This article gives an introduction into how private actors (e.g. individuals & businesses) can be incentivized by other private actors to provide public goods, allowing egoistic motivations to result in altruistic outcomes. A variety of matching funds and voting mechanisms have shown promise in theory and are already being used today. If you are researching or working in this field, I would love to discuss and write about your work. Feel free to reach out via: firstname.lastname@example.org
A related article was just published today on the EA Forum, glad to see more people actively thinking about this.
Incentives matter: when something gets more profitable, people produce more of it. When it gets less profitable, people produce less of it. Granted, incentives play a bigger role in some people's lives than others, and in some stages of life than others, but they play a role nonetheless. However, the incentives which govern most people are not aligned with global well-being. When a company sells a product, their aim is essentially to fulfill the preferences of a single buyer, not of society as a whole.
It doesn't have to be this way. By building institutions with new incentives, we can change the world. Many mechanisms exist which can induce both the public and private sector to better serve the world. The most important and biggest "mechanism" to date to resolve market failures is the government. Among other failures, they are the main tool we have to deliver public goods. Public goods are those which are non-excludable and non-rivalrous in consumption, such as clean air, national armies and dry educational blog posts. These two characteristics create two corresponding challenges when a free market tries to provision them:
- The free rider problem is that no one is willing to pay for the public good.
- The information problem is that no one knows the optimal quantity of the public good.
Lot's of proposals exist for enhancing how governments provision public goods, including adjusted patent law, innovation inducement prizes and performance-based pay for government employees. Although some of these improvements are promising, there exists a radical opportunity to get rid of certain parts of the government entirely. Economic theory (and every visit at your local municipality office to renew your passport) teaches us that governments are not the most efficient organizations, and yet government spending as a percentage of GDP has risen steadily over the past centuries. In this post, I aim to clarify how mechanisms can incentivize private individuals or organizations to provision public goods, which may even reverse this trend.
Government spending in early-industrialized countries has increased steadily, but is that really optimal?
How can preferences be expressed at scale?
In order to provision public goods, we first need to know what people want. Most of the mechanisms I will discuss below make use of two primitives which allow people to express preferences: votes (a given right to express your preference) and money (an earned or bought right to an asset with which you can express your preference). There are proposals to mix the two regimes, in which case the mechanisms still apply.
There are countless other methods to learn people's preferences: discussions, debates, online fora, social media platform classification algorithms, etc. Although it is harder to use them to aggregate the preferences of large groups of people, it is important to keep using, improving and expanding these methods as they still play a hugely valuable role in society:
The actual goal is not to take current preferences and translate them into the right outcomes in some Coasean or Arrovian sense. Rather the goal is to encourage better and more reasonable preferences and also to shape a durable consensus for future belief in the polity. - Tyler Cowen
The list of mechanisms
Lots of academic literature exists on the optimal provision of private goods. Many of the characteristics of perfect competition would improve the provision of public goods just as well as they would improve the provision of private goods (given many assumptions, this is economic reasoning after all). For example, do you know of any method to improve people's access to true information? Great! That is an excellent method to improve both private and public good provision.
What about mechanisms which are specifically designed for public goods? Over the past decades, a variety of ideas have surfaced, most of which revolve around crowdfunding schemes. The ideas I could gather are as follows:
- Merging free-riders: when an organization owns a large piece of the economy's assets, this reduces the amount of externalities and incentivizes the organization to take care of all the assets. Unfortunately, this also directly exacerbates the problem of monopoly power so it is not a universal solution.
- Excluding people who do not contribute to public goods. This model, as well as many other case studies, explains how self-organizing communities such as unions and even your local neighborhood watch can be successful in collectively providing public goods.
- Assurance contract or provision point mechanism: this is meant to solve the free-rider problem. When a crowdfunding target has not been reached, donors receive back their committed funds
- Dominant assurance contracts: when a crowdfunding target has not been reached, donors receive back their committed funds plus a bonus.
- Money-back guarantees: when the result from a crowdfunding action doesn't deliver, donors receive back their funds
- Proportional rebate of excess funds: when a crowdfunding target is exceeded, donors receive back the excess funds.
- Crowdmatching: this is a new crowdfunding scheme which requires contributors to increase their donation amount when a project attracts a larger amount of contributors. The team behind this scheme is still actively ironing out the details.
- Matching funds: private donors can offer matching grants based on how many votes or donations a project receives. In recent years many large corporations have organized fundraisers which match people's donations by the same amount, until the fund is depleted.
The promise of matching funds
This last mechanism deserves a more thorough explanation. There are a variety of ways in which a matching fund can decide under which conditions a project would receive a match. A typical case is that people's donations are matched by a fixed ratio (as is commonly done by corporates). Governments essentially also provide large-scale matching funds, as donations are tax-deductible up to a certain amount in many countries. A matching fund could also match people's donations based on their votes (as is commonly done in prizes and award settings). One can imagine many other matching mechanisms relying on a combination of cleverly designed voting and donation systems, as well as expert judgement.
In a paper from 2018, Buterin, Hitzig and Weyl show how a so-called Liberal Radicalism or Quadratic Funding mechanism can be a unique general method to fund public goods. Public goods can be crowdfunded by using a matching fund which tops up a donor's contributions in proportion to not just the total donated amount but also the number of individuals contributing. Specifically:
where is the contribution of one donor. For example: a project receiving a single donation of €1000 would be matched with €1000 from the matching fund. However, a project receiving 1000 donations of €1 would be matched with €1 million! For more background you can check out this primer. Although the mechanism is vulnerable to collusion, it has already been tested extensively in the field and seems to have promising results.
Taking into account future people's preferences
These mechanisms can be extended to take into account the preferences of entities in the future. Although satisfying future people's preferences is inherently harder, as it involves an amount of prediction with slower feedback loops, this doesn't mean such mechanisms are worthless. On the contrary, getting better at predictions in order to ensure people take into account the long term impacts of their actions is exactly why such mechanisms can be very useful. That is exactly the challenge which public goods funders should be trying to tackle anyhow. Two examples:
- Long term assurance contracts can lock up people's funds for a long time period, allowing or requiring that future entities join today's donation proposals.
- People vote on past (e.g. 2 year old) projects, which are funded up-front by selling shares of the results of these deferred votes; by buying shares people would be both funding the projects and betting on which project would be viewed as successful in 2 years' time.
To summarize, future people's votes or donations can indicate which past shares, votes or donations should be "activated". Extending these mechanisms to 10-year or 100-year time horizons create many new interesting design challenges: what happens if a historic donor dies, are the donations passed to their family? What happens if the historic donor changes their mind on the project? There is a lot of legal and product designing to do.
From theory to practice: testing is key
So how can we figure out which of these ideas should be implemented? And where can people currently make a difference?
Understanding which mechanisms to implement with which parameters is a challenge. The optimal mechanisms, if they even theoretically exist, are heavily context dependent. Creating large scale randomized control trials may be prohibitively expensive. But we can try to take an evolutionary and entrepreneurial approach: keep experimenting and copy/expand what works based on people's own evaluations. Over time, more funds can be redirected to promising mechanisms. Fortunately, the mechanisms mentioned above have a simple parameter which they can optimize for: if they collect donations from a large number of people they must be worth something, and the chance increases that profit-maximizing entities deliver goods which are currently overlooked.