Extended and somewhat revised since initial planting.
With all due caveats about making inferences to individual behavior from aggregate statistics, I thought it would be interesting to look at the presidential and deputies votes in the Mexican states for patterns. This is barefoot empiricism, folks!
The presidential race is nationwide plurality. The deputies are elected in 300 individual plurality races, and the votes cast in those districts are aggregated nationwide to determine another 200 seats by proportional representation (in parallel fashion, other than caps on maximum permitted over-representation that were not triggered in this election).
We would expect that three-party politics would result in some ticket-splitting, with some supporters of the trailing presidential candidate (Madrazo of the PRI) choosing between the two leading presidential candidates, but sticking with the PRI for deputies. An indicator that is readily available is the state-level votes for the parties for president and chamber.
The PRI won the plurality of votes for president in no states. (Please pause for a moment and think just how historic that is!!) However, the PRI did retain a plurality for deputies in six states: Campeche, Durango, Hidalgo, Quintana Roo, Sinaloa, and Veracruz. (If you do not know Mexican states, check out the Clickable map of Mexico.)
Of the states in which the PRI won the chamber plurality, which of the PAN and PRD presidential candidates won the most votes? The states split 4-2, with AMLO of the PRD winning Campeche, Hidalgo, Quintana Roo, and Veracruz, while CalderÃ³n of the PAN won Durango and Sinaloa.
This breakdown really demonstrates the extent to which Mexico does not really have three-party politics, but two distinct regional party systems with different pairs of leading parties. (In an earlier draft, I said two distinct “two-party” systems, but that is not really right, as the party that came in third in the presidential vote has at least 20% in eighteen states.)
The four states that split their chamber/presidential votes PRI/PRD are located in the Yucatan peninsula or the Gulf Coast, except for Hidalgo (which lies between coastal Veracruz and the Mexico City area). On the other hand, the two states that split PRI/PAN are in the north: Sinaloa is along the Pacific coast south of the border state of Sonora, and Durango is Sinaloa’s neighbor to the east (and immediately south of the border state of Chihuahua).
To those who see the PRI having struggled to make it over 20% in the national presidential vote and then conclude that the party must be over for Mexico’s “dinosaurs,” I would point out that the PRI came in second in the chamber vote in nineteen states, or more than three fifths of them. (Plus, it came in first in six more, as already noted.) In ten of those states, it fell less than ten percentage points from the leader, and in five it was less than five points behind. These showings suggest it will remain competitive in congressional races (as well as state politics) in several states for some time to come. Despite the PRI’s poor performance nationwide, it does not appear to be in danger of collapse. It simply has many supporters who are willing to vote tactically in presidential races.
The other side of the coin of the PRI’s continued strength is that the PRD remains largely a regional phenomenon. It finished third in the chamber vote in eighteen states (counting the DF as a “state”). In fourteen states, it failed to reach even 20% of the chamber vote, and in only eleven did it surpass a third of the vote. Fortunately for it, many of the entities in which it performs relatively well are among Mexico’s dozen most populous: the DF, Estado de MÃ©xico, Oaxaca, MichoacÃ¡n, and Chiapas.
Finally, about those regional party systems. Let’s consider a state to have competition between two major parties if the leading party has no more than 1.5 times the vote share of the second party and if the second party has at least 1.5 times the vote share of the third party. By this definition (and other plausible definitions would affect the point only marginally), Mexico has the following incidences of the various competitive pairs of parties in the 2006 chamber vote:
PRD vs. PRI: three states
PAN vs. PRI: seven states
PRD vs. PAN: one state
The one PRD vs. PAN state that met the criteria was Tlaxcala.
Most of the remaining states are either dominated by one party that gets 45% of the vote or more with no other party close to it (e.g. QuerÃ©taro, Guerrero, and the DF), or have genuine three-party competition (e.g. Zacatecas, Hidalgo).
So, while the presidential candidates of the PRD and PAN may have finished within about half a percentage point of one another, PRD vs. PAN is not the primary axis of competition within very many states. The implication, of course, is that all three parties are likely to be able to survive for some time.