Not So Super Majorities

We all want to live under good constitutions… whatever good really means. But how do we make sure that our constitution is, in fact, good? In a post at the Volokh Conspiracy (part of a series discussing their book on originalism), John McGinnis and Michael Rappaport argue that that “stringent supe[r]majority rules provide the best way to make a national constitution.” A constitution, they say, should be enacted by a super-majority and placed beyond the reach of amendment, except again by supermajority. I do not think that they succeed in demonstrating this claim.

First, it is important to distinguish two questions which professors McGinnis and Rappaport seem to run together. One concerns the best way to enact a new constitution; the other concerns the best way to amend an existing one. To be sure, this distinction is somewhat artificial and can be difficult to make; at some point an amendment might be so fundamental as to be equivalent to the making of a new constitution. Nevertheless, there seems nothing wrong with the idea that, say, the unanimous agreement of formerly independent states is required to create a new federal constitution uniting them, which will subsequently be amendable without their unanimous consent. The opposite move ― the creation of a constitution with lower support than would be required to amend it ― might seem sneaky, but I’m not sure that it is actually wrong in all circumstances. In Canada, the Constitution Act, 1982, was entrenched with the support of nine provinces, yet it requires the agreement of all ten for some types of constitutional amendment. I am not aware of anyone having criticized it for that reason.

Be that as it may, professors McGinnis and Rappaport seem to think that both the initial creation of a good constitution and the making of good amendments thereto require supermajority agreement. They make four arguments in support of this view. First, [“s]upermajority rules … screen norms for substantial consensus and bipartisan support,” which, in turn, “creates legitimacy and allegiance as citizens come to regard the Constitution as part of their common bond.” Second, “supermajority rules … encourag[e] richer deliberation about the Constitution.” Third, supermajority requirements ensure that citizens know that they might not be able to change the constitutional rules when they apply to them, and hence are more likely to settle on rules that are fair to all and express the public interest rather than partisan advantage. And fourth, “[s]upermajority rules also generate constitutions that are more likely to protect minorities,” since minorities are able to block the enactment of rules that do not protect them.

I do not think that any of these arguments proves that supermajority rules are either necessary or sufficient to make good constitutions. To keep this post from being much too long, I will mostly focus on the first claim ― that supermajority agreement on the contents of a constitution makes it more legitimate. However, here some quick thoughts about the other ones. Supermajority rules can, indeed, encourage richer deliberation, but they can just as well encourage unprincipled compromise, such as the “trading of fish for rights” that preceded the enactment of the Constitution Act, 1982 ― or, say, the compromise that preserved slavery and even give slave-owning states disproportionate political power under the U.S. Constitution as originally enacted. For the same reason, supermajority rules do not guarantee that a constitution will respect the public interest ― they might simply encourage horse-trading between partisan factions instead. And while supermajority rules can protect minorities, they can also give them the power to extract disproportionate advantages or protections that are larger than necessary and unduly impede collective action.

To return now to the claim that the legitimacy of and citizens’ loyalty to a constitution depend on the breadth of the consensus on its contents, it seems to me that it leads to absurd consequences. Most obviously, it means no only that supermajority is better than simple majority, but also that unanimity is better than any other supermajority. Yet professors McGinnis and Rappaport are not arguing for unanimity. Perhaps that is merely because it would be impractical. Perhaps also there is, in fact, a diminishing return on additional support, at least past a certain threshold. Yet it is not clear where that threshold lies, and whether it does in fact lie in supermajority territory. For some purposes―including elections in the Canada as well as in the U.S. ―we accept a plurality, not even a simple majority as sufficient for the win. Indeed, it is possible for a party or a candidate to win such elections without even a plurality of the national popular vote. It is reasonable to demand that a constitution, expected to endure for decades and even centuries, enjoy higher support than a politician elected to hold office for four or five years, but it is by no means clear just how much higher.

However, there is an even more fundamental problem with the argument that constitutions are best enacted and amended by supermajorities because their legitimacy is a function of the consensus they generate. If a constitution (however enacted) can only be amended by a supermajority, then it is quite possible for a constitutional provision to remain in force even though a majority of the people come to think that it is a rotten one. The more stringent the supermajority requirement, the wider consensus the can be on the need for constitutional amendment without the amendment being realized. From the standpoint of a consensus on the substance of constitutional provisions, the constitution is in such circumstances quite illegitimate, and thus supermajority requirements make illegitimate constitutions more, rather than less, likely.

I am inclined to think that what makes a constitution ― or, for that matter, a government ― legitimate is not the breadth of substantive agreement with it, but agreement on the procedures that led to its creation. This is what explains the continuing legitimacy of constitutional provisions with which a majority of citizens ― but one not sufficient to overcome a supermajority requirement ― disagree with. The focus on the contents of constitutional rules imposes too high a threshold for their legitimacy.

What I have said so far means that supermajority rules are not sufficient to make a good constitution. They are also not necessary. It is, surely, possible to arrive at good rules (whether by this we mean rules that generate broad agreement, or are conducive to the public interest, or protect minorities) by other procedures too, whether simple majority, through the development of constitutional conventions, or through adjudication. (On this last point, professors McGinnis and Rappaport recognize, in a subsequent post, that in some cases, judicial “precedent now enjoys such strong support that it is comparable to that necessary to pass a constitutional amendment.”)

The question of what procedure, if any, is the best for making a good constitution is clearly a difficult one. I have not attempted to answer it in this post. I do think, however, that professors McGinnis and Rappaport have not succeeded in demonstrating that supermajority enactment and amendment is that procedure. Supermajority requirements have some benefits, but also considerable flaws.

In the Trenches

Here’s the second part of my comment on C.J. Peters’ claim, in a recent and very interesting blog post, that constitutional law consists of “rules that are both entrenched and secondary.” I argued yesterday that constitutional rules often are primary rules (in H.L.A. Hart’s sense) or at least have a “double aspect” and are both secondary and primary rules. I turn today to the question of the entrenchment of constitutional law.

Prof. Peters defines “‘entrenched’ rules” as those “that are especially difficult to eliminate or change,” “typical[ly]” because they cannot be amended by “the ordinary legislative process.” He also says that such rules need not compiled in “canonical written texts,” such as the U.S. Constitution (or Canada’s Constitution Act, 1867 and Constitution Act, 1982). The important thing is that the rules that define democratic law-making processes not be subject to easy modification by these same processes:

We cannot have a functioning system of democracy if we are continually debating the details of how laws get made, who has the authority to interpret and enforce them, and so on.

Even countries without what is usually (and sloppily) called a “written constitution,” such as the United Kingdom, have “constitutional law” (emphasis prof. Peters’), the rules of which

take the form of entrenched statutes, traditional institutions, and established procedures rather than provisions of a single canonical document.

Constitutional law can sometimes be incomplete or uncertain in the details, but it must exist and do at least a tolerable job of defining law-making processes.

That much I have no quarrel with, at least if we understand “constitutional law” broadly, as prof. Peters does, and in particular if we understand it to include what in Westminster-type constitutional systems are called constitutional conventions. I should note, however, that this is not the orthodox self-understanding in such constitutional systems. Most notably, the Supreme Court of Canada has explicitly held, in Re Resolution to Amend the Constitution, [1981] 1 S.C.R. 753 (a.k.a. the Patriation Reference)  that constitutional conventions are not law and are not judicially enforceable. As I explain at length in my paper on the subject, “Towards a Jurisprudence of Constitutional Conventions”, 11 O.U.C.L.J. 29, I think this was and is a mistake, but until further notice, it remains the law.

I also have no quarrel with saying that constitutional law, so understood, must be “settled,” to borrow a word prof. Peters uses in passing. But there is quite a difference between “settled”―which I take to mean relatively stable, not often challenged and still less often changed―and “entrenched.” Constitutional law, especially rules specifying who may make what laws and how they must go about it, must be settled for ordinary political process to happen without constant challenges to the legitimacy of its output. As prof. Peters puts it,

continual fights about the meaning of democracy would make the actual operation of democracy impossible.  It would be like trying to play baseball with the teams constantly debating the definition of a strike.

But I don’t think entrenchment is necessary (arguably, it is not even sufficient) to prevent this sort of instability.

That is certainly the case for formal entrenchment―legal mechanisms preventing the amendment of constitutional rules through the ordinary legislative process. The British constitution is the most obvious counterexample, since none of its rules is protected from amendment by an ordinary act of Parliament. That includes statutes (those, for example, defining the membership and powers of the House of Lords), rules of the common law (such as those defining the royal prerogative, which is always subject to redefinition by Parliament) and even constitutional conventions (so the Statute of Westminster, 1931, superseded the conventions that had been developed to govern the relationship between the imperial Parliament and the dominions). But even in a system which to a great extent relies on formal entrenchment, such as the U.S. Constitution, some rules that are constitutional, in the sense of defining the procedures for making and interpreting law, are not formally entrenched. One example of such a rule is that which provides that there are to be 435 members of the House of Representatives: this is a statutory rule, which Congress could change tomorrow it felt like it, and which it had changed in the past, though not (with a minor exception) for the past century. Another notable example is the rule providing that there to be nine Justices on the U.S. Supreme Court.

This rule illustrates, however, an important qualification. Although it is not protected from amendment by ordinary legislation by any formal legal rule, it is quite unamendable by virtue of a constitutional convention, as Franklin Roosevelt discovered to his cost when he tried to “pack” the court with friendly Justices by increasing its size, in order to have it overturn decisions hostile to New Deal legislation. So it is possible that some, perhaps many, rules which appear not to be entrenched at first sight really are entrenched by convention which, prof. Peters believes is, and I argue ought to be treated as, just another sort of legal rule.

But the same is not true of every constitutional rule. I don’t know whether a convention could be said to have frozen the membership of the U.S. House of Representatives at 435, but there certainly exists no such convention with respect to the membership of the Canadian House of Commons, which is being increased every now and then to reflect population growth. In Canada again, legislation has also restricted royal prerogative, giving Parliament (some, limited and arguably insufficient) control over decision- and rule-making formerly entirely reserved to the Crown (and by convention to the cabinet). In the U.K., an ordinary Act of Parliament threw hereditary peers out of the House of Lords. In some Canadian provinces, ordinary laws created fixed-date elections. And so on.

It is certainly true that the rules of the democratic game ought to be reasonably clear at any given point in time. It is almost certainly true that it is best that these rules not change too frequently or without some very good reasons. But it is not the case that formal, or even informal, entrenchment is necessary in order to ensure such clarity and stability.

What Sort of Rules Are in a Constitution?

A very interesting post about constitutionalism by C.J. Peters on his new MODblog makes a couple of claims that is worth addressing at some length. Constitutional law, prof. Peters suggests, consists of “legal rules that are both entrenched and secondary” (his emphasis).  “Secondary” rules refers to H.L.A. Hart’s category for rules that, unlike “primary” rules which impose duties to do or not to do something, do not directly impose duties but rather specify in various ways how primary rules are to be created, changed, and ascertained. I think both of prof. Peters’ claims are overstated. Constitutional rules are not necessarily entrenched, nor are they always secondary rules. I will leave the first claim to a separate post, probably tomorrow, and take up the second one here.

Prof. Peters writes that in contrast primary rules “that govern people’s everyday conduct,”

[r]ules about who makes law and how (e.g., bicameralism in Congress, the division of power between the federal and state governments) are examples of secondary rules; so are rules about whether and to what extent the government can regulate areas of conduct like speech and religion.  These secondary rules don’t directly apply to the conduct of most people living their everyday lives.  Instead, they govern the making and content of the primary legal rules that apply to everyday conduct.  Most constitutional provisions are examples of secondary legal rules.  (The only current exception in the U.S. Constitution is section 1 of the Thirteenth Amendment, which directly prohibits private actors from practicing  “slavery [or] involuntary servitude.”)

This might get the distinction between primary and secondary rules, as defined by Hart in The Concept of Law somewhat wrong. The fact that a rule governs “people’s everyday conduct” isn’t relevant―plenty of primary rules apply in very narrow circumstances or to very few people. The key point is not whether a rule applies to “everyday conduct” but whether it creates duties or powers (81). So Hart classifies rules governing the formation of contracts as secondary, not primary, as prof. Peters does.

In any case, prof. Peters is not alone in saying that constitutional rules are secondary rules because they govern the way in which other rules come into existence. Jeremy Waldron takes the same view in a paper called “Are Constitutional Norms Legal Norms?“.

I think that’s not quite right. Some constitutional rules certainly are secondary rules. Prof. Peters’ example of bicameralism is a good one. So are arguably, for example, rules of both the US and the Canadian constitutions that create or allow the creation of federal courts. But the case of some other rules is more complicated.

I start with constitutional rules that are not generally recognized, albeit mistakenly in my opinion, as legal―constitutional conventions. Let me quote at some length from a paper I have published, “Towards a Jurisprudence of Constitutional Conventions”, 11 O.U.C.L.J. 29:

Jeremy Waldron argues [in the paper linked to above] that conventions are secondary rules … His main example is the rule (which I will refer to as R1) that the monarch must assent to Bills that have passed both Houses of Parliament. R1, he writes, ‘is plainly not a primary rule of the British legal system, since it operates to structure the creation of law.’ [1706] But notice the way in which R1 does so operate. Before R1 came into being, the old legal rule (R0) provided that the monarch may, in her discretion, give or withhold assent to a Bill passed by the Houses of Parliament. R0 was plainly a power-conferring rule; it made the monarch an active agent in the creation of law by Parliament. R1 deprived her of that role. It does not confer any power on the monarch but on the contrary denies a power that would exist in its absence and imposes a strict duty specifying how the monarch is to behave. At most, we could say (borrowing an expression from Canadian constitutional law) that the distinction between primary and secondary rules is not always categorical, and that some rules, such as R1, have a ‘double aspect’—they can appear, from different perspectives, either primary or secondary. (Similarly, the prohibition on fraud can be seen either (as it is in criminal law) as a primary rule, a duty not to deceive, or (as it is in contract law) as part of the rules specifying the conditions of the formation of a valid contract, which Hart repeatedly provides as an example of power-conferring, secondary rules.) (31)

I think that the same could be said of many other constitutional rules, including those which are incontrovertibly legal. To return to prof. Peters’ example, the rules that divide power between federal and state or provincial legislatures impose duties on legislators not to act in certain ways (i.e. not to legislate ultra vires). So do rights-protecting constitutional provisions. The opening of the First Amendment of the U.S. Constitution (“Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof; or abridging the freedom of speech, or of the press…”; emphasis mine) makes this especially clear.

These rules structure the way laws are made―at least federalism rules do; I’m not sure that this is a very apt description of rights-protecting provisions, which do not define or structure how laws are made, but rather limit what content the laws can have. But as importantly (in the case of federalism provisions) or more importantly (in the case of rights provisions) they tell legislators and officials in the executive branch of government what to do and what not to do. They are thus primary rules, or at least have a “double aspect” and are both secondary and primary rules.