Discussion Post: Can Gambling Increase Savings? Empirical Evidence on Prize Linked Savings Accounts

Call to Action: In these threads, we attempt to further the discussion of a key problem in this category and evolve our understanding of the domain space where research work has not yet answered the specific problem or question being considered. These posts are living documents, and it is our hope that the community will continue to contribute to their structure and content.

Background

Prize-Linked Savings (PLS) accounts could offer a safe, potentially profitable alternative to gambling according to a recent paper in Management Science. “Can Gambling Increase Savings? Empirical Evidence on PrizeLinked Savings Accounts” suggests that, based on a case study in South Africa, PLS accounts do not necessarily cannibalize from regular savings accounts, while directly contributing to a reduction in gambling.

Variations on PLS accounts are also beginning to emerge from the Decentralized Finance (DeFi) space from companies like PoolTogether. Often these projects use incentive mechanisms to promote saving and dissuade token holders from engaging in value extraction that might hurt prices or promote unwanted behavior. The following discussion also contains the results of a primary investigation into PoolTogether.

Prize-Linked Savings (PLS): A savings account that adds a periodic prize that is randomly awarded to a recipient(s) at a regularly scheduled interval. PLS accounts may also offer annual percentage yields (APY) in addition to prizes, in this case, the PLS would be offering two separate incentive mechanisms.

Million a month accounts (MaMa): A savings account in which depositors are entered into a monthly “Million Rand Prize” for keeping their savings in a MaMa account.


PoolTogether: A Decentralized Finance (DeFi) project offering a variation on a PLS account, in which various cryptocurrency tokens are pooled into separate contests. Pools may offer differing APYs. Prizes may be awarded weekly or daily.

Key Problem / Topic Area

In “Can Gambling Increase Savings? Empirical Evidence on PrizeLinked Savings Accounts”, Shawn Cole, Benjamin Iverson, and Peter Tufano looked at PLS accounts in South Africa to assess the effects of the new type of account on the local population’s banking and gambling behavior. A case study prepared by BTIG, LLC looks at PoolTogether, an example of PLS accounts being used to determine the benefit of rewards on the saving habits of small accounts.

Specific Question or Problem Statement

Can prize-linked savings accounts offer a viable alternative to gambling? Do prize-linked savings accounts cannibalize from regular savings accounts?

Approach / Methodology

The following charts break down the impact of PLS on various aspects of account growth during the South African case study conducted by Cole et al.

Growth Rates of Standard 32-day Savings Before and After MaMa

|602x304.05038064040673

Savings Balances of Bank Employees: MaMa Users vs. Nonusers

Effect of Winning Prize on MaMa Deposits

Effect of Local Jackpot Prize Winner on Local MaMa Demand

Growth of the MaMa Product

The following charts break down the PoolTogether depositors and their accounts by pool.

Starting Deposits compared to Current Deposits by Pool

Percentage of Depositors per pool

Values of Highest Winning Accounts and Lowest Winning Accounts

Annual Percentage Yield (APY) By Pool

Mean of Winning Account Value Compared to Highest and Lowest Value of Winners by Pool

Reward X Winning Accounts X Number of Depositors X Value of Accounts

SUM of Average Reward Value (last 12) Type of Reward
APY (at time of recording) Number of depositors Pool Daily Weekly Grand Total

0

42

GUSD

167.42

167.42

42 Total

167.42

167.42

0 Total

167.42

167.42

1.68

1129

UNI

2749.88

2749.88

1129 Total

2749.88

2749.88

1.68 Total

2749.88

2749.88

1.77

389

COMP

8588.25

8588.25

389 Total

8588.25

8588.25

1.77 Total

8588.25

8588.25

5.95

1866

POOL

5077.8

5077.8

1866 Total

5077.8

5077.8

5.95 Total

5077.8

5077.8

6.08

2321

USDC

47749.11

47749.11

2321 Total

47749.11

47749.11

6.08 Total

47749.11

47749.11

8.62

5398

DAI

36193.11

36193.11

5398 Total

36193.11

36193.11

8.62 Total

36193.11

36193.11

11.69

3323

USDT

493.07

493.07

3323 Total

493.07

493.07

11.69 Total

493.07

493.07

15.93

574

SUSHI

4983.57

4983.57

574 Total

4983.57

4983.57

15.93 Total

4983.57

4983.57

Grand Total

493.07

105509.14

106002.21

Conclusions / Key Takeaways

The research suggests that PoolTogether may provide similar incentives to a traditional PLS account based on the growth of account balances over time. However, the gas fees associated with entering and exiting a pool may offset any earnings, but upcoming changes to the Ethereum protocol may affect that aspect of earnings in a way that would significantly improve net yield.

Research also indicates that although the weighting of prizes favors the larger accounts, it was fairly uncommon for the largest account in a pool to be awarded a prize more than twice in a row. This suggests that while the mechanism deciding winners is inherently weighted towards larger accounts, smaller accounts winning indicates that the pool’s fairness mechanisms are designed to benefit smaller and medium-sized accounts over the long-term instead of larger accounts in the short term. In this context, the smallest accounts benefit the least due to the odds effectively making it impossible for them to actually win.

There may also be a benefit to using DeFi PLS accounts that use pegged tokens as the definitive token for their pools. In the recent market downturn, accounts with USD-pegged tokens naturally resisted the loss of value associated with market volatility, while pools using unpegged tokens suffered losses of value directly correlated with the market.

CTA: Future Work / RFP

This work looked at a case study and conducted primary research on PoolTogether. Further research on DeFi PLS accounts and the successes or failures of the pools in helping depositors build savings could provide additional context and clarity.

5 Likes

PLS accounts look like the alternative to get people saving up instead of buying lotteries. It provides a stronger psychological incentive and would be more appealing.

I’m still waiting for more empirical evidence though. The crypto world is quite different from traditional finance. There are lots of instruments available here. Pool Together has a lot to compete.

I skimmed the paper and found one piece of evidence that would be in favor of Pool Together.

Compared to bank employees with a savings account in the bank that offered MAMA, bank employees WITHOUT one are much more likely to open a MAMA.

This means that PLS accounts can incentivize people to open an account pretty well. That is, it has some potential to escape from the competition, and create its market.

3 Likes

You are correct that we need to have more empirical evidence from the defi space before we can come to any conclusions about the potential for PLS accounts in the defi space. I did use primary research on Pooltogether, as the charts from Pooltogether had to be created from hard data. There is a benefit to having open access to the pools so that the data can be analyzed in contrast to traditional finance where their reports are quarterly and not specific on individual accounts.

All that said, the current data shows that PLS accounts can potentially be viable alternative to traditional lotteries as a mechanism to increase savings while reducing gambling.

4 Likes

This is an interesting paper, but I do question the strength of the evidence regarding the reduction in gambling claim being made. But in some ways, that isn’t really the big news here. The real value here is the incentivization of a positive behavior as opposed to the alternative to a negative behavior (at least from my understanding).

I found the fairness mechanism observation interesting.

If the goal here is to get more people saving (which it might not be), it seems that some other mechanics would work better. Or is the likelihood to win enough of an incentive to continue doing the desired behavior?

3 Likes

The reason they correlated the findings with “gambling” specifically is that the balances in the mama accounts would change relative to large national lottery jackpots. When there were large national jackpots, the mama accounts would have fewer deposits; but as the mama accounts increased their monthly awards relative to smaller national lotteries, they saw increased deposits. It is obvious you can’t draw conclusions about causation in the real world, but they were finding correlation in movement with mama accounts and national lottery prize sizes. This is also specified in the context of South Africa.

The mechanism you’re observing in the latter part was relative to “Pooltogether” and not the South African study. I don’t think the defi space has a good grasp on building incentive mechanisms quite yet. The original research on the South African case study did not include a chart relative to the state lotto, but it was alluded to in the paper. One of the reasons I did not go into depth into that part is because the author gets into an unnecessarily racial analysis of the data, and I was not inclined to open that can of worms in this discussion post.

I think the issue is not even about modifying behavior so much as the exposure to “prize-linked savings accounts” which amount to “risk-free gambling”. If more people were made aware that these types of mechanisms exist, they may be inclined to put their money into a lossless prize pool rather than risk a prize in a system in which they lose their entrance fee if they are not the winner.

This is a zero sum design compared to a positive sum design. We are talking about a complete paradigm shift concerning capital management. In a sense, a “lossless prize pool” is still “gambling”. This is why framing it as “pls vs. gambling” is not quite accurate. “Gambling” implies a zero-sum game, but as we can see with PLS accounts that is not always the case. The outcome of the game is determined by the rules, and in this case a PLS system does not have the same type of “losers” as a traditional zero-sum lottery. If you do not win in a PLS system, you do not “lose” anything. That is a fundamentally different type of game, but it’s still a type of “gambling”.

3 Likes

If anyone is interested in learning more about PoolTogether, please check out our recent panel on Governance Implementation, where founder Leighton Cusack spoke to Connor Spelliscy, Seth Frey, Scott Moore and @kelsienabben discussed community, incentives and DAOs, among other topics.

4 Likes

First of all, thank you for the clarification, particularly regarding the PLS vs. gambling framing.

Perhaps it is still too soon to comment, but do we have any indication about the impact that gas fees might be having on the value of earnings?

I also wanted to come back to your question:

This is an interesting question because it might be where DeFi can appeal to more people. As you pointed out above, there is some strong gamification happening here that is appealing to people. Maybe more importantly, there is some analysis that could be done to identify tipping points between seeking a standard yield versus an opportunity to win. So, coming back to your question, what would cannibalization data look like? Certainly there will be some movement in the market if PLSs are present, but how much adoption is necessary to determine if it is changing the expectations and/or behavior of people looking to save?

2 Likes

I had thought about these questions, and I believe a way to tell if gas fees were cannibalizing the savings would be to look at accounts that had exited and average the difference between the yields and the gas fees at time of exit. I think the gas fees at time of exit are more relevant than the “least fee possible to exit the pool at the optimal time,” because those that exit may have had an external pressure that forced them to get out of the pool and in some cases be forced to take an excessive gas fee.

Since we can’t look at the Pooltogether users’ bank accounts, one may be able to look at their other coin holdings to see if any of them had been transferred into Pooltogether. It might be relevant to do a study to see how many users switched from trading on Dexes where the trades were visible to using Pooltogether. There is some way to glean behavioral shifts if they do exist. In that regard it would only be possible on accounts that had a history of trading prior to joining Pooltogether. While it would take a considerable amount of work, I do believe it would be possible to determine if Pooltogether cannibalizes trading market account balances. While the term “cannibalize” sounds bad, the overall outcome would likely be a net positive movement for the market if Pooltogether PLS accounts cannibalized speculative trading accounts; but potentially an overall net negative outcome if the gas fees subsequently cannibalized the PLS accounts.

2 Likes

I absolutely agree that there is a potential net positive here for the market if speculative trading gets displaced by PLS accounts. I suspect this would bring some stability to the market.

Thanks for digging into these questions a little more. Based on my understanding now, it seems to me that PLS mechanisms might be something that projects that are seeking to attract stakers could or should be using. More analysis of those tipping points of account holding and likelihood to win prizes is needed, but it does seem like something that could be alluring for potential project investors. Maybe @tjd233 could bring some perspective to this as well.

2 Likes