Not all votes count the same. It has been said many times that some provinces offer “cheap” seats, because they are obtained with fewer votes. But the distribution can also provoke other carambolas, such as that thousands of votes in Alicante or Cádiz are irrelevant (if no deputy moves) and that a single vote in Murcia decides a seat. Our electoral system, like so many others, is also somewhat capricious.
Political scientists know its peculiarities, or at least knew it, because it is not clear if what was true for a bipartisan equilibrium continues to be true now. We know that the system has a rural bias that ends up being conservative. The less populated provinces distribute more seats per inhabitant, and since this “empty Spain” voted more for the PP, the conservatives started with an advantage. As Alberto Penadés explained in this newspaper, “if the PSOE and the PP had tied votes in the last elections – the rest being the same – the PP would have obtained nine more seats than the PSOE.” This could be changing, however, if, as it seems, the vote of the small provinces benefits the PSOE before Podemos. It is also unclear how the rural bias affects Ciudadanos, now that it is around 18% of votes, or Vox, which is unknown.
The other feature of the system was majority bias. Except in the big provinces (Madrid, Barcelona, Valencia, Alicante and Seville), the rest distribute few seats and that penalizes small forces of dispersed voting. That was the case of Izquierda Unida in 2015, for example, when it did not obtain even 1% of seats although it had 3.7% of the votes. It was common to use a shortcut and presume that state-level parties needed to exceed 15% of the national vote so as not to be seriously harmed in seats. However, those references may not be valid now.
To demonstrate this we have made a simulation. First we have taken the most recent surveys (which say that the sum of PP, Ciudadanos and Vox will be around 49% of votes). Then we have found a way to distribute the votes by province, as explained in the methodology. And finally, we have been changing the votes of Vox to observe its effects. The sum of seats on the right can be very different if Santiago Abascal’s party rises to 15% of votes or stays at 6%, although the three parties add exactly the same votes. This is so due to the non-proportional effects of the electoral system.
As seen in the first graph, if Vox did not exist and its votes were distributed by Ciudadanos and the PP, these two parties would be around 190 seats. They would have the majority almost insured.
However, if Vox is around 10%, as polls now say, the sum of the three parties on the right stays at 176 seats and has the majority in the air. A small loss of Vox votes, even if it was to benefit the PP or Ciudadanos, would make their sum lose seats. On the other hand, if Vox grows above 12%, they would have more options to exceed 176 seats. These simulations are only an approximation, and the real thresholds may change with other assumptions, but these results show that small changes in the vote of a party can be decisive when configuring majorities.
A key is the medium provinces
Why does the majority of rights falter when Vox moves between 6 and 9 percent of votes? For something that happens in some medium-sized provinces: the right-wing party loses several seats that go to the PSOE or Podemos in provinces such as Toledo, A Coruña, Asturias, the Balearic Islands or Cádiz. In our simulation of Malaga, for example, when Vox exceeds 9% of the vote it takes away a seat that would otherwise be for the PSOE.
The provinces that distribute eight seats are especially important in a situation like the one that the surveys now draw. With a 6% vote, Vox could be left without representation in Almería, A Coruña, Asturias and Las Palmas. On the other hand, with 10% of votes she would get four seats there.
Those eight-seat provinces demonstrate how difficult it is to make predictions with the relationship between votes and seats, because the effect changes direction several times. We have seen that the sum of the right benefits in the provinces of eight deputies if Vox grows from 6% to 10%, but falls again if the party continues to rise above that figure (Vox does not win new seats, but the votes that snatches the PP and Citizens if they subtract deputies).
These swings do not occur in the provinces that distribute many seats. The clearest case is Madrid, where 36 seats are distributed and the system behaves practically as a proportional one: more votes translate into more seats for both a party and a block. There it matters very little how the votes are distributed among possible allies. If Vox rises or falls, they will change their seats, but the sum with PP and Ciudadanos remains almost unchanged.
In all these calculations we have assumed that the sum of the vote on the right did not change and it was always 49% that the polls now point to. That could change and open different governance scenarios. The third graph represents four different cases, changing the total votes obtained by the right (PP, Cs and Vox) versus the left of PSOE and Podemos. These two parties could add a majority only if the polls gave an eight-point turnaround, although it would be worth recovering three points from PP, Ciudadanos and Vox to drastically reduce their options to govern with a formula like the Andalusian one.
Methodology. These numbers are an exercise in approximation. First, because polls always are; second, because they have not stopped moving in the last year; and third, because the electoral announcement could change the electoral dynamics.
Estimate of seats. The calculation of deputies consists of three steps: 1) we take the average of polls at the national level, 2) we distribute the total votes on each province, taking the 2016 election results as a reference, and 3) we calculate the seats in each province applying the D’Hondt method.
Average of surveys. The average takes into account dozens of soundings to improve its precision. The data has been collected mostly on Wikipedia. In the case of the CIS, an own estimate is included from its raw data. The average is weighted to give different weight to each survey according to three factors: the sample size, the polling house and the date. Right now that average shows the following results: PSOE 24.4% of votes, PP 20.7%, Citizens 18%, Unidos Podemos 15% and Vox 10.6%.
Projection of the national vote on each province. The vote of PP, PSOE, Ciudadanos Unidos Podemos and the rest of the parties is projected on each province taking the 2016 result as a reference. The only exception is Vox, as it is a practically new game. In your case we use the data from transfers from the other matches. We know from surveys (from the CIS, Celeste-tel, Metroscopia and IMOP) that around 60% of Vox voters come from the PP, 18% from Ciudadanos, 3% from the PSOE, 4.5% from Podemos and a 15% of abstention. We estimate the vote of Vox in each province from the vote of those parties in 2016, assuming that the flows are the same in all.
Deputies by province. For the simulation we have used the number of deputies by province that was in force in the last elections of 2016. Those figures may change slightly with the new electoral census before the elections on April 28. Provinces with 1-2 seats: Ceuta, Melilla and Soria. With 3 seats: Ávila, Cuenca, Guadalajara, Huesca, Palencia, Segovia, Teruel, Zamora. With 4 seats: Álava, Albacete, Burgos, Cáceres, La Rioja, León, Lleida, Lugo, Ourense, Salamanca. With 5 seats: Cantabria, Castellón, Ciudad Real, Huelva, Jaén, Navarra, Valladolid. With 6 seats: Almería, Badajoz, Córdoba, Girona, Gipuzkoa, Tarragona, Toledo. With 7 seats: Granada, Pontevedra, Santa Cruz de Tenerife, Zaragoza. With 8 seats: Asturias, Balearic Islands, A Coruña, Las Palmas, Bizkaia. With 9 seats: Cádiz. With 10-12 seats: Alicante, Malaga, Murcia, Seville. With: With 16 seats: Valencia. With 31 seats: Barcelona. With 36 seats: Madrid.