# How Sponsoring Giving Games Can Multiply the Impact of Your Donations

Would you like to multiply the impact of your giving? Giving Games may be just what you’re looking for. This post will explore how the economics behind Giving Games can help you leverage your donations.

I’ll examine both the costs and benefits of hosting Giving Games, and show that while the costs are minimal, the potential benefits are extremely impactful. You will likely find the implications of the number crunching below surprising. The numbers demonstrate that even if Giving Games inspire only one participant out of a hundred to give more effectively, they still provide significant leverage for your charitable donations.

**What would it cost you to sponsor a Giving Game?**

Not much. If you want to sponsor a Giving Game, here’s the good news: you can do so with money you already planned to give to charity. Instead of donating the money directly to charity, you can have Giving Games participants donate that same money on your behalf.

I’ll give you an example: Let’s say you plan to donate $500 and to split that gift evenly between your favorite two charities, Charity A and Charity B. The net result would be that each charity receives $250.

Alternatively, you could sponsor a Giving Game that would allow 25 people (likely more) to choose whether to give your $500 to Charity A or Charity B. The same total amount of money still goes to your favorite charities. The only difference is that the participants will determine how the money is allocated between Charity A and Charity B.[i]

Your cost from this Giving Game is basically a function of how much you care about the voting outcome and the likelihood that the voting will deviate from your preferred outcome. If you’re indifferent as to how the money will be allocated between the charities, you essentially don’t bear any cost. Sponsoring this Giving Game is free relative to your default.

We can use an extreme example to help illustrate the cost of sponsoring a Giving Game for someone who does care about the precise outcome. Consider a donor who thinks Charity A is highly effective but that Charity B has negligible impact. This donor’s preferred outcome would be for Charity A to get all $500, and each dollar Charity A did not receive would be considered wasted. Let’s assume the voting structure in this scenario is “winner take all” and each charity is equally likely to win. This donor would incur a cost of $250, half of their intended gift. This represents the expected value of a 50% likelihood that the $500 would be wasted. While this extreme example suggests a 50% cost, in practice most sponsors would have significantly lower costs.[ii]

In either scenario, the costs of sponsoring Giving Games is low, and for some people, negligible.[iii] In return for these costs, sponsors gain a remarkable opportunity for leverage: philanthropy education.

**The Economics of Philanthropy Education: The Basics**

In the previous post in this series, I described how each of the key characteristics of a Giving Game is tailored to encourage people to do more good with their giving, either by giving more or by giving more effectively, preferably both. Now I’ll lay out a model for thinking about the payoff from this “Philanthropy Education,” and the opportunities for leverage it creates for Giving Game sponsors.[iv]

Most Giving Game participants are college students, so we’ll model the reward from getting one of these students to give better. One of the reasons we target this demographic is because it offers the opportunity to influence an entire lifetime of giving. It may seem strange to think an hour-long experience could influence the way someone gives decades later. But donors are quite habitual and people typically give to the same charities year after year. [v] So if we can influence young (or prospective) donors to give better at an early stage of their giving, there’s a good chance they’ll continue with this behavior.[vi]

I encourage you to read the following section, which will lay out the details of the model. But I’ll summarize the conclusions here, in case you’d prefer to skip the math. The reason that philanthropy education is so appealing is that a typical college student will give tens of thousands of dollars over their lifetime (in today’s dollars, i.e. adjusted for the effects of inflation). Even if we further adjust for the fact that future donations are less valuable than current donations, and assume that we might only influence a portion of this person’s giving, we still have the opportunity to influence thousands of dollars of giving if we succeed. This dwarfs the magnitude of the costs associated with Giving Games, suggesting that we’d only need to convert a small number of participants in order to create a substantial net benefit.

**The Economics of Philanthropy Education: The Details**

Our model requires a number of assumptions, and unfortunately we’re missing some key inputs. We don’t really know how many people Giving Games cause to give “better”, or how much better these “Converts” will give as a result of the experience. Since this program is quite young, not enough time has passed to measure the long-term behavioral changes we hope to create (though we’re working to improve tracking of both short-term and long-term results.)[vii]

Since it’s so difficult to estimate these numbers accurately, we instead try to show how many people we would need to convert to better giving in order to break even, i.e. the conversion rate at which the costs and benefits would be equal. If our break-even conversion rate is lower than what we could plausibly believe our *actual *conversion rate is even under conservative assumptions, then we know sponsoring Giving Games is an attractive proposition.[viii]

Below I lay out the variables in our model and calculate the break-even conversion rate using a set of very conservative assumptions. I invite you to test your own assumptions using this calculator. (For best results, download the document from Google Drive in order to read the instructions and use the interactive components.)

**How much incremental money does a Convert donate to highly effective charities?**

Our breakeven conversion rate will only make sense in the context of a particular definition of “conversion”. I prefer to conceptualize a Convert as someone who makes a moderate shift in giving behavior that leads to more money going to high impact charities. We’ll define this Convert as someone who gives a minimum of an incremental $1000 to effective charities each year as a result[ix] of participating in a Giving Game.[x] This threshold is meant to approximate a “moderate” change in giving.[xi]

This $1000 threshold is more of a definition than an assumption; as such, it’s neither conservative nor aggressive. We need this definition in order to attach meaning to the conversion rates we’ll calculate. However, I’ll introduce a significant amount of conservatism by treating $1000 not as a minimum but as an average in these calculations. In other words, our model will ignore the benefit of any giving above the minimum threshold.[xii]

Note that this threshold, and in fact this entire model, is in terms of today’s dollars. I assume that participants will increase their giving in line with inflation (which lets us ignore inflation from here on out).

**We prefer present donations to future donations, but by how much?**

Would you rather your favorite charity get $100 today, or $100.01 five years from now? Even though there’s more money in the future, you’d probably prefer the donation be made now. How about $100 now vs. $10,000 five years from now? In this case, I’m sure you’d be willing to wait.

As these examples show, there are tradeoffs we can make based on how we value time and money. All else being equal, we’d prefer donations be made sooner rather than later. But our preferences depend on how much time is involved and the relative amounts of money.

A “discount rate” represents the rate at which we devalue donations because they take place in the future. An annual discount rate of 10% implies that we’d be indifferent between $100 given today and $110 given one year from now.[xiii] This rate compounds over time: at a 10% discount rate we’d be indifferent between $100 today and $121 given two years from now.

The benefits of philanthropy education come from improvements in future giving. The discount rate is therefore a very important variable in the model, since it dictates the value we place on those future donations. I’ve used a 10% estimate in an attempt to be highly conservative,[xiv] but since discount rates are difficult to estimate, and relatively subjective, I suggest you test out a range of your own assumptions.

**When does a Convert start “giving better”?**

I’ll assume our participant is a college student in their second to last year of school, which would be pretty typical of our experience. If we expect they’ll start giving once they graduate and get a job, that will be two years after they participate in a Giving Game.

**How long does a Convert give better?**

I’ll use 50 years of lifetime giving as a conservative assumption. That would be consistent with our participant graduating and starting to give in their early 20s, giving into their early 70s, and leaving no bequests or other large end-of-life donations (meaning we’ll ignore a large source of potential giving).

**How does a Convert’s incremental giving vary over time?**

We’ve already made an assumption about average incremental giving per year. But realistically, people are likely to give less than that average early in their careers and more than average later in life. Since future donations are less valuable than present donations, it’s important to account for this dynamic.

I’ve assumed a Convert’s giving starts at $0 and ends at $2,000 (twice the average yearly gift), increasing linearly between those points.

**What is the cost per participant?**

The previous assumptions will all be used to estimate the benefits of converting someone to improved giving through philanthropy education. In order to calculate our break-even conversion rate, we also need to make an assumption about our cost per participant.

I’ll assume our cost (total donation amount divided by number of participants) is $25 to be conservative. In our experience, the cost per participant is generally more in the $10-$20 range.

As discussed earlier, this “cost” isn’t like most costs. The money will go to charity, and possibly to the same charities a sponsor would otherwise give to. But for these purposes, we’ll ignore this benefit and consider the money used to incentivize participation in a Giving Game to be a true cost.

**What do these assumptions imply?**

When we crunch the numbers, we find that even using these conservative assumptions, it would be hard to believe that Giving Games aren’t providing significant leverage. **We only need to get .76% of participants to give moderately better--just one participant out of 132--in order to break even**.[xv] If we can convert people at a higher rate, even more money will flow to effective charities. And given the conservatism built into our estimate, our real breakeven rate is likely even lower.[xvi]

I can’t yet tell you what the actual conversion rate is for The Life You Can Save’s Giving Games program. But I’m quite confident it’s over 1%, and I’d be shocked if our conversion rate didn’t rise going forward.[xvii] The body of evidence we do have access to suggests that we’re reaching people, and it’s hard believe we’re reaching so few that it would render the effort unproductive. I’ll discuss this evidence more in a subsequent post, but for now consider the following data points:

- Roughly 80% of the participants we’ve surveyed self-report that they expect the experience to change their future giving behavior.So we’d exceed our breakeven if just 1 out of 80 of these people follow through.[xviii]
- Giving Games “facilitators” have given overwhelmingly positive feedback, in both words and action.After leading one Giving Game, facilitators often look for opportunities to run more.John Sturm, the co-president of Harvard Effective Altruism, says Giving Games “have been by far our best way to generate interest in Effective Altruism.”Experienced professors, including those who have specifically taught philanthropy, are among the most enthusiastic advocates for Giving Games.
- Giving Games are spreading through word of mouth.We regularly hear from new Effective Altruism groups who want to run Giving Games after hearing other groups describe their experiences.[xix]
- The Life You Can Save has gained valuable volunteer help from several people as a direct result of their participation in a Giving Game

Considering what we know about Giving Games and the incredibly low bar for success, it’s easy to see why we’re so excited about their ability to create leverage.

**Putting it all together**

I hope this blog post gets you excited about the idea of sponsoring a Giving Game. And if you are excited about the prospect of multiplying your donation, it’s quite reasonable for you to want to know how big that multiplier effect might be. To project this figure, we’ll need to make a specific estimate of our actual conversion rate, something I’ve resisted so far for the reasons explained.

Let’s assume our actual conversion rate is 2%. That’s a plausible rate, though I certainly wouldn’t be shocked if it were a bit higher or lower. A 2% conversion rate would imply that one out of 50 participants, or one out of 40 who self-report they expect the Giving Game to change their perspective on giving, would give moderately better.

Using these assumptions, a 2% conversion rate implies you could leverage each dollar you give by 2.6 times by sponsoring a Giving Game. That figure assigns no value to the donation the participants will make with your money; if you’re as happy with their decision as your own, it would raise your multiplier to 3.6.[xx]

Maybe Giving Games don’t offer that kind of multiplier -- maybe they only double the impact of your gift. Even so, how would you rather double the good that comes from your charitable giving? By spending twice as much money, or by helping teach the next generation of donors how to give better?

*If you’d like leverage your giving by sponsoring Giving Games, please contact givinggames@thelifeyoucansave.org to learn more about this opportunity.*

[i] Some Giving Games use a “winner take all” voting style, while others split the donation between the charities in proportion to the votes. In this example, the sponsor would likely want to use a proportional voting system, as the results would more closely mimic the sponsor’s default 50/50 allocation.

[ii] Someone sponsoring a Giving Game featuring a highly effective charity and a “worthless” charity presumably believes this structure will be of more educational value than a structure featuring two high impact charities. So it may be more realistic to view the ex-ante probability of Charity A being selected as more like 70%, rather than our naïve guess of 50%. In this case, a donor would bear a probabilistic cost of $150-- 30% of their $500 total donation.

[iii] Some sponsors may also bear overhead costs. If you sponsor a Giving Game that you organize yourself, you should include any time or money you put into executing the event as costs. However, if you fund a Giving Game organized by The Life You Can Save, you’ll bear no such overhead- we’ll use your full donation to sponsor Giving Games.

The Life You Can Save’s Giving Game program does have non-donation costs (primarily employee salaries), and we include these in our internal analyses. However, it’s important to understand how that employee time was used. Significant time went into producing Giving Games resources that should provide lasting improvements; our resource library is one example. Due in large part to these investments, we expect the ratio of donation costs: non-donation costs to rise over time as we become more efficient at running Giving Games.

[iv] Since this section looks at the payoff from philanthropy education, it’s not specific to Giving Games (aside from the cost per participant) but can be generalized to any approach to improving giving behavior. Giving Games are our preferred method of philanthropy education because donors can fund them through existing donations, we believe the structure of Giving Games is particularly well-suited to promoting better giving, and we think the model is more scalable than other approaches.

[v] The Money for Good study found that “almost 80% of all gifts made are ‘100% loyal,’ meaning that there is a virtual certainty that these gifts will be repeated next year”. One of the recommendations of this report was to “invest in the lifetime potential of donors, not just this year’s potential.”

[vi] Other non-profits already try to benefit from this dynamic. For example, universities are known for investing fundraising resources into trying to get young alumni to give small gifts, knowing that they’ll likely give more over time. A study by Meer (2013) suggests this is a winning strategy, finding “the large magnitude of the effect of being a frequent giver when young suggests that nonprofit organizations in general and universities in particular should give serious consideration to devoting additional resources to raising participation rates among young potential donors. Even if the benefits are far in the future, the effects are large enough to justify incurring some losses in the pursuit of gifts in the present.”

[vii] We’re in the midst of our most systematic survey collection to date. We’re also actively seeking research collaborators to conduct a randomized, longitudinal field test assessing Giving Games’ impact.

[viii] Determining *whether* Giving Games can leverage donations is also a more relevant question than *to what degree *Giving Games can provide leverage. It doesn’t really matter if Giving Games can double the impact of a gift or triple it; in either case, a strategic sponsor should fund them.

[ix] By “incremental”, I mean relative to a counterfactual world where someone didn’t participate in a Giving Game. Note that the Giving Game itself does not need to cause all the behavioral changes involved, it merely needs to start someone on a path that leads to the different outcomes. For instance, a Giving Game might only convince a participant to sign up for a mailing list for charity recommendations, but being on that mailing list may lead to them giving better over time.

[x] This figure is a *net* amount flowing to effective charities. $1000 of additional giving going to effective charities or $2000 that would have otherwise gone to a charity half as good would each count as $1000 in incremental money.

$1000 is the minimum average annual gift for a Convert, but they might give less than that in some years and more in others. For example, someone who gives $900 in Year 1, $1000 in Year 2, and $1100 in year 3 would classify as a Convert since their average annual gift would be at the $1000 threshold. See “How does a Convert’s giving vary over time” section for further elaboration.

[xi] Some reference points for interpreting $1000 in annual giving:

- In 2013, per capita giving by U.S. adults was $1,016 (Source: Giving USA).Note that as presumptive college graduates, we’d expect most Giving Game participants to earn and give more than average.
- $1000 would be 1% of income for someone earning $100,000 a year.The average person at that income level gives ~3% of their income to charity, so 1% would represent a moderate shift.

[xii] The other reason to use this methodology (besides conservatism) is that defining a minimum threshold makes it much easier to conceptualize what our conversion rates actually represent. Using the average amount given by a Convert rather than the minimum amount would make the math tie out more neatly, but it would horribly muddle our mental models of what makes someone a Convert. Any model that can capture and assign probabilities to the vast range of possible outcomes for a Giving Game participant will be inherently complex. I prefer to use a simple threshold method that’s easier to understand and adds conservatism.

[xiii] This also implies we’d prefer $100 given today to $109 given in a year and we’d also pass up $100 given today in favor of $111 given one year later.

[xiv] It may be helpful to think of a discount rate as having three components: inflation, opportunity cost, and “other”.

Inflation doesn’t figure into our discount rate, since we’ve assumed people will increase their giving in line with inflation.

Our opportunity cost accounts for the fact that instead of donating today, we have the opportunity to invest that money and then donate more in the future. Long term real yields in the US are currently only ~.7%, so there’s presently limited expected return from low-risk investments.

The “other” component is a catchall. We can use it to account for things like uncertainty in the future, or the desire to provide help as soon as possible to jumpstart positive knock-on effects (e.g. malaria-free children growing up to be more productive workers).

When we look at the components, a 10% overall discount rate seems quite conservative. Our opportunity cost is probably under 1%, and even if we assume that we can increase our investment returns by incurring risk, that might only account for a few percent (and that strategy by definition comes with risk).

Similarly, the “other” bucket is unlikely to add more than a few percent to our discount rate. To illustrate this point, we know a 10% adjustment is likely much too high as it leads to some weird implications. Imagine an investor who could somehow guarantee returns of the risk-free real yield + inflation + 9%. Wouldn’t you want this person (who would basically be the greatest investor in history) to invest their money and donate the proceeds? Most people would think that’s a no-brainer, a 10% “other” adjustment would suggest you’d rather this person donate all their money today.

For a more thorough discussion of discount rates in giving, see here.

[xv] Since the discount rate variable is so critical (and so difficult to estimate), I find it helpful to look at results under a variety of discount rates assumptions:

[xvi] In addition to the sources of conservatism already mentioned, this analysis completely ignores the value of non-monetary effective giving, as well as the possibility of participants contributing some of the money at stake. Sponsors of this sort of Giving Game can use their donation as a seed or a matching gift to incentivize participants to contribute.

[xvii] The more Giving Games we run, the more we learn how to run them effectively. We’re also planning some specific initiatives to improve our conversion rate which I’ll outline in a future post.

[xviii] Some sample comments from a recent Giving Game survey:

- “Impact is the most important to me.Whichever charity impacts lives the greatest amount.”
- “It really opened my eyes to researching before you just give your money away, you want to make sure that it is being effective.”
- “I now realize the scope of many problems, and feel obligated to act on them… I would look for a charity that is most effective.People helped, cost efficiency, and severity of problem.”
- “I’ll definitely put more thought into charity effectiveness in the future...Probably organizations that help internationally.Important factors: severity of problem and how donations help.”

[xix] Roughly 1/3 of the Giving Games sponsored by The Life You Can Save are facilitated by this sort of EA group. In many cases, if these groups didn’t run a Giving Game they’d perform some other form of outreach. So what we’d really like to know is the impact of a Giving Game relative to the alternative. We’re collaborating with Giving What We Can to get a sense of this relative impact, through a study that will compare the efficacy of Giving Games to speaker events.

While this analysis ignores the counterfactual behavior of these EA groups, we also ignore the offsetting effect of Giving Games on their other outreach events. For instance, if Giving Games help a group grow, and that growth makes all their outreach efforts more effective, we ignore that benefit.

[xx] You can examine the implied leverage multipliers under different discount rate assumptions in the table below: