Abstract
Standing institutions have a continuous existence: examples include the United Nations, the British Parliament, the US presidency, the standing committees of the US Congress, and the Environmental Protection Agency. Intermittent institutions have a discontinuous existence: examples include the Roman dictatorship, the Estates-General of France, constitutional conventions, citizens' assemblies, the Electoral College, grand and petit juries, special prosecutors, various types of temporary courts and military tribunals, ad hoc congressional committees, and ad hoc panels such as the 9/11 Commission and base-closing commissions. Within the class of intermittent institutions, one may distinguish periodic from episodic institutions. The former come into being on a schedule set down in advance, while the latter come into being at unpredictable intervals. The Electoral College is a periodic institution, while the Roman dictatorship is an episodic one. This article attempts to identify the benefits and costs of intermittent institutions, both as a class and in their periodic and episodic varieties. The largest goals are to state some general conditions under which intermittent institutions prove superior or inferior to standing institutions, and to illuminate the temporal dimension of institutional design.
Standing institutions have a continuous (although not necessarily permanent) existence: examples include the United Nations, the British Parliament, the US presidency, the standing committees of the US Congress, the Environmental Protection Agency, and the International Criminal Court. By contrast, intermittent institutions have a discontinuous existence; they come into being when legal rules or standing institutions trigger their existence and go out of being when they have fulfilled designated tasks or when a preset time limitation has expired. Examples include the Roman dictatorship, the Roman censorship, the Estates-General of France, constitutional conventions, citizens' assemblies, the Electoral College, grand and petit juries, special prosecutors, various types of temporary domestic and international courts and military tribunals, ad hoc congressional committees, and ad hoc panels such as the 9/11 Commission and base-closing commissions.
As is immediately apparent from the heterogeneous character of this list, there are several useful distinctions to be made within the class of intermittent institutions. The main distinction I shall explore is between periodic and episodic intermittent institutions. The former come into being on a schedule set down in advance, while the latter come into being at unpredictable intervals. The Electoral College is a periodic institution, while the Roman dictatorship is an episodic one.
Despite the heterogeneity of the examples, intermittent institutions pose common questions and problems, as do the subcategories of periodic and episodic institutions. When institutional designers get the temporal dimension wrong –when they use continuous institutions to solve problems that would have been better handled by intermittent ones or vice versa, or use periodic intermittent institutions to solve problems that would have been better handled by episodic ones or vice versa– the consequences can be grave. As we will see, corruption in the jury system, the politicization of economic policy, and even political dictatorship have all been said to flow from mistakes in the temporal dimension of institutional design. Accordingly, my major aim is to identify the costs and benefits of intermittent institutions in varying contexts, and thus to isolate some conditions under which intermittent institutions, of either the periodic or episodic variety, prove superior or inferior to standing institutions.
Moreover, understanding the benefits and costs of intermittent institutions will shed light on the larger and more familiar category of continuous institutions. I attempt to collect and evaluate some unusual institutional forms and proposals (such as proposals for ‘continuous elections') that give us a usefully critical perspective on extant institutions, showing that extant institutions themselves rest on contestable choices about the temporal dimension of institutional design. Once we appreciate that designers often face material choices on the temporal dimension, our institutional repertoire will be greatly enriched, or so I will try to show.
Section 1 offers a brief taxonomy of the temporal dimension of institutions. Section 2 explores justifications for making institutions intermittent, rather than continuous. Section 3 explores justifications for making intermittent institutions periodic or, instead, episodic. Section 4 discusses some problems and costs that are common to intermittent institutions. The Conclusion attempts to identify some general conditions under which intermittent institutions do or do not prove both feasible and desirable, as compared to continuous institutions.
1. The temporal dimension of institutions
I begin with an analytic taxonomy of the life histories that institutions may follow. These life histories are inevitably stylized, but they are useful for the analyst and in many cases reflect real problems that confront political actors – especially designers of institutions, who cannot help but make some choice or other about the temporal structure of the institutions they create.
As a first cut, I will lump intermittent institutions together and distinguish them from standing institutions. As a second cut, I will distinguish periodic intermittent institutions from episodic ones and explore the differences. To illustrate these distinctions, consider a standard electoral system and two variants. The baseline electoral system, let us suppose, requires elections for a given office on a predetermined schedule, such as the four-year election cycle for the presidency in the USA. Elections are intermittent, but periodic. Against this baseline, one variant involves continuous elections, in which the incumbent is ousted mid-term if her support falls below a predetermined level. Another variant involves randomly timed elections, in which there is a constant x percent chance that some independent body will authoritatively declare that an election will be held in the current year. The first variant makes elections a continuous rather than intermittent institution, while the second variant makes elections episodic rather than periodic.
In addition to the baseline electoral system and the alternatives of continuous or randomly timed elections, a variety of hybrid arrangements are possible. In standard parliamentary systems, the government remains in office until it declares new elections, loses a vote of no-confidence, or a preset number of years has elapsed, whichever comes earliest. This system combines elements of periodic elections, because there is an n-year maximum term; continuous elections, because the government will lose power if its support falls below a certain level; and randomly timed elections, because on the first day of the government’s term the length of its tenure is uncertain. 1
Having laid out these distinctions, they must immediately be qualified. The main qualification is that the distinction between standing and intermittent institutions is a matter of degree. Even standing institutions do not meet in continuous session. A Martian observer might believe that the House of Representatives blinks in and out of existence; during congressional recess or summer vacation, the buildings that house the institution are all but deserted, and the House (unlike the Senate) has no standing rules, but adopts new rules at the beginning of each new Congress. Moreover, although turnover in the membership of the House is usually gradual, because it is rare for members to retire or lose their seats, there are occasional upheavals in which a large fraction of the membership changes. In such a case, the Martian observer might question whether the new House is or is not the same institution as the old; perhaps it is best understood as a new institution that simply uses the old one’s name. There are philosophical puzzles about institutional identity lurking here, as in the Ship of Theseus, all of whose planks were replaced one by one.
I will skirt these conceptual pitfalls for strictly pragmatic reasons; taken too seriously, they would prevent any discussion of a set of real problems and choices that confront real actors in real historical cases. Starting with the cases and approaching the subject from the bottom up, there is a continuum between standing institutions and discontinuous ones. An extreme case at the standing end of the continuum is the Senate, whose rules are conventionally said not only to make the Senate a continuing body, but also to entrench that status against change. 2 At the other end of the continuum are single-shot institutions, exemplified by ad hoc investigative panels such as the Iran/Contra Commission and the 9/11 Commission; by the lawgivers of the classical city-states, Solon and Lycurgus, who performed their lawmaking task a single time and then withdrew from public life; and by single-shot ‘citizens' assemblies', made up of randomly selected groups of citizens who convene to propose constitutional referenda and then permanently disband (see Warren and Pearse, 2008). There are also radically discontinuous institutions such as the Estates-General, which met repeatedly, but unpredictably, sometimes with lapses of hundreds of years. Although there is a region of hard cases in the middle of this continuum, the poles are sufficiently clear cut to motivate the analysis.
Similarly, I will sidestep the conceptual problem that different institutions can be described at different levels of generality, and that the choice of description can affect whether they are seen as continuous or intermittent. The jury system is a continuous institution, whereas any particular petit jury is a radically episodic and thus intermittent one, convened for only one case. Here, too, the sorting out the conceptual issues is difficult, but not as interesting as the substantive arguments for, and against, the relevant institutional forms and practices.
Many other lumps and splits might be thought relevant; I will ignore some and invoke others as the nature of the problem requires. Let me take note of some of the main distinctions here. One might distinguish ‘institutions' narrowly understood from the functions or activities of institutions. On this construal, the Census Bureau is a continuous institution that periodically carries out the decennial census as an intermittent function. Likewise, we might describe quadrennial presidential elections as an intermittent function carried out by a complex of standing public and private institutions, such as the Federal Election Commission, the Electoral College, and the League of Women Voters. I will lump together institutions and their functions, taking a catholic view of what counts as an institution; on substantive grounds, the problems seem sufficiently similar to justify common treatment.
A more important set of distinctions involves various modes in which intermittent institutions may come into being or else be terminated. At the front end, the distinction between periodic and episodic intermittent institutions is a distinction between different modes of creation or triggering conditions. In the periodic case, some provision of written law or custom sets an ex ante schedule for bringing the institution into being, such as the four-year cycle of the Electoral College. In the episodic case, law or custom empowers, but does not require, actors in the system to summon the institution into being at a time of their choosing, as when the attorney general appoints a special prosecutor, the French monarch summoned the Estates-General, or Roman consuls and senators agreed on the appointment of a dictator.
In the latter case, the requirement of consent by in-system actors makes the appearance of the episodic institution at least somewhat unpredictable. However, nothing prevents institutions from moving from one category to the other as legal rules or customary norms and practices change over time. In the early centuries of the Roman republic, the appointment of a censor occurred on no fixed schedule, but in the last two centuries of the republic’s censors were (barring some minor lapses) regularly elected on a five-year cycle (Lintott, 1999: 116).
At the back end, intermittent institutions can be terminated either when a fixed deadline lapses or (more commonly) when they have fulfilled a task designated in advance. In some Roman sources, the dictatorship is described as having a six-month term (Abbott, 1911: 183; Lintott, 1999: 110). Other sources, however, describe the dictatorship as lapsing when the dictator had accomplished a given task, such as defeating an external military threat, quelling internal sedition, holding a special election, or carrying out a special religious ritual. 3 In general, ‘magistrates elected to perform special functions – from the dictator down to the commissioners who distributed land, founded colonies, or dedicated temples – only held office as long as their appointed task required’ (Lintott, 1999: 117). Likewise, special prosecutors and temporary tribunals typically have no fixed period of office, but are supposed to go out of existence after accomplishing discrete tasks, such as the resolution of a class of investigations, cases, or claims.
The first mode of termination, illustrated by the dictator’s six-month term, is a rule; the second, in which the institution lapses after a designated task has been completed, is a standard. The former produces legal certainty and a measure of resistance to political manipulation, but may be poorly tailored to the problem that brought the institution into being. The latter is vague and politically manipulable, and creates a serious conflict of interest if the intermittent institution is charged with deciding upon its own termination. Its appeal, in theory, is that it can avoid the twin problems of terminating the special institution too early, before the task is complete, or too late.
In important cases, designers of institutions vest the decision to terminate not in the intermittent institution itself, but in some independent oversight institution, attempting to gain the benefits of the standard while avoiding the risk of self-dealing. In the USA, at various times after Watergate, federal law authorized the attorney general, acting with the permission of a special panel of federal judges, to appoint an Independent Counsel to investigate abuses by high-ranking executive officials. The Independent Counsel had no fixed term of office, but could be terminated either by its own motion or against its will by the same judicial panel that created it, upon a determination that the Counsel’s designated tasks were substantially complete. 4 Needless to say, this distinction between self-termination and termination by an external agent has no legal analogue at the front end; intermittent institutions cannot legally bootstrap themselves into being, but can only be summoned up by an existing actor.
Despite all these complexities, I will attempt to show that intermittent institutions display important common features at the first decimal place, although important differences at the second. These claims are substantive, not merely taxonomic, so they cannot be proven in the abstract. Rather, the proof will be found, if at all, in the subsequent discussion of concrete examples.
2. Intermittent versus standing institutions
I turn now to normative justifications for intermittent institutions. These justifications are not offered as explanations for the genesis of such institutions. Even if democratic elections, for example, arose as a device whereby elites committed to redistribution in order to preempt revolution (see Acemoglu and Robinson, 2006: 136), political actors and analysts typically justify elections in quite different terms. Although reasons are sometimes causes, the normative justifications for an institution need not have any explanatory force (see Elster, 2007: 14–15).
I will explore several different types of justification for making institutions intermittent. Not every such justification can be or has been invoked for every intermittent institution, but the list is comprehensive as far as I am aware, in the sense that every intermittent institution can be justified by one (or more) of the following arguments. Where possible, I give pride of place to justifications that the political actors concerned have themselves offered in context, but in some cases the justifications are purely external.
2.1. Burdens
The simplest justification for making institutions intermittent is to avoid unreasonable burdens that continuous institutions may impose. To make this issue concrete, I will focus on proposals for continuous elections (see, for example, Knutsen, 2003). The leading version takes the form of a continuous recall. Voters deposit votes in an account for the incumbent or another candidate, and may switch their votes at any time. If the incumbent’s share falls below a preset share of the total, the other candidate takes office. Many of the most obvious objections to the system could be dealt with through technical refinements. Vote shares could be computed as an n-month rolling average in order to prevent ephemeral public passions from having an outsized effect (Knutsen, 2003). Reformers argue for the system on grounds of increased accountability, pointing to cases in which elected officials become unacceptable to large majorities, yet serve for a fixed periodic term and cannot be ousted midway (Knutsen, 2003).
On what grounds, if any, might one oppose continuous elections? Systems of this sort more or less require advanced technology, implying that in an earlier era continuous elections were not technically feasible. Today, the main objection to continuous elections must be that they are undesirable. The main ground for thinking them undesirable is that they are simply too burdensome. If the internal morality of voting (civic duty) requires that voters make efforts to inform themselves about politics, a system of continuous voting implies that voters must constantly consume political news, even against their wishes. Under periodic elections, civic-minded voters must learn a great deal in a limited time, whereas under continuous elections, they must learn less at any given time, but without letup. If the aggregate informational burden of the latter regime is greater than that of the former, continuous elections may impose deadweight losses.
On this account, periodic elections solve a coordination problem among voters. Under a regime of continuous election, even if many voters wish for a breathing spell, whether they will have one depends on the choices of other voters, anticipating the choices of other voters, and so on. The result is multiple equilibria and political uncertainty. Under periodic elections, by contrast, voters have a fixed target (election day) that creates a focal point for the coordinated exercise of civic duty, allowing all concerned to ignore the horse-race elements of politics between elections (except as a hobby, rather than an obligation). Highly permissive rules of advance voting, voting by mail, absentee balloting, and similar electoral practices undermine this coordination benefit of election day and amount to a small step toward a system of continuous elections. 5
On similar grounds, when Woodrow Wilson described the Supreme Court as a ‘constitutional convention in continuous session’, 6 the description was not intended as praise. Part of Wilson’s point was doubtless that an ongoing convention staffed by judges is of dubious democratic legitimacy, yet a complementary point is that such a convention would simply demand too much time, energy, and engagement from citizens. In one picture, constitutional politics occurs in ‘constitutional moments' of extraordinary higher lawmaking that punctuate long stretches of ordinary politics (see Ackerman, 1993: 19–22). Although higher lawmaking can occur either through formal constitutional amendment or through decisive elections that informally determine the shape of the constitutional regime for years to come (Roosevelt’s re-election in 1936 is the most plausible example (Ackerman, 1993: 53)), it is not conceptually possible or pragmatically desirable for every moment to be an extraordinary constitutional moment.
2.2. Distrust
Another reason to make institutions intermittent is distrust of standing institutions or their consequences. Here there are two important sub-cases: distrust of the standing institution itself or else distrust of third parties, for whom the standing institution may represent a fixed target for rent-seeking and corruption.
2.2.1. Distrust of standing institutions
Notoriously, many of the founding generation in the USA feared standing armies, distrusting them as instruments of despotism. The new Constitution of 1789 (Article I, Section 8) thus granted Congress the power to ‘raise and support Armies', but provided that ‘no Appropriation of Money to that Use shall be for a longer Term than two Years'. This sunset provision makes the armed forces a periodically renewed intermittent institution, rather than a standing one.
Yet there are other, less familiar cases that pose similar questions of distrust. Should the executive be a standing office or an intermittent office? The main argument for the latter choice is fear of excessive executive power. At the convention held in Virginia in 1788 to decide whether to ratify the proposed US federal constitution, Patrick Henry argued against the standing presidency, contrasting it unfavorably with the intermittent and episodic dictatorship of the Roman republic: This government is so new, it wants a name. I wish its other novelties were as harmless as this … we had an American dictator [Washington] in the year 1781. We never had an American President. In making a dictator, we followed the example of the most glorious, magnanimous, and skilful nations. In great dangers, this power has been given. Rome had furnished us with an illustrious example. America found a person for that trust: she looked to Virginia for him. We gave a dictatorial power to hands that used it gloriously; and which were rendered more glorious by surrendering it up. Where is there a breed of such dictators? Shall we find a set of American Presidents of such a breed? Will the American President come and lay prostrate at the feet of Congress his laurels? I fear there are few men who can be trusted on that head. The glorious republic of Holland has erected monuments of her warlike intrepidity and valor; yet she is now totally ruined by a stadtholder, a Dutch president. (Henry, 1891)
Henry’s argument rested on skepticism that, in the long run and over many cases, presidents would be as public-spirited as Washington, and as willing to surrender power. A standing presidency must, by the law of averages, sometimes be held by self-interested and power-hungry individuals, whereas an intermittent dictatorship can, in theory, be awarded only to the best types – although Henry does not indicate how Washingtons can reliably be identified ex ante.
Henry contrasted the standing presidency with the Roman dictatorship, to the latter’s advantage. Even the intermittent dictatorship, however, was hedged about with safeguards whose effect (and perhaps intention) was to ensure that the dictator did not convert himself into a standing officer – the dictatorship for life that Caesar assumed after the collapse of the republic. Distrust influences not only the main choice between intermittent and standing magistrates, but also the detailed rules by which intermittent institutions operate.
In Rome, the standing magistrates were constrained by procedural and substantive law, and by institutional design: magistracies were always shared by more than one officer, even at the top of the hierarchy – the two-man consulate. In emergencies, a single dictator with broader powers could be created, yet even the dictator’s powers were hedged about with safeguards. The main pattern was for the consuls to nominate and the Senate to appoint two officials, the dictator and his chief assistant, the magister equitum. The sources are somewhat ambiguous about the length of the dictator’s term, but most suggest a six-month limitation, roughly the length of the campaigning season. Even within that period, the dictator’s powers lapsed when a designated task was accomplished (see Lazar, 2009: 126).
The dictator inspired distrust because he was exempt from many of the usual restraints on the consuls, although the precise scope of the dictator’s powers was a recurring subject of political contention. In Livy’s account, when Cincinnatus was made dictator for the first time, preceded by his lictors [official attendants] he was then escorted to his residence through streets lined with great crowds of common folk who, be it said, were by no means so pleased to see the new Dictator, as they thought his power excessive and dreaded the way in which he was likely to use it. (Livy, 1971: Bk. III, Ch. 26, p. 213, emphasis added)
Distrust must have been the paramount motive for limiting the dictator to a six-month term; the choice of a rule rather than a standard suggests that the political coalition behind the appointment was willing to incur (or was forced to accept) a risk that the emergency the dictator was appointed to solve would still be extant when the term expired. However, because the six-month term approximated the campaigning season in the local conflicts of the early republic, that risk was a small one, and the rule was reasonably well tailored to the exigencies of the military environment. Moreover, beyond the explicit legal rules governing the office, the dictatorship was surrounded by a web of informal political norms and politically useful mythologizing, celebrating dictators who stepped down and returned to a humble private life, especially if they did so before the end of their appointed term. Having repelled a military threat, Cincinnatus famously ‘resigned after holding office for fifteen days, having originally accepted it for a period of six months' (Livy, 1971: Bk. III, Ch. 29, p. 230).
While the dictatorship was an episodic intermittent institution, distrust can also be a rationale for preferring periodic intermittent institutions to standing ones. The Roman censors, who served as a pair, had broad powers to deprive citizens of their voting rights, property, or place in the Senate, and to impose public stigma on those the censors deemed immoral. Given these powers, the history and features of the office were pervasively shaped by public distrust. Even when elected on a periodic five-year schedule, the censors were ‘unusual among Roman city magistrates in having no fixed date for demitting office’ (Lintott, 1999: 116). The censors' duties were held to be complete once they had taken a census of all citizens, altered the list of senators according to legal and moral criteria, and performed a ritual sacrifice, after which the censors would leave office. As these tasks were typically accomplished well before the five-year cycle for electing censors ran its course, the norm was that the censors would serve only a year or two.
Although our evidence is uncertain, at a later date this norm seems to have morphed into an enacted legal rule restricting censors to a maximum term of 18 months. The impetus seems to have been an incident in which ‘a pair of censors argued that they had the right to stay in office until the next election of censors, which led to accusations of tyranny’ (Lintott, 1999: 117). In this example, an attempt to convert a periodic intermittent office into a continuously occupied office provoked grave public distrust. The 18-month maximum term did not at all imply that elections for the censorship would be held every 18 months. Indeed, the point of the norm-transmuted-into-law seems to have been a public demand that for the last several years of each five-year census cycle, there should be no censor in office at all. Here the distrust justification and the burden justification for intermittent institutions overlap: one can imagine that distrust of the censors' sweeping powers was combined with a desire to throw off, at least temporarily, the burden of their brooding moral supervision.
As the example also illustrates, the psychology of distrust characteristically runs together two distinct issues: distrust of an office and distrust of its occupants. In some cases, it might be desirable that a powerful office have continuous existence, so long as there is a sufficient rate of turnover of the officeholders. It is logically consistent to approve both of the ‘imperial presidency’ (Schlesinger, 2004) and also the twenty-second amendment of the US Constitution, limiting any particular individual to two presidential terms. Indeed, one might approve of the former institution because, and only because, the latter safeguard is in place. Yet distrust tends to blur these distinctions.
Although the causes of this phenomenon are unclear, one may speculate that a political heuristic is at work: the public is quick to infer that anyone who accepts a powerful office is motivated by lust for power, so that distrust of the office transfers automatically to distrust of its occupants. In an attempt to negate this inference, both dictators under the Roman Republic and emperors elected by soldiers or senators under the Empire made a ritualized show of refusal before taking office, portraying power as a burden thrust upon them. Before the Senate installed Cincinnatus as dictator a second time, he argued (unsuccessfully?) against his own appointment (Livy, 1971: Bk. IV, Ch. 13, p. 305).
2.2.2. Distrust of third parties
In the previous cases, a reason for making institutions intermittent was to prevent them from becoming too strong. In other cases, however, one might distrust standing institutions because they are too weak, and will prove unable to resist capture, rent-seeking, or corruption by private or public third parties. The standing institution can be a sitting duck, whereas an intermittent institution is a more elusive target.
One justification for the peculiarities of the jury system, for example, is a fear of corruption or intimidation by private parties. Petit juries sit only for one case, in part because of the burdens on jurors of putting aside their private concerns. Another plausible concern, however, is that private parties (especially repeat players) will have greater opportunity and incentive to corrupt or intimidate a standing panel of decision-makers than a panel that disperses after each decision. That concern was important in founding-era debates over the jury system: If bribery was the mortal enemy of impartiality, then a network of local juries rotating anonymous persons through its ranks was far more bribery-proof than standing panels of known judges could ever be. As one Anti-Federalist put it, ‘Judges once influenced, soon become inclined to yield to temptations, and to decree for him who will pay the most for their partiality … [By contrast,] it is not, generally, known till the hour the cause comes on for trial, what persons are to form the jury’. (Abramson, 1994: 32; quoting Storing, 1981: Vol. 2, p. 321)
The intermittent character of the petit jury can also be cited to show that the petit jury will be less subject to influence from prosecutors than will a standing institution such as the judiciary. The Supreme Court implied as much in Duncan v. Louisiana (391 US 145 (1968)), describing the petit jury as a safeguard against ‘corrupt or overzealous prosecutors' and contrasting the jury favorably with ‘compliant [or] biased judges' (391 US at 156 (1968)). In other words, distrust of standing public actors such as the prosecutor and judiciary, on the one hand, and distrust of private litigants, on the other, are mutually compatible concerns: the petit jury’s ephemeral character makes it at least somewhat independent of all the forces that would attempt to suborn it.
2.2.3. Distrust and expertise
The notorious price for this independence, however, is the petit jury’s lack of expertise, as the Supreme Court recognized in Duncan v. Louisiana when it contrasted the ‘common-sense judgment of the jury’ with the ‘more tutored but perhaps less sympathetic reaction of the single judge’ (391 US at 156 (1968)). On one view, more information generally implies more bias (see Levmore, 1998: 2100–1), so that minimizing bias by making petit jury panels intermittent precludes such panels from learning over a series of cases. This retardation of institutional learning is a general concern about intermittent institutions, as I will discuss below.
A different view, however, holds that information is power, so that a jury that sits on a series of cases and develops a measure of expertise will in fact enjoy more independence from the prosecutor, who would otherwise hold all the informational cards. Thus the longer life span of the American grand jury, which typically considers a number of prosecutorial applications for indictment over many months, has been justified as a necessary prerequisite to questioning the prosecutor’s judgment (see Brenner, 1995: 90–1). The opposite argument about the grand jury is that the repeat-play relationship causes grand jurors to defer more heavily to the prosecutor than petit jurors would (Fouts, 2004: 328–9). The latter argument parallels the justification for intermittent petit jury panels that disperse after a single case. While the premises of these two arguments are in some tension with each other, as a logical matter the arguments are not mutually exclusive: as compared to petit juries, grand juries may simultaneously have greater capacity to challenge the prosecutor’s judgments, yet less willingness to do so.
2.3. Self-defeating continuity
The most general justification for intermittent institutions is that in some circumstances continuous life would be self-defeating. In a strong form, some institutions are defined by their extraordinary character, so that continuous life would be conceptually incoherent. In a weaker form, continuous life would pragmatically undermine or defeat the purposes for which the institution is created. Analytically, either argument is sharply distinct from an argument that a standing institution would have pernicious collateral effects, because it would be too burdensome or too powerful or too easy a target for corruption. In practice, however, the arguments are often hard to disentangle, in part because political actors often attempt to fold concerns about collateral effects into the definition of the institution’s purposes. Nevertheless, I will attempt to isolate a few clear cases.
One example involves Livy’s suggestion that at least in some cases, the true purpose of the Roman dictatorship was not so much to repel external enemies as to cow the plebians. The impetus for Cincinnatus' second appointment as dictator was that Spurius Maelius, an eques (wealthy cavalryman), was distributing grain among the starving poor in what the patricians feared was a bid for monarchy. The consuls argued that they were not to blame for permitting the situation to get out of hand: for, fettered as they were by the Appeal laws, deliberately designed to break their authority [by giving citizens an appeal to the popular assembly before magistrates could inflict punishment], they had less power than will to punish an offence of that sort in the summary fashion it deserved. What was needed, was not merely a resolute man, but a man who was also free from the net of legal controls … [T]he following day [the dictator] picketed the city and made his appearance in the Forum, where the unusual and surprising sight turned all eyes upon him. Maelius and his fellow conspirators knew well enough that it was against themselves that the powers of that exalted office were directed, but all who were not in the plot asked in bewilderment what sudden rising or unexpected threat of attack could have rendered necessary so drastic a step – a Dictator in all his majesty. (Livy, 1971: Bk. IV, Ch. 13, p. 305, emphasis added)
Although Livy does not spell out the details, his narrative here and elsewhere implies that the dictator’s awful presence – emphasized by political symbolism, such as the fasces (ceremonial rods with axes) borne before the dictator by official attendants – exerted a kind of in terrorem effect on the seditious plebians. If familiarity breeds contempt, however, then the terror inspired by the dictator presupposes that the dictatorship will be highly intermittent. Quite obviously, a standing dictatorship of indefinite duration would, in the Roman political schema, simply recreate kingship and thus violate the central political norm of the republic. Less obviously, Livy implies that continuous life would leach the dictatorship of its extraordinary political aura and dilute its effect on the crowd, thereby defeating the ends for which the aristocracy created the office.
In a similar fashion, some intermittent institutions or practices possess an expressive value or social meaning or informational content that arises only by virtue of an extraordinary departure from the norm. If declarations of emergency, or of an increased threat level in a terrorism-warning system, become too frequent, each successive declaration becomes less and less informative. The Roman triumph, in which a victorious general would enter the city gates and parade through the Forum, was a deliberate breach of one of the central political norms of Roman life, a norm forbidding holders of a military imperium (command) from entering the sacred civil space surrounding the city. The triumph was thus, on one interpretation, a ritual that created ‘a hole in the augural space’ (Brennan, 2004: 42). Just as a hole is defined by the contrast with what surrounds it, a triumph is defined as a breach of the ordinary; a continuous triumph would be an oxymoron.
3. Episodic versus periodic institutions
So far, I have focused on justifications for intermittent as opposed to standing institutions. Within the class of intermittent institutions, there are several justifications for choosing episodic over periodic institutions, or the reverse. Subsections 3.1 and 3.2. examine those justifications. Completing the set of logically possible comparisons, Subsection 3.3 will note some cases in which the question is whether episodic institutions can perform better than standing institutions at accomplishing certain tasks, including the design of new constitutions and the reform of existing policies.
3.1. Episodic problems
Just as the simplest argument for intermittent institutions is that standing institutions impose excessive burdens, so too the simplest argument for episodic institutions is that certain classes of policy problems are themselves episodic; creating a periodic institution to cope with such problems would be unnecessary. Lord Bryce criticized the system of periodic elections for the US presidency on the ground, among others, that ‘the presidential election, occurring once in four years, throws the country for several months into a state of turmoil, for which there may be no occasion … [I]f issues do not exist, they have to be created’ (1908: 53). Similarly, one might imagine a counterfactual version of the Roman republic in which a dictator came into existence on a regular schedule, yet such a system would be bizarre if serious military threats or internal sedition did not occur on a predictable cycle. The problem of distrust is relevant here as well: calling a distrusted institution such as the dictatorship into being periodically increases the risk that the institution will be put to uses other than those intended by its creators.
As against this, one might prefer periodic rather than episodic institutions on the ground that in-system actors will be too unwilling to trigger episodic institutions when they are truly necessary. Although most discussions of the dictatorship emphasize the Type-I error (that the dictator will take power when there is no true emergency or hold onto power after the emergency has passed), there is also a Type-II error to consider, in which incumbent consuls and senators fail to appoint a dictator when they should. The latter error might arise because the appointment of the dictator would at least temporarily diminish the power of the very officials doing the appointing. The very checks that minimized the Type-I error, namely the customary (although not invariable) requirement of agreement by consuls and senate, might also have exacerbated the Type-II error. Indeed, the dictatorship fell into abeyance during the full flowering of the Senate’s dominance after about 200
3.2. Strategic behavior: periodic versus episodic institutions
A related problem is that periodic institutions may cause actors to engage in socially wasteful strategic behavior. Because the periods are predictable, the actors' time horizon will be decisively shaped by the institutional cycle, rather than by the merits of the underlying problems of policy. A topical example involves ‘redistricting’ in American states, which is typically tied by law to the decennial census. As redistricting becomes imminent, parties notoriously engage in strategic behavior in order to anticipate and shape the redistricting fight. Anticipating the 2000 redistricting cycle in Texas, Democrats filed a preemptive lawsuit in Austin, a liberal enclave, while Republicans filed a preemptive lawsuit in Houston, a conservative bulwark; each party thereby hoped to have the case decided by a local judge who would share its politics. 8 A randomly timed redistricting cycle might obviate these difficulties. 9
A more serious example involves the so-called ‘political business cycle’ (Nordhaus, 1975). Because presidential elections in the USA occur on a fixed four-year cycle, the incumbent presidential party has an incentive to pursue policies that result in high unemployment and low inflation during the first part of the electoral term and low unemployment and high inflation during the second part. Where long-term aggregate welfare would be better served by a longer period of high unemployment and low inflation, the political cycle induces an economic distortion. A stock objection is that the cycle pictures politicians as strategic, but voters as myopic; it assumes that voters do not form rational expectations and do not take into account the post-election consequences of the incumbent’s behavior in the period before the elections. Here I will simply assume the existence of the political business cycle and examine one possible remedy for the problem: randomly timed elections.
In one model, randomly timed elections would dampen or even eliminate the political business cycle. 10 Behind a veil of uncertainty about when elections will occur, incumbents can do no better than to choose policies that treat the present and future in an impartial fashion. To be sure, no real-world liberal democracies explicitly use randomizing devices to trigger national-level elections, so it is difficult to find evidence on the validity of this mechanism. However, as discussed above, parliamentary systems, which contain elements of both periodic and continuous elections, also contain elements of the randomly timed election. From the parliamentary government’s point of view, the n-year maximum term of office is liable to be cut short by an exogenous shock that more or less requires early elections to be called, and this uncertainty may at least somewhat dampen or smooth the government’s incentives to manipulate unemployment over time, compared to systems with a fixed cycle. As it turns out, there is some evidence that the political business cycle is indeed more pronounced in systems with periodic electoral cycles than it is in parliamentary systems. 11
This does not necessarily imply that randomly timed elections are a good idea overall, because their collateral costs must be taken into account. Many objections to such elections might be advanced, but I will confine myself to the one that is most relevant for present purposes. Randomly timed elections might induce a truly permanent electoral campaign, in which the incumbent treats the whole (expected) electoral cycle as though it were the last year of the cycle. If this occurs, randomly timed elections might perversely increase the net amount of strategic behavior. In a weaker version of the same concern, random elections might cause the incumbent to engage in relatively less wasteful strategizing during the final year, because the incumbent will not know it is the final year, but in relatively more during the early year(s) of the administration, which will not be known to be such. The net amount of undesirable behavior under randomly timed elections could then be less than, roughly equal to, or greater than under fixed elections.
The advantage of the fixed electoral cycle, on this perspective, is that it at least guarantees the incumbent a breathing space of a few years, in the first part of the electoral cycle, in which to pursue optimal policies. While randomly timed elections might produce better results, they might also produce worse ones, and are thus a risky choice from the social point of view. These possibilities are clearly speculative – the results of the two systems will depend on the incumbent’s rate of time discounting, risk preferences, control over the economy and policy, and a host of other variables – but they at least suggest that randomly timed elections are not obviously superior.
The final example involves revision of the constitutional framework itself. Jefferson (1958) argued that constitution writing should be a periodic enterprise occurring on a fixed schedule, with the constitution lapsing every 19 years, in order to prevent rule by the dead hand of past generations. As against this, Madison (1958) argued, in effect, that periodic ‘sunsetting’ of the constitution would induce socially wasteful strategic behavior: Would not a Government, ceasing of necessity at the end of a given term, unless prolonged by some Constitutional Act previous to its expiration, be too subject to the casualty and consequences of an interregnum? Would not a Government so often revised become too mutable and novel to retain that share of prejudice in its favor which is a salutary aid to the most rational Government?
Would not such a periodical revision engender pernicious factions that might not otherwise come into existence, and agitate the public mind more frequently and more violently than might be expedient?
3.3. Strategic behavior: standing versus episodic institutions
In these examples, strategic behavior arises from the periodic character of the intermittent institution, because the predictability of the institutional cycle shapes the incentives of the relevant actors. The comparison is thus between periodic and episodic intermittent institutions. In another class of cases, there is an important comparison between standing and episodic institutions.
At least since 1776, most constitutional design or amendment in most polities has been entrusted to standing institutions, usually legislatures that combine constitutional and ordinary functions. In the USA, constitutional amendments have invariably begun with congressional approval, although this is only one of several legally permitted paths. While the US federal constitution was initially drafted by a freestanding constitutional convention, such conventions are somewhat anomalous in historical and comparative perspective. Typically, constituent assemblies also possess a concurrent power to legislate (Elster, 2010).
In the classical world, by contrast, constitutional design or large-scale reform was sometimes accomplished by a single-shot lawgiver. In extreme cases, the lawgiver would then withdraw from politics or the polity: the Athenian lawgiver Solon exiled himself for 10 years, while the Spartan lawgiver Lycurgus disappeared into obscurity. Although these episodes are murky and partially mythologized, there are many other lesser-known cases (see Gagarin, 1989: 58–62), and little reason to doubt that something of the sort actually occurred. The practice can best be justified on the ground that the single-shot lawgiver, who has no concern for re-election or reappointment, will have no self-interested or strategic incentive to distort his judgments on the merits of constitutional rules. Whereas Madison argued that legislators are constrained to behave impartially by the anticipation of living as ordinary citizens under the laws that the legislators themselves create, 12 the classical practice suggests that the maximally impartial lawgiver is one who will not be affected at all by his own laws.
The obvious downside is that the single-shot lawgiver may be insufficiently accountable for his choices. Indeed, in Plutarch’s account (1914: Vol. 1, p. 123), Solon withdrew from Athens partly in order to avoid criticism and demands by his fellow citizens that he give reasons for his choices. By contrast, the ‘citizens' assemblies' that have recently been entrusted with agenda-setting power to trigger a referendum on constitutional reforms, in two Canadian provinces and elsewhere, are composed of citizens chosen by lot who then return to ordinary life (see Warren and Pearse, 2008). Such citizens exemplify the Madisonian approach of impartiality and accountability generated by the lawmakers' anticipation of living on equal terms with fellow citizens, rather than the Solonian approach of impartiality conceived as maximal detachment from fellow citizens. The two approaches, however, share the premise that lawmaking by single-shot or highly episodic institutions can minimize the distortions arising from strategic behavior. Whether the lawgivers leave the jurisdiction altogether or return to the body of the citizenry, they do not pursue re-election to a standing institution and a permanent political career.
In a less extreme form, it has been argued that constitutional conventions, which assemble for the single task of writing a constitution and then permanently dissolve, are more likely to be impartially motivated than are hybrid constituent assemblies that combine their constitution-writing functions with the standing functions of an ordinary legislature (Elster, 2010). Members of the latter bodies may seek re-election and strategize to obtain it, and are thus likely to display excessively short time horizons or to pander to narrow interests. By contrast, single-shot lawgivers and citizens' assemblies face no re-election constraint and thus evade bad accountability to interest groups and other repeat political players. 13
Similar trade-offs arise when the question is whether to channel policymaking through standing institutions, such as standing congressional committees, or through episodic institutions, such as ad hoc congressional committees or special blue-ribbon commissions, for instance the 9/11 Commission. Legislatures sometimes create military-base-closing commissions, and empower them to propose take-it-or-leave-it packages of cuts, precisely in order to loosen the grip of regional and economic interests who have excessive influence on standing legislative committees and who would otherwise stymie reform. In general, single-shot policy committees or investigative committees, whose members need never be reappointed or re-elected, can partly or wholly evade bad accountability. By the same token, however, their members may be less subject to good accountability to constituents or to the citizenry at large. However these trade-offs are best resolved in a given setting, the main argument for single-shot or episodic institutions is that their fleeting character puts them in a better position than either periodic or standing institutions to carry out difficult tasks of constitutional design and large-scale policy reform.
4. Problems
Having elicited some justifications for institutions that are intermittent rather than continuous and for institutions that are episodic rather than periodic, I now turn to problems with intermittent institutions as a class. Intermittent institutions display common defects that regularly recur across contexts. This need not imply that they are bad, all things considered. In some contexts, the benefits will outweigh the costs.
4.1. Interrupted traditions and many minds
Under certain conditions, continuous institutions might embody a kind of accumulated wisdom developed and constantly adjusted by ‘many minds' over time (see Sunstein, 2006b, 2009). The argument for the latent epistemic value of continuous institutions comes in many shadings and variants; in one version or another, it is a long-standing favorite of political conservatives. Burke’s version
14
is the most famous, but hardly the first. Cicero’s rendition of the many-minds argument runs as follows: Cato used to say that our constitution was superior to others, because in their case there had usually been one individual who had equipped his state with laws and institutions [for example, Solon and Lycurgus] … Our own constitution, on the other hand, had been established not by one man’s ability but by that of many, not in the course of one man’s life but over several ages and generations. He used to say that no genius of such magnitude had ever existed that he could be sure of overlooking nothing; and that no collection of able people at a single point of time could have sufficient foresight to take account of everything; there had to be practical experience over a long period of history. (Cicero, 1998: Bk. II, Section 2, p. 35, emphasis added)
The argument from the epistemic benefit of continuous institutions is shaky on its own terms, and in my view holds only under fragile and rather precious conditions (Vermeule, 2009a, 2009b). Putting that general issue aside, however, intermittent institutions create a separate set of problems for epistemically inflected conservatism. Intermittent institutions represent a kind of interrupted tradition that undermines the premises of the many-minds argument. 15
In most of its standard renditions, the emphasis of that argument is on incremental and continual small adjustments, accomplished either through intentional tatonnement or through some sort of evolutionary process constrained to take small steps (‘gradient-climbing’, as Elster (2007: 111) put it). Such processes may get stuck at local maxima rather than global peaks (Elster, 2007: 111), but even where they can reach global peaks, they can only do so through continuous adjustment. By contrast, an interrupted tradition, in which actors decide to trigger an episodic institution, amounts to a discontinuous leap from one set of institutional arrangements to another. This requires not only an intention to break with the immediate past, but also the sort of overall vision of the situation that Cicero and later epistemic conservatives deride as beyond the capacities of any one person – or even, Cicero adds, any given generation.
In an epistemic framework, interrupted traditions are a strange beast, a kind of epistemic camelopard. Such traditions can amount to antiquarian revivals of old institutions in a changed environment, in many cases after a long lapse of time. An extreme example is the French Estates-General, which met at long and irregular intervals between 1302 and 1789, typically in times of crisis for the monarchy, when it needed consent to new taxes from the main status groups of the realm. When they met for the last time in 1789, only to dissolve themselves permanently, the Estates had not been convened since 1614 – a time lag roughly equivalent to holding a federal constitutional convention in the USA today. 16
Two consequences followed. The first was massive uncertainty about the rules governing the Estates (I take up this problem shortly). The second consequence was that the Estates met in an environment that had changed out of all recognition. The corporate particularism of the medieval world had given way to a world of nationalism, yet the Estates were still organized along corporate lines and the relationship between constituents and their deputies was one of fees for services rendered, rather than one of genuine political representation within the framework of a national constitutional order (Ulph, 1951: 230–1). One historian speculates that payment of deputies by the crown might have given deputies a national, as opposed to parochial, perspective, but notes that Since the estates general were usually summoned when the royal treasury was empty, the crown was incapable of introducing the practice of royal payment of deputies on its own initiative. No escape from this impasse was possible. The estates general lay dormant for one hundred and seventy-five years, and any national sentiments which the orders might have possessed were awakened too late to save the institution. (Ulph, 1951: 231)
The example suggests that the very conditions of crisis that induce actors to convene episodic institutions may also prevent needed adaptation of those institutions. At a minimum, this rules out the sort of continual adjustments that usually underpin claims about the latent wisdom of old institutions. In the most extreme cases, episodic institutions may simply break down when they are called into being in crisis-ridden environments radically different than the ones in which they were created. In 1789, neither the delegates nor other actors were capable of carrying out, all at once, the wrenching changes needed to adapt the Estates-General to the new environment. By contrast, as De Tocqueville (1856) pointed out, many standing institutions of the ancien régime showed surprising persistence, reappearing in new guises after the Revolution.
A less extreme example, from the USA, involves military tribunals and commissions. Such bodies come in a bewildering variety of forms, but appeared in one version or another in the American Revolution, in 1847 (the Mexican–American War), during the Civil War and Reconstruction, in World War II and its immediate aftermath, and after the terrorist attack of 11 September 2001. In the interstices of these major episodes, military tribunals appeared occasionally in a limited way, such as in the Philippines after the Spanish–American War, but their use was not widespread. Counting only the major cases, military tribunals are a highly episodic institution. After Reconstruction, ‘seventy years passed before military commissions came into widespread use again’ (Vagts, 2007: 41), while after World War II, some 50 years elapsed before military tribunals were revived after 9/11.
Given these long periods of dormancy, it is odd to complain that the Bush administration’s 2001 version of military tribunals represented a departure from the ‘carefully carpentered’ procedures of past versions (Vagts, 2007: 36). The military, political, and legal environments in which tribunals have been used have differed so greatly, over time, that it would be mindless (pathologically rigid) to use the same procedures over and over again. Interrupted traditions such as the military tribunal do not embody the kind of incremental and continuous adjustments that might have a claim to embody latent collective wisdom, at least in a relatively static environment. Equally oddly, Bush’s 2001 order was quite similar to Roosevelt’s 1942 order, 17 which had been upheld by a unanimous Supreme Court ( Ex parte Quirin, 317 US 1 (1942)). As a matter of policy, the right objection to the 2001 order was not that it departed from the latent wisdom of the past, but that it mindlessly imitated the past despite changed circumstances. 18
Furthermore, military tribunals have no fixed form, but, with the conspicuous exception of the 2001 Bush order, have historically changed their procedures and rules to some degree in each new instantiation (see Glazier, 2005). The ‘tradition’ of military tribunals is thus a peculiarly self-negating type of tradition, one that itself embodies abrupt revivals and discontinuous change. Cicero himself, the arch exponent of many-minds constitutionalism, supported a law granting unprecedented military powers to Pompey with the argument that ‘our ancestors have always yielded to precedent in peace, but expediency in war, and have always arranged the conduct of new policies in accordance with new circumstances' (Brennan, 2004: 35). The twist is that Cicero ascribes this policy to the ‘ancestors'; the argument is not simply for innovation, but instead for a second-order tradition whose practice is to throw tradition overboard when expediency dictates. 19 I suggest that this captures the peculiar institution of military tribunals, and more generally that it captures a characteristic feature of intermittent institutions, especially episodic ones. Although such second-order traditions of punctuated innovation may have many benefits, they cannot be justified by reference to the putative epistemic virtues of continuous and incremental change.
4.2. Instability and political controversy
I turn now to a complex of problems involving instability and intractable political controversy. Once created, standing institutions in some sense go of themselves, whereas intermittent institutions require further action to bring them into being. If the intervals between appearances are long, the result is political conflict about the definition and ground rules of the institution itself. The less frequently a given institution comes into being, the more unstable it becomes; in extreme cases, its very identity becomes elusive. Such conditions give rise to strategic behavior that creates deadweight costs – politics in the most pejorative sense.
The Estates-General again supplies an illuminating case. From the beginning, ‘the elections to the Estates-General and their internal organization followed no set pattern … [Moreover,] the internal rules of the orders were as ill-defined as the rules of the Estates themselves' (Elster, 2009: 9, 12). At several key moments, such as in 1614, political struggle over the ground rules caused the Estates to collapse in acrimony. When the Estates were summoned for the next and last time in 1789, basic and highly consequential rules of procedure were thus unsettled. The main one was whether the Estates would vote by order (in which case, the clergy and nobility might outvote the commoners of the Third Estate) or, instead, by head (in which case, the double size of the Third Estate, a concession granted by the king under political pressure, might be decisive). The impasse over this question proved irresolvable; the issue only disappeared when the Third Estate seceded along with some of the clergy and nobility, declared itself a National Assembly, and abolished the Estates-General for good. Throughout the history of the Estates, ‘there were no fixed rules of the game, only self-serving claims that this or that principle was a fundamental law of the kingdom’ (Elster, 2009: 15).
The (repeated) collapse of the Estates-General illustrates a chronic problem that is latent in the less extreme cases. In my view (a view that is philosophically controversial and that I will not defend, but simply state), when the rules of an institution are essentially contested by the actors concerned, the institution does not exist. Institutions are just equilibria, temporarily settled networks of mutually adjusted expectations. If actors cannot agree on the minimal operating procedures that define an institution, then there is no conjuncture of expectations, an equilibrium is lacking, and there is no fact of the matter about the institutional rules. Moreover, even if there were a fact of the matter, its existence would have no pragmatic importance for action, in light of the actors' intractable disagreement (compare Waldron, 1992). These points hold even if all the actors concerned erroneously believe their disagreement to be genuine, rather than ersatz. The problem is particularly acute for episodic institutions that tend to appear at long intervals. In such cases, politically motivated disagreement about the institution’s nature and procedures often becomes so acute that the institution itself dissolves.
An implication is that some recent arguments, in American legal theory, over whether to convene a new federal constitutional convention are incomplete. In these debates, proponents of a new convention point to the (indisputably) obsolete and (arguably) undemocratic character of much of the federal Constitution (see Levinson, 2006). Opponents respond that despite its imperfections, the Constitution works adequately well, and that a new constitutional convention might simply produce a document that is worse overall (see, for example, Sunstein, 2006a). Neither camp considers the very real possibility that such a convention would simply collapse into politicized wrangling over procedural ground rules. As the experience of the Estates-General suggests, when episodic institutions are reconvened after a lapse of centuries, it can turn out that the institution’s rules, procedures, and indeed its very existence have an elusive or illusory quality.
5. Conclusion
In the foregoing, I have attempted a thematic survey of justifications for intermittent institutions as opposed to standing ones, of justifications for casting intermittent institutions in a periodic or, instead, episodic form, and of some general problems with intermittent institutions. I will conclude by drawing together the threads of the discussion in order to state some comparative statics – some conditions under which intermittent institutions (and their respective subclasses) are more or less appealing.
Frequency of appearance. As intermittent institutions appear more frequently, whether on a periodic schedule established ex ante or, instead, on an episodic basis, they tend to become more stable and successful. Frequent and repeated appearances seem to clarify and stabilize the ground rules of the institutions, dampen strategic behavior and rent-seeking, and lower the stakes for all parties. Highly intermittent institutions that appear at very long intervals, such as the Estates-General, are typically the least stable and least successful.
On the other hand, as we have seen, continuous institutions (the limiting case of frequency of appearance) can be too burdensome, generate too much distrust, or defeat their own ends. The upshot is that while there are, in some institutional and political environments, good arguments for intermittent institutions, they should usually not be very intermittent. Roughly speaking, considerable frequency of appearance seems to optimize the cross-cutting considerations and is thus the most promising approach. An implication is that periodic institutions should be given relatively short cycles, while episodic institutions should be relatively easy for in-system actors to create, and will thus (in expectation) appear with greater frequency.
Rapidly changing environments. Holding constant the frequency with which intermittent institutions appear, and the duration of their appearance, the rate of environmental change matters as well. As change becomes rapid, intermittent institutions are more likely to appear in political contexts remote from their creation, putting greater strain on the actors who must quickly attempt to adapt them to the new conditions. Continuous institutions, by contrast, may do better as the rate of environmental change increases; even if their elasticity is low, it cannot be worse than that of an intermittent institution that reappears in an antiquarian form. Accordingly, intermittent institutions tend to show to best advantage in political environments that change relatively slowly. The Roman dictatorship, whose intermittent form reduced the risk of tyranny and alleviated public distrust, plausibly functioned most successfully in the fairly static environment of Italian city-state conflict during the republic’s first centuries.
Ex ante predictability. When political actors can predict the appearance of an intermittent institution and have the capacity to take socially harmful action or to engage in strategic behavior in light of that prediction (as in models of the ‘political business cycle’), periodic institutions are less appealing than episodic ones. Randomly timed elections are a pure case of the latter, and might dampen incumbent politicians' strategic behavior. This conclusion is sensitive to many assumptions about actors' risk preferences and other factors, however. In any event, periodic institutions are preferable where predictability is a benefit, not a cost, as when a fixed schedule of elections gives voters a fixed target for political involvement and thus frees them up for other pursuits.
Overall, mid-level generalizations of this sort are the most that can be expected. Intermittent institutions are one useful institutional form among others, and deserve a place in the toolkit of institutional design, but their costs and benefits are highly context-specific. The only stronger conclusion that seems justified by the cases is that the benefits of intermittent institutions plausibly exceed their very real costs only in a fairly narrow range of environments. Despite the many examples of intermittent institutions in political history, then, there may be some rough justification for the brute fact that in any given polity the great majority of institutions are standing institutions.