The limits of generality for constitutional design

Abstract

James Buchanan was a fervent advocate of a non-discriminatory politics. However, he translated his views on constitutional political economy into (de jure) constitutional design in an insufficiently thoughtful way. Simply writing non-discriminatory politics into a Constitution is unlikely to have the desired effect. All Constitutional language is open to interpretation and political entrepreneurs will be ready to push interpretation in their favoured directions. The history of US Constitutional law bears this out. This does not necessarily discount Buchanan's quest for constitutionalized non-discriminatory politics. However, it does mean that it must be tempered by realistic concerns regarding constitutional design. With this in mind, I suggest that focusing on procedural, rather than substantive, Constitutional provisions may be more fruitful.

Introduction

The idea of a generality principle features prominently in constitutional political economy (Buchanan, 2000 [1975]; Buchanan and Congleton, 2003 [1998]; Congleton, 2004; Salter and Young, 2019). For Nobel Laureate James Buchanan, it provides the normative bridge between individual and collective action: when ‘carried out only after general agreement, collective action is essentially voluntary action’ (Buchanan, 1959: 134).¹ A generality principle underlies the argument for unanimity (or something very close to it) in the choice of a de jure (or ‘big-C’) Constitution; likewise, it motivates ‘Article V’-type rules that insist upon large supermajorities for Constitutional amendments.² In terms of a positive theory of governance, generality characterizes situations where the benefits and costs are, respectively, enjoyed and borne broadly.

A generality principle informs not only Buchanan's views on how a Constitution should be adopted and amended, but also his preferences for Constitutional content. For example, Buchanan (2005) argues that generality, in its broadest sense, should be explicitly incorporated into the US Constitution by means of a ‘non-discriminatory politics’ amendment. He references his 1978 interview of Friedrich Hayek, during which the latter argued for adding the following Constitutional text: ‘Congress shall make no law authorizing government to take any discriminatory measures of coercion.’ Hayek (and by implication Buchanan) believed that this amendment would ‘make all the other [Constitutional] rights unnecessary’.³

However, Buchanan has a tendency to neglect the ‘nitty-gritty’ of real-world constitutional design. In his (2005) essay for Cato Unbound, he advocated three specific amendments to the US Constitution, including the non-discriminatory politics amendment. This was a rare instance of Buchanan stating what his Constitutional preferences might actually look like in Constitutional text. The essay was published with three responses, including two by distinguished legal scholars (Amar, 2005a; Kozinski, 2005; Niskanen, 2005). The responses are fascinating. All of those authors (including Niskanen, an economist) at times seem non-plussed by Buchanan's perfunctory consideration of how Constitutional provisions are put into effect, interpreted, and subsequently exploited by political entrepreneurs.

The above authors are largely concerned with critiquing Buchanan's advocacy of three specific amendments to the US Constitution. However, other authors – many writing in the public choice tradition – embrace Buchanan's advocacy of generality-based Constitutional reform specifically, or more generally believe that Buchanan's contributions are directly relevant to Constitutional design (recently, e.g. Congleton, 2014; Dougherty and Edward, 2011; Dove and Young, 2019; Ginsburg, 2010a; Lehto and Meadowcroft, 2021; Meadowcroft, 2020; Voigt, 2020). These latter scholars take Buchanan's contributions – including those with coauthors, e.g. Buchanan and Tullock 1962; Brennan and Buchanan 1980 – as serious guides for Constitutional design. In this paper, I aim to highlight ways in which Buchanan's neglect of the ‘nitty-gritty’ of Constitutional design is problematic in ways that are ignored by many scholars.

Before doing so, I note one potential problem that I will not concern myself with in this paper: to wit, the possibility that Constitutions are simply irrelevant. Many scholars are dismissive of what James Madison referred to as ‘parchment barriers’. For example, Wenzel (2010: 65) ‘emphatically reject[s] the notion that good constitutional parchment is sufficient for successful constitutionalism’. This is a fundamental objection: Constitutions do not matter or are, at the very least, an order of magnitude less important than a society's informal constitutional conventions and norms.⁴ And recent studies on the ‘de jure-de facto gap’ (or ‘constitutional underperformance’) provide empirical evidence that this objection needs to be taken seriously, at least in regard to many countries (e.g. Chilton and Versteeg, 2016; Gutmann et al., 2022; Law and Versteeg, 2013; Voigt, 2021).

However, Buchanan is clearly concerned with the US Constitution and that document – albeit imperfectly – does bind political agents.⁵ Why is this the case in the US and in many other countries? The most compelling explanation put forth is that a Constitution can serve as a coordination device (e.g. Hadfield and Weingast, 2014; Hardin, 1989; Ordeshook, 1992; Weingast, 1997). By codifying rules in a shared document, a Constitution can provide ‘a focal solution […] so that citizens gain the ability to act in concert and police their government’ (Weingast, 2005: 105). In similar fashion, it can coordinate political agents in policing one another (Young, 2021).⁶

Of course, a Constitution cannot coordinate citizens and political agents around just anything. A constitutional order is ultimately shaped by shared beliefs (conventions and norms) regarding the rules of the game for political agents. On the one hand, codifying rules that are blatantly at odds with individuals' expectations will be for naught. On the other hand, codifying rules that everyone already accepts is redundant. Reality, though, is always somewhere between that first hand and the other: the extent to which beliefs are shared will vary. A Constitution can coordinate agents around a specific subset of the feasible norms and conventions (Salter and Furton, 2018; Young, 2019).⁷ By providing a common point of reference, the codification increases the extent to which that specific subset of beliefs is shared.⁸

Even while accepting that Constitutions can matter, important problems remain with Buchanan's belief that generality can be straightforwardly written into them. These have to do with the inescapable ambiguities of language – the essential means of codification – and the endogeneity of Constitutional provisions to political entrepreneurship. Due to these ambiguities and endogeneity, provisions may be adopted that are ultimately inconsistent with generality. Furthermore, those inconsistencies can become entrenched over time. As such, a Constitutional provision aimed at ensuring generality may actually result in durable rules that facilitate discriminatory politics.

In section ‘The limits of generality in constitutional design’, I outline the limitations in trying to straightforwardly codify any principle into Constitutional text. I pay especial attention to the limitations in codifying generality. In section ‘Legal scholars respond to Buchanan’, I then illustrate how those limitations are inherent in the critical responses to Buchanan's (2005) call for a non-discriminatory politics amendment. Based on the limits of generality, I argue in section ‘Substance versus procedure’ that an emphasis on the procedural aspects of constitutional design is more likely to promote generality than substantive Constitutional text. Concluding remarks are in section ‘Conclusion’.

The limits of generality in Constitutional design

A generality principle suggests that Constitutional provisions should facilitate governance for which the costs and benefits are borne and enjoyed broadly. If generality is a lodestar for Constitutional design, then Buchanan and Hayek's non-discriminatory politics amendment (‘Congress shall make no law authorizing government to take any discriminatory measures of coercion’) is prima facie desirable. However, in this section, I discuss reasons why such a de jure amendment may fail to promote generality or, worse, end up entrenching rules that are decidedly inconsistent with it.

In what follows I will discuss Constitutional text and its interpretation generally; however, the arguments apply specifically to Buchanan's quest for a non-discriminatory politics amendment.

Ambiguity of rules

Thomas Reid (1852: 1) began his Essays on the Intellectual Powers of Man by stating: ‘There is no greater impediment to the advancement of knowledge than the ambiguity of words.’⁹ This is certainly relevant to Constitutional design. While not concerned with the advancement of knowledge per se, the ambiguity of words is an unavoidable – which is not to say immitigable – constraint on effective codification of any convention. Any Constitutional provision, therefore, must be interpreted, both when adopted and then over time.

Constitutional interpreters include not only judges (in societies with judicial review) but also the citizenry and their political agents. All of these interpreters are relevant as to how certain words translate into the de facto constitutional order. For examples, US legislators make choices regarding the laws they will pass; presidents then make choices regarding which of those laws they will veto rather than sign, and also which of them they will enforce ex post; and all of those choices are based, in part, on the relevant agents' understandings of what lies within the four corners of the Constitution.¹⁰ Furthermore, in making those choices, all of those agents are subject to pressure from their constituents, whom have their own understandings of the Constitutional text.¹¹

Inconsistencies in those understandings will inevitably arise – across both types of agents and time – because the conventions which drafters attempt to codify in a Constitution are unavoidably characterized by what Friedrich Waismann (1945) felicitously terms ‘open texture’.¹² Such conventions ‘are not delimited in all possible directions’ (p. 122). Trying to provide text that exhaustively conveys the same thing to all minds is futile; ‘we can never exclude altogether the possibility of some unforeseen situation arising in which we shall have to modify our definition’ (pp. 122–123). Reading the examples provided by Waismann, it is clear that the ‘unforeseen situation[s]’ are in reference to looking across individuals as well as time.¹³ The ambiguity of words means that drafters have a Herculean task in providing a convention with an exhaustive definition, either interpersonally or intertemporally.¹⁴

To illustrate the above points, consider Williams (2017)'s comparative examples of ‘freedom of expression’ provisions in the Icelandic and Polish Constitutions.

Freedom of expression may only be restricted by law in the interests of public

order or the security of the State, for the protection of health or morals, or for

the protection of the rights or reputation of others, if such restrictions are

deemed necessary and in agreement with democratic traditions. (Iceland; Article 73)

versus …

The freedom to express opinions, to acquire and to disseminate information

shall be ensured to everyone. (Poland; Article 54)

Beyond the straightforward difference in the textual length between the two provisions, Williams emphasizes the use of adjectives/adverbs (e.g. ‘necessary’), conditional conjunctions (e.g. ‘if’; ‘and’; ‘or’), and indeterminate modal verbs (e.g. ‘may’) in the Icelandic case. Greater use of such ‘indeterminate parts of speech’ (Williams, 2017: 1) introduces vagueness and makes interpretation more difficult and/or open.

The Icelandic and Polish provisions also fundamentally differ in that Iceland's takes the form of freedom can only be restricted when; as opposed to Poland's freedom shall be ensured. The Icelandic provision provides an itemization of specific cases when restriction is permissible, leaving latent the general freedom shall be ensured (i.e. not restricted). Alternatively, the Polish provision makes the general protection of expression explicit while remaining silent on specific exceptions, which undoubtedly exist. (Some are found elsewhere within Poland's Constitution [e.g. Article 13 prohibits ‘political parties and other organizations […] whose programmes or activities sanction racial or national hatred’] and some are based on the legislature's interpretation of what exceptions remain implicit [e.g. sections of the criminal code that extend Article 13-type prohibitions to individuals].)

An implication of this differential constitutional language is that the de jure role of Iceland's government is relatively vague. For example, it ‘may only’ restrict expression in certain cases (whereas Poland's simply ‘shall not’); when such restrictions are ‘necessary’ and/or ‘in agreement with democratic traditions’ (whereas Poland's provision contains no such conditionality). Furthermore, such language opens up broad possibilities for constitutional interpretation de facto. As Williams (2017: 4) notes: Poland's ‘protection of free speech is far less elastic and liable to elite dominance’ than Iceland's. Stated more generally, Iceland's language provides greater latitude for political agents and their constituents to offer plausible interpretations of the provision. This makes the constitutional constraint itself endogenous to that interpretation in practice.

The ambiguity, both de jure and then de facto, opens up possibilities for constitutional political entrepreneurship (Salter and Wagner, 2018; see section ‘Political entrepreneurship’ below), the results of which are likely to be inconsistent with generality. This is where Williams (2017) is concerned with ‘elite dominance’, but it applies to any organized special interest. A constitutional constraint (or, for that matter, mandate) provides for generality only to the extent that it cannot be perverted and captured by special interests. Alternatively, if it is de jure ambiguous and therefore de facto endogenous to how agents interpret it, then generality cannot be ensured.

Worse, constitutional provisions designed ostensibly to ensure generality may lead to decidedly contrary outcomes. Consider the Icelandic constitutional provision on free expression from above. Not only is it positive and conditional relative to the Polish provision – expression can only be restricted when rather than expression will be ensured (not restricted) – but its use of vague language (adjectives/adverbs; conditional conjunctions; indeterminate modal verbs) opens up broad avenues for interpretation.

Time-invariance of rules

Constitutional provisions are likely to contain words or phrases, the understood meanings of which are inherently subject to change as the environment – technological, economic, social, and/or political – changes. When the meanings of these words and phrases change, this reshapes the Constitutional rules in which they are embedded.

Consider the much-debated Second Amendment to the US Constitution: ‘the right of the people to keep and bear Arms, shall not be infringed’. To begin with, the term ‘Arms’ meant something different in the late eighteenth century than it does today. This difference in meaning was unavoidable with the passing of time. For one thing, technology changed: even limiting ‘Arms’ to mean ‘guns’, the underlying technologies are today vastly more accurate and deadly. (Alternatively, a Constitutional right to own and use muskets specifically would seem patently absurd today.)

Even holding technology constant, social changes over time have led to a different understanding of ‘to bear Arms’. The term had a decidedly military connotation to eighteenth-century individuals. (As such, the ‘well regulated Militia’ in the first clause of the Second Amendment did not then seem incongruous with the switch to ‘people’ in the second.) To someone in the eighteenth century, ‘a deer hunter or target shooter carries a gun[;] he does not, in the strictest sense, bear arms’ (Amar, 2005b: 322). The loss of this connotation contributed to the development over time of a new, decidedly individualistic, understanding of the ‘right to bear Arms’.¹⁵

The time-invariance of the Second Amendment's meaning does not necessarily have much directly to do with generality. However, consider another example: the Commerce Clause of Article I, giving the US Congress ‘the power to regulate commerce with foreign nations, and among the several states, and with the Indian tribes’. As Epstein (1987: 1387) notes:

[It is] the clause in the Constitution to which most federal power can be traced in today's general welfare state. The labor statutes, the civil rights statutes, the farm and agricultural statutes, and countless others rest on […] a construction of the commerce clause that grants the federal government jurisdiction so long as it can show (as it always can) that the regulated activity burdens, obstructs, or affects interstate commerce, however indirectly.

But the tracing of present-day federal powers to the Commerce Clause is tied to the evolution of the meaning of ‘commerce’ as US society and the economy changed over time.¹⁶

Epstein (1987) marshals evidence regarding the US Constitutional text and how contemporaries (to its adoption) would have interpreted the word ‘commerce’. In particular, contemporaries would have deemed ‘commerce’ to be ‘closer to the idea of ‘trade’ than to other economic activities’ (p. 1389; also see Barnett, 2001). Consistent with this, Chief Justice John Marshall's opinion in Gibbons v. Ogden (1824) referred to it as ‘intercourse’.¹⁷ As Barnett (2001: 128) notes, Samuel Johnson's Dictionary of the English Language (6th edition, 1785) defined ‘intercourse’ as ‘1. Commerce; exchange’.¹⁸ The meaning ascribed to ‘commerce’ in the late eighteenth and early nineteenth centuries can be contrasted with how it today is equated to economic activity broadly. For example, consider the US Department of Commerce's stated mission (‘to create the conditions for economic growth and opportunity for all communities’) and its strategic goals (‘Innovation, equity, and resilience’).¹⁹

This evolution in the meaning of ‘commerce’ corresponded to a substantive change in the interpretation of the Commerce Clause. In Gibbons, the issue at hand was state-granted monopoly rights to navigation of interstate waters. The Marshall Court ruled against those monopolies to prevent state governments from interference with interstate trade. (Note the two specific words: interstate and trade.) Combined with the Supremacy Clause (Article VI, clause 2), the Marshall Court determined that Congress' power to regulate interstate commerce amounted to states' lack of power to interfere with it.²⁰ The Commerce Clause was then a shield against state governments creating an unlevel economic playing field. Furthermore, decades later, ‘[d]uring the Progressive Era, the Supreme Court rejected a broad conception of commerce […] in favor of a more limited conception […] as “trade and exchange”’ (Barnett, 2001: 129). This limited the federal government's own role to defining the level playing field.

Fast-forward three more decades and ‘the Commerce Clause […] in its New Deal incarnation expanded the powers of the federal government far beyond any level that it had previously held’ (Epstein, 2014: 167). How? It was at this point that ‘commerce’ became interpreted to mean essentially any economy activity. In watershed cases like Nebbia v. New York (1934) and Wickard v. Filburn (1942), the Supreme Court established its deference to Congress regarding legislation when ‘the public interest [is] deemed to require’ it, and/or when legislation affects an industry that ‘exerts a substantial economic effect on interstate commerce’ (Barnett, 2004: ch. 11; Epstein, 2006: 64–81).²¹ Unsurprisingly, given deference to Congress' (subjective) evaluation of public interest and substantial effect, it enjoyed essentially carte blanche throughout the latter part of the twentieth century and into the twenty-first.²²

Political entrepreneurship

All Constitutional provisions are open to interpretation, at least to an extent. As elaborated on above, this is necessarily so because, first and foremost, language is inherently ambiguous; furthermore, interpretations of any Constitutional provision are likely to change as the environment (technological, economic, social, political) changes. For any provision, then, the range of interpretation provides a window of opportunity for political entrepreneurs who stand to gain via constitutional change.

As Salter and Wagner (2018: 281) recognize, ‘societies rarely entail universal agreement about constitutional provisions’. It follows that ‘there will exist margins of contestation where political entrepreneurship is active in seeking support for alternative constitutional interpretations’ (p. 281). A political entrepreneur's contestation of status quo interpretations may, of course, be disingenuous and self-serving; but to affect constitutional change his alternative interpretations must be compelling to a broad swath of the society.

Theorizing in constitutional political economy typically begins from a two-stage framework. First, the rules of the game for political agents are chosen. Second, political agents proceed to operate within those rules. To wit, there is a ‘constitutional moment’ which establishes the framework within which ‘ordinary politics’ then plays out over time (Buchanan, 1975; Buchanan and Tullock, 1962; Brennan and Buchanan, 1980). While the stylized two-stage framework is often very useful, that does not change the fact that ordinary politics is tied up with the constitutional contestation in any society (Salter and Wagner, 2018: 282; also see Runst and Wagner, 2011).

Hence, the potential for special interest capture (Stigler, 1971; Tullock, 1967) is relevant at the constitutional – as well as the ordinary – level of politics (Boudreaux and Pritchard, 1993; Landes and Posner, 1975). Given an unchanging Constitutional text, political entrepreneurs, seeking to further the special interests that they embody or represent, can promote change in constitutional norms and conventions. They can, by exerting direct or indirect influence on judicial review, lead to changes in the interpretation of unchanged text. While no formal Constitutional amendment occurs, political entrepreneurship can lead to constitutional capture in favour of certain special interests (Salter, 2017; Salter and Furton, 2018).²³

Consider the Seventeenth Amendment to the US Constitution. In the original document, selection of US senators was via states' legislatures. (This can be contrasted to membership in the House of Representatives, which was based on popular elections.) The insulation of the Senate from direct popular input into its membership was designed to prevent the excesses of democracy, including calls for federal government growth at the expense of individual liberties (Garrett et al., 2010; Holcombe and Lacombe, 1998). However, interstate special interests pushed to amend the Constitution such that both houses of Congress were directly answerable to popular election (Schleicher, 2014; Zywicki, 1994). The Seventeenth Amendment served those interstate special interests because individual voters were more amenable to them than state legislators. The latter were more likely to be focused on the shared interests of those within their states. Alternatively, voters were more apt to be swayed by interstate special interests that appealed to them individually (e.g. a truck driver in California may identify more with interstate trucking than the intrastate interests of Californians).

In the case of the Seventeenth Amendment, political entrepreneurship actual led to a change in the text. However, now consider the more recent example of political entrepreneurship's role in affecting the constitutional status of abortion rights in the US. That status has had two watershed moments in Supreme Court judicial review (Roe v. Wade 1973 and Dobbs v. Jackson Women's Health Organization 2022) and important Court decisions in between them (e.g. Planned Parenthood v. Casey 1992; Gonzalez v. Carhart 2007).²⁴ Through all of the above, the text of the US Constitution did not relevantly change.²⁵ Yet special interests have been active – both during and before the 1973–2022 period – to influence the courts' interpretations of that text in relation to abortion rights (Soper, 1994; Staggenborg, 1991).²⁶

The activities on both the pro-choice and pro-life sides have been varied, ranging from grassroots movements, to co-opting both broader and narrower interest groups, and to procuring the services of lobbying groups, lawyers, and legal scholars.²⁷ Their efforts have been both direct (e.g. having lawyers file test cases within the court system) and indirect (e.g. lobbying for state legislatures to pass abortion laws that are the bases for the test cases). Importantly, counted amongst the indirect efforts are those aimed at electing a US president who will appoint judges favourable to certain views (and also Senate members who will have a say in whether a president's nominees are confirmed).

The direct and indirect efforts alluded to above all have – and will continue – to play a role in the constitutional status of abortion rights in the US. They are driven by political entrepreneurs of various stripes (grassroot organizers; leaders of organized political interest groups; constitutional lawyers; etc.). All of these efforts ultimately have relied on the interpretability of a constant Constitutional text via judicial review and the agents of ordinary politics who make such review come into play.²⁸

Entrenchment of non-generality

I have emphasized the endogeneity of constitutional government role above. This is obviously relevant for any Constitutional text which is added towards an ideal of generality. Such text may be interpreted in ways that, perversely, favour special interests. On the other hand, Constitutional text can also be reinterpreted over time to promote generality in ways that it had not done so before. The former is obviously more relevant for those pursing a Buchanan-esque non-discriminatory politics amendment. But both scenarios are plausible and I will start with a case of the later.

Wu (2018: 548) sates that ‘The First Amendment was a dead letter for much of American history.’ While surprising to many, early US history evidences little Constitutional challenge and Supreme Court silence in response to government abridgements of political expression. The Alien and Sedition Acts of 1798 are a case in point.²⁹ It was only in response to new Espionage and Sedition Acts during WWI that First Amendment judicial challenges began to occur. A famous example is Debs v. US (1919), associated with the indictment of Eugene Debs for giving a speech in violation of the 1917 Espionage Act.³⁰ Interestingly, this case was plausibly motivated by special interest (in particular, Deb's special interest in not being imprisoned) but stood to further generality (by making political speech Constitutionally protected for all). In the event, Deb's conviction was upheld by the Supreme Court. Indeed, interpretations of the First Amendment as providing broad protection of political speech first appeared in dissenting opinions. It was not until the 1950s and 1960s that Supreme Court majorities established the core protections that are today taken for granted.³¹

Now I turn to the possibility of changing interpretation leading away from generality. This possibility is more concerning for those who favour Buchanan's advocacy of Constitutionalizing non-discriminatory politics. The US Commerce Clause example provided above is pointedly relevant. While all must admit that the original text was open to interpretation to begin with, the Marshall Court (1801–1835) clearly established an interpretation that limited the federal government's powers to (1) its enumerated powers and (2) striking down state-based actions that conflicted with those powers (the so-called ‘Dormant Commerce Clause’). The Marshall Court's interpretation tended towards supporting a generality norm and also the benefits of a market-preserving federalism (Weingast, 1995). However, this interpretation was turned on its head by the Hughes and Stone Courts; subsequently, the US Congress' carte blanche for expansive and discriminatory powers has been entrenched.

Legal scholars respond to Buchanan

Turning back more decidedly to Buchanan's (2005) proposal of a non-discriminatory politics amendment, the responses of two prominent legal scholars (Amar, 2005a; Kozinski, 2005) are informative of the concerns outlined above.

Akhil Reed Amar is one of the foremost scholars of the US Constitution and the history of its interpretation and application (e.g. Amar 2005b, 2012). He seems puzzled by Buchanan's proposal, asking: ‘What, exactly, is this trying to add above and beyond existing Equal Protection Clause doctrine?’ The Equal Protection Clause is generally associated with the first section of the Fourteenth Amendment and the due process guarantee of the Fifth Amendment.³² On the one hand, Amar is suggesting that ‘Equal protection’ under law and ‘non-discriminatory politics’ are prima facie the same thing. On the other hand, his subsequent discussion proceeds along the lines of (paraphrasing): Ok, let's call it ‘nondiscriminatory politics.’ Why do you think that will, in practice, be interpreted any differently?

For example, Amar (2005a) points to Buchanan's (2005) discussion of ‘uniform’ taxation. Buchanan claims that a non-discriminatory politics amendment would not imply a repeal of the Sixteenth Amendment (allowing for a federal income tax) but, rather: ‘offer the basis for the replacement of the complex tax structure by a uniform rate of tax that is imposed on all income, without exemptions, deductions, credits, or other special treatments.’ Amar (2005a) notes that the original (pre-Sixteenth Amendment) text:

[D]irect taxes shall be apportioned among the several States […] according to their respective Numbers […] (Article I, Section 2); No capitation, or other direct, Tax shall be laid, unless in Proportion to the Census or Enumeration herein before directed to be taken (Article I, section 9)

was actually interpreted, in a ‘sensible decision’, to uphold ‘progressive’/‘discriminatory’ taxes by the Supreme Court of the 1790s.³³ It was only a century later that the Supreme Court provided a repudiation of that precedent. That repudiation was an important contributor to the push for the Sixteenth Amendment. Given that amendment is now entrenched, Amar ponders: would Buchanan's non-discriminatory politics amendment be one that ‘prohibits even a standard [income tax] deduction (say, for income below the poverty line) – [which] marks a truly radical break with American constitutional history’. (Moving away from taxation, Amar muses that ‘aid to impoverished orphans might [also] run afoul of nondiscrimination as Buchanan understands it’.)

I infer from his text that Amar is in favour of discriminatory taxation and other policies in certain instances. Buchanan would disagree, but that is not the point. Amar is pointing out that, from the point of view of a constitutional law scholar, (1) what Buchanan calls ‘non-discriminatory politics’ was arguably already included in the Constitutional text; (2) to the extent it was, that text was interpreted in decidedly different ways by different Supreme Courts; and (3) why would we expect otherwise, if a Buchanan-esque change to the text were adopted, moving forward?

Alex Kozinski (2005), a constitutional scholar and a long-time judge on the US Ninth Circuit Court of Appeals, raises related concerns in a different way. He states plainly: ‘[A] constitutional constraint [of the type that Buchanan proposes] would pretty much hand over the running of the federal government to the judiciary.’ He continues, poignantly: ‘All government action is inherently discriminatory: The penal codes discriminates against those who commit certain acts [… and] Taxing and spending are inherently discriminatory acts[:] No matter how uniformly a tax is laid, it will fall more heavily on some than on others[.]’³⁴

Given that scholars of constitutional law are taken aback by Buchanan's straightforward proposal of writing non-discriminatory politics into the US Constitution, we need to understand why. I argue that they are bemused because they understand that Constitutional text is not the be-all and end-all: (1) that text is always open to interpretation and (2) political entrepreneurs will always be present, seeking opportunities to move interpretations towards the special interests that they embody or represent.

Substance versus procedure

Is the lesson of this paper that Buchanan was wrong to promote de jure Constitutional change towards non-discriminatory politics? I do not believe that is necessarily true. However, this paper points towards an emphasis on procedural versus substantive Constitutional provisions (Wagner and Gwartney, 1989). Substantive provisions are those that, for example, state specific rights that the government will either not infringe upon or positively act against infringements upon. Procedural provisions, alternatively, are those that layout the structure of the government, providing the checks and balances between different branches of that government.

Vanberg (2011) argues that procedural provisions have advantages over substantive ones. His arguments are based on the idea that procedural provisions, relative to substantive ones, ‘tap into focal understandings that that emerge out of a shared political history [and therefore] may offer better prospects of successful constitutional governance’ (p. 317). This is consistent with the above-cited studies of Hardin (1989), Ordeshook (1992), Weingast (1997), and Hadfield and Weingast (2014). Vanberg's argument is essentially that procedural provisions are more effective focal points than substantive ones.

We can think about this in terms of the sources of constitutional drift (Salter, 2017; Salter and Furton, 2018) discussed in section ‘The limits of generality in constitutional design’ above. First consider the Ambiguity of Rules and, relatedly, the Time-Invariance Rules. These concerns can be relevant to both substantive and procedural provisions. However, they are relatively more important for the former. For example, something like ‘the right of the people to keep and bear Arms’ (US Constitution; Second Amendment) is open to wide interpretation from the get-go of adoption; even more so as the technology of ‘Arms’ evolves. Alternatively, a procedural provision such as the ‘Senate […] shall be composed of two Senators from each State, elected by the people thereof, for six years’ (US Constitution; Seventeenth Amendment) is more clear-cut and, over time, it is difficult to see how technological, socioeconomic, or demographic changes would change the provision's interpretation. As such, the substantive provision may be a less effective focal point than the procedural one.

A similar point can be made in terms of Political Entrepreneurship. Again, as Salter and Wagner (2018: 218) note, there will always be ‘margins of contestation where political entrepreneurship is active in seeking support for alternative constitutional interpretations’. Consistent with the points immediately above, those margins of contestation will be less exploitable in relation to procedural versus substantive provisions. Because the former have less wiggle room in terms of interpretation – both immediately and over time – political entrepreneurs will tend to channel their efforts on the latter.

The Seventeenth Amendment may illustrate this point. Prior to that amendment, Article I, Section 3 dictated that each state's senators be elected by its legislature. Despite the clear-cut nature of that provision, there were informal end-runs that interstate special interests could exploit. For example, prior to 1913 a number of US states approximated direct election via state legislator pledges, public canvases (i.e. when senators publicized the state legislators that would vote for their selection), and direct primaries (Riker, 1955; Zywicki, 1994). But these practices did take hold uniformly. For example, while they were common in western states, they were avoided by southern states (where one-party control allowed legislatures to consistently return senior senators to Washington [Zywicki, 1994]). At the end of the day, the Article I procedural provision was difficult for interstate special interests to effectively exploit. The margins of contestation were too meagre. This made the movement for the formal Seventeenth Amendment critical despite the informal end-runs.³⁵

For the reasons above, the Entrenchment of Non-Generality is less of a risk when it comes to procedural versus substantive Constitutional provisions. If there is less wiggle room for a provision to be perversely (from its authors' point of view) interpreted as favouring special rather than general interests, then there is less of a risk that such a perverse interpretation will be entrenched by court decisions and/or subsequent development of convention based on it.

Additionally, consider Alexander Hamilton's response to objections to the original draft of the US Constitution that it ‘contains no bill of rights’. He argued in the Federalist 84 that any substantive right of value would be provided by procedural checks and balances; and in the absence of such, any codified substantive right provides a mere ‘parchment barrier’ (borrowing James Maddison's terminology) that would not be enforceable.³⁶ In other words, substantive rights are implicit in the definition of veto players within the governance structures (Tsebelis, 2002). Those veto players – who are particular governance providers – will self-interestedly defend the balance of their powers versus those of other players.³⁷ Since the checks and balances are implicit in the balance of powers between veto players, the implied substantive rights are not dependent on wishful parchment.

If procedural checks and balances imply substantive rights, then a corollary is that the removal of the former tends to weaken the latter. The Seventeenth Amendment to the US Constitution – briefly discussed in section ‘Political entrepreneurship’ above – is a case in point. While this amendment obviously did not change any substantive enumerated rights, it does appear to have weakened checks and balances by making Senate and House of Representative makeup and behaviour more similar. For example, Crook and Hibbing (1997) report that, post-amendment, senators were more likely to have had previous experience in government and, more generally, Senate and House elections tended to mirror one another more. (Engstrom and Kernell [2007] more recently echo the latter point.) Likewise, Meinke (2008) reports that senators exhibited ‘behavioral shifts toward mass audience’ (p. 445), including increased sponsorship of bills and roll-call participation, and Gailmard and Jenkins (2009) report that senators both became more responsive to the mass electorate and gained leeway to cater to special interests.

What about adding procedural checks and balances? The US case is a difficult one to evaluate: the Bill of Rights was approved by Congress the very year the Constitution went into effect, and then ratified two years later.³⁸ However, recent work by Voigt and Gutmann (2013) is consistent with the importance of procedure versus the codification of substantive rights.³⁹ They find that the inclusion of substantive property rights into a Constitution is not strongly associated with their provision; alternatively, they find that the effect becomes more meaningful when conditioned on strong independence of the judicial branch.⁴⁰ Granted, the independence of the judiciary is, itself, not entirely a de jure matter.⁴¹ Still, Voigt and Gutmann's work is suggestive.

Recent work also points to constitutional checks and balances in medieval Europe as foundational to the development of substantive rights, both political and economic. Obviously, medieval polities did not have Constitutions, but they did have meaningful checks and balances that implied certain substantive rights (e.g. Salter, 2015; Salter and Young, 2018, 2019). Furthermore, given the documented gaps between de jure Constitutional provisions and de facto practice on the ground (e.g. Chilton and Versteeg, 2016; Gutmann et al., 2022; Law and Versteeg, 2013; Voigt, 2021), Vanberg's arguments and Voigt and Gutmann's evidence clearly call for further work to empirically assess the importance of substantive codification versus procedural realities.

Conclusion

James Buchanan was a fervent advocate of a non-discriminatory politics; in other words, he saw value in a generality norm for governance. The author of this paper tends to agree with him.

That being said, Buchanan attempted to translate his views on constitutional political economy into (de jure) Constitutional design in an insufficiently thoughtful way. Writing non-discriminatory politics into a Constitution is unlikely to have the desired effect. All Constitutional language is open to interpretation; and political entrepreneurs representing special interests will be ready to push interpretation in their favoured directions. The history of US Constitutional law bears this out.

Does this mean that Buchanan's quest for constitutionalized non-discriminatory politics must be abandoned? Not necessarily. It does mean that it must be tempered by realistic concerns regarding constitutional design. It also means that focusing on procedural, rather than substantive, Constitutional provisions may be more fruitful.