07 March 2011

The Entrenched Politician-Voter Disconnect

It has been said that democracy is the worst form of government except all the others that have been tried.

- Sir Winston Churchill, British politician (1874 - 1965)

Churchill's wisdom has much to be said for it, except that it doesn't do much to tell us which kind of democracy is to be preferred to the alternatives.

An important flaw of the American system, and some similar systems, is that it deeply and fundamentally entrenches a process that naturally produces an unstable equilibrium that is at always odds with voter desires in one direction or the other, rather than a stable process that consistently produces elected representatives who seek to maximize the popularity of the decisions made by the government.

Enik Rising (Seth Masket's blog) reports research that he and Hans Noel have done with a clever experimental design that establishes empirically and rigorously the empirical wisdom that:

There are actually plenty of moderate Assembly districts in California; there are basically no moderate Assembly members. Virtually every Democrat in the Assembly is more liberal than her district; virtually every Republican member is more conservative than her district. . . .

We also find that members of the majority party tend to deviate further from their districts than members of the minority party do. Time out of office, we suggest, causes the minority party to try to moderate to win back the majority.

The study focused on California, but its conclusions almost surely hold for all but a couple of states in the United States (Nebraska and Louisiana may be exceptions).

Actual partisanship in legislative districts has a bell curve distribution. Elected legislators have a bimodal distribution, like a two humped camel. The political middle is systemically underpopulated by elected officials, despite the fact that the vast majority of voters are in the political middle between the typical elected Democrat and the typical elected Republican on the spectrum of ideology from the political left to the political right.

The study is focused, appropriately, on the facts, rather than the causes for those facts. But, political theory would suggest that this outcome is a very natural and direct consequence of our election laws and legislative process.

The Electoral Bias Against Moderation

Single member plurality district election system naturally gravitates towards having two dominant political parties in any one geographic area. In that system, adding a new credible candidate to the mix hurts that candidate's allies and helps that candidate's enemies. In order to mitigate mutually self-destructive behavior, we have entrenched, in the United States, a two party system in which the political left and the political right choose nominees in partisan primaries and then face off against each other.

The single member plurality district system doesn't necessarily have to create a two party system. It can support regional parties that have majority support in a particular area, like a Quebec Nationalist party or Irish Republican Party. It also doesn't require that the party of the right be the same everywhere, or that the party of the left be the same everywhere. Canadians have one party of the right to the West of Ontario, and another one, the Tory's to the East. One U.S. States have a Democrat-Farm-Labor Party rather than a Democratic party. The U.S. historically used to have, de facto, Dixiecrats in some parts of the country, and Democrats in other partys of the country, both against a common Republican party.

But, for distinct regional parties to emerge, something has to give that region a strong identity. The more homogeneous a region is politically, the less prone it is to develop a distinct political party.

A Majority Requirement As a Weak Fix

A weak solution to this problem is to use French style elections, which have been adopted in Louisiana, Denver and in a less pure form, in a handful of states, where the winner of the election needs to get a majority of the votes cast to win in the first round, with the top two vote getters facing off in a second round.

While this isn't immune to the pathology that similar candidates tend to undermine each other, it eliminates the requirement that a candidate wanting to make it into the second round must have a plurality coalition made up entirely of partisan voters on the right, or partisan voters on the left. A coalition of unaffiliated and bipartisan moderate voters can also secure a candidate a second round slot.

More generally, in its pure form, it is more more neutral towards the number of political parties competing in the district than a traditional two party system. Not surprisingly, Louisiana has historically been the source of moderate candidates in both the Democratic and Republican political parties.

Proportional Representation As A Strong Fix

A stronger solution to this problem would be a true proportional representation system that doesn't penalize two political parties for having similar views without not actually merging. In a system where each voter picks on political party and that party gets a number of seats proportional to the share of the vote received by that political party, the sensible thing for a voter to do is the pick the party most closely aligned with their views and a political party doesn't face a penalty for being small unless its market share approaches the treshold minimum percentage necessary to secure representation in the legislature (typically in the low single digit percentages). A proportional representation system also allows representation of parties that have a majority in no particular district.

This kind of true multi-party system, which naturally gravitated towards roughly four to eight major political parties, leads to unequal sized clusters of voters with much more homogeneous views.

The Legislative Bias Against Moderation

When Are Coalitions Made? How Stable Must They Be?

In an idealized two party system, coalitions need to be assembled before the election. In an idealized multi-party system, coalitions are made after the election.

In a parliamentary system, a governing coalition forming a legislative majority, need to hold together on all major issues for the political system to function. The legislative leader becomes the head of government (i.e. the Prime Minister) who runs government on a day to day basis. If the coalition that put the Prime Minister in place fails to secure majority support for any major proposal, the Prime Minister will typicall then lose a "no confidence" vote in short order, and new elections will typically be held.

Somewhat complicating the matter of coalition building is that American legislative coalitions don't need to be very stable across all or most issues. Since we have a strong President whose tenure doesn't hinge upon maintaining a legislative majority (indeed our current President doesn't have one), the executive branch can continue to function and carry out the business of government even when one coalition of legislators may prevail on one set on issues, and another coalition of legislators may prevail on a different set of issues.

Historically, for example, the U.S. had a de facto "two and a half" party system at the national Congressional level. One domestic economic issues Democrats generally united against Republicans. But, on matters of national defense and social issues like race, Southern Democrats often allied themselves with Republicans against Northern Democrats. The Democratic Presidential nomination race reflected the vacillation between these two factions.

This three party system allowed the party that was moderate on any issue where there was no consensus among the three parties to prevail, moderating the process as a whole to some extent.

Today, after a process called "realignment" that has largely run its course, this isn't the case any more. There are few notable blocks of "moderates" in either the Republican Party or the Democratic Party in Congress who deviate from their party in a systematic way on a particular way. The Blue Dog faction in the Democratic Party has withered, as has the faction of Northeastern social moderate Republicans. We have a relatively pure party of the political left and a relatively pure party of the political right, and they are quite evenly matched on the national political scene.

The Majoritarian Legislative Process Has A Hidden Bias Against Moderation

Of course, some of the partisanship we see at the legislative level is a product not just of a two party electoral system, but is inherent in the legislative process itself. Legislative decision making generally involved for or against voting on politically controversial issues. Partisans get the results most to their liking when they favor proposals the secure majorities, but only the narrowest majorities possible. It is natural, in this context, for legislators to informally rank themselves from the political left to the political right, and for proposals that require legislators to form an allegiance with one side or the other to be common.

Thus, even in multi-party proportional representation system, governing coalitions that are clearly right leaning, or clearly left leaning are the norm. Governing coalitions like the Penteparti system of Italy, in which five mainstream parties formed repeated unstable and short lived coalitions in the political center in a calculated effort to exclude old school communists on the left, and old school fascists on the right, ar the exception, except in times of national peril from an outside force like the unity governments of the United Kingdom during World War II.

Moderates are more favored in a place like the U.S. Senate, that is not strictly majoritarian, with most proposals from the left or the right requiring supermajority support, than in the U.S. House, where majority rule prevails.

In order to favor the kind of moderate policies that voters in the political center favor, you need not only an electoral system that doesn't naturally generate a bimodal distribution of legislators, but also a legislative process that has stronger incentives to find supermajority, rather than mere majority support for legislation.

Supermajority Requirements Come At The Risk of Deadlock

Favoring supermajorities, however, comes at a cost. Any requirement that there be more than majority support for a measure in a single legislative body with ultimate say over the matter creates a risk that the system will be overconstrained and fail to produce action in situations where there is no acceptable status quo.

In the United States, recess appointment power and a strong Presidency prevent the status quo of all important positions in the government going unfilled for lack of a resolution to deadlocks between a nominating executive and a ratifying Senate. But, at both the state and at the federal level, it remains possible for a lack of an ability to reach consensus in a deterministic way on a budget bill to produce a government shutdown, something that House Republicans on one side, and Senate Democrats and the President on the other, have been dancing with in the lame duck session of the 2010 Congress and the current Congress.

Probably the most notable recent example of an overconstrained set of legislative rules in recent memory is that of the Iraqi Constitution, which requires supermajority support to choose a President and Prime Minister, at a time when there is stark disagreement between well defined factions that individually lack that supermajority (whose moderate unanimity the constitution attempted to coax). The result has been many months of government deadlock due to a lack of the required consensus. Afghanistan's constitution, notably, in contrast, was carefully crafted to allow the President to impose a budget and appointees in the absence of constructive disagreement by a legislative majority on an alternative (despite the fact that Afghanistan has less well defined factions in any case).

Supermajority Processes In Practice

The best examples of supermajority incentives in the legislative process are probably the adminsitrative law notice and comment process, and the Quaker meeting process, each of which hinges on having a "honest third party" receive input and try to formulate from the input a maximally popular (in the case of administrative rule making) or maximal consensus (in the case of Quaker meeting decision making) stance, by taking into account the details of the concerns expressed.

Monarchs, because they are well positioned to be that honest third party, can, if they are not politically tone deaf, secure support in the fact of partisan elected bodies, precisely by favoring the middle against partisan majorities, and favoring solutions that would secure supermajority support over those that would secure mere majority support.

I think that the Founders probably hoped (largely in vain) for that kind of dynamic to play out in the American political system, something that failed largely because the Presidential election system got locked into the two party dichotomy. It isn't implausible to think that if the Presidential election process could be redesigned to decouple it from partisan politics and instead have this office elected in a way that favored moderates, that a similar dynamic could be established, even without reform at the legislative level, through the veto power.

Indeed, I think it is a fair hypothesis that Governor Ritter's experience with consensus oriented politics as the norm from his days as a special interest advocate for Colorado's District Attorneys' Association may help explain a great deal of his approach to dealing with the Colorado General Assembly, in which he frequently cited process rather than substance oriented objections to legislation in veto messages.

Good legislators in the existing legislative process frequently try to craft such agreements in "smoke filled room" negotiations involving interested parties with a goal of reaching "kumbaya" (i.e. a consensus of the interested parties to support a bill). But, this kind of deal making and mutal accomodation reaching is ill suited to the formal legislative process.

Consensus or supermajority oriented politics are actually much more common than most people would think, indeed, it is the norm. This is because partisan caucuses and special interest groups much each reach consensus or something close to it in order to move boldly on legislative proposals. They tend to stay silent and refrain from acting in the absence of consensus.


In the absence of divided government (something that both the U.S. Congress and Colorado General Assembly are experiencing right now), there is little incentive to stretch beyond a majority supported compromise.

So, in American politics, voters are left between insisting that government function with deals negotiated between the left and right in exchange for a risk a damaging deadlock, within divided government, and government by a right of center, or left of center consensus with little to encourage it to be inclusive.


Dale Sheldon-Hess said...

What do you think of, as an "intermediate"-strength fix, consensus-seeking election methods, such as approval voting? Because, as you mention, majority requirements have been tried, and have had little effect.

I understand that there's a group in Colorado trying to work toward approval voting now.

Andrew Oh-Willeke said...

I think majority vote requirements would make third party candidates much more viable, and a third party Governor can greatly change the debate overall. Another option would be to turn the upper houses whose distinct role was obliterated after Baker v. Carr (one man, one vote), into proportional representation bodies.

I don't think approval voting is very helpful, because it relies too much on the good faith of those agreeing to second best options.

In the legislative process, a more "dictatorial" role for the bill sponsor in the committee process, with the sponsor allowed to amend the bill as desired up until a final committee vote, rather than using a Robert's Rule of Order process, might be helpful.

mw said...

Well - I don't know where the first half of this comment is, hopefully in your moderation cue or spam folder. This is the conclusion:

There is a simple fix which you point to in your conclusion, that can be enabled by even a small subset of the true independent voters. Instead of a dilemma, I'll restate your conclusion as an assertion:

Independent voters have the power to insist that government function with deals negotiated between the left and right by always voting for divided government. Every time. Without exception.

Divided government is not a bug. It's a feature.

Even if only 5-6% vote this way in federal elections, it is enough in our partisan polarized electorate to determine the outcome of federal elections. It is an organizing principle that is kind of like a 3rd party, but instead of functioning as a spoiler you actually get the policy moderation desired by those voters.

The beauty of this idea is that it requires no candidates, no leaders, no platform, no conventions, no party, no election rule changes, no constitutional changes, and no contributions. Just communication and agreement among a small subset of the independent electorate to first vote for divided government before all other considerations.

This is a tactic, not a strategy. But it can be implemented with no major effort in every election. It does not preclude any of the grand strategy, "centrist", 3rd party, electoral reform, constitutional amendments, etc. etc. initiatives that Indies so love. Hey - knock yourself out. Just - some of you need to vote for divided government first.

And Andrew - Welcome to the United Coalition of the Divided.

Dale Sheldon-Hess said...

Glancing over DW-NOMINATE rankings, it doesn't appear to me that Louisiana's congressional delegation is any less (or any more) partisan than any other state. On what do you base your claim to their moderateness?

And, if there were three equally-competent (but ideologically diverse) candidates in an election, you don't think that the extremes would agree "in good faith" that the moderate choice was second best? Wouldn't approval voting be the moderate's best chance of success?

DLW said...

What if we tended to use more multi-seated 3 (or 5)-seated elections for "more local" elections and then used two (or three) stage elections for "less local" elections? As you note w. Baker v. Carr, this was very common earlier in our country's history. I think that if third parties can win some seats then they can give moderates more viable exit threats from the two big parties, which would then force them both not to deviate from the de facto center too much.

The use of small multi-seated elections in state assembly (and municipal city council) elections would support the existence of two major parties per state, with the two parties not necessarily being the same in each state. And if the electoral college were revamped to be the second stage of a two stage election then we could trust that we'd elect (dynamically) centrist presidents in far less negative elections.

I personally believe that if we used multi-seated state assembly elections that we could entrust our assembly persons to elect our US senators again, possible with an approval or range voting system, which I think work best with more-informed voters whose votes are public information (not unlike as Range-Voting is used in Olympics contests).

Thanks for your post, it's not often I read such a well thought out and researched posts on this topic.


DLW said...

The "top two primary" would be scarily easy to "fix" so as to be more reliable in producing more centrist winners. This was the motivation behind my idea for a "top three IRV". It takes into account the common criticisms made of "top 2 IRV", while recognizing that winner-take-all elections inevitably do tend to winnow down the number of feasible candidates/parties...


Andrew Oh-Willeke said...

"This is a tactic, not a strategy. But it can be implemented with no major effort in every election."

Coordinating independent voters who are systemically less interested in politics to start with is generally infeasible. Without coordination, people will split tickets in opposite ways and cancel themselves out.

Even if coordinated, the systemic problem with divided government is that failure to agree on must pass matters, like budgets and basic executive branch appointments, causes government shutdown, which is typically not what anyone would have as a preferred outcome. Iraq's long period of deadlock is an example of this problem.

Andrew Oh-Willeke said...

"What if we tended to use more multi-seated 3 (or 5)-seated elections for "more local" elections and then used two (or three) stage elections for "less local" elections?"

Unless people get less than one vote per available seat or there is some kind of proportional representation, this kind of election, when conducted on a partisan basis, has a strong tendency to be winner take all for the entire partisan slate. The effect then, is almost equivalent to simply making the size of the legislative body smaller. It doesn't have a strong moderating effect.

DLW said...

They could commit to vote quasi-strategically together to increase their chances of being the swing voters.

The problems of Iraq illustrate the need to use both winner-doesn't-take-all and winner-take-all elections. The former could be done with relatively few contested seats to reduce the proliferation of parties or to make them local parties that advocate for localish issues to limited effect.


DLW said...

AOW:Unless people get less than one vote per available seat or there is some kind of proportional representation...

DLW: the Least Remainder (LR) Hare system can be used with 2 or 3 seats so that it works almost exactly like 1-seated LR Hare or First-Past-the-Post.

O.w., you are correct.

Andrew Oh-Willeke said...

"it doesn't appear to me that Louisiana's congressional delegation is any less (or any more) partisan than any other state. On what do you base your claim to their moderateness?"

See e.g., Charlie Melancon (LA-3), Blue Dog Co-Chair for Communications (ran for Senate in 2010, lost); Billy Tauzin (LA-3) (became a Republican in 1995); J. Bennett Johnston (D); John B. Breaux (D) (conservative Democrat); John Bennett Johnston, Jr. (conservative Democrat); Mary Landrieu (political moderate). Review of runoff elections will show more extreme candidates routinely losing in the first round.

John Bennett Johnston, Jr.

Andrew Oh-Willeke said...

"They could commit to vote quasi-strategically together to increase their chances of being the swing voters."

Voters are utterly incapable of acting like this in practice, even if highly disciplined.

"The problems of Iraq illustrate the need to use both winner-doesn't-take-all and winner-take-all elections. The former could be done with relatively few contested seats to reduce the proliferation of parties or to make them local parties that advocate for localish issues to limited effect."

Iraq's problem is that supermajorities are required to do anything of consequence. In a two party system with reasonable party discipline you get the same supermajority effect by requiring bipartisan agreement.

Andrew Oh-Willeke said...

"the Least Remainder (LR) Hare system" is a prototypical proportional representation system.

DLW said...

utterly incapable so far...

A group can vote strategically, yes? Then why not set up more than two possible settings for how they'd vote based on how well the candidate(s) take on their issue(s)?

Our democracy is evolving as our habits evolve. If it is inevitable that we have relatively few major parties then we can expect that some folks are going to have to vote strategically. This is a way they could commit themselves to leverage their vote for more influence.

LR Hare is a prototypical PR, but with 2 or 3 contested seats it is similar to our typical election rule with one candidate per party and one vote per voter. Thus, one can have the benefits of a closed ordered list form of PR without the lists, since the list would be tacit.


DLW said...

I don't think one can deal with might-bes and avoid a certain amount of advocacy...for mere observation inevitably changes things.


Andrew Oh-Willeke said...

Ah, quantum physics meets political science. Asmiov predicted it first.