Local Constitutional Design

This is the latest post in my constitutional design series. Up until now I’ve discussed constitutional design primarily at the national level. In this piece, I want to discuss how much of what I’ve described holds on the local level.

Local Politics

First I want to talk about the different dynamics that hold on the local level as compared to national politics.

Accountability > Consensus

The main difference is that in my view accountability democracy is much more applicable on the local level and consensus democracy much less so.

This is not a law of nature, but politics on the local level is usually less partisan. Because everyone’s views are less strong and concrete, it is less important that their preferred policies are exactly adopted. And since voters have more of an open mind, they are then able to assess a politician’s results more objectively.

Accountability democracy is also better on the local level because in my view the upsides of local policy are higher and the downsides lower. This is for two reasons. First, people can easily move between localities, so if a mayor really ruins a city people can just move away. Of course that’s easier said than done - it’s especially difficult for lower income people - but it’s a real effect. The second reason is that of laboratories of democracy. When a locality adopts good policy with positive results, other localities can learn from it and implement those same policies. When a locality adopts bad policy, other places can also learn from it as what not to do.

Lastly, accountability democracy is better on the local level because the risk of democratic decline is much lower. There just don’t tend to be nearly as many cults of personality on the local level that lead to authoritarianism. And if a region ever does become undemocratic, the national government can always step in and override them, as the national government in the US did with Jim Crow laws. (Granted that took a long time).

Parties

In most places, the parties that dominate on the regional level are the same ones that dominate on the national level. In some of these countries, the party that dominates in an area in national elections is also the one that almost always wins local elections. This is the case in the current day US and in the UK. In some countries they’re almost entirely uncorrelated, like in the US 20 years ago or in Canada. Most countries are somewhere in between.

Game theory says that there should be very low correlation, because in a left-leaning area the right-wing party would moderate to the point where they started winning elections. Either that or a different more moderate party would take over as opposition (like happened in the Saskatchewan region of Canada).

I think there are a few potential reasons this doesn’t hold sometimes. One is that votes in local elections are taken as a proxy vote for national elections. This may be because voters don’t want to look too closely into a candidate and so vote the same way they would nationally. Or it may be because local election results have real or perceived symbolic importance; when Nigel Farage started doing well in local elections for example, that was used as a sign by national Conservative politicians that Brexit had a real constituency.

Another potential reason is that strategically moderating is a difficult thing to do, so voters correctly see the candidates as largely holding the same stances as their national politicians. Democrats running in red areas used to hold very different views than the Democrats running nationally, but now because of the general polarization of politics this is much less true.

Lastly, it also seems possible that the governing party in many regions is actually closely aligned with the median voter in those areas. In that case, voters may never have any reason to vote for the opposition party, almost no matter how much they moderate. In the US, when state parties try to pursue more radical actions, they are in fact sometimes punished by voters, even in deep blue or red states. This suggests maybe more moderate governors are actually acting with restraint tolerated by the median voter.

The upshot of all of this is that you have to be intentional if you want local democracy to be about local issues and accurately represent the public on these issues.

Local Leader

On the national level, I am in the parliamentary democracy camp, read about that here. On the local level, I could go either way.

The big reason to have the leader chosen directly instead of indirectly (through legislative elections) is that often your desired winner - for example, the Condorcet winner (read here for a description of what that is) - is from one of the major parties but is not the most popular within that party. I think choosing a Condorcet winner is more important on the local level than nationally because the alternative is often letting one party take power forever rather than the alternation that takes place on the national level.

A Condorcet winner could in theory run independently on their own ticket in a parliamentary democracy and become the swing vote in the legislature, as they would nationally. As stated earlier, however, the parties on the local level tend to be the same ones present nationally, and the one dominant on the national level in the region tends to be the one dominating locally, which suggests this doesn’t happen much. Whatever the reason for this, if you want to elect that person leader you’re probably going to need to adopt a single-member electoral system that selects for moderate winners.

Powers

On the local level, I think you should give the leader much more power to push through their agenda in order to facilitate accountability democracy. This is especially useful on the local level because there is a lot less public awareness, so the mayor of a city may be the only figure that anyone really knows.

To this end, I think if you do implement presidential democracy you should give the leader some number of votes in the legislature, maybe 20% of the size of the legislature. If the opposition party wins the leader spot, as long as they have at least 40% support in the legislature they get a majority. If the leader is from the majority party, this allows them to face less intra-party opposition, although they can still be stopped if the opposition party and a decent chunk of their own party opposes them.

I actually think this is a good mechanism in parliamentary democracy as well. The way this would work is that after an election, the legislature chooses a leader by majority vote, and the leader then gets some number of votes going forward, to be used even in votes of no confidence against them (except after new elections where you want the legislature to be able to replace the previous leader by majority vote). Since the opposition party will never win the leader spot, this solely serves the purpose of letting the leader overcome intra-coalition opposition. I also think this is a good mechanism for discouraging the norm that the largest party in the legislature gets the leader spot instead of the median party ideologically, since it would often actually give them a majority. I think this is a bad norm on the local level for the same reason I think you should consider parliamentary democracy: the alternative to choosing the median/Condorcet winner is often just one party (the plurality party, the one dominant nationally) being in power forever.

The one issue is that you need some mechanism to ensure the legislature actually picks a leader after an election and doesn’t just keep the interim leader. The normal way this is done is through snap elections, but if you were to do this on the local election they would not coincide with the national election, which would lead to a an electorate dominated by very engaged voters. I would just not have snap elections on the local level and have local elections correspond to regularly scheduled national elections¹.

I would give the leader a lot of executive. One option is to allow them to fire and appoint any members at will. Alternatively, you can force the legislature to approve all appointments and firings, but then have the multiple votes count thing count toward these nominations as well.

Electoral Systems and Party Candidate Selection for Local Leader

For these next few sections, I would recommend reading my posts on Single-Member Electoral Systems, Multi-Member Electoral Systems and Party Candidate Selection for descriptions of systems and discussion on the national level, as I’ll reference them a lot.

If you have parliamentary democracy, most of what I discussed in my piece on party candidate selection holds. If things are really polarized on national lines, you might consider holding direct primaries along with the general election as I talked about in that piece (people vote for a party and alongside vote for the leader), so as to minimize the importance of party candidate selection which gives engaged voters more power.

If you have presidential democracy, I wouldn’t bother with first-past-the-post (FPTP) or ranked choice voting (RCV). As I described in my post on electoral systems, the point of these systems is to produce polarized two-party systems, just use parliamentary democracy with a two-party legislative electoral system if that’s what you want.

Instead, I would adopt score voting or a Condorcet ranked method like Minimax. The one concern I have is that on the really local level (like medium-to-small towns) Minimax may just be too complicated to implement. The other thing to note is that because there is much less polling on the local level, score voting is much less likely to produce the Condorcet winner and somewhat more likely to choose the utilitarian winner.

Whichever of these you choose, I would have a top 4 system and allow people to run as any party label they want (with maybe a restriction that they be members of that party), instead of having parties nominate a single person. The whole point here is to reduce the importance of party candidate selection so that figures who are not the most popular in their party can win the leader spot.

Legislative Electoral Systems

On the local or regional level, I would not adopt direct multi-member systems like single transferable votes (STV). Having STV on every single level of government is just overkill, voters are going to have way too many options and will likely not know what to do with them, leading to potentially poor rankings. It is also potentially too complicated to implement on the local level. The only such system I think is potentially ok is my proportional multi-vote representatives (PMR), since it is quite simple to use. If you do go with that system I would use districts of size ~5 and allow anyone to run as any party label they want.

If you want to give voters more control, I think the best system to do this is open list PR: by default the party gets to choose, but if voters feel particularly strongly about a single person they can have a say. Alternatively, you can go with mixed member proportional (MMP), which gives voters a bit more say in some ways and a bit less in others.

Closed list PR is also ok, which in polarized environments means relying on candidate selection (for leader and legislature) to ensure legislators’ views on local issues represent voters’ well.

As in national elections, the choice of whether to adopt thresholds and plurality bonuses depends on if you want a two-/few-party system or a many-party system. If you have local politics really polarized along national lines, I would have low/no thresholds and no plurality bonuses, as they’re usually just going to lead to the dominant party in an area holding a majority forever.

Legislative Party Candidate Selection

On regional levels, you can probably do what I described in my piece on national party candidate selection.

On really local levels however, I think you have two problems. The first is that this system I set out to maintain contiguous sublist proportionality is probably too complicated to implement. The second is that electing special delegates to choose local candidates is kind of overkill; it’s hard enough to find good candidates, electing people who are willing to participate in candidate selection for a small number of candidates seems hard and pointless.

One option is just to have local elites select the candidates. The local elites in this case would either be the party leader or some local party non-legislative leadership selected by members. I don’t love giving local elites so much power, but it’s an option. You can also let elites select only some of the candidates as a compromise between elite and public selection.

The other option is to have voting by party members. If you have single-member districts like in mixed member proportional (MMP), you can have primaries in each district using one of the single member methods. Funnily enough, I might actually prefer first-past-the-post (FPTP) in this case just to maximize the randomness in candidates selected to try to counter the tendency of single member districts to let the largest faction dominate.

If you have party lists, the only way of selecting them by party members is to have local party meetings. One other advantage of these meetings is that aspiring candidates can give speeches and people can discuss, so you might even want to do this for single-member districts.

The contiguous sublist proportionality stuff I talked about in my party candidate selection piece is probably too complicated to implement at these meetings. You may just have to sacrifice proportionality and adopt the “sequential voting” system often used in practice: everyone in the room votes for the top spot on the list using FPTP, then for the second spot on the list, then the third spot, etc. I do have one more idea, however, on how to modify this system to make things a little bit more proportional. My idea is to have some candidates chosen the same way by FPTP, but then some candidates chosen through a top three system where everyone vote for one person and the three candidates with the most votes get a spot (in order of how many votes they got). This system is meant to balance FPTP, which lets the largest faction dominate, with this top three system that has the opposite effect of splintering the field even when one candidate may have many more votes than the others². Maybe you could alternate back and forth between these two methods to give you more of that contiguous sublist proportionality. I think this is a simple enough system that it can actually be implemented even at the local level.

Term Limits

Because of lower risk of democratic decline, I am opposed to term limits even for leaders on the local level. Same goes for party leader.

Funnily enough I think the case for term limits for legislators on the local level is stronger than that for leaders. On the local level, public knowledge of people's legislators is much lower, which can lead to much higher incumbency advantage. If this is something you're really concerned about, you might be attracted to legislative term limits. I'm still opposed to them, but I see the logic. I wouldn't implement them on a really local level though due to a lack of good candidates: once you find a good legislator, you really want to keep them around.

Direct Democracy

If you think local elections are insufficiently about local issues, or that local elections are insufficiently democratic because of low public knowledge, you might want to adopt elements of direct democracy so people can directly weigh in on local issues. Direct democracy might also be attractive on the local level because of the lack of polling.

On the other hand, direct democracy undermines accountability because it doesn’t allow those in power to push through their preferred policies so they can be evaluated for them. Direct democracy might also be worse on the local level because the unrepresentativeness of those who partcipate can be more severe, particularly for things like referenda held not coinciding with national or local elections.

Other than these factors, most of what I talked about in my piece on direct democracy still holds.

1. And potentially have national snap elections in between these regularly scheduled ones

2. Note that if the most popular candidate has way more votes than the others some voters may strategically choose someone else to get them in the top three. This is why those fancy electoral systems exist, to get around this.