Peru Election 2006

The archived version

Analysis of Congressional results at 99.904%

without comments

Thanks again to Rici for his excelent analysis!

As of May 31, the ONPE web report is still not complete. Only Apurimac and Lima are still missing mesas; Apurimac is at 99.904% (1 mesa left to process) and Lima is at 99.123% (292 mesas left). Although it is mathematically possible for the Lima results to change, it seems fairly unlikely. So the final results are as I reported on May 26 (and as El Comercio reported on May 29).
The attached PDF contains the complete results. The first page shows the Cifra Repartidora (C.R.)and the quotient of each party’s vote divided by the C.R.; the second page shows the actual number of seats won. Each electoral district is colour-coded according to the party which received the most votes. With the D’Hondt method, which Perú and most South American countries use, the number of seats won is the integer portion of the quotient; the fraction is discarded no matter how large it is. This tends to favour parties with larger votes.
Close results in some districts
In the first page, I’ve coloured in green the “near-misses”; the parties which would have received one more seat had they improved their result in that district by less than 2%, so you can see that Apra narrowly missed out on two seats –one in Apurímac and the other in Cusco– and Restauración Nacional (RN) narrowly missed a second seat in Lima. Diógenes Alva will be disappointed; despite having position 1 on RN’s list, he came second in the preferential ballot in Lima, losing to Alda Mirta Lazo Ríos (#5) by 28,982 to 24,235. Máximo San Román (#2) came fifth in the preferential vote.
RN’s current vote tally is slightly more than 4% of the projected total valid vote, so even if they get no votes in the 293 uncounted mesas, they will have passed the threshold. If, on the other hand, they get 7% of the votes in the uncounted mesas (that is, half again as much as they are polling in Lima), then Sr. Alva would win his seat. That is not mathematically impossible, but it would count as a miracle.
Overall Congressional results
Women in Congress
Of the 36 women elected (30% of Congress, a record for Perú), 11 represent Lima (no women were elected in Callao). In the 2001 congressional elections, 22 women were elected (18.3%) with 12 representing Lima and Callao. So the number of women representing “the provinces” has increased from 10 to 25. See the papers at for an interesting analysis of the interaction between quotas, the preferential ballot, and representation of women in Perú. In particular, Gregory Schmidt argues that the particular form of the preferential ballot in Perú — the unusual Double Optional Preferential Vote — has the unanticipated consequence of increasing representation of women, compared with closed-list or single preferential vote systems.
The effect of the electoral threshold
It is worth asking whether the 4% electoral threshold have any effect on congressional results. Looking at the results themselves, it would seem to have had very little effect; only one potential congresista was eliminated as a result of their party not making the threshold (Cesar Acuña, running for Alianza Para El Progreso in La Libertad, received 62,007 preferential votes, giving him the second-highest preferential count in La Libertad.)
It could be argued that the existence of the electoral threshold might have convinced parties to form electoral alliances rather than running independently. However, that is not borne out by the evidence either. There were a total of 24 parties and alliances in the election, compared with just 13 who contested the 2001 congressional election. While it is true that Somos Perú and Acción Popular probably only managed to reach the threshold by virtue of their alliance as Frente de Centro, it seems unlikely that the alliance was motivated by that reason; they probably expected to do much better. Meanwhile, the various splinter parties of the left, who might conceivably have reached the threshold had they formed a single alliance, failed to do so.
Other aspects of Perú’s electoral system should also encourage alliances, particularly the small size of most electoral districts and the use of the D’Hondt highest averages method; the major alliances in Perú were probably formed as a result of these influences. (In D’Hondt elections, an alliance is either neutral or beneficial; with Saint-Laguë, it is possible for an alliance to receive fewer seats than the parties’ would have received separately.)
If the threshold has any effect at all, then, it will be seen in the next congressional election as a result of the 17 parties who failed to make
the threshold being deregistered.
The proportionality of Perú’s proportional vote
I computed the proportionality index of the election based on two popular measures of disproportionality. The Loosemore-Hanby index is approximately 25.7% and the Gallagher Index is approximately 14.1%. These are very high; they are well out of the expected range for a proportional voting systems. The figures for the 2001 congressional election were 16% and 9.26%, respectively; still high, but not quite so dramatic. In part, this reflects the large number of votes for the 17 parties which received no seats (14.64%); in 2001, 11 of 13 parties won at least one seat, and only 3.3% of the electorate voted for a party which did not win a seat.
The disproportionality is made quite clear by the results for UPP and Apra; although they received very close to the same total vote (21.2% to 20.6%), UPP won 45 seats to Apra’s 36 (There is a similar but less marked disproportionality between UN and AF, see the table above.)
Disproportionality is introduced through three mechanisms:
1) The use of the D’Hondt algorithm, which tends to favour parties with larger votes.
2) The small size of many electoral districts, combined with an over-representation of small districts.
3) The electoral threshold (as noted above, this contributed only slightly to the electoral results.)
In European countries with PR systems, these factors tend to favour right-wing parties; clearly, that is not the case in Perú. It is possible that right-leaning political analysts in Perú will now revise their opinions on the value of these measures.
Article 187 of the Peruvian constitution specifies that “En las elecciones pluripersonales hay representación proporcional, conforme al sistema que establece la ley.” (I believe this was copied word for word from the 1979 constitution, article 65.) The interpretation given by the Constitutional Tribunal in its judgement of February 2, 2006 indicates that the thinking of that Court is that a system which “tends towards proportionality” is acceptable, given that perfect proportionality is impossible. That judgement cites a similar decision by the Constitutional Tribunal of Spain from 1985, also ruling on an electoral threshold (albeit a much more gentle one); both seem to be of the view that some deviation from proportionality is justifiable on the basis of other important considerations. Nonetheless, one would assume that there must be some limit to the disproportionality of an electoral system which conforms with the constitution.
Download file

Written by Michael Ha

June 1st, 2006 at 1:28 pm

Spam prevention powered by Akismet