Fifty Shades of Republic | Part 3: constitutional amendment rules

This post is part of Fifty Shades of Republic, a series of overviews of US political institutions at the state level

While the Federal Constitution is arguably the world’s oldest national constitution, the states were the ones that began the practice of having ‘written’ constitutions – entrenched laws with a higher status than regular laws, establishing the main features of the system of government. Since I’ve been doing some work on constitution amendment rules (of both national and US state constitutions) for my dissertation, I thought I’d do this topic next (it is also closely related to the topic of a podcast I am currently preparing for Leviathan’s Couch).

The amendment procedure has far-reaching effects. John Burgess, one of the 19th century pioneers of political science, argued it to be the most important part of a constitution. Constitutional amendment procedures entrench written constitutions, making them harder (or, at least, different) for politicians to change than regular laws. Entrenchment is meant as a means to provide the system of government with stability and to protect basic rights from change by temporary majorities. However, it also often leads to the constitutional entrenchment of various policy measures that do not pertain to the state’s basic institutions, either as a result of opportunistic coalitions seeking to protect legislation from future change or because restrictive provisions in the constitution make the incorporation of contradictory provisions into the constitutional text the easiest way of passing certain policies. Lastly, entrenchment is also an important driver of the power of the judiciary (to whom enforcement of the constitutional hierarchy is delegated).

Most US states have two or more of the following tracks to initiating a constitutional amendment: by the legislature, by a voter initiative, or by constitutional convention. The final stage is almost always ratification by voters in a referendum.

Map 1: legislative and voter initiative tracks to amendment – click for full size

Proposal

Supermajorities – 28 states today require a supermajority for passage in the state legislature. 22 only require regular majorities (simple or absolute) before going to referendum.

Of those only requiring approval by one legislature:

  • 10 states (turquoise) require a majority in both houses
  • 9 states (blue) require 3/5 in both houses
  • 16 states (dark blue) require 2/3 in both houses.

Of those requiring intervening elections:

  • 11 states (pink) require a majority in both houses before and after the elections
  • 3 states (red) require a 2/3 supermajority at one stage but not both. In Tennessee, the votes after the election must be by 2/3 in both houses; in Vermont, the amendment must be supported by 2/3 before the election, although only in the state Senate. South Carolina (dark red) requires 2/3 in both houses, after which the amendment is put to the voters at the following election. The referendum, however, is not final; it the amendment is approved, the newly-elected legislature must ratify it by majority vote in both houses.
  • Delaware (green) requires 2/3 in both houses both before AND after the intervening election, with no referendum requirement.

Intervening elections – 15 states require approval by two legislatures separated by intervening elections (shown on the map in various shades of red). A few of those states allow this requirement to be bypassed by supermajority support in the legislature (shown by a blue asterisk, with the shade of blue corresponding to the required supermajority (in states without intervening election requirement – see below) – only exception being Connecticut, where ¾ is needed to bypass the intervening election requirement). Since main idea behind this institution is often said to have been to allow voters input on constitutional amendments through the election process, many states that had this feature in earlier constitutions removed it when introducing ratification of amendments by referendum, but many have kept it nonetheless.

Initiative – Currently, 17 states allow voters to propose amendments to referendum by petition (indicated by a capital letter I on the map). The exact signature requirements vary widely from state to state. This provision originated in the Progressive era, although a few states introduced it later.

Ratification

Referendum – the referendum requirement became a near-universal feature by the time of the Civil War. In states that did not have it to begin with, it often replaced requirements for intervening elections and/or assembly supermajorities. Referendums today require a majority to ratify in almost all states; the denominator is sometimes simply the number of valid votes on the respective amendment question, but in some states it is the more demanding number of people voting at the election, so that ballots left blank count as a ‘no’. In a few states a supermajority is required for some or all amendments, e.g. in Colorado, 55%, in New Hampshire, 2/3.

Legislative vote – today, Delaware (in green) is the only state not to require voter approval for constitutional amendments – the second round of 2/3 vote in the legislature makes an amendment part of the constitution. As noted above, South Carolina does require voter approval, but an amendment approved by voters is still subject to a final (majority) vote by both houses.

Map 2: constitutional convention track to amendment (or replacement) – click for full size

Legislative convention call – In 6 states, the legislature can directly order the election of a constitutional convention (without needing voter approval) by either majority (light green) or supermajority (green) in both houses.

Legislative convention proposal – In 30 states, a convention call by the legislature must be ratified by voters. In some states the proposal can be made is by a majority (light purple), in others by supermajority (purple) in both houses. In Pennsylvania, there is precedent for this, even though the state constitution does not provide for it.

Ambiguous, referendum required – In 3 states, the constitution requires the summoning of a constitutional conventions to be approved by voters, but is silent on whether the legislature can propose this (pale blue).

Automatic ballot question – In 14 states, a convention proposal must be placed before voters at least once every certain number of years. This is indicated on the map by the specific number of years in each case.

Initiative – In 17 states, voters can initiate the summoning of a convention, either by the constitution’s explicit provision or by virtue of the possibility of proposing one using the procedure for initiating amendments. This is indicated by a capital letter I on the map.

No provision – in 11 states, the constitution does not provide for the calling of a constitutional convention (grey) nor does the state have any established practice. De facto, each of these state legislature could summon a convention by means of the regular amendment procedure.

Constitutions are usually silent on the conditions for conventions’ proposals to be made or ratified, with the exception of requiring a referendum as on ‘regular’ amendments. A few noteworthy exceptional provisions include Illinois’ requirement for 60% voter approval for any kind of convention proposal to be successful, New Hampshire’s requirement for amendments to be proposed by 3/5 vote of the convention, and Minnesota’s requirement for 60% voter approval to ratify any amendment proposed by a convention.

International comparisons

While explicit constitutional recognition of a convention route is somewhat unusual, state constitutions’ regular amendment procedures in state constitutions are very comparable to those of many national constitutions around the world. Here are a few essentially exact parallels:

  • Minnesota, Rhode Island – Ireland (majority + referendum)
  • Texas, Maine – Japan, Romania (2/3 + referendum)
  • Oregon, Arkansas – Switzerland (majority + referendum, initiative option)
  • New York, Virginia – Denmark (majority + election + majority + referendum)

However, there are some clear differences, as well. Around the world, amendment procedures that do not require direct voter approval are far more common than among US states. In fact, most democracies today do not require a referendum at all. Meanwhile, many other democracies have a referendum as just one potential method of ratification – an alternative to a legislative supermajority (so not an absolute requirement), or required for some changes but not all.

Meanwhile, it’s internationally rare for referendums to be combined with an intervening election requirement (especially when further combined with supermajorities). I just gave the example of Denmark, but I think it’s the only one, at least for the main amendment procedure; Spain has a procedure for 2/3 legislative vote before and after an election, followed by a referendum, but this procedure is reserved for amendments to the chapters on fundamental rights and the Crown.

Fifty Shades of Republic | Part 2: presidential (gubernatorial) veto powers

This post is part of Fifty Shades of Republic, a series reviewing US political institutions at the state level

One of the most significant aspects of any presidential system is the extent of a president’s legislative power. Despite being known as the main example of a “separation of powers system”, few presidential systems really separate the classic “powers” (judicial, executive, and legislative) – instead, as Neustadt (1960) puts it in Presidential Power and the Modern Presidents, presidentialism typically creates a separation of offices which share powers.

The US constitution features a package veto subject to a two-thirds override. ‘Package’ means that when presented with a bill or resolution passed by Congress, the President can only agree to the proposal in full or veto it in full; by contrast, many presidents around the world, and governors of many US states, possess some version of an ‘amendatory’, ‘partial’, or ‘final offer’ veto. These give presidents more active power to intervene directly in the lawmaking process. In the case of a partial veto where the president can enact parts of an assembly bill directly into law (cutting through compromises agreed to in the legislature) are even arguably akin to decree powers in their scope for active lawmaking by the executive. By comparison, package vetoes are often described as a ‘reactive’ power.

In practice, however, the difference is not necessarily so clear-cut. As Moe & Howell (1999) argue, American presidents are able to engage in a great deal of unilateral policymaking (de facto lawmaking), and their ability to do so stems in large part from their supposedly ‘reactive’ veto power. Any actions (or rule-making) a president undertakes may be opposed by Congress, but any bill or resolution Congress passes to oppose presidential actions can be vetoed by the president. Given the two-thirds threshold, overrides are unlikely to be successful absent broad bipartisan support; the only remaining recourse is the courts[1]. Executive orders are often portrayed as much more constrained than the decree powers available to other presidents (which are sometimes also formally limited to what existing legislation allows, e.g. Russia), but we should always keep in mind that such a distinction is only as good as its enforcement.

Presidential veto powers in the states

All US governors (who, as directly-elected chief executives which a fixed term, are presidents) currently possess a legislative veto. North Carolina used to be an exception, as its governor had no formal say over legislation until its constitution was amended to provide for one in 1996. All states have a package veto, which is subject to a varying override threshold[2]:

  • 37 states require a two-thirds vote to override (in 27 it’s 2/3 of the total membership, in 10 2/3 of those voting);
  • 7 states require three-fifths (of the total membership in 6, of those voting in the remaining state);
  • 6 states require an absolute majority vote (always a majority of the total membership).

Where the state legislature is bicameral, override votes require the threshold to be reached in each house separately, with the exception of Alaska, where override votes are held in a joint sitting of both houses.

Many state constitutions offer the legislature some possibility of referring bills to a referendum. In at least one state (Oregon) it is fairly clear this means the legislature can bypass a governor’s veto by submitting a proposal to referendum (the standard override procedure in Iceland and de facto in Weimar Germany). However, most constitutions I have looked through are not clear on this point, as their language on ‘referendum and initiative’, present in about half the states’ constitutions, is usually ambiguous and contradictory (often in the same ways, as many states copied each other’s provisions in the early 20th century). If anyone knows more, please let me know in the comments!

Map of gubernatorial veto powers (click for full-size image)

Partial and ‘final offer’ veto power

The vast majority of states (44) grant their governors a ‘line-item veto’ – the power to enact some appropriations proposed in a bill while disallowing others (indicated on a map with a dollar sign). In 12 of those the governor can also reduce appropriations without deleting them entirely (indicated with a minus before the dollar sign). In all states, the override threshold is the same for line-item vetoes as for regular package vetoes – with the exception of Illinois, where an absolute majority is sufficient compared with the usual three-fifths.

As far as I am able to tell, no governor has the power to partially enact any other type of bill. Some states’ constitutions also clearly limit the power of deletion to items which appropriate money. In some states where the provisions are more ambiguous they have sometimes been interpreted more broadly. Infamously (and apparently uniquely), Wisconsin governors for many years successfully exploited the ambiguities in their state’s constitution to stretch their veto authority. Governors would edit bills by deleting sentences, words, digits and even letters to enact completely new legislation, with sometimes very different effects from those intended by legislature. This was mainly done to appropriations, but often also in other policy areas, since any bill with at least one appropriation item counted as an “appropriation bill”. Successive Wisconsin Supreme Courts affirmed governors’ interpretations of the rules, and the state’s partial veto was trimmed only gradually through two constitutional amendments in 1990 and 2008, and last year by a Supreme Court ruling seeming to reverse course on its previous judgements on the topic[3].

6 states add another feature to their gubernatorial veto, commonly called an “amendatory veto” in US sources (hence it is indicated on the map with an A), although it is best understood as the ability for a president to present the legislature with ‘final offer’. It means that the president can return to the bill to the assembly with some proposed amendments – the assembly can then choose to accept the bill as amended (by regular majority/majorities), in which case the amended bill becomes law, or insist on its original version of the bill, which requires whatever the usual majority or supermajority for (package) veto override is. This ‘final offer’ feature is very common across Latin America (e.g. Colombia, Mexico, Peru).

Effectively, any assembly can accept a president’s conditions for approval, even when only the package veto is allowed. The amendatory or ‘final offer’ feature formalises this possibility, but effectively also simplifies the procedure to do so, which might otherwise require restarting consideration of the bill and having to pass all three readings again before being presented to the governor. Besides anything else, therefore, this feature makes the veto procedure more forgiving of any errors in communication between president and assembly.

Veto power regarding resolutions and ‘legislative vetoes’

As I suggested above, a president’s power relative to the assembly depends in large part on the ability to veto not just bills, but also resolutions (which may not count as a “bill” or proposed legislation), specifically ones meant to counter executive action. It may also mean that whenever the legislature delegates authority to the executive branch or the bureaucracy by statute, the majority party’s ability to review executive or administrative decisions will be be limited by the president’s ability to veto such review[4].

As with the referendum, this was difficult to figure out precisely from constitutional texts. However, this much is clear: many (perhaps most) state constitutions, like the federal constitution, explicitly state that resolutions can be vetoed. Some states (e.g. Louisiana, North Carolina, Montana), on the other hand, clearly place legislative resolutions outside the reach of a gubernatorial veto. Idaho was one state constitution where I was able to find more detail about legislative delegation: the legislature can reject administrative rules, an action not subject to the governor’s veto.

Now, I almost certainly missed more examples and relevant provisions, and moreover I suspect the reality on the ground may be at least as affected by statutes and judicial interpretation as by constitutional provisions. Again, if anyone knows more I would love to hear about it in the comments!

The ‘weak governor belt’

This post is already fairly long, so I will conclude with just one observation: there is a sizable share of the states where the governor is comparatively weak, since they cannot join with a minority in the legislature to block a bill. In most of these states (indicated on the map in yellow), the governor may use the line item veto to undo a compromise she disagrees with on appropriations, but in other areas, the governor’s veto will mostly serve as a test of the legislative coalition which passed the bill. If the coalition was fragile, or if its initial success owed too much to abstentions, the override may fail, but there is no need to appease a minority for an override to succeed. In principle, legislative deadlock cannot result from divided government between president and assembly, only from things internal to the assembly or its parties.

Interestingly, these ‘weak governor states’ happen to all border each other; moreover, most of the states with a 3/5-override veto border these states as well. I haven’t really looked into the history of this geographic pattern, so I don’t know if it was the result of some manner of institutional diffusion – maybe some of our readers know more.

At any rate, what is clear is that there is meaningful variation on this point within the United States (variation which often seems to go unnoticed – multiple articles I’ve seen about a veto override in Kentucky or Arkansas failed to mention that the override takes just a majority in these states). Looking to these states’ experience with executive overreach and other consequences of the usual veto power should provide fruitful for research, and would be easier for reformers to emulate than any foreign example.


[1] Courts which are appointed by the president. Democratic and Republican presidents may disagree on many things, but they all favour increasing presidential power, which influences their choice of judges.

[2] At least two states have an override threshold of three-quarters in a few selected areas: in Alaska this applies to money bills; in Arizona this applies to some specific cases (e.g. amendments to enacted popular initiatives, emergency bills, raising taxes) where the minimum requirement is already a two-thirds majority.

[3] Although the ruling’s ambiguity (caused by the multiple separate opinions offered) and its potentially partisan motivation (it has been a fairly conservative court, and the incumbent governor is a Democrat) suggest it may not be the last word on the matter.

[4] Though this ultimately depends on what kind of review the constitution allows the legislature to write into statute (or the judiciary’s interpretation of the constitution on this point – which in the case of the US federal government has certainly been very narrow).

Fifty Shades of Republic | Part 1: State legislative terms of office

The US House of Representatives stands out internationally as the having the shortest term of any national legislative chamber, being the only chamber with all seats up for re-election every two years. With that electoral cycle, it also holds elections more often than almost any elected chamber (a few other chambers hold elections every two years, but with staggered elections for a longer overall term).

[Note: For context, you might want to read the brief introductory post to this series.]

The vast majority of states follow the federal level in electing their lower house for a two-year term. Many states, especially in the original 13 and in the Northeast, used to have annual election, but those all switched to two-year terms at various points during the 19th century. Of the six states lower houses with four-year terms, two (North Dakota and Nebraska’s unicameral) elect half the seats every two years.

By contrast, the federal Senate’s 6-year term is emulated by no state upper house[1]. Most states Senates have a four-year term, which is usually staggered, with half the seats up every two years. Seven states alternate to accommodate the redistricting cycle, with every decade seeing two four-year terms and one two-year term[2]. However, almost a quarter of states (twelve) have biannual terms for both houses.

What does this mean for any reforms? Well, maybe the main thing is simply that there is a lot of work to do! Biannual terms are almost universal, and I suspect they help lower turnout and accountability to voters (though I’m sure interest groups love them) in addition to lowering the government’s effectiveness and time horizons. Another observation is that the staggered elections that exist in about two-thirds of the states could make PR harder to implement in those houses than in other places, at least without disrupting the existing electoral cycle. When staggering puts half the districts up for election, those districts are not necessarily geographically connected, which is a practical necessity if single-seat districts are to be merged to form multi-seat districts which can support PR; if staggered elections mean half the seats in each district are up each time, this means the districts are already larger and harder to make the argument for making even larger. But maybe it’s not a bad idea to combine the move to PR with an abolition of staggering (as well as of bicameralism, as I will no doubt explore in a future post) – my hunch is that most voters probably find it confusing; politicians are probably more likely to feel attached to it, but then again politicians probably have much bigger issues with PR than this one…

What do you think? Do you have any thoughts about legislative term lengths in the states and what it might mean for reform? Or do you have any suggestions for future posts in this series? Please let me know in the comments!


[1] Although I was able to find a partial exception in Maryland’s Senate, which had five-year terms from 1776 until 1838, and then had 6-year terms with 1/3 elected every two until the term was shortened to 4 years in 1851.

[2] The fact that the other states don’t do this can often effectively mean district boundary changes leave a few voters without direct representation in the upper house for up to two years at a time!

Introducing: 50 Shades of Republic – a review of political institutions in US states

The topic of US state constitutions comes up on this blog from time to time. Naturally, they form an obvious comparison to the federal government. They share many similarities with the federal constitution, but also differ from it, and from each other in various ways. On the other hand, state institutions are also easily dismissed, as their design, in practice, varies relatively little from each other and is often mired in antiquated constitutional models. Moreover, given the extremely nationalised of American politics today, it’s not surprising reformers’ focus is more often than not on the federal level. Perhaps partly for these reasons, state institutions often appear not so well known, both by us comparativists and reform proponents, as well as on the part of journalists[1], whom we occasionally ridicule for their apparent ignorance and parochialism when they report about politics outside the US.

Here are two reasons why it is actually important to be more familiar state-level political institutions, specifically for reform-minded Americans. Firstly, institutional reform at any level (but especially at the highest level) in the US is only likely to gain traction once it is shown to work in the United States. So electoral and other reform at the state level should not be a secondary priority – instead, it will probably be crucial to get PR and other reforms on the agenda at the federal level. Secondly, and related to the first reason, some states have institutions that will make reform easier to pass – and these states should be the ones reform advocates should probably focus on. Moreover, some states already implement some political institutions worth emulating, yet these get scant attention either as proposals for reform in other states or in discussions for federal constitutional reform.

These are some of the reasons motivating my new series of posts on this blog: Fifty Shades of Republic – a review of political institutions in the states of the USA. Every few weeks, I will present a different dimension of state political institutions with a map which shows the distribution of different institutional variations across the states. Feel free to copy and distribute these maps – the more the better.

All seriousness aside, I think this will be a fun exercise, and I hope it will prompt some interesting discussions! Let me know in the comments what you think, and especially if there are any specific institutions you would like me to post on.

My first post in the series should be up momentarily…


[1] I’m thinking, for example, of an article I saw recently about a Kentucky bill which had been vetoed by the Governor and whose fate now rested with legislators. Its author wrote at length about various political factors affecting the legislature’s consideration of whether to pass the bill, and the various constituencies and factions involved, but nowhere was there any mention of the fact that in Kentucky, overriding the Governor’s veto takes merely an absolute majority of each house, compared with usual supermajority requirement.