Exploitation and Disadvantage

Transcription

1 Exploitation and Disadvantage Benjamin Ferguson July 3, 2014 (10569) Abstract Some accounts of exploitation, most notably the degradation-based account provided by Ruth Sample (2003) and Robert Goodin s (1987) vulnerability-based account, argue that persons are exploited when the benefit they receive from a transaction falls below a certain threshold value. According to these accounts, the duty of the advantaged party to constrain their advantage in light of the other party s disadvantage is independent from the historical cause of that disadvantage. Rather, it is justified by a direct appeal to the other party s disadvantage. They argue that a failure to discharge these duties amounts to exploitation. Because the duty of constraint in these accounts does not depend on the reason for the disadvantage the duty is what I call a come-what-may duty. I show that come-what-may duties create moral hazards that can be exploited by the disadvantaged parties. In such cases these accounts of exploitation are either self-frustrating or over-demanding. I conclude that there is a tension between come-what-may duties and fairness. This tension leaves accounts that contain come-what-may duties unable to provide a coherent analysis of exploitation when the concept is understood to involve taking unfair advantage. 1

2 1 Introduction It is widely agreed that exploitation involves taking unfair advantage. 1 Of course, as Richard Arneson points out, this is not a particularly helpful analysis of the concept since there will be as many competing conceptions of exploitation as theories of what persons owe to each other by way of fair treatment (Arneson 1992: 350). Informative analyses must go further. Some accounts do so, in part, by arguing that exploitation involves a failure to exercise constraint in light of others disadvantage regardless of the cause of disadvantage. 2 According to these accounts, our duties to protect the disadvantaged and vulnerable are what I will call come-what-may duties. In this article I will argue that this approach is misguided; accounts of exploitation based on come-what-may duties are either self-frustrating or over-demanding. I consider two possible solutions to these problems and conclude that the most promising is to omit self-caused disadvantage from the scope of these duties. That is, we exploit others when we fail to exercise constraint in light of others disadvantage only if this disadvantage is the result of an injustice for which the disadvantaged is not responsible. 3 *Versions of this paper were presented at the LSE; Aarhus University s Vejle II workshop; the Decisions, Games and Logic workshop at KTH, Stockholm; and the University of Pavia. I thank those present for their comments. I also thank Paul Bou-Habib, Conrad Heilmann, Sebastian Kohler, Alice Obrecht, Serena Olsaretti, Thomas Scanlon, Hillel Steiner, Roberto Veneziani, Nicholas Vrousalis, my colleagues at the LSE, an an anonymous referee for Economics & Philosophy for helpful suggestions. 1 See for example, Elster (1982), Goodin (1987), Sample (2003), Steiner (1984, 1987), and Wertheimer (1996). A dissenting view is proposed in Wood (1995). In addition, in this paper I will focus on exploitation that occurs in transactions between individuals. My arguments do not necessarily apply to exploitation understood as a systemic phenomena affecting a particular class of individuals. 2 Examples here include Sample (2003) and Goodin (1987). 3 I intend my preferred restriction of the scope of duties to the disadvantaged to be neutral with respect to accounts of personal responsibility. More specifically, there are ordinarily many conditions necessary for holding an agent responsible for an action. For example, we might require the agent knows the possible consequences of her action, the likelihoods of these consequences, that these consequences are themselves just in the sense that the agent would accept them from behind a veil of ignorance, that the agent s identity remains stable across 2

3 In the following section I outline two accounts that include versions of these come-what-may duties: Ruth Sample s (2003) degradation-based account and Robert Goodin s (1987) vulnerability-based account. In section three, I present a case that creates problems for both accounts. In section four I consider two possible solutions to these problems, argue that omitting self-caused disadvantage from the domain of exploitable disadvantage is the best solution, and finally I address a general tension between fairness and come-what-may duties. 2 Come-what-may Duties Before outlining two approaches to exploitation that include come-what-may duties, I want to briefly clarify a potential confusion about the use of the word advantage. There are three senses of advantage implied by the claim that exploitation involves taking unfair advantage, each of which denotes a different concept: advantage as use, advantage as lack of vulnerability, and advantage as gain. When A exploits B, A takes advantage (i.e., uses) B s disadvantage (i.e., vulnerability) to acquire some advantage (i.e., gain) for himself at B s expense. Using advantage to denote each of these features quickly becomes confusing. Thus, in the arguments that follow I will use advantage and disadvantage to refer only to advantage as lack of vulnerability. I will use gain (or occasionally benefit) to refer to advantage as gain and use to refer to advantage as use. 4. Additionally, I will assume that the slogan exploitation involves taking unfair advantage is equivalent to the expanded slogan, exploitation involves unfairly using another s disadvantage for gain. time, etc. Different accounts of responsibility will include and emphasise different conditions. But in what follows I simply suppose that whatever conditions are necessary to hold the agents responsible, these conditions are in fact met. 4 Two of these three forms of advantage are noted by Allen Wood. My use of advantage corresponds to Wood s advantage exploitation our relation to that which gives us a hold or advantage over the person. My gain or benefit corresponds to Wood s benefit exploitation the benefit we receive from advantage exploitation (Wood 1995: 8) 3

4 While many of the exploited are also disadvantaged 5, the following two accounts of exploitation distinguish themselves from more restrictive accounts through their claims that mere disadvantage can generate exploitation, regardless of the particular source of disadvantage (Goodin 1987: 187). For these accounts, the historical source of disadvantage is to some degree irrelevant to whether a transaction is exploitative; for some transactions non-historical properties of a transaction itself are sufficient to ground exploitation claims Goodin s Account Of the two accounts, Goodin s implies the widest scope for our duties to the disadvantaged (as we will see below, Sample s come-what-may duties are already restricted to a sub-domain of disadvantage). According to Goodin, we exploit a person when we unfairly use of one of their attributes and, he claims, exploitation in general involves playing for advantage[-as-gain] in situations where it is inappropriate to do so... [it] consists in playing games of strategy in circumstances which render them somehow inappropriate (Goodin 1987: 184, emphasis in original). Furthermore, all cases where playing for strategy is inappropriate are manifestations of one particular kind of wrong, namely, a failure to honour the moral norm of protecting the vulnerable... regardless of the particular source of their vulnerability (Goodin 1987: 187). Thus, for Goodin, exploitation occurs when we use others vulnerabilities to secure for ourselves some form of benefit in games of strategy, no matter the source of these vulnerabilities. Taken at face value, Goodin s norm of protecting the vulnerable seems to make exploitation pervasive. It appears to render exploitative any transaction 5 Disadvantage may take many forms, but Goodin and Sample are primarily concerned with disadvantage as impoverishment, vulnerability, and disability. As discussed above, I use disadvantage as an umbrella term for vulnerability and these similar states. 6 For Sample, the historical source of disadvantage is not always irrelevant. It is only irrelevant in cases involving a failure to meet basic needs, as discussed in section

5 in which one transactor uses his or her advantage for gain. This seems odd, since (i) nearly all economic transactions involve unequal bargaining power and (ii) unequal bargaining power, ordinarily, seems to be a form of vulnerability. Yet, (iii) many ordinary transactions do not appear unfair. It seems that Goodin should provide an explanation of this oddness. He might either address (iii), showing why we should consider most transactions unfair, or provide a response to (ii), explaining why the conception of vulnerability that he has in mind is narrower than our ordinary notion of vulnerability and thus explaining why mere inequality in bargaining power is not sufficient for his notion of vulnerability. That is, though he claims that any form of vulnerability is potentially exploitable, perhaps Goodin s account of vulnerability is less inclusive then our ordinary conception. Goodin provides an account of vulnerability in his book Protecting the Vulnerable, where he writes Vulnerability is a relational notion: a full specification will tell us who is vulnerable to whom with respect to what.... [Vulnerabilites are] also relative. A is more vulnerable to B (1) the more control B has over outcomes that affect A s interests and (2) the more heavily A s interests are at stake in the outcomes that B controls (Goodin 1986: 118). Thus, for Goodin, B is vulnerable to A with respect to x when A has the ability to control x and x has an impact on B. The scope of this conception of vulnerability does not appear narrower than the scope of vulnerability in ordinary use. In lieu of a restriction of the scope of vulnerability, Goodin must provide a convincing argument for why we should believe most transactions are exploitative. And indeed, Goodin does provide such an argument, albeit one that I will argue is not fully convincing. However, before considering Goodin s 5

6 defense of his relatively expansive conception of exploitation (in section 4.2) I want to consider the come-what-may duties that appear in Sample s account and an outline of the problems these duties create for both accounts. 2.2 Sample s Account According to Sample s account of exploitation as degradation, exploitation involves interacting with another being for the sake of advantage[-as-gain] in a way that degrades or fails to respect the inherent value in that being (Sample 2003: 57). She claims these failures of respect take three general forms. We can fail to show proper respect for degrade the value of another by: D1 Taking advantage[-as-use] of an injustice done to him. D2 Neglecting what is necessary for that person s well-being or flourishing. D3 Treating as a fungible object of market exchange, an aspect of that person s being that ought not to be commodified (Sample 2003: 57, lettering added). The first form of degradation, D1, involves disadvantage caused by historical injustice: If a person is in a weaker bargaining position because of a past injustice, we stand to gain disproportionately in virtue of that injustice (Sample 2003: 82). For example, if C steals B s wallet and A uses the disadvantage created by this theft to gain more than he otherwise would have in a transaction with B, A exploits B. The second form of degradation involves a form of disadvantage that is ahistorical. Here Sample claims that well-being or flourishing are best captured by what she variously calls basic needs and capabilities. 7 7 Sample draws on Martha Nussbaum s extension of Amartya Sen s capabilities approach to wellbeing. See Nussbaum (2003). 6

7 She argues that transactions can be degrading when the [impoverished] capabilities of our interactors are ignored in the pursuit of our own advantage[- as-gain] (Sample 2003: 81). 8 Though these first two forms of degradation often coincide, neither implies the other. While B s basic needs may be unmet because of a past injustice committed by another person, D2 implies that failing to properly consider B s basic needs may generate exploitation even when the reason B s basic needs are not unmet does not involve a past injustice. Note that the existence of D2 as a separate source of exploitation-generating degradation requires cases that fall within the scope of D2 and outside the scope of D1. Finally, Sample s D3 implies that the exchange of certain goods may be intrinsically degrading. Here too there may be overlap with the first and second forms of degradation. Prostitution is often considered a paradigmatic example of this third form of transaction. When A pays for sex with B, B may transact only because she has been previously unjustly disadvantaged. A may also fail to give proper weight to B s well-being or flourishing. But Sample s inclusion of this anti-commodification condition as a separate condition implies that even if D1 and D2 are not satisfied, the transaction may be exploitative. Again, in order for D3 to operate as a separate source of degradation, there must be cases of D3 that do not fall within the scope of either of the first two domains. Like Goodin s account, Sample s account faces some face value difficulties, in particular, a version of what Rawls calls the priority problem (Rawls 1971: 8). The presence of separate degradation-generating principles means that it is theoretically possible for these principles to conflict. If these principles also conflict in practice, then absent either a single principle to guide action or a weighting between the three forms of degradation, Sample s account of exploitation appears to provide conflicting guides to action. However, unlike the 8 Note that Sample does not mean ignored in the literal sense, but rather not given proper weight. In order to succeessfully extract value from B, A must be aware of B s (lack of) capabilites, if only to use this disadvantage for his own gain. 7

8 prima facie problem of scope for Goodin s account, Sample s priority problem is not caused by the presence of a come-what-may duty. Nevertheless, as we will see the inclusion of a come-what-may duty causes its own problems for Sample. Come-what-may duties enter Sample s account through her inclusion of the basic needs condition, D2. She claims that when A transacts with B and fails to ensure B s basic needs are satisfied, he degrades and exploits her. Conditions like D2 are threshold conditions; conditions that claim exploitation occurs when B s post transaction outcome falls below a certain threshold level. 9 In Sample s case, B is exploited when post-transaction her basic needs remain unmet. Threshold conditions, when not embedded as a subset of more restrictive conditions like D1, generate come-what-may duties. When B s basic needs are unmet pretransaction, if A chooses to transact with B, then he must satisfy the constraints implied by D2 and this is so regardless of the reason that B s pre-transaction basic needs were unmet. 10 In the next section I present a case that exemplifies the over-demandingness problem for Goodin s account and that forces a conflict of duties for Sample s account. I show that both problems can be resolved by abandoning the broad come-what-may duties contained in each account. However, before doing so I want to briefly comment on Sample s non-commodification condition, D3. 9 This does not mean that exploitation is a binary concept. There may also be degrees of exploitation. Nor does this mean that all levels of exploitation require the same form of response. For example, some exploitation may be strong enough to justify state enforced prohibition, while weaker forms of exploitation do not. Rather, the claim is that exploitation (caused by a failure to respect D2) begins when we neglect others basic needs. Greater neglect of these basic needs may amount to greater exploitation. 10 Sample does not specify a criterion for determining the duties that D2 implies for A. One possibility is that if B s basic needs are unmet pre-transaction A must, ensure they are met post-transaction. But this is surely too strong. More likely is a criterion according to which A must forgo the gain he would receive from transacting, either up to the point that B s basic needs are met, or up to the point at which he would make a loss from transacting, whichever comes first. This is consistant with Sample s caveat that if the only beneficial interaction possible is one in which [basic] needs cannot be satisfied, then such interaction is not exploitative (Sample 2003: 75). 8

9 The arguments that follow pertain to the problems that arise from Sample s inclusion of cases that lie in D2, but not D1. D3 is excluded from the criticism because, without denying that they may be degrading or disrespectful, I am unconvinced transactions falling only within Sample s D3 are exploitative. When A buys sex from B, if the transaction is degrading for only the reason that sex ought not be commodified, then it seems that both A and B are degraded by the transaction. Neither is relatively disadvantaged with respect to the other. Yet, if exploitation involves taking unfair advantage-as-use of another s vulnerability, then it must be the case that one of A or B is relatively disadvantaged. While it seems natural to say that A the buyer is the advantaged party in cases of prostitution, the belief that A is unfairly advantaged is likely due to our belief that it is unfair (for reasons stemming from either D1 or D2) that B is in a position in which she must sell her services as a prostitute. If neither D1 nor D2 hold and A and B are equals in terms of power, resources, etc, and B is not disadvantaged through past injustice, it is not clear that either party is relatively disadvantaged. 11 If neither is relatively disadvantaged, there can be no exploitation. Indeed Sample makes just this point when she writes, I reject the idea that a person could be exploited if no vulnerability is made use of (Sample 2003: 83). She claims that when D1 and D2 are not violated but a person nonetheless chooses to transact in a way that would violate a putative restriction on exchange, [they are] not exploited... however, [they] may be degraded if the appropriate case against commodification can be made (Sample 2003: 83). These arguments, along with Sample s own claims, indicate that D3 is not a separate source of exploitation, but rather falls within D1 and D2. Its inclusion as a separate form of degradation plays only a rhetorical role. 11 Of course, in addition to this central concern about the inclusion of Sample s third condition, as Sample herself acknowledges, it is difficult to reach a consensus on the objective account of value that underpins D3. For additional concerns about this case see Wertheimer (2007). 9

10 3 Two Problems for Come-what-may Duties In some contexts Goodin s and Sample s claims appear plausible. In sweatshops, labour firms fail to constrain their bargaining advantage when employing the severely disadvantaged. These are paradigmatic cases of exploitation and their being so is not, so it seems, dependent on the source of these workers disadvantage. When workers are extremely vulnerable, using this vulnerability for self-enrichment seems to be a clear case of exploitation, regardless of how the workers became vulnerable. Despite this initial plausibility, as we saw above, both accounts appear to face certain difficulties and these difficulties appear more clearly in the following case. 3.1 The Sequential Game Case Consider the following sequential game, Γ 1. Stage 1: Consider the following two lotteries, L 1 and L 2. Suppose that A is presented with a choice between the lotteries, but B is (for whatever reason) committed to L 2. L 1 (p = 1 2, $100; 1 p, $0) L 2 (p = 1, $50) Stage 2: A and B face each other in the labour market. The wealthier is the employer, the less wealthy the employee. If they are equally wealthy they may work cooperatively, splitting revenue equally. Suppose the values for revenue, wages, and profit are as follows: R = $11, ω = $5, π = 6. The revenue from cooperating is $11 2, or $5.50. Finally, suppose the following constraint, embodying Sample s particular approach to D2, holds: 10

11 BN If a player chooses to transact with another, then both players outcomes must be $10. Suppose $10 is what is necessary for an individual s well-being. Constraint BN operationalises Sample s D2, the claim that we degrade others when we neglect what is necessary for their well-being. Sample argues that if we can interact with persons so that their basic needs are taken into account through the transaction, we ought to (Sample 2003: 75). Thus, employers should pay their employees enough to ensure their basic needs are met. In Γ 1, this amount is set at $10, but the problem that the case presents does not depend on the particular value given to BN. 12 In the strategic form of Γ 1 shown in figure 1 below, A and B denote the players, 0 a chance node, and C, W, and E represent the options cooperate, work for, and employ respectively. When B is committed to L 2, which lottery should A (rationally) choose? We can answer this question using backward induction. First, consider the bottom branch of figure 1, supposing A chooses L 2. In this case, the players will be equally wealthy, each receiving $50. They may choose to form a cooperative venture or refrain from cooperating, but clearly it is in the interest of each to cooperate, so the outcome of A choosing L 2 is ($55.50, $55.50) This is true in the sense that the basic needs level may be set at any point. The case does depend on BN being $10 given the other figures in the game. That is, the relative differences in the values of the payoffs associated with L 1, L 2 and BN matter. 13 I represent outcomes as: (A, B) and action profiles as A, B. The presentation and discussion that follow assume A and B are risk neutral over monetary outcomes of the lotteries. Though this assumption is admittedly unrealistic, it would be possible to construct cases for which the currency was utility rather than money. Risk-neutrality with respect to utility outcomes is less controversial. However, I have opted to present the problem in money-space because an example that relied on utilities would sacrifice simplicity for (unnecessary) realism. Finally, note that Γ 1 has two odd features. First, the production process which results in gains of only $11 is such that it is impossible for both individuals to satisfy BN without drawing from their stock of wealth. Second, these small profits mean that satisfying the BN requirement if A loses the lottery requires B to forgo a large proportion ( 10 ) of the gains from 11 production. These features do not dissolve the criticism I will outline below, they may raise 11

12 L Win 1 2 Loss A A E E W W $106, $55 B W $100, $50 $100, $50 E $10, $51 B E $0, $50 B L 2 A W $0, $50 L 2 C C $55 1 2,$551 2 A C $50, $50 B C C $50, $50 A C $50, $50 Figure 1: Γ 1 in Extended Form Figure 1: 1 (BN Holds) Suppose A chooses L 1. Then, there are two equiprobable possibilities: either A will win the lottery, or A will lose the lottery. Suppose A loses. Then after the lottery A will be poorest. He can choose to work for B or to remain unemployed and B can choose whether to employ A, but it is in the interest of both to choose employment. Since post-lottery A has $0, BN holds. B must pay A $10 in wages, but she may keep $1 for herself. The outcome of a lottery loss is thus ($10, $51). Suppose A wins the lottery. Here A is the wealthiest player. And again, it is rational for both players to choose employment. This time A is the employer and, since both players holdings are greater than $10, BN doesn t apply. A will pay B $5, keeping $6 for himself in profits. Thus, the outcome questions about the scope of the criticism. However, the problematic features for Sample s and Goodin s accounts that I will identify below remain regardless of whether the production process generates profits of $11 or, say, $111. That is, regardless of the profits generated by the productive process, come-what-may duties imply that A is able to take unfair advantage of a moral hazard. 12

13 of a lottery win is ($106, $55). Since a lottery win and a lottery loss are both equally likely, the expected value of choosing L 1 is 1 2 the outcome of a win and 1 2 the outcome of a loss, or ($58, $53). Since the expected value of L 1 is greater for A than the expected value of L 2, A will rationally choose L 1. B B L 2 L 2 (BN) A L 1 $58.00, $53.00 L 2 $55.50, $55.50 (No BN) A L 1 $55.50, $55.50 L 2 $55.50, $55.50 Figure 2: Expected values in Γ 1 with and without BN Now consider a variation of Γ 1 where BN is absent. Without BN, the outcome of A choosing L 1, losing the lottery, choosing work, and B choosing employ changes to ($5, $56), making the expected value of L 1 ($55.50, $55.50). Thus, if BN does not hold, the expected value for both A and B of A s choice of either L 1 and L 2 is equivalent. The tables in figure 2 summarise the expected value of each lottery for A and B with, and without, BN. When BN holds, if A chooses L 1, A does better and B worse. In particular, A gains $2.50 of expected value at B s expense. With BN in place, A has an incentive to choose L 1 over L 2. Without BN the outcomes for each player are the same regardless of whether A chooses L 1 or L 2. With these facts in mind, let us return to the two accounts of exploitation outlined above, this time beginning with Sample s account. 3.2 Sample s Self-Frustration Problem Consider Γ 1 in the context of Sample s account. According to Sample, if A chooses L 1 and loses, then if B chooses to employ A, but does not respect BN, B exploits A. When B fails to ensure A s basic needs are met when she interacts with him, B engages in the form of degradation specified by D2, and 13

14 thus exploits A. However, pre-theoretically it also seems natural to say that the reverse is true, that is, A exploits B. There are two senses in which A extracts benefits from B in Γ 1. First, A gains ex ante from BN at B s expense through the increase in expected value that BN gives to A s choice of L 1. Second, when A chooses L 1, loses, and then works for B, he receives ex post benefits in the form of the BN subsidy: his wages increase from $5 without the subsidy to $10 with the subsidy. If either of these gains are also unfair, then in a pre-theoretical sense, A exploits B. Note that because A receives these benefits at separate times, the locus of the exploitative act (assuming the gains are, in fact, unfair) depends on whether the ex post or ex ante perspective is adopted. This leaves us with three questions: Which perspective on gain ex post or ex ante is morally relevant for exploitation? Are the gains received in either perspective unfair? If so, is the resulting exploitation problematic for Sample s account? Unfairness. It is easiest to begin with the second question: are either ex ante or ex post gains unfair. First consider A s ex ante gains. There are three features of A s ex ante gain that, taken together, motivate the claim that it is unfair. First, A s pre-choice feasible set contains the option in B s (L 2 ), plus the risky option L 1. According to Sample s own preferred capabilities-based approach to welfare, A s pre-choice welfare is greater than B s. 14 Second, the presence of BN 14 This is so because B s choice set is a strict subset of A s; consequently, the capabilities enabled by A s larger choice set are greater than those enjoyed by B. A s welfare is, of course, greater than B s on many other accounts of well-being as well. 14

15 means that if A chooses L 1, regardless of whether he wins or loses the lottery, B will do worse than she would have done had A chosen L 2. If A wins the lottery, then in the second stage B will work for A and B s outcome will be $55; if A loses the lottery, then B will employ A but, because she must respect BN, her outcome will only be $51. Both outcomes are less than the $55.50 B would receive if A chose L 2. Finally, because BN increases the expected value of L 1 it creates a moral hazard that incentivises risk-taking. This occurs because, for A, BN is costless. Although the constraint requires B to subsidise A s losses, it does not require that A share his gains with B. In other words, BN requires B to provide A with insurance for which A does not pay. When A, who is already better off than B, chooses L 1 he unfairly uses the moral hazard created by BN in order to (attempt) to enrich himself at B s (certain) expense. It seems clear that A s ex ante gain is unfair to B. 15 Now consider ex post gain. Suppose A chooses L 1 and loses. Although he was better off than B before the lottery, he is now worse off. Having lost the lottery, is his ex post benefit his actual receipt of the subsidy really unfair? Sample clearly thinks not. Respecting other persons should, so the argument goes, commit us to ensuring that our interactors basic needs are in fact met, not that they might have been met had things gone differently. To claim the latter offers no consolation to the person who is starving now. Although this line of reasoning is, perhaps, initially attractive, it should be resisted. If A s ex ante gain is unfair, then A s ex post gain must also be unfair. This is so because the unfairness of the ex ante gain derives from the fact that A is in a position in which he could receive a subsidy for which he does not pay. If it is unfair for A to knowingly to put himself in a position where this might happen, it must 15 I should emphasise that the primary motivation for the intuition that A treates B unfairly is that A takes advantage of a moral hazard, and not that his welfare is, pre-choice, greater than B s. However, this latter point may be a minor contributing factor to the intuition of unfairness. 15

16 also be unfair for him to actually accept the subsidy. Therefore, both A s ex ante and his ex post gains are unfair. Two Problems. Consider the third question: when A gains unfairly at B s expense in either the ex ante or ex post sense does Sample s account encounter problems? The short answer is yes. To see why, let s start again with ex ante gain. A receives unfair ex ante gain of B when he chooses L 1. If exploitation involves the impermissible use of others disadvantage to gain unfairly from them, then it is impermissible for A to choose L 1. The reason L 1 is impermissible is that it is unfair to B. Further, this obligation to refrain from choosing L 1 does not depend on how valuable L 1 is to A. Suppose A s choice is not between L 1 and L 2, but between L 3, where p = 0.99, $500; 1 p, $0, and L 2. L 3 is a very attractive lottery from which A stands to gain much ($500) with little risk of calamity (1%). I argued three facts make it unfair for A to receive ex ante benefits when he chooses L 1 : (i) pre-choice A is better off than B, and yet (ii) his choice of L 1 harms (the less well-off) B and (iii) the requirement that B subsidise A with BN means that A receives insurance (BN) at B s expense for which he does not pay. These three facts apply also to L 3. Indeed, under L 3 (i), the inequality in pre-choice welfare, is much greater. Not only is it unfair for A receive ex ante benefits from B when faced with the L 1 lottery, A must refrain from taking a risk that leads to his being subsidised by B even when this risk is very prudent. Therefore, if the relevant form of gain for exploitation is ex ante gain, Sample s account is over-demanding. But perhaps ex ante gain is not the form of gain involved in exploitation. Suppose then, that exploitation involves unfair ex post gain. Here too Sample s account faces a problem. If the relevant form of gain is ex post gain and, as I have already argued, such gain is unfair, then A exploits B when he receives 16

17 the BN subsidy. However, according to Sample s own account, B exploits A if she fails to pay the subsidy: B s failure to ensure A s basic needs are met violates D2. If A exploits B when he receives BN and B exploits A if she fails to pay BN, then in order for each to avoid exploiting the other, they must allow themselves to be exploited. And conversely, in order to avoid exploitation, they each must allow the other to exploit. In this case, Sample s account is self-frustrating because it implies conflicting obligations. In a discussion of a related case, Hillel Steiner writes, both A and B are each obligated to employ all available measures necessary to remove the obstacle constituted by the other person s [not exploiting]. And this obligation persists until the point where either (i) one of them has actually prevailed... and the other is incapable of further attempting, or (ii) both are [so] incapable... (Steiner 1994: ). Sample s account provides no way of determining who should concede. There is no way to adjudicate between either party s attempt to not exploit. If ex post gain is unfair, then Sample s account implies a self-frustrating moral code that, again quoting Steiner makes [non-exploitative] performances obligatory and, at the same time,... requires (in certain circumstances) the deliberate frustration of any [non-exploitative] performance for the sake of no other principle or value (Steiner 1994: 199, emphasis in original). 16 Note that self-frustration is not the same as mutual exploitation. The claim is not that both A and B exploit the other. Rather, A s exploiting B and B s exploiting A in Γ 1 are mutually exclusive actions. The performance of one action precludes the other. So, Sample s account encounters problems regardless of whether the form of gain relevant to exploitation is ex ante gain or ex post gain. In the case of 16 Note that in one sense, this conflict can be avoided: A and B can simply avoid any kind of economic interaction. Because the obligation to transact non-exploitatively is a conditional obligation that arises only if one chooses to transact avoiding transactions also entails avoiding exploitations. But this does not dissolve the self-frustration problem for Sample s account, for we want to know what transactors obligations are, should they choose to transact. Sample claims that she is And here Sample s account implies conditional obligations to both φ and φ. 17

18 the former, her account is over-demanding; in the case of the latter it is selffrustrating. Before I address the problem that Γ 1 creates for Goodin s account, I will consider the final question: which form of gain is relevant to exploitation. Moral Relevance. Suppose A chooses L 1 and receives unfair ex ante gain at B s expense. If exploitation involves unfair ex ante gain, then having chosen L 1, a uses the disadvantage BN creates for B to extract unfair gain at B s expense. That is, A has exploited B. But now suppose that though A initially intends to accept BN, upon losing the lottery he has a change of heart and rejects the subsidy. Though A has received unfair ex ante gain, it does not seem that any exploitation has occurred because A has not received any actual benefit. A s choice of L 1 is akin to attempted exploitation. 17 Although his disregard for the unfair effects of his choice of L 1 on B likely makes this choice wrongful, the choice of L 1 is not itself exploitation. The issue here is not one of permissibility, but of classification. Since (actual) maldistribution is necessary for exploitation and this only occurs when A accepts BN, the form of gain morally relevant for exploitation is ex post gain. And, this, along with my arguments above, means that the problem Sample s account faces is one of self-frustration. Revisiting the Priority Problem. In my outline of Sample s account I indicated that absent either a metaprinciple or a weighting of the three forms of degradation Sample s principles could conflict. While I trust the self-frustration problem I have identified is itself straightforward, it is, perhaps less clear how self-frustration is related to the priority problem. There are two possible sources of confusion. The first concerns the way the priority problem is related to self-frustration. 17 But note that A s choice of L 1 is only akin to attempted exploitation. It cannot actually be described as attempted exploitation since in choosing L 1, A does not intend to receive BN. Rather, he hopes to win the lottery; his receipt of BN is only a possible (and forseeable) outcome of his pursuit of the welfare maximising outcome. 18

19 The self-frustration problem involves the following conflicting claims: F1 B exploits A if she fails to provide A with the BN subsidy. F2 When A chooses the risky option, loses, but nevertheless accepts the subsidy from B, A exploits B. While D2 implies F1, it is not clear that any of Sample s existing three forms of disrespect imply F2. In this case, Sample s account cannot suffer from an internal self-frustration problem. The conflict would instead involve D2 and an external principle. If this claim is true it does not spell the end for Sample s problems. A s actions in F2 are a paradigmatic form of exploitation that Sample should want to capture. If her account cannot account for F2, then it seems that it is incomplete because it fails to capture a case of taking unfair advantage: free-riding on moral hazards. Regardless of whether the problem is an internal problem of self-frustration or an external problem of scope the case presented in Γ 1 is problematic for Sample s account. However, of these two possibilities, I believe the internal conflict is a more accurate, and indeed, more charitable description of the problem that Sample s account faces. While F2 may not be prohibited by any of the existing three forms of disrespect, it does appear to be implied by Sample s general claim from which she derives these three forms of disrespect. Sample writes exploitation involves interacting with another being for advantage[-as-gain] in a way that degrades or fails to respect the inherent value in that being (Sample 2003: 57). Surely using others commitment to obligations like BN for self-enrichment at the others expense involves a serious failure of respect both for their person and for the spirit of the moral commitment. If this is so, then a fourth form of disrespect is implied by Sample s general claim that exploitation involves disrespect or degradation and in this case, her account faces a problem of self- 19

20 frustration: D2 implies F1 and the fourth form of disrespect, D4, implies F2. 18 The second possible source of confusion concerns the way Sample s principles conflict. Initially it might seem that conflict is not possible. Perhaps, D1, D2, and D3 are all (along with some structural conditions) sufficient for exploitation. If multiple forms of degradation were present in a transaction, then the exploitation would simply be over determined; if A were guilty of D1 and B of D2, then they would mutually exploit each other. It is true that D1 D3 need not conflict, but it is not true that they cannot conflict. If B desperately sells a kidney to A because all of her possessions were stolen and after the sale her basic needs remain unmet, all three forms of degradation are satisfied. However, the problem presented by the conflicting claims F1 and F2 is different. Cases like that just described, all forms of degradation coincide with one person. But in cases like Γ 1 one person is degraded by one of Sample s domains of disrespect and the other by another domain. It is in this type of case that potential conflicts can arise. In order to avoid exploiting B, A must ensure he does not disrespect her and vice versa. But if these avoidances of disrespect require mutually exclusive actions, then exploitation cannot be avoided and the account is self-frustrating. 3.3 Goodin s Over-demandingness Problem I now turn my attention to the problems Γ 1 creates for Goodin s account. Recall that for Goodin exploitation involves a failure to respect the duty to protect the vulnerable, regardless of the particular source of their vulnerability (Goodin 1987: 187). We exploit others when we use their vulnerabilities to secure for ourselves some form of benefit in a game of strategy. The vulnerable are those whose interests are strongly affected by our actions and choices (Goodin 1987: 18 Of course, as mentioned above, Sample can attempt deny F2 involves any form of disrespect, but in this case faces the incompleteness problem. 20

21 191). Further, Goodin argues that the responsibility to protect the vulnerable falls upon whoever has the last clear chance to prevent harm... regardless of the causal history up to that point (Goodin 1986: 129). According to Goodin then, if A chooses L 1, loses, and decides to work for B, A is vulnerable to B. If B fails to protect A when she transacts with him, she exploits him by failing to honour the moral norm of protecting the vulnerable. On Goodin s account, determining whether B has protected A is less straightforward because, unlike Sample, Goodin does not appeal to a particular baseline (such as basic needs) by which we can judge whether B has discharged her duty to protect A. Nevertheless, Goodin is clear that A s vulnerability places some constraints on B; she cannot simply ignore A s plight. In the absence of an explicit protection criteria, I will assume these constraints are captured by the level stipulated by the BN condition, though in section four I will consider alternative operationalisations of what it means for persons to protect others. Using the BN constraint as a baseline for B s protection of the vulnerable A, it follows that when B does not respect the BN constraint, B exploits A. This will strike those who are convinced by my prior claims about the fairness of A s gains as the wrong result. Rather than B exploiting A, it seems that A exploits B. A case involving a spiteful A who hoped to lose the lottery in order to worsen B s outcome makes an even stronger case for the claim that it is A who exploits B. However, Goodin s position not only entails that B exploits A when she fails to subsidise him, it also entails that A exploits B in Γ 1 when he chooses L 1. If, as Goodin claims, the vulnerable are those whose interests are affected by our choices, then B is vulnerable to A with respect to A s choice of L 1, since A s potential loss will negatively affect B s interests. Goodin s account allows us to claim that when A chooses L 1, A exploits B Although in my discussion of Sample s account I concluded that A s choice of L 1 cannot itself be exploitation, this result is not supported by Goodin s vulnerability account. If exploitation involves a failure to protect the vulnerable, then since A fails to protect B when 21

22 Though it is possible to conclude from Goodin s account that A exploits B and that B exploits A, his account is not self-frustrating. The account implies only that A exploits B when he chooses L 1 and that B exploits A at a later time, when she fails to subsidise him. Since these acts are temporally distinct and not mutually exclusive, Goodin s account avoids the self-frustration problem. Nevertheless, there is something odd about the application of Goodin s account to the case. When A loses the lottery and works for B, A is vulnerable to B because he is now impoverished this kind of vulnerability is fairly typical in cases of exploitation. However, B s vulnerability to A s choice of L 1 is an odder sort. It is a vulnerability that is itself created by Goodin s come-whatmay duty to protect the vulnerable. Since B must constrain his actions towards A if A loses the lottery, this obligation makes him vulnerable to A s choice of the risky lottery. And were A to choose the certain option (L 2 ), B would not be required to constrain his actions towards A because A would not be vulnerable to B. Thus, it is the presence of the unconditional duty to protect (here operationalised as BN) that generates B s vulnerability. So, although Goodin s account is not strictly self-frustrating, it is self-undermining. It prohibits an act (A s choosing L 1 ) because a possible outcome of this act (A s loss) implies an obligation (BN) that the theory, at an earlier stage, repudiates. That a theory is self-undermining is not necessarily a reason for abandoning it, though in Goodin s case it erodes support for an obligation that is already suspect. Setting this issue aside though, the primary problem for Goodin s account is that it is over-demanding in the way that Sample s account would have been, had it implied A s choice of L 1 was impermissible. As with Sample s account, the features of Goodin s account that make A s choice of L 1 exploitation apply also to any case in which A faces a gamble between a choice that offers him an attractive outcome and one that leaves him vulnerable to B. For even in choosing L 1 he exploits her under Goodin s conception. 22

23 the case of the prudent lottery L 3, B is vulnerable to A s choice. If A chooses L 3 and looses, then Goodin s duty to protect the vulnerable regardless of the source of their vulnerabilities requires that B constrain herself when transacting with A. Since B is vulnerable to A, then according to Goodin, A has a duty to do whatever [he] can to prevent anyone (ourselves or others) from taking unfair advantage of [B s] vulnerabilities (Goodin 1987: 189). Thus, A must refrain from choosing the (very prudential) lottery L 3. Indeed, in many cases quite reasonable and morally uncontroversial actions (driving, hunting, playing sport) carry very small probabilities of catastrophic vulnerable outcomes that oblige others to do whatever they can to help us and thus, make these others vulnerable to our choices. Not only is the prohibition of these prudent risks over-demanding, the self-undermining property of the account means that this prohibition is motivated by the obligation to avoid a consequence that Goodin s own theory obliges. 4 Solutions and Responses Both the over-demandingness of Goodin s account and the self-frustration of Sample s are caused by the approaches come-what-may duties. Broadly speaking, both Sample s and Goodin s accounts claim: S1 certain states of disadvantage generate duties 20 to aid and S2 exploitation involves a failure to discharge these duties. When the states of disadvantage in S1 are unconstrained and imply obligations like BN, then the claim in S2 that a failure to discharge these obligations amounts to exploitation encounters either the self-frustration problem or the over-demandingness objection. These two problems can be solved by altering 20 In Sample s case these duties are conditional on A s choosing to transact with B. 23

24 either S1 or S2. We might abandon Goodin s claim that any source of vulnerability can generate duties by limiting the states in S1 or the duties they imply. Or we might accept the accounts claims about the obligating states and the reach of come-what-may duties, but deny S2 s claim that a failure to discharge these duties amounts to exploitation. In this section I will consider these possibilities alongside Goodin s defence of come-what-may duties. 4.1 Abandoning Self-caused Disadvantage If exploitation involves taking unfair advantage, then in order for A to exploit B, A must have some kind of relative advantage over B; that is, B must be relatively disadvantaged compared to A. In a broad sense, there are three possible sources of disadvantage. B may be disadvantaged because of her own actions (self-caused disadvantage), the actions of others (other-caused disadvantage), or a natural event involving no human action (natural disadvantage). Though there are three historical sources of disadvantage, there are many different manifestations of disadvantage. B may be disadvantaged because she lacks material goods, because she is malnourished, because she has stronger (or weaker) preferences than A, and so on. The states of disadvantage in S1 that generate duties to others can be limited by restricting the historical sources of disadvantage, by restricting the manifestations of disadvantage, or by restricting both. Goodin s account limits neither the sources, nor the manifestations of disadvantage. His claim that exploitation involves a failure to protect the vulnerable regardless of the particular source of their vulnerability is an explicit rejection of limits on the sources of vulnerability (Goodin 1987: 187). That Goodin also places no limits on vulnerability s manifestations follows from his account of vulnerability. As we saw above, Goodin claims that B is vulnerable to A with respect to x when A has the ability to control x and x has an impact on B. Goodin s account of exploitation places no explicit limits on x. Exploitation in- 24

25 volves a failure to protect others from any manifestation of vulnerability. Unlike Goodin, Sample does provide a restriction on the manifestations of disadvantage. Her D2 condition restricts the exploitable manifestations of disadvantage to cases in which persons basic needs are unmet. However, Sample does not limit the sources of disadvantage and her restriction on the manifestations of disadvantage does not succeed in avoiding the self-frustration problem. Perhaps the most obvious solution to both self-frustration and over-demandingness is to omit self-caused disadvantage as a possible source of exploitation. If self-caused disadvantage did not generate obligations, then in Γ 1 B would not be required to provide A with some form of subsidy and would thus not be vulnerable to A s choice of L 1. If B is not vulnerable to A s choice of L 1, then on Goodin s account A is no longer obliged to refrain from choosing the risky option. And when A is no longer obliged to refrain from choosing the risky option, the over-demandingness objection disappears. Similarly, Sample s claim that disrespect may be caused by a failure to ensure a transactor s basic needs are met, regardless of the reason that they are unmet, includes self-caused disadvantage as a form of exploitable disadvantage. In the context of Sample s account, B s failure to subsidise A in Γ 1 is a case of exploitation because B degrades A by failing to ensure his basic needs are met when she transacts with him. Recall that self-frustration involves the following conflicting claims: F1 B exploits A if she fails to provide A with the BN subsidy. F2 When A chooses the risky option, loses, but nevertheless accepts the subsidy from B, A exploits B. If self-caused disadvantage is excluded as a form of exploitable disadvantage, then F1 is false. B is not obliged to respect the BN constraint when A is responsible for his basic needs going unmet. In this case, B does not exploit A 25