How to Detect Election Fraud: the Bolivian Example
A basic challenge for any democracy is to accurately count votes, both to ensure the right outcome and to promote shared trust in the electoral process. There is a perennial temptation for politicians to try to manipulate vote counting, which has led to the development of sophisticated methods to detect election fraud. The recent election in Bolivia, which led to waves of protest and the resignation of President Evo Morales, provides an illustration of how we can detect election fraud and prevent the corruption of democracy.
Corruption, Fraud, and Statistics in Bolivia’s Disputed Presidential Election
On October 20th, 2019, Bolivian president Evo Morales ran for a contested fourth term. His candidacy came after losing a 2016 popular referendum on term limits, packing the constitutional tribunal with supporters, successfully arguing that term limits violate his human rights, and appointing the officials who oversee elections. The opposition parties voiced concerns about fraud early and often. The Organization of American States (OAS) and the European Union (EU) sent teams of election observers, as they had in past elections.
Bolivians voted in a process that appeared free and fair to observers. However, around 8 pm, the nonbinding rapid vote count shut down. Protests broke out across the country and the system came back 24 hours later with Morales winning by .1% above the threshold necessary to avoid a run-off. Under pressure from opposition and protesters, Morales invited the OAS to audit the election.
The OAS audit found forged ballot counts, ballot stuffing, a secret server, a dangerously vulnerable system, and statistical evidence of manipulation. The November 10th audit report concluded that the OAS could not verify the official results and that Bolivia should hold new elections. Morales called new elections but the military and civil society groups that had previously supported him joined the opposition’s calls for him to resign. Amid mounting resignations from his Movement towards Socialism (MAS) party’s senators, mayors, governors, and his own cabinet, Morales resigned.
How Can Experts Detect Electoral Fraud?
Auditors use several techniques to detect electoral fraud. In Bolivia, auditors interviewed Electoral Tribunal officials, reviewed documents and procedures, inspected electoral system infrastructure, and analyzed data from the rapid and official counts. The audit detected fraud and mismanagement throughout the electoral system. For example, the embassy and consulates in Argentina reported 38% more votes than eligible Bolivian voters in Argentina. Additionally, 333 polling station documents reported that over 99% of voters voted for the incumbent, which is highly unlikely in any political system; 23% of these documents had forged signatures, reported 100% participation of eligible voters, or had other glaring signs of fraud. Auditors focused on the most egregious evidence of fraud and noted that they expected to uncover more with more time.
Second, auditors use electoral forensics, which are statistical tools that detect patterns in voting data that are unlikely under free and fair conditions. Auditors and scholars have applied electoral forensics methods to elections in Kenya, Mexico, Argentina, South Africa, and Bangladesh, among others. Francisco Cantú analyzes electoral data in Mexico and Argentina and his findings of fraud in the 1988 Mexican election were strikingly similar to evidence of fraud in Bolivia in 2019, with political appointees overseeing a vulnerable system and forged documents with 100% participation from the incumbent’s strongholds. The OAS auditors used distributional assumptions and analyzed discontinuities to demonstrate that the vote count in Bolivia was highly suspicious.
A third indicator of electoral fraud is when precincts report vote totals. Polling stations’ reporting times should follow a normal distribution under free and fair conditions, as should the verification times from the Electoral Tribunal. In the Bolivian elections, polling stations’ reporting times followed a normal distribution but approval times were bimodal, with the Electoral Tribunal approving 5% of the initial ballot documents 24 hours after the first 95%. These 5% were enough to change the outcome of the election. OAS statisticians demonstrated that the 5% behaved suspiciously compared to the first 95%. For example, Morales’s vote share increased by 15% while his main rival’s vote share decreased by almost the exact same amount; a highly unlikely dynamic in an election with 9 candidates and many blank or null votes. Morales’s vote share had very slowly increased over the first 95% as more rural votes came in but his share jumps sharply in the last 5%, though the votes come from six states and are not dramatically different than the votes immediately preceding. The OAS concludes that these results are possible but highly unlikely under free and fair conditions.
There are additional techniques for detecting electoral fraud that the auditors did not use. Given more time, the auditors could have conducted a detailed geographical comparison of the votes in this election versus the last election, which could reveal major discrepancies. Additionally, with more time the auditors could build more detailed statistical models or formally test hypotheses. The election data are on the Electoral Tribunal website and these analyses remain for researchers interested in electoral forensics.
Corruption Set the Stage for Fraud Allegations
Morales’s disputed election took place in a country rife with corruption scandals and allegations. Bolivia has persistently underperformed in anti-corruption efforts, according to Transparency International. Bolivia currently scores 29 (where 0 is highly corrupt and 100 is very clean) on the TI scale, well under the regional average of 44 and rubbing shoulders with the worst performers in the region, Nicaragua, Haiti, and Venezuela.
Data from the Latin American Public Opinion Project shows that individual perceptions of corruption have increased considerably. In 2019, 46% of respondents reported that corruption among public officials was very widespread, compared to 32% in 2014. Personal experiences with corruption also increased: the percentage of surveyed Bolivians reporting they paid a bribe to a police officer or government employee in the last year has increased from 22% in 2014 to 32% and 30% in 2017 and 2019 respectively. At a staggering 40%, Bolivia has the highest rate of corruption victimization in Latin America. Corruption scandals occur regularly and few officials face consequences. This fuels Bolivians’ perceptions of corruption as rampant and sows distrust in officials and in the political system.
The 2009 constitution abolished campaign subsidies, making Bolivia one of two countries, along with Russia, to abolish subsidies, according to our data. Our research on campaign subsidies across 175 countries finds that they reduce corruption by diluting the impact of private donations. In Bolivia and Russia, removing public subsidies hurt the opposition parties but not the ruling party. Morales and his MAS party used state resources to campaign in 2009, 2014, and 2019. Opposition parties denounced MAS’s use of state resources throughout the campaign as unfair and corrupt.
Corruption erodes interpersonal trust and the legitimacy of the political system, according to research by Christopher Anderson and Yuliya Tverdova as well as Mitchell Seligson. Given the impact of political corruption on institutional trust, it is not surprising that electoral fraud became the last straw in a series of corruption scandals involving MAS officials.
A Wave of Retribution Complicates Reforms
The next step for Bolivia is to rebuild its democratic and governing institutions after 14 years of populist rule. The Bolivian constitution requires interim President Jeanine Añez to hold new elections within 90 days of taking office. The OAS and the EU have pledged resources and technical assistance to revamp electoral infrastructure. This is a gargantuan task: the system is vulnerable to fraud, hacking, and data loss at every step of the process, seven of the nine departmental electoral tribunals were burned in protests, and many of the system’s staff resigned or were arrested after the election.
Añez has overseen retribution against MAS supporters. Police and military opened fire on pro-MAS protesters in Sacaba on November 15th and Senkata on November 19th, killing 17 civilians and wounding scores more. The Añez administration has arrested dozens of former MAS officials since November 10th. Retribution further erodes trust and legitimacy in officials and institutions for the many Bolivian voters who support the MAS.
Bolivia’s leaders must reform the electoral system and call elections with the participation of the MAS in order to restore legitimacy to a political system in crisis. To rebuild trust in institutions, the next president must address corruption and build broad-based pacts with civil society and a fractured party system — two extremely difficult tasks.
Calla Hummel is an assistant professor of political science at the University of Miami and author of several articles on Bolivia. @CallaHummel
V. Ximena Velasco-Guachalla is a Ph.D. candidate at the University of Colorado and studies corruption and political protest. @vximenavg