The deal that was brokered to end the crisis in Haiti over whether there would be a presidential runoff was a classic example of what in Bolivia would be called a salida–a way out that satisfies the main players, even if it bends or breaks the formal rules of the game. I suppose there is a Haitian Creole word for the same phenomenon, but for the time being I will go with French sortie.

The LA Times on 19 Feb reports that the ‘Belgian Option’ Helped Avert Crisis in Haiti. Election officials huddle through the night and:

“They thrashed through the different proposals and eventually settled on a formula for handling blank votes that is applied in Belgium…”

Other than that the deal was over apportioning “blank” ballots, and that doing so took Rene Preval from just under to just over the 50%+1 threshold to win without a runoff, the Times is not very clear on what the ‘Belgian Option’ is.

Thanks to Google, I turned up the following from Indybay. Apparently, the deal concerns an interpretation of what is a “null” ballot vs. what is a “blank” ballot and how to consider them in the denominator for purposes of calculating whether a candidate has cleared the threshold. This distinction is made in several Latin American elections laws, with null referring to ballots on which the voter intent can not be discerned (“overvoted,” improperly marked, spoiled, etc.), as opposed to ballots that actually contain no mark for the office in question (“undervoted”).

The following is excerpted from the Indybay entry by Brian Concannon Jr., Esquire, directs the Institute for Justice & Democracy in Haiti:

Electoral officials have discarded 147,765 votes, over 7% of the total, as “null.” Article 185 of the Electoral Code allows officials to nullify ballots if they “cannot recognize the intention or political will of the elector.”

[...]

But 147,765 voided votes is a high number, suspiciously high since the decision to nullify was made by local officials handpicked by an Electoral Council that had no representation from Preval’s Lespwa party or Lavalas. Overly strict criterion (such as requiring an “x” to be completely within a candidate’s box), even if neutrally applied, would have had a disproportionate impact on poor voters, who are more unused to filling out forms than their better-heeled compatriots, and therefore more likely to make mistakes.

Another group of votes, 85,290, or 4.6% of he total valid votes, are classified as blank ballots. These votes were actually counted against Preval, because under the election law they are included in the total number of valid votes that provides the baseline for the 50% threshold. This is a potentially reasonable system, just unreasonably applied to Haiti. In principle the system allows voters to show their displeasure with all the candidates by voting for no one, which can make sense in places where voting is easier. In practice the system makes no sense in Haiti- it is absurd to think that 85,000 people, many without enough to eat, would leave their babies, their fields and other work and spend hours walking or waiting in the tropical heat just to say they did not like any of the 33 candidates. A more likely explanation is that some voters got confused by the complicated ballots and marked nothing. Again, this problem would disproportionately affect poor voters likely to vote for Preval.

[...]

The null votes could have been rechecked through a procedure that applied consistent rules across the country. The null ballots are supposed to be segregated in a separate envelope, so it would be easy to go through the envelopes from a few Bureaus, to ascertain whether there were enough improperly nullified ballots to justify a comprehensive review. If Preval could have added 22,500 votes to his lead from the 147,000 null votes, this alone would have put him over the top.

[...]

The Chosen Solution

The negotiators, instead of correcting the tabulation, decided to change the rules for the calculation of blank votes. They allotted blank votes to the candidates’ totals proportionately to each one’s existing vote share. So Preval got 48.7% of the blank votes, Manigat 12%, etc., which pushed Preval up over the 50% bar. This solution does make sense- it assumes, probably correctly, that the blank votes resulted from confusion, and allocates the votes accordingly. The result is the same as if the CEP simply discarded the blank votes, and treated them the same as null votes.

The Indybay report–which is really thorough and interesting, and recommended to anyone interested in this election–goes on to comment:

But what is sensible is not always what is legal.

Indeed. That is the very essence of a salida/sortie.

Some countries, such as Colombia, attempt to get around this problem by actually requiring a voter to make a mark on the ballot to vote “en blanco”–essentially a “none of the above” option. These votes are counted in the calculation of the threshold or quota for the given election, while spoiled (null) ballots are not. (See the Colombian ballot for the upcoming legislative election for an example.*)

Quite apart from the oxymoron of having to mark a ballot in order to leave it “blank,” the Colombian ballot design makes sense.

Of course, Haiti could have avoided a runoff crisis altogether by adopting a rule in the first place that allows a plurality to suffice if the runner up is far back in the field, as I discussed previously, in comparison to the Costa Rican election (two candidates each just above 40%, yet no runoff likely).

But as Indybay notes, ideally rules should be changed before elections, in anticipation of sensible ways to prevent crises, not to resolve crises after them.

* h/t Steven Taylor, personal communication.