Mixed-member systems and regionalism

Miguel asks a good question about the extent to which mixed-member systems might encourage regionalism and thus discourage nationwide coalition-building. He was responding to an earlier (and since revised) planting of mine on Mexico, and offering lessons from the Bolivian experience. Regarding Bolivia, Miguel suggested that in 1997-2002 there appeared to be an emergence of separate regional party systems, similar to what appears to have happened in Mexico. He says:

I actually suspect this is a byproduct of MMP systems, since it encourages the localization of candidates (to win SSD seats) — which is part of my skepticism for MMP in new democracies, if it encourages regionalist splits & discourages nation-wide coalition-building.

First off, Mexico does not have MMP. But we might expect that MMM would have more localizing tendencies than MMP (because the districts actually matter to the overall outcome in terms of the balance of seats held by parties). Such tendencies could generate party systems that are distinct in different regions, possibly with two major parties in one region that are not the same two parties that are most competitive in some other region–national multipartism, regional bipartism.

Does MMP (or MMM) discourage nationwide coalition building and emphasize regionalism? As always in comparison of electoral systems, one needs to ask, relative to what? If the allocation of PR seats to parties is national (as it is in Mexico, though not Bolivia) then it obviously encourages parties to think nationally to a degree that a system of just single-seat districts would not. In Bolivia, it is possible that the adoption of MMP increased both personalism and localism because the antecedent electoral system was pure closed-list PR. In fact, on the interparty dimension–how seats are awarded to parties–Bolivia changed little. The old system was PR with each province as a self-contained multi-seat district. The new one is also PR, but of the MMP variety. There is no national allocation.

What there is, with the change, is a series of individual races by plurality. With MMP, that means a change primarily on the intraparty dimension, in that some legislators are being elected on their own “nominal” votes instead of by party ranking. As a result parties may care about the qualities of the candidates being nominated to a greater degree than on the lists, and those candidates might seek to develop personal connections to the districts.

So, should we expect MMP in Bolivia to be more localizing and personalizing than what went before? Probably, and because there is no provision for overhang seats (the addition of seats to compensate for some party’s getting more seats in the SSD competition than its proportional share entitles it to), there should be more premium on parties’ putting effort into the SSDs than if winning these had no impact on the partisan balance of the legislature. In other words, the MMP system provides some new incentives on both the interparty and intraparty dimensions for parties and candidates to exert effort in winning local races. Given that some parties may have an edge over others in any given geographic region either programmatically or in terms of the types of candidates they can recruit, then this should promote more “localization (to win SSD seats)” than the system that went before it.

As for the expectation that I allude to above that MMM would generate greater emphasis than MMP on the single-seat districts (and hence localization and at least local bi-polarization) the only problem with that is that while the rank order of the parties differs across Mexican states, in only a minority of the states is the third party far behind. So, Mexico would seem not to support the hypothesis that MMM (or mixed-member systems more generally) promote local 2-party politics.* The continued strength of the third party (whichever it may be) in districts that it has no chance to win, even under MMM, is partly a result of the single vote. That is, every party that hopes to win PR seats has a strong incentive to nominate 300 candidates, whether or not viable. Even with this consideration in mind, however, we have to ask what the mixed-member system adopted in Mexico is being compared to. Does it promote localization or nationalization, relative to what?

For Mexico, where at one time the system was all SSDs (and all dominated by one party), the addition of a nationwide PR allocation clearly has helped nationalize politics to a degree that could hardly have happened under a pure SSD system.

In assessing the impact of adopting MMP or MMM (or any other electoral system), we have to ask what variables are changing relative to the former system. And the different starting points, as much as the differences in the systems adopted, affect our expectations about whether mixed-member rules would increase or decrease the extent of regional electoral competition in any given polity.

* I realize that Miguel is responding to the original post, where I employed the “slip of the keyboard” and referred to regional two-party systems, when what I meant was simply that the second party is the PRI almost everywhere and thus local competition is usually PRD vs. PRI or PAN vs. PRI, but rarely what it was in the presidential election: PAN vs. PRD.

0 thoughts on “Mixed-member systems and regionalism

  1. Thanks for the correction on Mexico’s MMM (also a slip of my keyboard). And I appreciate the thoughts on this question. So far, I think I agree.

    I would add, however, that in Bolivia, the MMP races did more than simple promote local bi-partisan races. In many SSDs, previously minor parties emerged as single/dominant party systems.

    As for comparison “to what” is a tricky question, since Bolivia’s constitutional structure is so unique.

    In my mind, even though Bolivian seats are awarded on a per-department (provinces are subdivisions w/in departments) basis, there was a strong incentive in the constitutional structure for a national party system. This is because w/ no presidential candidate winning a majority, parliament elected the president. So parties had incentives to be concilliatory, since their campaign opponents might be potential coalition partners.

    But MMP altered that, since anti-systemic parties were able to capture seats in SSDs w/o any PR seats. IU (which would later become MAS) won 4 critical seats by SSD districts in 1997 & set the model. To win those seats, such parties radicalized their discourse. In the end, in some SSDs (rural Altiplano, the Cochabamba valleys) the more radical, anti-systemic rhetoric ensured victories. This meant that as anti-systemic parties were winning SSD seats (primarily in the Andes), which also increased their presidential ballot votes (the coat tail effect seemed to work backwards in 1997-2002), the traditional parties (MNR, ADN, MIR) could only rely on votes in the lowlands. And that left a tough choice: To compete for Andean votes might risk lowland votes, to solidify their lowland voters would further alienate Andean voters.

    Since 2003, the country is now clearly polarized along regional identity issues, more than ideological or policy issues.


  2. Miguel, that is interesting. Of course, SSDs often promote one-party-dominant regions. There is no necessary reason why there “has” to be two-party competition for SSDs. (Duverger gave us a theory, not a description of reality!) And the more homogenous a local area (as often is the case of rural districts), the more likely it is to become a “safe” district. Additionally, there is no reason why a district has to be safe for a major national party. It can be a regional party. SSD systems are full of such parties: Scottish Nationalists, Bloc Quebecois, etc., many of which probably would be far less significant under a PR system.

    Good points on the nationalizing and conciliatory effects of the Bolivian presidential-election process, at least when combined with pure PR (even with department-level allocation).


  3. The other major institutional change that predates regionalism in Bolivia was a massive decentralization. The result has been a predom ination of local-oriented parties contesting only municipal elections (which was made even more common by the recent reforms allowing non-parties to seek local government office as well). The result, to me at least, seems to mirror Chhibber and Kollman’s basic logic: candidates and parties became oriented around local office ands the resources it provided for building bases of support while upstart parties could use local government to gain a foothold and legitimacy and a base that was indepdent of the party. This provides the necessary tools to form indepdent parties in pursuit of the SMD seats while shifting the logic of competition away from national alignments toward local ones. Recent studies on career paths in Mexico suggests that a similar dynamic may be at work in Mexico.


  4. Actually, from my yearl-long frield research into this (the central topic in my dissertation) … the change to SSD had an enourmous impact on Bolivian political parties. Of course, so did the introduction of local, municipal elections. Both had the predictable result of tying parties & political elites more to local races/issues, rather than national ones.

    More importantly, in a country w/ a long tradition of populism & w/o institutionalized political parties (only the MNR comes close to being an “institutionalized” rather than a personalistic party), the local municipal elections created a new group of local political elites w/ resources (municipal funds, local constituencies, etc) they could mobilize in campaigns.

    What’s interesting, is that SSD delegates had little influence in the party — which is opposite of what one would expect. In part, the large parties in 1997 focused more on the presidential list candidates, and put up “fluff” candidates for the SSD seats. There was also very little media coverage of those races. And since even the SSD seats are held by parties, not individuals, those representatives also had to respond to their party. Plus, w/ the exception of “safe” districts (only a handful), most SSD seats were won w/ about 20% pluralities (!!), so parties didn’t have to worry about their cosntituents, just so long as they could generate enough votes to beat their opponents. The only parties that took advantage of SSD seats, were small anti-systemic parties.

    In 2002, the major parties paid more attention to SSD seats (as did the media), but again the candidates were selected by the parties, not local people. There were even many accusations that some candidates didn’t live in their disctricts. Studies by Bolivian think tanks showed that there was little if any difference iin how SSD or plurinominal deputies behaved in parliament. Again, since parties picked candidates for SSD seats, representatives had to ingratiate themselves w/ the party, not constituents. Still, the parties did pick local notables (often from municipal politics) who could win locally.

    So it’s a complicated story. It was parties — not candidates — that began to alter strategies into regionalistic campaigns. And they did so in order to win seats in “safe” regions — not necessarily districts. That is, because of the reverse coat-tail effect, parties began nominating local notables in SSD districts in order to push up presidential votes in those regions to gain plurinominal & senate seats. And as the national vote for parties spread thinner & thinner at the national level (from about 4.3 effective number of parties to about 5.5), in many regions it actually shrunk into two-party contests. The winner was the MNR, which w/ barely 22% of the vote in 2002 was able to win a third of the seats (and almost a majority of the senate).

    So long as the traditional parties (MNR, ADN, MIR) could win the majority (if not all) the seats in the media luna departments, and pick up a few seats here/there in the Andes, they’d win the presidency. Which is what happened in 2002.


  5. Surely a key part of the change to MMP in Bolivia was the fact that voters were given a second vote. In the PR system, their only vote was for a President and a list; despite the fact that the lists were regional, the primary voting decision would have been president/party, i.e. a national decision.

    The MMP system introduced a second vote, giving voters the ability to vote both on national alignment (president/PR list) and on local alignment (SSD). That’s an enormous change.

    One of the aspects of that change must be a change to the internal power relationships within each party. With a single-vote and closed lists, power within the party is highly centralized; elected representatives owe their positions to the party. Adding SSD-elected representatives to the mix gives those representatives considerably more influence within the party. Presumably, this was part of the motivation for introducing MMP, but I wonder if the traditional parties were actually aware of the consequences, and able to adapt to them.


  6. “It was parties — not candidates — that began to alter strategies into regionalistic campaigns.”

    Just as Shugart and Wattenberg said it would be! A move from closed-list PR to MMP is a change on the intraparty dimension. That is, parties should start nominating candidates whose vote-pulling ability can increase party votes.


  7. “Parties should start nominating candidates whose vote-pulling ability can increase party votes.”

    Absolutely, and whether they do that by decree or by holding local nominating meetings doesn’t make much difference — either way, they end up with candidates who reflect local politics, and ultimately with elected representatives who reflect local politics.

    I don’t know much about the particular case of Bolivia, I’ll freely admit. But it seems to me that elected representatives, in most cases, do have quite a bit of influence on internal party decision-making. (This might be less of a factor in Mexico, where re-election is forbidden.)

    The fact that SSD and plurinominal representatives exhibit similar voting behaviour does not in itself cast doubt on the hypothesis; the plurinominal representatives are motivated to toe the party line. Also, some of them may have run as SSD candidates, or may be planning to do so in the future.

    One consequence of a power shift towards elected representatives might be internal pressure to do candidate selection at local nomination meetings.


  8. Rici, ¡exacto! Voting behavior in the legislature is a rather poor place to look for evidence of different personal-vote or local-service incentives of SSD and list members (or members at different list ranks, etc.).


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.