How the Arce/Choquehuanca ticket reclaimed the pre-2016 Morales majority

On October 18, 2020, Bolivia’s Movement Towards Socialism–Political Instrument for the Sovereignty of the People (MAS-IPSP) party roared back into a majority at the ballot box, winning a solid 55.11% of the vote and nearly doubling its nearest rival, Carlos Mesa’s Citizen Community (CC)’s 28.83%. The presidential ticket of Luis Arce Catacora and David Choquehuanca easily surpassed the simple-majority threshold for avoiding a runoff vote, and had an ample 26.28% margin of victory. Their victory came just 364 days after the disappointing performance of Evo Morales, who garnered 47.08% with a 10.57% margin over Mesa. It is all the more impressive since Arce and Choquehuanca campaigned without the benefits of incumbency and under the cloud of political persecution imposed by the temporary government of far-right interim president Jeanine Áñez.

This post takes a quantitative look at which parties gained and lost votes between these two elections and how the MAS-IPSP majority has evolved in size and geography since 2005. In the past year, the most consequential shift was missed by the headlines: the collapse in support for third-party candidate Chi Hyun Chung. I will also consider what these shifts reveal about Evo Morales’ second-round chances in 2019 (much better than expected), and Carlos Mesa’s fateful decision not to negotiate a prompt second round. A year ago, I looked at where and how Evo Morales lost his majority in 2016 and 2019; in the final section, I extend that analysis and see where Arce and Choquehuanca gained back supporters.

I draw on a statistical analysis of votes shifts by Diego Aliaga and colleagues, municipality-level vote analysis by Arián Laguna, citing and sometimes questioning their conclusions, and vote data from the Plurinational Election Organ (2019|2020). The analysis from 2019 to 2020 is simplified by the fact that nearly equal numbers of valid votes were cast in the two elections, meaning that we can work with raw vote counts as well as vote shares.

How did votes shift among the parties from 2019 to 2020?

Looking at the national vote totals we can see the a handful of major shifts.

  • The MAS-IPSP (dark blue in the graphics) won with 504,693 more votes than in 2019.
  • Creemos (red) received 601,870 more votes than Bolivia Dice No
  • CC (green) received 464,967 fewer votes than in 2019.
  • Chi Hyun Chung (yellow) switched parties & got 443,826 fewer votes
  • 144,444 voters shifted from non-competing smaller parties, while the PAN lost 8,061 votes from its already marginal performance in 2019.

Put simply, the parties to the left of zero largely gave their votes to the parties on the right. (For purposes of understanding these shifts, I’ve made two simplifying assumptions: thinking of Chi Hyun Chung’s two candidacies as the same even though he switched parties, and treating the Bolivia Dice No “21 February” alliance in 2019 as the base for Creemos in 2020. (See a full list of the 2020 candidates and their alliances.) The latter is something of a provisional assumption, but it appears that vote flows out of this BDN/Creemos block were minimal, so the choice is validated by the later analysis.)

There are two easy narratives to explain this. If you focused on the topline of the election, you might assume that Arce/MAS-IPSP gained at Mesa’s expense. On the other hand, evangelical Chi Hyun Chung & the MAS-IPSP compete for the same rural, indigenous voters. So it’s possible to imagine a world where votes flow between Chi and the MAS-IPSP, and separately between Mesa and the right-wing parties.

Diego Aliaga and Ana Lucía Velasco advance the latter hypothesis here and also share vital quantitative work analyzing those voting tables that were used in both elections. By disaggregating the vote into individual precincts and the tables within those precincts, we can get a pretty clear sense of how many voters switched from which party to which other party. They propose: “What happenened in the election? The CHI Factor” Put another way, Chi’s 2019 voters gave the MAS-IPSP a majority.

Principally, the 7.3% of the vote which Chi lost in 2020 went in greater measure to the MAS and smaller measure to the CC. The second evident change is that Creemos recovered the vote of Bolivia Dice No (21F), a bit of Chi’s vote, and an even smaller bit of the MAS’s vote, but managed to attract an important percentage from the CC.

So the MAS-IPSP was fortified by Chi voters. Mesa was weakened by defectors to Camacho. A 10% margin widened to a 26% chasm.

But I wanted to know how many voters went each way, so I dug into Aliaga et al.’s cluster analysis which breaks down voting tables into five clusters, based on their 2019-to-2020 shifts. You can see the clusters geographically on this cartogram (which enormously magnifies cities and shrinks rural areas). Metro Santa Cruz is the yellow and green ++CRE and +CRE patches The highland cities are +CC light blue El Alto & the Altiplano are ++MAS dark blue Much of rural Bolivia is +MAS orange. (A helpful tool for wrapping your brain around this cartogram is the side-by-side map here.)

By looking at the vote shifts in these five clusters, we can see exactly how votes got from the vote-losing parties to the MAS and Creemos. The net result:

250K votes from Chi Hyun Chung
98K votes from small parties
40K net votes from Mesa (CC)
and lost 2K votes to Creemos.

Creemos got 
71K votes from Chi Hyun Chung
435K net votes from Mesa (CC)
29K votes from small parties
on top of 260K from Bolivia Dice No.

In areas where there is data, the “Chi factor” made up 64% of the MAS-IPSP vote gain over the past year. Put another way, new support from Chi Hyun Chung’s voters alone was enough to put Arce and Choquehuanca into the lead. Assuming that this 64% share is representative, Arce/Choquehuanca would have cleared the 50% threshold by just five thousand votes if all of Chi’s voters had stayed loyal to him in 2020.

Chi Hyun Chung gave 66% of his vote loss to the MAS , 19% to Creemos, and 15% to CC. (The dataset only investigates 76% of the MAS-IPSP vote gain, 88% of the Creemos vote gain, since there were some rearrangements of voting tables.) This is a very significant share, but far less than the Aliaga and Ana Lucia Velasco’s claim that “That is to say, the Chi’s losses manage to explain 13 of every 14 votes gained by the MAS.” This overstates the case. Also, where Chi’s vote went depended on region, flowing to the MAS in the West, Creemos in the East, and the CC in the central and western cities (minus El Alto).

It turns out that Chi’s 2019 voting tables were places where not just he, but also other parties had vote switchers to MAS in 2020. Where the MAS gained votes from Chi Hyun Chung, in El Alto, other urban peripheries of the Altiplano, and much of rural Bolivia, it also won over nearly 98 thousand votes from the small parties, 57 thousand from the CC, and 30 thousand from the PAN. This reconsolidation of the MAS vote cut across all of the other parties, but was geographically specific. The MAS reconsolidated support in its high-altitude heartland.

Carlos Mesa gave up even more voters than Chi did between the two elections. The cluster analysis shows that 88% of these votes went to Luis Camacho and Creemos rather than to the MAS. (For every vote the MAS-IPSP gained from the CC, Creemos gained 7.5.) This too was regionally concentrated in Santa Cruz, here Camacho’s block consolidated 45% of the vote. Divisions between the CC and Creemos allowed the MAS to become the first place party, padding its senate majority.

The voting clusters

If you want to see where these votes came from read this annex. If not, skip to the next heading.

I used the cluster results by Aliaga et al., but I had to recalculate the vote shifts in these clusters a bit since I treat Creemos as the successor to Bolivia Dice No. Since there is almost no vote loss from this combined group, this simplifies the results.

The bulk of the MAS’s gains, 76%, are explained by the two clusters identified here. And an even greater share of Creemos’ relative gains, 88%, came in their two clusters. You can see the vote flows here. Everything to the right of zero flowed in, small gains in the Creemos clusters were shared by the PAN and the MAS (but these are nearly cancelled elsewhere).

Keeping the horizontal scale the same, we can see that Mesa’s CC party did pick up votes in some places (apparently the cities besides El Alto and La Paz, as well as rural Beni, per the light blue shading on the cartogram above). These 80,284 votes gained were far less than those lost to Creemos, so Mesa ended up with a far smaller vote share.

Departmental results: While the MAS-IPSP lost ground everywhere in 2019, its rebound was regionally limited

Breaking down the MAS-IPSP vote share over the past 5 presidential elections (plus the 2016 term limit referendum) by department shows a few key patterns. Just as Evo Morales suffered broad-based losses (in all nine departments) between 2014 and 2019, the Arce-led MAS-IPSP made broad-based gains (everywhere but Beni) by Oct 2020. While they gained everywhere, the big increases came in La Paz and Oruro, and secondarily in Cochabamba and Potosí. These were former MAS-IPSP voters coming home to them.

Where MAS vote shares were tightly clustered in the high 60s in these four departments plus Chuquisaca as recently as 2014, it suffered 20%+ vote losses in Potosí and Chiquisaca in the 2016 referendum. In 2020, MAS support in these two departments remained well below its 2014 levels, while the party had nearly recovered its vote share in La Paz, Oruro, and Cochabamba. Elsewhere in the four departments known as the media luna (crescent moon), Arce and Choquehuanca were unable to recover more than 2% of voters.

While La Paz, Oruro, and Potosí remained at the center of MAS support they are also the departments that swung the most in terms of that level of support. As I noted last year, the shifting fate of Evo Morales is a consequence not of a stronger opposition in the East, but less loyal support in the West. That is still true, and now we know that one-time MAS-IPSP voters who temporarily deserted the party largely went to Chi Hyun Chung and the minor parities.

Given its much larger population, La Paz department—and really the twin cities of La Paz and El Alto—have become the critical swing region between the MAS and its opponents. The defection of many in La Paz and El Alto from the coalition in 2016, 2019 was decisive, as was their return in 2020.

Arce/Choquehuanca led the MAS-IPSP to its highest vote share since 2014, but they haven’t quite recovered the dominance it held back then, in any department. (Obviously, they campaigned under an adverse government and without incumbency this year.)

Over time, there has been a great geographic evening out of MAS-IPSP support: less astounding dominance in the Altiplano, more backing in the lowlands.

Municipal results: Where exactly did these shifts happen?

For this, I turn to impressive work by Arián Laguna at @cemees_mx. Laguna explains that the 2019 to 2020 shift occurred predominantly in the large cities. The huge vote gain in El Alto (+142k / +22% vote share) is 29% of all votes gained nationwide by the MAS. Add in the MAS-IPSP’s rebounds in Cocabamba city (+33k), La Paz (+31k), Sacaba (+15k), Sucre (+14k), and Quillacollo (+10k) and you have 245 thousand votes, nearly half the national vote gain by the MAS-IPSP. Altogether, cities over 100,000 inhabitants contributed over 301 thousand new votes to the party.

Meanwhile, Laguna calculates that the rural heartland of the MAS-IPSP (including medium-sized cities Villa Tunari, Achacachi, Viacha, Puerto Villarroel, Yapacaní, Sipe Sipe y Caranavi) increased its strong backing for the party, raising its vote share from 76% to 89%. This brought in 155 thousand new votes. (Laguna’s calculations were made with a half-percent of the vote still uncounted.) For Laguna, “There is no doubt, then, that the MAS owes its victory fundamentally to the radicalization of the rural vote and of the medium-sized cities of the Altiplano (and the colonization zones [such as northern La Paz and the Chapare], as well as the El Alto vote.”


Aliaga and Velasco see the rise and fall of Chi Hyun Chung as key to the electoral comeback of the MAS-IPSP. Laguna locates just where the MAS swelled its support. As it turns out, the Chi factor, the radicalization of the rural MAS vote, and its bounceback in El Alto and other cities are the same story, just with a reversal of figure and ground. Either way, the arena of partisan contention remains the same.

The sheer number of MAS-IPSP voters who came home to the party in 2020 suggests that Evo Morales would have had a very strong chance of winning the 2019 runoff. He needed little more than a third of the voters that Arce and Choquehuanca won back to do so. Evidently Chi Hyun Chung’s voters were neither too Christian nor too conservative to back the MAS-IPSP again.

If so, a variety of parties gravely miscalculated their actions in the 2019 crisis:

  • Evo Morales should not have been so eager to avoid a runoff, and either proclaim or insist on a first-round victory. Negotiating a runoff early on would have de-escalated tensions and kept him in the Palacio Quemado.
  • Carlos Mesa also had a vital interest in bargaining for a second-round runoff during this period. His proclamation that he “had nothing to negotiate” with Evo Morales could not have been more wrong. Mesa had a far better chance as a presidential candidate in a December 2019 runoff than in 2020 re-run of the election. Instead, the CC lost 3 senate seats and 11 deputy seats.

Separately, we now know that the halting efforts of the right-wing parties to negotiate an effective alliance in 2020 were all for naught. Surely, these parties were hurt by the failures of Áñez and her government, but to a large extent their path to an electoral victory required winning over indigenous voters from the MAS-IPSP or from Chi, something they never seriously tried to do. Likewise, the pleas of Áñez and Tuto Quiroga for their supporters to coalesce behind Carlos Mesa went unheard. Apparently a large number of Mesa voters defected to the right, likely because they were confident that they could enlarge the Creemos vote in the first round (and its parliamentary delegation) and help win Mesa the presidency in the second round. They were mistaken. Áñez and Quiroga gave up their parties’ place in parliament and got nothing in return.

Luis Fernando Camacho heads the only party on the right to emerge stronger than 12 months before, but 14% of the vote, and little over a tenth of the legislature is hardly a secure grasp on power. Despite the visibility of the Santa Cruz-led right wing in the past 15 months, voters backing the MAS-IPSP remained a consistent if minority presence in Santa Cruz. Creemos seems likely to consolidate departmental control in Santa Cruz in the upcoming regional elections, with CC in the pole position in Beni and Tarija.

More generally though, the MAS-IPSP’s electoral dominance is based on its universal presence in every department and every municipality. The opposition parties are regional and disproportionately urban. (Listen to Vladimir Torrez on this last point: from 44m onward.) Sure of widespread victories, the party sets numerical goals in regional elections: this time it is going for 300 of the 343 municipalities and seven of nine departments, though these targets could be a stretch. Until an opposition coalition emerges that can bridge urban and rural, East and West, and receive the support of disaffected MAS voters in larger numbers, it will be unable to defeat the renewed MAS-IPSP.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s