26 May 2021, Iris Proff
If you could, how would you change your neighbourhood to make it more livable for everyone? Would you add a fountain on the square? Set up a weekly meet-up for the elderly? Or install a park bench at the water side?
In Amsterdam, as in many cities around the world, residents have a say in how the budget for their neighbourhood should be spent. “Participatory budgeting” is an instrument of deliberative democracy: Everyone living in a given town or city district can formulate their own ideas about what a fixed budget should be spent on. The residents deliberate publicly about the proposals and finally vote on them. The winning projects will be implemented by the administrative institution that set up the process. Usually, this procedure is repeated every year.
What might sound like a straightforward process to empower citizens and non-citizens holds many challenges in practice. For instance: How to combine everyone’s opinion in the fairest way possible to select the winning projects? The Computational Social Choice research group at the ILLC tries to answer this question. Without ensuring to be fair to all voters, participatory budgeting can hardly fulfill its original purpose.
How participatory budgeting travelled around the world
Participatory budgeting was first invented in 1989 in Porto Alegre, a city in southern Brazil, after the end of the military dictatorship. The goal was to strengthen democracy, overcome notorious corruption and decrease inequalities – a means to achieve a profound social transformation of the city. Participatory budgeting quickly spread throughout Brazil, then through other Latin American countries with left-wing governments and finally over the whole globe.
In 2012, it was estimated that 1,500 municipalities on all continents employed a form of participatory budgeting, from rural towns to metropolises such as New York City. Some scholars argue that, as participatory budgeting spread across the globe, it lost much of its ideological value and became more of a handy public administration tool, embraced by politicians on all parts of the political spectrum.
The “buurtbudget” in Amsterdam
Amsterdam implements a participatory budgeting scheme since 2019, also called “buurtbudget” (“neighbourhood budget”). Rocco Piers, member of the political party GroenLinks, is responsible for the “buurtbudget” in Amsterdam South. In the last year, this district of 150,000 inhabitants got assigned a budget of half a million to be allocated by residents. Piers is enthusiastic about the process that encourages people to engage actively with their neighbours and neighbourhood. During last year’s votes, there was a turnout of 16 percent – a decent amount for a procedure of this kind.
Many of the residents’ proposals aim to make the district greener by planting trees or exchanging parking spots for patches of green. One of last year’s winning projects was an initiative to help people install green roofs on their sheds, which turned out to be a big success. Social projects, such as setting up meet-ups or support groups are not frequently under the winning proposals, says Piers.
The big fountain on the square or the bench at the waterside
A question that naturally arises is how to select the winning projects from peoples’ votes. Should one favour big projects – think of the fountain on the square – that many people like but that also cost a lot? Or should one opt to implement many small projects – think of the park bench on the water side – each of which is supported only by a small number of people?
“Imagine each neighbourhood wants a fountain on their square, but you can only pay for one,” says Piers. “Should you only build the one with the most votes, while all other neighbourhoods get nothing? That doesn’t seem fair.”
This is where the Computational Social Choice research from the ILLC comes into play. Ulle Endriss, Arianna Novaro, Simon Rey and Ronald de Haan analyze different voting rules for participatory budgeting and compare in which ways they are fair or unfair. “In parallel to the development of participatory budgeting in the real world, our research is looking at it in a more principled way,” says Ulle Endriss.
Arianna Novaro, who was involved in an analysis of participatory budgeting in Toulouse, says: “Especially when there are very expensive and ambitious projects, we found that outcomes differed between voting rules.”
Why is that? Let’s take a look at two examples.
The greedy approach
Most commonly applied is a rule called “greedy approval voting”. Proposals are first ranked with regard to the number of votes they received. The project with the most votes is approved first, then the project with the second most votes is approved – until a project is reached which would exceed the budget when added to the already approved projects. Now, one needs to look further down the list to see if there are cheaper projects that still fit within the budget. This way, the budget is used up as much as possible.
This voting rule favours expensive projects. It favours the fountain in the park over the small park bench, because the fountain will tend to attract more voters. Therefore, the voting rule bears the danger of disregarding voters that are not in favour of these popular but expensive projects.
Voting that leaves no one behind
An alternative set of voting rules which address this problem is called “Phragmén voting”, named after the Swedish mathematician Lars Edvard Phragmén. It works pretty much like this: Imagine all voters are standing lined up besides one another, each one with an empty bucket in front of them. The mayor comes by and puts the same amount of money in everybody’s bucket. If a group of voters now comes together and votes for a project that can be realized with the combined money in their buckets, it will be approved. A group of 100 voters with 10 euros each in their buckets can pay for a park bench that costs 1,000 euros, but not for a fountain that costs 50,000.
This voting scheme favours cheaper projects, as it is difficult to gather a cohesive group large enough to pay for the fountain in the square. As many people as possible get the chance to get a project they like approved. Phragmén voting is thus “fair with respect to diverse groups of voters, such that no one is left aside”, says Simon Rey.
With a limited budget however, it might still happen that some voters are left empty-handed. But, as Simon Rey outlined in a recent paper, it is possible to consider fairness over multiple consecutive years. Put simply: Voters who did not get any wish approved will be prioritized in the future. Any money left in their buckets in one year will be carried over to the following year.
While Phragmén voting might be more fair, it also has a downside: It might prevent bigger and potentially more impactful projects from being realized.
How to prevent residents from strategizing
Another aspect the researchers considered is the initial stage of the participatory budgeting process, where residents propose projects. Organisers commonly remove proposals that are not feasible within the budget or given regulations and sometimes combine similar projects. The researchers wanted to determine if it is possible to devise a rule for the proposal stage that is “strategy-proof”: where residents do not get any advantage out of making strategic proposals that do not correspond to their true preferences.
It turns out that strategy-proofness is very difficult to achieve. “People might strategically propose a second alternative fountain to the first fountain. That might prevent any fountain from getting realized, because half of the people will vote for either,” says Simon Rey. One way of addressing this, for organisers of a participatory budgeting initiative, is to limit the number of similar projects that might compete with each other.
The different facets of fairness
Even a perfect voting rule cannot ensure that participatory budgeting is fair to all residents. Groups of voters who do not engage in the process in the first place will be left aside.
In Amsterdam, Rocco Piers noticed that most proposals originated from richer neighbourhoods. “I think those people are better at organising themselves,” he suspects. “They can invest a lot of time and print flyers to mobilize others.” Residents with a lower income might not have the time or money to engage actively in the decision process. Language barriers and limited internet access pose other obstacles.
Rocco Piers is determined to address these challenges. For the coming edition of the “buurtbudget”, his team wants to reach out to people on local street markets. The organisers also found that splitting up the process in smaller neighbourhoods proves more successful: Here, people have the chance to collaborate more directly and discuss ideas in person. It also prevents larger or richer neighbourhoods overruling smaller or less privileged ones.
Overall, participatory budgeting might have lost its ideological character on its travel across the globe. But when implemented in a careful, inclusive, and fair way, it still remains a powerful instrument. An instrument to get the local population engaged with their neighbourhoods, to strengthen democracy and promote dialogue between people who might not interact otherwise.
Curious for more on Computational Social choice? Check out this video that explains why no voting rule is fair.