When you try to promote the notion of limited government in front of an audience, academic or otherwise, you are invariably confronted with two objections:
1. How dare you propose that the government should not intervene to redistribute money in favor of the poor and provide them with health and education?
2. We obviously need the government to intervene in order to correct market failures, the most important one being environmental externalities, and in particular global warming.
At this stage, you are in trouble. The first objection (often phrased in a borderline ad hominem formulation) makes you look like an unpleasant person. The second one makes you look like an idiot.
Of course none of these objections change anything to the fact that the government implements involuntary transfers under the threat of force. From a libertarian perspective, there is no difference between environmental taxation and Greenpeace committing armed robbery in order to finance the cleaning of a river. Or between public welfare and a poor man getting 100 $ from you at gun point. Indeed, utilitarians should condone such muggings, as long as the marginal utility of wealth is larger for the criminal than for the victims.
Utilitarians know these difficulties and the reason why they lend so much legitimacy to government is because they dress its interventions under a pretense of consent. A naive concept of consent is the will of the majority, by which it is legitimate for two wolves and a lamb to vote over what to have for dinner. This parable illustrates that majority voting is not a source of legitimacy but (at best) a lesser evil, and that it makes sense only in places where a constitution imposes substantial restraints on the scope of government intervention. A less naive notion of consent is that of the “veil of ignorance”, by which we all pretend not to know who we are when making decisions. In the extreme, everybody would have the same preferences, which would conveniently turn out to be identical to some utilitarian social welfare function, with the weight on a person’s utility being equal to the probability that one is incarnated into that person’s type. If only people voted under such a veil, the allocation of resources would always maximize that social welfare function. All externalities would be addressed, and optimal redistribution would take place. In fact, such redistribution would rather be interpreted as insurance against being born poor. And all these policies would be consented to unanimously by the polity. Except that, in reality, there is no unanimity; instead there are salient conflicts of interest, which is prima facie evidence against people voting under a veil of ignorance. Therefore the question of consent seems untractable for utilitarianism. The only entity which voluntary submits to the policy prescribed by utilitarianism is an abstract being who pre-exists history and thus can vote under the veil of ignorance, or, worse, Rousseau’s “general will”.
Libertarians have exactly the opposite problem. By insisting that any transfer of resources must take place under the consent of both parties, they run into the issue that obvious gains from trade involving the coordination of a large number of agents (such as solving environmental issues) may not take place. Even enforcing private contracts requires resources, and people would free ride by trying to undercontribute to the enforcement infrastructure. It looks like operating society under consent, unless you believe that all the functions of government, including contract enforcement, can be delegated to the private sector, is a practical impossibility.
In principle, though, that is not the case. There may be a coordination problem in creating the infrastructure which is necessary for the conduct of the collective decision-making process. But once such an infrastructure exists (and historically it has been provided by the coagulation of violent armed gangs into actual governments), there is no reason why public policy should violate the natural rights of private individuals. The reason is that if a policy (such as clean air) indeed improves social welfare, it can be implemented in such a way that all individuals are better-off. In other words it should command unanimous support, and therefore will not be implemented against the consent of any citizen. Indeed, imposing such discipline would compel authorities to systematically come up with a satisfactory scheme of transfers in order to compensate the losers from their policies. And if they cannot come up with such a scheme, it means the proposed policy cannot increase social welfare.
Of course, this remains theoretical. There are plenty of informational problems involved in eliciting the costs and benefits of policies; and some voters will lack the cognitive ability to make the correct choice. Nevertheless this suggests that, if we take consent seriously, there is considerable legitimacy in imposing that public interventions should be supported by a supermajority rather than simple majority. For example, a constitution should prescribe that any policy which involves imposing coercion on individuals needs approval of, say, 75% of the electorate.
Such a supermajority rule would compel policy makers, when proposing laws that solve an externality, to actually implement a side-transfer scheme that would compensate the losers.
For such a rule to respect natural rights, it is of course necessary that the default option – the option that is implemented absent a 75% majority in favor of a proposal — obeys liberal principles and has minimal government coercion. If the initial status quo is such that the corporation of shoemakers is highly subsidized, and if shoemakers amount to 26 % of the population, a supermajority of 75 % will be insufficient to remove the subsidy. A liberal constitution should make sure that the default option remains the one with minimal government intervention. In other words the default option should be different from the status quo. If the status quo were the default option, under the influence of shocks such as technical progress, one may end up with a situation where neither consent nor natural rights are respected. Suppose, for example, that a car toll for entering London is in place, so as to reduce pollution. Suppose that clean, non polluting cars based on nuclear energy are invented. The car toll should be removed. But if it is designed in such a way that 26 % of the relevant constituency benefits, it won’t go away. Or it could go away in exchange for large transfers in favor of those beneficiaries, which is unfair (it is fair to compensate you for a violation of your property rights; it is not fair to compensate you for losing a rent based on violation of other people’s property rights). In other words, public interventions should not be designed as entitlements, but as renewable schemes that must pass the 75% majority test each time they have to be renewed.
Where does that leave the need for “social justice”? For one thing, simple majority rule itself has no reason to deliver “social justice”. Indeed the most important items of the welfare state (health, education, and pensions) cater to the median voter. It is unclear to me why, in a modern affluent society, these people could not pay for their own health, education, and pensions (which they do, to a large extent, through the taxes they pay, except that it funds other people’s health, education, and pensions). As for the welfare state items that benefit the poorest members of society, since they benefit only a minority of people, they must either be the outcome of altruism or of the objective of avoiding “social unrest”. As for altruism, it has no reason to be mediated by the government: people can and do contribute to charities. Furthermore, what is considered as poverty in advanced economies is not poverty but inconvenience. For example, in the US, 95 % of households own a car, and of the 5 % that remain, many are affluent urban families for whom owning a car is just impractical. The fraction of households that are truly needy must be small enough for voluntary transfers to take care of them. As for the concern for “social unrest”, it is an externality and therefore there must exist a redistributive scheme for tackling it that should pass the test of supermajority.