A Test of Dominant Assurance Contracts

The free rider problem is a challenge to the market provision of public goods. In my paper on dominant assurance contracts I use game theory to show how some public goods can be produced by markets using a special contract. In an assurance contract people pledge to fund a public good if and only if enough others pledge to fund the public good. Assurance contracts were not well known when I began to write on this topic but have now become common due to organizations like Groupon and Kickstarter which work on this principle (indeed, I have been credited with the ideas behind Groupon although sadly for my bank account I don’t think that claim would stand in a court of law). Since no money is paid unless the total pledges are high enough to fund the public good, assurance contracts remove the fear that your contribution will be wasted if other people fail to contribute. However, if you don’t believe that others will pledge there is no reason for you to pledge and if you don’t pledge others have no reason to pledge either so your beliefs will turn out to be accurate. With even low transaction costs the no pledge equilibrium seems quite likely.

What a dominant assurance contract adds is that the entrepreneur agreeing to produce the public good if k or more pledge also agrees that if fewer than k pledge he will pay a prize to those who did pledge. Pledging is now a no-lose proposition–if enough people pledge you get the public good and if not enough pledge you get the prize. A contract like this makes it a dominant strategy to pledge and so the public good is funded. (See the paper for details).

Recently Jameson Quinn tried the idea out with a campaign on Quora. The public good was to contribute to The Center for Election Science. Each pledger agreed to give $60 to the Center if 20 or more similarly pledged and Jameson agreed that if fewer than 20 pledged he would give each person who did pledge $5 and they would have to pay nothing. (This is the essence, Jameson actually modified the dominant assurance contract to offer slightly different terms at 19 and over 20. He has some interesting ideas on this score, see the post for details). So what happened?

Success! It was a nail-biter as the final 3 pledges came in only in the last half hour. Remarkably, the lure of $5 for nothing helped to produce a public good which the pledgers all wanted but might not have produced had the incentive to free ride not been counteracted.

Kickstarter has made assurance contracts familiar, perhaps the next evolution of funding mechanisms will do the same for dominant assurance contracts.


An "assurance contract" is a cool concept ... but wouldn't such a contract suffer from high transaction costs, since they require a third party to enforce its terms?

"wouldn’t such a contract suffer from high transaction costs, since they require a third party to enforce its terms? "

I'm missing the logic of your argument.

It doesn't take more than an escrow account and a website to do an assurance contract.

Is there any special reason why escrow accounts and websites are immune to competition?

Kickstarter handles this by only paying transaction costs if the project is successfully funded. So transaction costs are higher when funded, but are zero when not funded. So again, no downside to pledging.

Two of those last three nail biter pledges may have been Jameson Quinn himself? Else he loses $90 anyways.


Why not make the reward by drawing lots, and the winner gets ... oh, say, a ticket for a lottery where he might win a bazillion bucks, but where if he doesn't win the ticket price anyway goes to good causes?


You should have patented your idea then sued them after they got their business set up for infringement.

Maybe he could have done this, but I'm not so sure if he should have done it.

Would something like this be workable for action on global warming? Non-binding UN resolutions notwithstanding.

This was my fundraiser. I can assure you that I was not one of the pledgers. There were 13 pledges from 12 people I've never met and who have never to my knowledge supported or even heard of the organization in the past; 2 from 2 people I've never met who are already involved with the organization; and 5 from 5 members of my own family (cousins and such). There was also one extra pledge 8 minutes after the deadline from a person I've never met. I understand that you might be suspicious about the family pledges, but I regard them as legitimate.

In my opinion, the "dominant assurance contract" functioned in three ways here to help a successful outcome. In order from least to most important:

-It encouraged people to pledge by showing that I myself was serious about wanting to support the organization. Not so much "wow, I could get $5 for free"; more "wow, if he's willing to send me $5, he must be serious". Note that the latter is still, I'd guess, a lot more effective than "wow, if he's donating $5 of his own money to the organization, he must be serious". The possibility of $5 going into your pocket makes that money realer for you than the certainty of it changing hands between two third parties.

-It was a successful "gimmick", making this fundraiser more interesting and fun than it would have been otherwise. The various posts I wrote on Quora about the fundraiser got upvoted by several dozen users, and the ones that focused on this gimmick were more popular in general than the ones that focused on the target organization.

-It got me off of my ass to look for donors. For instance, I was too embarrassed to write to my extended family, until I already had 10 donors and it was clear that I would lose money unless I did so.

The "must be serious" signal is the only reason I can think of for warranties on standard consumer products (that most purchasers could afford to replace) to make sense.

Build model airplanes, says the fairy. Next thing you know, money is missing from the drawer and your daughter's knocked up. I've seen it a hundred times.

Based on my experience with this fundraiser, I think that this idea has its upsides and downsides.

On the upside, it does help you raise more money than you would otherwise, if you set your target well.

On the downside, it could misfire badly if you are raising a poorly-defined amount of money from a poorly-defined group.

One scenario I can imagine it working very well is an open source/shareware phone app. Consider this plan:
-Keep statistics on your entire user-base
-Take a sample of 5% and ask them to name the highest price they'd be willing to pay to fund continued development. Remind them that it's OK for them to be honest because you'll never ask the sample participants for money, and so the higher the number they honestly say the more money you'll be able to raise.
-Set your goal using this data to estimate the other 95%. If that's enough to fund development plus risk (ie, an insurance pool with similar DAC initiatives) then run a DAC to hit your goal.
-Ideally, the payout for the "D" part of the DAC would be some digital good you control, rather than $5 cash.
-My experience also shows that it would probably be worthwhile to have the payout decay in some way. For instance: the first 100 people to pledge get 30 days free of netflix, the second 100 get 29 days free, the third 100 get 28 days free, etc. The decay factor would ideally be nearly continuous, as in that example, but could also be discontinuous (T-shirt for the first 200, nothing for the rest). That would create a "hurry up" psychology that would boost donations.

There's nothing illegitimate about getting pledges from close associates or making pledges yourself. NFL teams do this all the time by, say, buying the last few thousand tickets remaining for their home game so it won't be blacked out on TV.

Isn't an alternative to an assurance contract simply an election or referendum for the purchase of a public good.

Consider: if I and 50% of my colleagues vote to support a bond for a school, the school gets built.

Public goods manage to get built via the election process as well. I don't see why this would supplant an election.

I can see this for private goods where there are externalities and you could not use a public election process, but is it any different than a wealthy person committing to contribute $x dollars if so many people contribute to their public tv or radio station?

Elections are good for spending other people's money for public goods. An assurance contract is good for getting people to voluntarily spend their own money for public goods.

Some of us think that voluntary actions are nicer than forcible ones.

eddie, The definition of a public good is one which has positive externalities for everyone, but where there is an incentive to free ride.

An assurance contract, in the way you interpret it, is more akin to a club, where the participants are the beneficiaries and there are no positive externalities to others.

Some of us think that there are free riders for public goods, while others think that a club will just serve them because there are no spillovers--but then, that is something that is not a public good by definition.

Look up the definition of public goods.

Alex specifies that it is a public good

And, therefore it is what whatever Alex says it is. Is that the way one thinks?

Look below in my response to Andrew as to what is a public good, and how it is defined by others.


You sure do tell people to look up a lot of stuff you have no business doing.

Andrew', So you do not have to look up the definition of a public good, here it is, delivered as a public good to you:

"In economics, a public good is a good that is both non-excludable and non-rivalrous in that individuals cannot be effectively excluded from use and where use by one individual does not reduce availability to others.[1] Examples of public goods include fresh air, knowledge, lighthouses, national defense, flood control systems and street lighting. Public goods that are available everywhere are sometimes referred to as global public goods."

Yes, Bill.

I already knew the definition was completely different than what you thought it was.

"Elections are good for spending other people’s money for public goods"

eddie, I'd correct your statement to read "Elections are good for spending other people’s money" as there is no assurance whatsoever that the 50% Plus Bill theory results in any public goods whatsoever.

Also note that we don't even have direct democracy. So, almost nothing Obama has or will do even meets the 50% Plus Bill threshold.

Taxation is considered a necessary but insufficient condition for producing the optimal level of a public good, although Alex falsifies even that mistaken notion.

The basic "assurance contract" or "soft assurance contract" idea is indeed similar to "matching donations" such as in public radio fund drives, but there are a couple of differences:

1. It doesn't artificially divide people into wealthy "matchers" and everyday "donors". Everyone is on the same footing, which makes more sense from a game theory point of view because there's no need to waste time worrying over which way is a more effective donation.

2. It can be structured to give any multiplier, not just 2x or 3x. The tradeoff is that the higher the multiplier, the more closely you have to hit your goal. I structured my fundraiser so that there was a 10x multiplier if there were 19 or 20 donations, and around a 2x multiplier if there were 21-25 donations.

Sure, it's an alternative, as long as all the participants agree to abide by the majority's decision and pay even if they personally didn't want it. But it may be tricky to get them to agree on that. Some might downplay their desire for the public good, in hope that the others get together and pay for it without them.

Another issue is that people's desire for a good might vary. The public good might be worth a lot more to some people than to others. In that case it's good for participants to be able to pay closer to what it's worth for them, rather than a fixed amount. That wasn't an option in this case, but I suspect Tabarrok's scheme can be extended nicely to support it - a binding vote certainly doesn't support it.

Tabarrok, how about that? If instead of fixed, equal amounts, everyone can pledge what they like, and if the goal isn't reached they get paid 1/20 of what they pledged. What assumptions in your original paper break down if you extend it this way?

I think the scheme would have to allow for variable payments in order to be a viable kickstarter-like service. Without different pledge levels, I think many successful kickstarter projects would have failed.

"Isn’t an alternative to an assurance contract simply an election or referendum for the purchase of a public good. "

What? No, not even close. An assurance contract is purely voluntary and everyone who agrees to it, voluntarily commits their resources.

In an election, the winning side forces everyone to commit resources.

"In an election, the winning side forces everyone to commit resources. "

Correction: in an election, the winners force the losers to commit resources.

All it takes to prove to yourself that elections are the best externalizing force is for one to open their eyes. Our government is almost completely out of the public goods business.

Ah, everyone has a 50% Plus Bill theory and this is yours.

Take an extremely naive political model, throw out everything known about public choice and then you'd be right.

No, wait. You'd still be wrong. Why would your hypothetical 50% Plus Bill care about something being a public good? Why wouldn't they just vote themselves largesse as is plainly the case?

A commitment to pay provides an incentive against rational ignorance among participants.

The poor, uneducated voters of my municipality have gone on a binge for unnecessary projects financed by 30 year bonds. How many voters understand that the principal plus interest will be paid with their, or their children's, taxes?

The poor, uneducated voters of my municipality have gone on a binge for unnecessary projects financed by 30 year bonds. How many voters understand that the principal plus interest will be paid with their, or their children’s, taxes?

It's not their money, it's someone elses money.

Their children are the beneficiaries...they are educated in that school.

Ask yourself this question: does IBM issue a bond to pay for capital goods that will be used over time?

Actually there already is a school. The district will move it to a brand-new structure a few miles away. There was nothing wrong with the existing structure.

The capital is property taxes. The school and its employees are present consumption. Pay is above average while student outcomes are below average.

In school districts where I have lived, votes for school board members and bond issues are held on special elections days -- not on primary days, not on general election days. The voters who participate are mostly parents of students in the schools. The rest of the electorate is either unaware of or uninterested and does not participate.

The "no pledge equilibrium" seems to be refuted by Kickstarter's track record. While the concept makes sense, perhaps in practice it's simply not relevant, similar to the way that adverse selection has not put the insurance industry out of business.

Perhaps there are some projects on the margin that would benefit from offering an incentive to pledge even if the funding drive fails, projects where the project is neither nearly certain to succeed nor nearly certain to fail. I suspect that most projects aren't like that.

Also, strategically, using this as a gimmick is perhaps the wrong way to go about getting your project funded. The most important part of a Kickstarter is the marketing effort you put into your Kickstarter. Projects get funded because they attract attention, excitement, and word-of-mouth. You want EVERYONE to be thinking "wow, this is a great idea, and I really want it to succeed, and I want to tell everyone I know about it because they'll think it's a great idea too." Adding the "... and if it doesn't get funded you win a prize" clause would generate some buzz the first few times simply because it's an unusual idea, but it would be the WRONG KIND of buzz, because it would cause people to think about all the reasons why your project might NOT get funded. It would be like Coke putting up billboards saying "... and if you don't like the way our soda tastes, that's okay, we understand, that's a perfectly reasonable thing, not everyone likes it after all, so we'll pay you a dollar. Be sure to tell all your friends." You will NOT sell many sodas that way.

Wouldn't a dominant assurance contract result in very small projects because of the risk faced by the entrepreneur?

Say there's a 10 million dollar project to which many people contribute 9 million. The entrepreneur either has to pay the prize or kick in the million.

That is quite a bit of risk for the entrepreneur. More people would pledge to the projects that exist but there would be fewer projects.

Totally agree. I'm not at all understanding why someone would do this for the reasons you state plus the point eddie makes about this being a negative, not a positive association with a campaign.

Entrepreneurs typically don't have lots of extra cash sitting around to take chances with: that's why they're fundraising.

Last point: unless the campaign is soliciting a specific limited donation amount per donor, offering $5 to each donor doesn't make much sense. I'd offer $1 to every single campaign on your website if that's how you're going to set up the rules.

So are pledgers not effectively betting that non many other people will pledge? I don't know what the consequences of that are, but it seams strange.

This is bullshit.

If it is indeed a public good, then not pledging is still always better if you believe others are going to pledge, since you still get the good (because by definition everyone gets a public good) and pay nothing.

The only ways to fund a public good are with compulsory taxation, by somehow making it into a non-public good or by somehow arranging things so that people pledge because it makes them feel good.

BTW, most things on Kickstarter are NOT public goods (or at least they aren't assuming no copyright violation): they are simply goods that benefit from economies of scale, and Kickstarter removes the risk of sales volume being too low to benefit from them.

Also, most people don't have the huge capital that it would take to give out free money if their campaign fails and are not willing to take the risk.

In the paper, Tabarrok shows that when N=K it's dominant and when K<N it's Nash.

Haven't read the paper, but it must be either wrong or have additional assumptions.

Most simply, if I know that there enough other people that are going to pledge, then not pledging is my strategy, since I get the public good anyway and I don't pay for it.

In other words my_utility(I don't pledge, others pledge) > my_utility(I pledge, others pledge), so (I pledge, others pledge) is not Nash equilibrium and (I pledge) is not a dominant strategy.

What happens in the (I pledge, others do not pledge) case where the reward is proposed is in fact totally irrelevant.

Someone who has not read the paper should not be calling bullshit so readily.

Your comment is bullshit.

Taxing does not cause boondoggles to magically become public goods. Our government taxes the shit out of everything and funds almost no public goods.

Having people commit to taxing themselves is far closer to the ideal than the 50% Plus Bill plan.

It may be closer to the ideal but it's much farther from the realistic or possible.
This is where libertarianism fails. Most libertarians I know are smart, thoughtful, ethical people, but they assume everyone else is that way as well. In a world full of smart ethical people, a purely voluntary society of millions might work. In the world we all live in, someone's gotta be in charge, and there gotta be rules and institutions and taxes.

I do appreciate trying to move those people and rules and institutions in a libertarian direction, but there's no chance of a pure libertarian society any more than a pure communist one.


You have correctly observed that the world contains some bad people. And that there are other people who have less cognitive ability than others. But I am puzzled by your therefore statement;

"In the world we all live in, someone’s gotta be in charge, and there gotta be rules and institutions and taxes"

Given your observations, surely the therefore statement is;

"We must minimize the risk that these bad and/or cognitively challenged people get put in a position of power on us, by moving to a non-monopolistic (or market determined) solution rather than a government/technocrat defined solution wherever possible." The example I like to use is Singapore and Cambodia. Both south east asian countries so similar cultural factors. The Government directed solution worked out well for Singapore, but not so well for Cambodia. There is a very significant risk with having very strong governments, even if sometimes it works out.

By the way, libertarians are not the same as anarchists, so support "rules and institutions and taxes". They just want them to be minimized rather than maximized.

Hell, anarchists are fine with rules and institutions. It's the state they're down on.

The real question, of course, is how can PACs employ assurance contracts to get more campaign donations.

You'd need to be careful with the reward size for those who donate on a non-completed project. It would have to be considerably smaller than the "reward" in the form of the service being provided on a completed project, otherwise you'll draw in donors hoping to game it for money.

This made no sense to me until I realized that the analysis is restricted to *excludable* goods.

If it is excludable, it is by definition, not a public good, despite Alex's statement: "...I use game theory to show how some public goods can be produced by markets using a special contract."

It is more like the economics of a club.-

It's not excludable. Why would you think that? The example given, for example, contributing to the Center is not excludable.

If it is not excludable, then I can get a free ride.


And if everyone else realizes that they can get a free ride, then...

I think Kickstarter figured out something better than dominant assurance contracts. They figured out Pareto Assurance Contracts.

This assuming the entrepreneur didn't miscalculate in their reward promises - which has happened before.

In this contract, the people who want the "public" good (in the case of kickstarter, the availability of a private good on the marketplace for purchase) - they can choose to kickstart the project and get the advantage of being given an extra reward for funding the good. And the entrepreneur is willing to produce and sell the good at the lower than retail price if they get enough orders. Everybody wins, and you don't have to worry about the existence of arbitragers who will step in at the last minute to collect rents from unpopular dominant assurance contracts.

Comments for this post are closed