Thursday, June 14, 2007

Run-Off Voting: Marginally Better

Certain quarters, including mlq3, have called for a switch to run-off voting in order to ensure that incoming presidents have majorities and therefore stronger mandates. The main drawback here is cost: depending on the particular variant used, we'll be holding at least two elections every time we choose a president. Considering the costs involved in just a single election, and considering that we'll probably end up having two campaign periods as well, this would be very difficult to justify, more so since the advantages of a stronger mandate are hard to quantify. Should this proposal come up for serious consideration, political and economic conservatives would likely oppose it.

As cost is unfortunately an object, the variant we'll probably be considering is the top-two runoff method. In this, only two elections are required, with the top two candidates from the first round going on to the second round. The other two variants, elimination and exhaustive runoffs, often require more than two elections, with the former depending on the number of candidates to be eliminated, and the latter depending on the candidates' coalition building and/or poaching skills.

Still, there's a problem with the top-two runoff method. Essentially, the voters are being asked to decide between the top two first-choice candidates; no consideration is given for getting lower-preference votes (second choice onwards). Thus, there is a significant possibility that a candidate, although not very popular as a first choice, is overwhelmingly popular as a second choice compared to the rest of the field. Such a candidate then could conceivably defeat either of the top two candidates in a two-way race. The rules, however, would eliminate him in the first round, which is a pity since it would be more likely he's a moderate acceptable to a large majority of the electorate, and would probably be able to work more productively with the legislature.

The elimination method, which removes the lowest scoring candidate in successive rounds, would be better in selecting a winner, as it takes into account lower-preference votes. Indeed, it could select a different winner from the top-two method, one more likely to defeat all other candidates in a two-way race. The caveat, as I mentioned above, is that it would require more than two elections. In fact, the number of elections needed is equal to the total number of candidates minus one. Given the number of candidates fielded in recent elections, this method, despite its advantages, is not really economically feasible.


To illustrate these two methods, consider a hypothetical race with four candidates: FPJ, Lacson, Roco, and GMA. They get 35%, 25%, 21%, and 19% of first-choice votes respectively. Using the top-two runoff method, Roco and GMA are immediately eliminated, and voters are forced to choose between FPJ and Lacson. Of the 40% total who voted for Roco and GMA, assume 18% are willing to support Lacson either as second or third choice, 9% would vote for FPJ, and 13% would abstain. Results for the second round would thus be 44% for FPJ and 43% for Lacson. If we don't count the 13% who abstained, that would be 50.6% for FPJ and 49.4% for Lacson. Using the top-two method, FPJ wins.

The elimination method yields a different result. In the first round, GMA is eliminated. It's not farfetched to assume that most of her supporters would have Roco as their second choice. Thus we can postulate that of her 19% base, 15% would go to Roco, 2% to Lacson, 1% to FPJ, and 1% would abstain. The second round results would thus be 36% for FPJ, 27% for Lacson, and 36% for Roco, eliminating Lacson. Of Lacson's 27% total, perhaps 2% would abstain, 13% would go to Roco, and 12% would go to FPJ. Final round results would thus be 48% FPJ and 49% Roco. Leaving out the 3% total who abstained, we'd have 49.5% FPJ vs 50.5% Roco. With the elimination method, Roco wins after three rounds. This is arguably a more accurate reading of the electorate's collective intent.


The top-two runoff method is unfortunately only marginally better compared to plurality voting. The sole advantage gained is the certainty of a majority for the winner. As we noted, it fails to take into account voters' second and third choices (as well as all subsequent preferences). By combining an emphasis on first-choice votes with a high cutoff requirement (as only the top two are selected for the second round), candidates are rewarded for staking out relatively extreme positions that appeal to a sufficient portion of the electorate who can give them their first-choice votes. Moderates, while acceptable to a larger portion of the electorate (as second or third choice candidate), would lack a large enough base of first-choice votes to advance to the second round. They are inherently disadvantaged by this system.

In any case, the advantage of a majority is dubious anyway if turnout for the second round falls significantly. Is it still a strong mandate if a large number of voters found both top-two choices unacceptable?

The elimination method, by getting rid of only one candidate at a time, allows a greater range of voter preferences to come into play. Second and third choices, even fourth or fifth, can affect the outcome significantly. There is a relatively higher tendency for the more moderate, more generally acceptable candidates to be selected. If we consider that such people would likely be able to work with Congress more effectively, then this may be the better method. Unfortunately, there's the cost. If only there was a way to do this more cheaply...

Next, we look at instant run-off voting...

Monday, June 11, 2007

Plurality Voting: Has It Outlived Its Usefulness?

The current voting system we use is called plurality voting. It's the most widely used system in the world at the present time. Its main advantage is simplicity: it doesn't take a rocket scientist to figure out how it works. And it works fine most of the time, IF there are only two candidates. Unfortunately, that's the exception rather than the rule in recent Philippine presidential elections. When there are three candidates or more, the limitations of plurality voting become readily apparent.

In a two-way contest, one candidate would always get the majority of the votes. He or she comes into office with an unquestioned mandate: it can't be disputed that more than half the citizens who voted preferred this particular candidate for this position. When you add another candidate to the mix, however, things can get sticky. Imagine an election in which FPJ, GMA and Raul Roco all ran for president. Suppose FPJ got 45%, GMA got 35%, and Roco got 20%. Under the rules of plurality voting, FPJ would be proclaimed president even if he didn't get a majority. A plurality (45% in this case) is all that's required. Consider what this means: 45% prefer FPJ as president, but 55% would prefer somebody else first. Clearly, this mandate won't be as strong as one accompanied by a majority.

Consider further if, hypothetically, 90% of those who voted Roco had GMA as their second choice while 10% had FPJ. Thus, if Roco were not to run, his votes would be split between FPJ and GMA, with most going to the latter. In a two-way race, GMA would have gotten 53% [35% + (20% x 90%)] while FPJ would've gotten 47% [45% + (20% x 10%)]. Given a simple choice between FPJ and GMA, the voters would choose GMA, but the addition of a third candidate to the race (Roco) sufficiently alters the dynamic to allow an FPJ victory. The result is an outcome unacceptable to 53% of the electorate, a paradoxical result.

In a real world example you might be familiar with, the 2000 U.S. presidential elections, there was a third candidate, Ralph Nader, appealing mostly to Democrats who would otherwise have gone for Al Gore. Nader was able to garner enough Democratic votes that Gore lost narrowly to G. W. Bush. If Nader hadn't run, Gore would easly have won the presidency, and with a majority at that.

Closer to home, the 1998 presidential elections saw de Venecia, Roco, Osmena, Lim, de Villa, Defensor-Santiago and Erap, among others, contesting the presidency. Erap won with 39.6% of the vote. De Venecia got 15.9%; none of the others got more than 14%. Conventional wisdom at the time had it that the same demographic was being courted by all the major candidates except Erap (who relied on his "masa") and Osmena (who represented the Cebuano vote). Had de Venecia been able to count on the votes that went instead to Roco (13.6%), Lim (8.7%), de Villa (4.8%), and Defensor-Santiago (2.9%), Erap might never have become president. As it was, Estrada was sworn in with 60.4% of the electorate preferring someone else. More importantly, perhaps as much as 45.9% (combined total of de Venecia, Roco, Lim, de Villa and Defensor-Santiago votes) thought him unacceptable for the post. Might this have contributed to his downfall three years later? Possibly.

Perhaps plurality voting is not the ideal system to use in a polity such as ours. Numerous interests as well as personal ambitions conspire to erode the influence of parties, and restoring the de facto two-party system of the pre-Marcos years remains a pipe dream. Indeed, special interest groups are quite happy with the multi-party circus we have now, and some quarters are even calling for a partyless democracy. Without political duopolies to focus diverse contending interests into two-way races, we will continue to have multiple candidacies for each election, and we will continue to run the risk of our plurality voting system producing paradoxical results.

Next, we look at run-off voting...

Sunday, June 10, 2007

Remembrance of Things Past

Surfing the net looking for data on past presidential elections, I came across this archived article from December 1997 in Asiaweek. It's a primer on the run-up to the '98 elections. Reading it, I was struck by all the names still in the game today. A few interesting quotes...

"Ramos's choice was hailed by many Lakas members. 'Joe will make a good president,' says Manuel Villar, congressman for Las PiƱas. 'He has vast experience in economics, foreign affairs, even security. He has shown his skill in handling talks with [Muslim and Communist insurgents].' Fellow congressman Gary Teves agrees: 'Joe has talent interacting with people. He's very patient and a very good conciliator.'"
"Other potentially tough opponents for de Venecia include Senators Gloria Macapagal Arroyo and Miriam Defensor Santiago (the latter was narrowly defeated by Ramos in 1992), who rank No. 2 and No. 3 respectively in popularity surveys. De Venecia's rating has largely languished in single digits, though one recent Metro Manila poll put him ahead. He sees an advantage in the number of opposition hopefuls. 'The opposition will be split,' he reckons. 'In that event, we are sure to win.'"

Wishful thinking, Joe.

Hard to believe it was almost 10 years ago. We're still living the aftermath of that particular electoral exercise.