many of these cases, it might be argued that we have de facto monopoly pricing.
63
Is exploitation unjust? This is disputed territory, and I do not resolve the issue here. For a
64
different view, which draws a distinction between exploitation and coercion see David Zimmerman,
Coercive Wage Offers, 10(2) PHIL. & PUB. AFFAIRS 121 (1998). Thanks to one of the editors for
raising the issue.
102
10.2202/1938-2545.1032
But there is a more principled reason why someone might think that
permanent labor tying contracts should not be binding. Part of being a free person
is having the ability to use one’s powers and set and pursue one’s own purposes.
Permanent bondage contracts create a class of people who are unable to use or
regain their capacity to set their own ends;; moreover, they give this power to another
person. But no one should have that kind of power over another person: even in
ideal circumstances. I’ll elaborate on this point below when I discuss the ways that
labor bondage does and does not resemble slavery.
In practice, in the non-ideal conditions in which bondage appears, bonded workers
lack the freedom to disobey their employer’s commands, no matter how arbitrary,
humiliating, or personally costly such commands are. Workers are expected to
answer to their employer’s demands “around the clock” both in the fields and in
their own homes. Although poor peasants retain some formal control over their
bodies and their labor, in bonded labor arrangements they most often lack any
meaningful substantive control. Like the worker in a company town, the bonded
laborer lives in situation of complete dependence on his employer, vulnerable to the
employer’s whims and abuse.
Labor bondage is not only a contractual phenomenon, but it is also a psychological
one. In a study of workers held in debt bondage in northeastern Brazil, subject
workers referred to their employers as men (homens) and to themselves as goats
(cabras) perhaps indicating their social subordination.65 Researchers have also
reported on the role of social norms in preparing women and lower caste men
for lives of submissiveness and compliance. Women who have been abducted as
sexual slaves, and children who have been repeatedly sold, sometimes return to
their owners when freed.
Consider the example of Baldev, a bonded laborer who managed to free
himself through a windfall inheritance from a relative. Two years later, lacking any
preparation for freedom, he reentered bondage. In an interview, he explained:
After my wife received the money, we paid off our debt and were free to
do whatever we wanted. But I was worried all the time—what if one of our
See Schaffner, supra note 3, for a discussion of this. However, Prof. Mariana Mota
65
Prado cautioned that the language in this example may have other explanations, based on race and
on culture.
103
children got sick? What if our crop failed? What if the government wanted
some money? Since we no longer belonged to the landlord, we didn’t get
food everyday as before. Finally, I went to the landlord and asked him to
take me back. I didn’t have to borrow any money, but he agreed to let me be
his halvaha [bonded plowman] again. Now I don’t worry so much;; I know
what to do.66
Baldev places little value on his rights, exhibiting a condition that Tom Hill
has referred to as “servility.”67 A servile person is not simply a person who refuses
to press his rights in certain cases, but does not see himself as having rights in the
first place. Even when he is able to walk away, his mind is unfree, shaped by a
world in which others set ends for him.
When workers have been bonded for decades, when they know no other
reality, when their employers constantly reinforce their inferiority and the pre-
ordained nature of their bondage, it can be frightening for them to contemplate
disobedience or flight.68 Bonded workers are often purposively isolated from non-
bonded laborers and so they do not even know their rights. But even when they
know their rights intellectually, they may lack the capacity to really see themselves
as independent beings with rights and standing.
These four aspects of contemporary labor bondage are reasons not to enforce
or support such arrangements even where they arise through an agreement and
represent welfare improvements. The state has good reasons not to lend its support
to arrangements that depend on the exploitation of the vulnerabilities of the most
vulnerable, permanently bind one person to another, give one person inordinate
power over another, or undermine the capacities of laborers to stand in society
as equals. Arguably, several of these reasons are relevant to the desperate and
non-ideal circumstances under which bondage arises. But my argument also calls
attention to the need for society to produce and reproduce certain capacities in its
members, a need which would hold even under more idealized circumstances.
Every society depends on its members having the capability to behave in
ways that realize and reproduce it. Democratic societies, in particular, depend on
66
Cf. Kevin Bales, The Social Psychology of Slavery, SCIENTIFIC AMERICAN (Apr. 24,
2002).
67
See discussion in Thomas Hill, Servility and Self Respect, 57 MONIST 87 (1973).
68
Schaffner, supra note 3 argues that employers attempt to manipulate the psychology of
their employees to reduce the costs of enforcing the bondage relationship. To the extent that an
individual’s preferences are conditioned by the behavior of those that he meets daily, then employers
have an incentive to restrict the access of their servile workers to others whose presence might widen
their horizons and thereby change their sense of what kind of relationships at work are possible. Id.
at 247ff.
104
10.2202/1938-2545.1032
the ability of their citizens to operate as equals. This means not only that in such
societies people have equal rights, but also that they see themselves as having equal
basic rights;; understand and act on the requirements of justice;; and see themselves
as self authenticating sources of claims. I have already alluded to the ways that
such considerations give us an interest in how children are treated in my discussion
of libertarianism. But the examples of Baldev and other bonded laborers show us
that we have a related interest in how adults are treated.
Employment occupies a large fraction of most adult individuals’ time
and attention. Acting as a bonded laborer may lead to a failure to develop and/or
an erosion of a person’s capacities for independence and equal relations among
others. The long term experience of social subordination, dependence and servility
may dampen a person’s ability to see him or herself as a source of interests and
rights that matter and matter equally to those of others. Such a person may find
it more comfortable to allow others to make decisions for him. If labor bondage
has such effects, then democratic societies have a special interest in discouraging
or prohibiting the practice of bondage. (Indeed, they may also have an interest in
continuing the skill education for adults, as well as restructuring some forms of
non-bonded labor.69)
Although the idea that labor markets have potentially powerful effects on
the way that human beings are shaped is most often associated with the writings of
Karl Marx, it was noted as well by the classical economists, including Adam Smith.
In the Wealth of Nations, Smith wrote:
The employment of the great body of the people comes to be confined to a
few operations;; frequently to one or two. But the understandings…of men are
necessarily formed by their ordinary employments. The man whose whole
life is spent performing a few simple operations of which the effects too are
perhaps always the same … has no occasion to exert his understanding or
to exercise his invention in finding out expedients for removing difficulties
which never occur. He naturally loses, therefore, the habit of such exertion
and generally becomes as stupid and ignorant as it is possible for a human
creature to become … [he is incapable] of forming any just judgment
concerning many even of the ordinary duties of private life. Of the great and
extensive interests of his country, he is altogether incapable of judging...70
It is not enough, in thinking about these effects, to call attention to the
endogeneity of preferences I alluded to above. Our evaluative frameworks need to
go further if we are to understand what is objectionable about bondage. Not only
69
This argument may also have implications for the long term incarceration of prisoners.
70
ADAM SMITH, WEALTH OF NATIONS 2-781 (1976).
105
are preferences created through bonded laborer, but also and perhaps primarily,
certain skills and capabilities are enabled while others are hindered or eroded.71 If
egalitarians are to embrace the aim of ensuring that people have the ability to stand
in society and relate to one another as equals, then they cannot be neutral about
the effects of institutions on those skills and capabilities. Nor can they be neutral
about relationships that subject one person to the complete domination of another.
Slave contracts, bonded labor contracts, and severe inequality are objectionable
because they render one person utterly dependent on the whims of others. The
project of creating and sustaining a democratic society itself depends upon there
being the kinds of people who care about functioning as equals. Our evaluation of
institutions, including markets, then, also needs to consider their possible effects on
human motivations and aspirations.72
It might be argued that all labor contracts involve at least one of the parties surrendering
some aspects of control over her self. Furthermore, many unskilled workers who
retain the formal ability to exit from their employers have no realistic alternative
options. Moreover, it is unclear that the practice of bonded labor inherently hostile
to the capacities needed for independence and equal social relations. Aren’t there
forms of contract that while enacting “specific performance” or restrictions on a
worker’s ability to exit, are perfectly compatible with equality of social relations?
What about professional sports players tied to a particular team or soldiers in a
volunteer army?73
Undoubtedly, many of the problems associated with bonded labor have
to do not with the bondage per se, but with extreme poverty, lack of education,
imperfect information, incomplete credit markets and lack of decent alternatives
for the poor. My claim in this Article, however, is that poverty, information
71
Thanks to Seana Shiffrin for suggesting that I emphasize my expansion of the vocabulary
of endogeneity and underscore its relationship to skills as well as preferences.
72
The idea that works is a source of personal development has received support in
experimental studies. Studies have measured the effect of work organization on worker’s capacities
for independence;; attitudes towards conformity;; self-concepts;; and sense of moral responsibility.
Indeed, if workplaces are in fact schools for the development of cognitive and moral capacities,
then this has implications beyond bonded labor. See MELVIN KOHN & CARMI SCHOOLER, WORK AND
PERSONALITY: AN INQUIRY INTO THE IMPACT OF SOCIAL STRATIFICATION 55ff (1983).
73
Michael Blake raised this objection to my argument in discussion. But see BRAD SNYDER,
A WELL-PAID SLAVE: CURT FLOOD’S FLIGHT FOR FREE AGENCY IN PROFESSIONAL SPORTS (2006) for an
argument that even high pay may not compensate for the right to exit. Military service is exceptional
in ways that perhaps justify its hierarchical and authoritative structure.
106
10.2202/1938-2545.1032
problems, and inefficiency do not exhaust our potential concerns with employment
relationships, including those of contemporary bondage. Some contracts are
affronts to the equal standing of the agents by their overt content. In this category
would likely fall employment contracts that give an employer an unbounded control
over the worker’s entire days and nights, allow the employer to sexually harass his
employees, or give the employer the right to flog or otherwise physically discipline
his employees. Other contracts may be objectionable because of their scope. While
people frequently enter into agreements that bind their future selves (that’s what
most contracts do, after all), I have argued that we have reasons to reject permanent
labor tying contracts. To the extent that we value allowing people to form, revise
and act on their conceptions of value, we have strong reasons to allow people to
retain the legal right to exit (at some point) from their employment relationships
with other adults. To the extent that we seek to place limits on the power that one
person can exercise over another, and to maintain the conditions for autonomy, we
have strong reasons not to enforce permanent labor tying contracts even when they
offer fair renumeration and are the products of an agreement.
I believe that this is true even of labor tying contracts that limit the power
of employers on the job because I think that work has a special role in a person’s
conception of their good. While not everyone pursues meaning through their work,
many do.74 Because the development and exercise of one’s capacities and skills is
such an important element of human well-being, and because they are so intimately
tied up with our notions of self-respect, we have reasons to protect the opportunity
to have a range of work choices, even to people who turn out not to value having
that opportunity.
Labor movements have long recognized the centrality of concerns about
self respect. Workers have struck not only over wages and hours, but also for
the right to organize, and for the establishment of limits on the non-job related
discretionary power of their employers. More generally, liberal societies have
typically placed limits on the discretion that one person has over another: they
do not enforce “specific performance” in the majority of contracts;; they do not
enforce contracts whose terms are considered “unconscionable,” and they do not
throw people into prison for non-payment of a debt. Such societies also recognize
the right of people to divorce one another, even after they have pledged life long
fidelity.75 The prohibition on such contracts gives expression to a richer idea of
74
For discussions of the role of work in sustaining personal welfare, self-respect and social
and civic status see CYNTHIA ESTLUND, WORKING TOGETHER (2003), Vicky Schultz, Life’s Work, 100
COLUMBIA L. REV. 1881 (2000).
75
Some of the restrictions on length of contract can be explained on informational
grounds: tomorrow’s hunger cannot be felt today, etc. Thus, some restrictions have a welfare based
rationale.
107
The two dominant schools of liberal economics—Paretian welfarism and libertarian
public choice—are committed in theory to an ideal market without limits. The
normative properties of the neoclassical general equilibrium system depend on there
being markets in everything (including futures, uncertainty and the like.) The very
different contributions of the libertarian Chicago school are derived from assuming
that commodity-like relationships determine outcomes in arenas not customarily
thought of as economic, including voting and family life. Neither theory would
condemn bondage contracts—nor indeed, voluntary slavery contracts—in principle.
Instead, such arrangements are condemned insofar as they generate externalities,
reflect imperfect information, are the effects of incomplete markets, or are based on
physical force or theft.
Ironically, neither theory has the resources to fully appreciate capitalism’s
advance over its feudal predecessor. Capitalist labor markets are not facts of nature,
but are shaped by social preconditions, including underlying entitlements and social
norms. In order to transform feudal relations, capitalism had to limit property
rights and transform the ways that people related to one another, conceived of one
another and, indeed, conceived of themselves. Currently, millions of people in the
developing world have no rights against employer violence, little political voice or
civil rights, and are routinely exploited, sexually abused, traded to others, or simply
disposed of. Bonded labor thrives in societies without compulsory education (either
in law or, more typically, in practice), with weak rule of law, weak formal credit
or labor markets, weak rights of exit, and where caste-like social divisions and
conflicts are common.77 That’s not capitalism, that’s feudalism.
Banning bonded labor—or at least failing to enforce its terms—may be an
important step in creating the capacities and norms that people need for egalitarian
social relationships with one another. Even then, banning such markets is unlikely
to do that alone: creating genuinely free capitalist labor markets requires also
developing the formal credit sector of the economy, enforcing restrictions on
monopoly and monopsony, strengthening education systems, combating caste and
76
Convention Concerning the Prohibition and Immediate Action for the Elimination of the
Worst Forms of Child Labor, (ILO No. 182), ¶ 182, 2133 U.N.T.S.161, entered into force Nov. 19,
2000 emphasizes the special wrong of labor bondage a perspective that resonates with the approach
I adopt here.
77
Cf. MYRON WEINER, THE CHILD AND STATE IN INDIA (1991).
108
10.2202/1938-2545.1032
gender subordination, and perhaps having the state serve as an additional source of
credit for poor families.78
In my view, our evaluation of labor bondage contracts needs to take into
account a number of parameters missing from the standard libertarian and welfarist
economic theories of markets: the extent to which bondage depends on exploitation
of the vulnerabilities of the most vulnerable;; the extent to which bondage places one
person wholly under the power of another, the unlimited duration of contemporary
labor bondage, and finally the effect of bondage on the capabilities people need to
operate in society as equals. We also need to attend to the nature of and relation
between the contracting agents, especially when one of the agents is a child.
Both Paretianism and libertarianism operate with a very abstract idea of what
a market is. Yet, if different markets have different effects on our capabilities—
effects that differ in kind from those produced by other markets—then it is a
mistake to treat all markets in the same way. In particular, a democratic society has
a strong interest in supporting institutions that cultivate egalitarian motivations and
capacities and withholding support from institutions that cultivate subordination
and servitude, even if these institutions are not strictly illegal.79 A certain level of
income is undoubtedly needed to function as an equal in society, but we should
not focus only on income. To the extent that we endorse the idea that members of
society should interact as equals, we will have strong and principled reasons to limit
the scope of markets in labor.
78
One innovation that may help governments intervene in cushioning the vulnerability of
the poor involves group lending. Group lending seems to reduce the costs of collecting information
and enforcing repayment. The idea was pioneered by the Grameen Bank in Bangladesh and spread
to many other countries. Currently about 10 million households are served by micro-finance
programs.
79
An example would be the state’s withholding of tax exempt status from racially exclusive
or segregated private schools.
109