Thanks for your thoughtful summary and questions. I would like to give you solid answers but first, two comments and a few requests for clarification.
Comment 1: converting money to GLO is not a donation, but a currency transfer.
Comment 2: the future stuff you describe is broadly possible and, IMO, desirable—but I just want to be clear that GLO the company is not making any specific commitments to this as a roadmap; rather, our post serves to state our big picture intent at this juncture, and our intended first step on that path.
Requests for clarifications
do you mean ‘keep as dollars rather than treasury bills’?
I’m pretty sure whatever numbers we put in, a plan that distributes money to the truly needy (GiveDirectly) and that has a much lower probability of the currency reaching $0.00 (as might happen with any uncollateralized crypto) is going to be higher EV for now. But did you have something more formal in mind? Frankly, I’m not totally sure what parameters we should choose, nor how to estimate their values. I will ponder this but I think that something fuzzy/directional is probably as good a map of the territory as something more structural.
Do you mean: if some large accounts got frozen — let’s say that a whale who holds 50% of all extant GLO turned out to be Someone Who Is Very Bad — would we need backup, uninvested reserves to be able to keep liquidity?
I’m not sure what you mean by ‘downside across the inflection point’—like, as we’re converting, risks of that conversion? There are many possible paths here, including introducing a new currency with a different name and a different model, converting one to the other, or anything in between. I personally have some thoughts about pros and cons of different strategies there, but I’m afraid they’re going to be so speculative as to be uninformative.
I think the reserves let you provide dollars to anyone wanting their GLO back as dollars, and you structure your investments and nonprofit expenses so that you can access those dollars on demand. Whether that means you keep cash on hand, or invest it all, I don’t know.
I thought your long-term plan was to use GLO as UBI, but you have a very different plan for now, converting GLO to dollars and giving dollars to Give Directly, right? You are using expected value to form your goals, so I thought it was a natural question.
Yes, another possibility is that you plan safeguards to protect you from money launderers or fraud in the first place, and will not structure your reserves to account for regulatory actions that could reduce GLO liquidity, because you feel you are protected from the worst-case scenarios by your safeguards.
Right, the risks, so rather than the graph shooting up after the inflection, it shoots down. It’s good you have ideas to handle those risks, or are thinking about them.
I am still curious about your nonprofit’s take on the cryptocurrency industry in general, and how to be good stewards of your currency, if you get it into common use.
Our general plan is to only invest in highly liquid assets, and to fund the operations of our non-profit separately (re: through donations). So while I I hope that we will in time build dashboards that provide complete transparency into how our treasury is distributed/stored, in the meantime, we think that treating our treasury investments as “cash-like” is a fair assumption.
Makes sense, thanks. So—ballpark estimate here—we expect the yield from treasury bonds to be about 2%, and we were aiming for an 8% yield in the seigniorage model. But I personally think that the stablecoin model has at least a 5X greater chance of going somewhere rather than nowhere—of not being driven into the ground by a speculative attack, for instance—so in the short-term, I think EV has gone up by at least 20%. In the long run, we expect to get to that 8% once we meet the conditions described in the post. Comparing the EVs of getting there at Time 1 vs Time 2 would require pinning those dates down, and frankly, that would be straight guesswork.
We will work with exchanges that perform KYC to obviate money laundering, and we’ll build our own infrastructure to support this if it’s absolutely necessary. But in general, we prefer to ‘orchestrate rather than create’, so we’ll hope to build on existing infrastructure, or incentivize its creation.
It is definitely possible that our shift over to a new currency will lead to a devaluation, but it’s also possible that we’ll launch a separate currency to supplement our existing stablecoin. Another potential idea is to peg one coin to the other; perhaps that will be standard protocol by that point rather than in ill repute. This is far enough in the future that it’s hard to forecast. But we would only make the shift when we feel confident that it’s reasonably likely to work.
We’re definitely thinking about this! I don’t have anything definitive to say yet on behalf of the company, but personally, I think a standardized, regular framework for crypto companies to operate in would be welcome. Tyler Cowen has an interesting recent column touching on this.
all reserves will be in treasuries. That was what I guessed, but I wanted to check. My confusion about stablecoins is whether there are in fact reserves equivalent in dollars to the full amount of stablecoins available. In the case of GLO, there will be less than a dollar of reserves for every GLO stablecoin, is that right?
I actually am wondering about the expected value to Give Directly UBI recipients of offering the interest yields vs allowing them to receive direct $$ donations. Your action to incentivize currency exchange rather than donations could reduce total UBI donation amounts to Give Directly. After all, you plan to funnel people with charitable interests into buying your currency rather than giving directly to their preferred causes. As the value of your treasury investments fluctuate, there will be fluctuations that could include actual reserve losses. Three month treasuries dipped into negative territory in the last couple years, for example.
There is also the matter of the transaction fees, which could be meaningful in some calculations of expected value.
OK, so in the situation that the KYC fails, will you have contingency plans involving use of your reserves to handle any fallout?
Hm, I am curious about how you intend to learn from other’s mistakes. The trouble with algorithmic stablecoin (based on my very limited understanding) is their vulnerability to a spiral of death: sell-offs, reserve value losses as sell-offs continue, and never enough reserves to recover. For example, with the arrangement of Terra’s stablecoin and its Luna counterpart there was vulnerability to:
attacks (because of Terra’s choice of bitcoin as reserves, shorts took advantage of bitcoin sell-offs as Terra’s stablecoin lost its peg)
bank runs once fears rose about available reserves or loss of the stablecoin peg.
unsustainable interest rate offerings to entice currency buyers. (let me know if I got anything wrong or missed anything)
I read the article you referenced. I didn’t find any suggestions about specific regulations, however I do see a framing of the issue as how to better regulate cryptofinance when costs and friction involved in issuing securities “on the blockchain” are less than with traditional securities.
Based on what little I know about this area, the biggest claims of web3 benefits, decentralization and blockchain-ensured security, don’t show up in practice for users who keep private wallets on their computers and who rely on a few companies to handle their web3 transactions. The underlying technology that handles their transactions is mostly centralized web2. This makes me doubt the difference in friction and costs that crypto companies can provide versus more traditional securities vendors who can also use web2.
I believe that most in finance have a security in mind for folks to trade that does not involve the blockchain or cryptocurrency, no matter who those folks are, and the only question is if Wall Street can get past regulators and develop that market (for example, put an app in somebody’s hands, and get them to commit dollars in a marketplace for the security). That’s just business as usual. I wonder how you would improve on it in case of web3 and cryptocurrency. I guess that’s a question you’ll have to answer later.
Thanks, Seth, for your time. It’s refreshing to talk to someone who is sincere in wanting to do good in what seems to be a wild west full of speculators and information asymmetries. There’s so much money moving through cryptocurrencies, I see the rationale for wanting to do good through the technology.
Hi Noah, a pleasure to talk to you as well! My apologies for the delay, I have been traveling.
there will be less than a dollar of reserves for every GLO stablecoin, is that right?
GLO will be fully backed! This means 1-to-1 between cash and equivalents and GLO in circulation.
RE: reducing donations to GiveDirectly—we see a purchase of GLO and a donation to GiveDirectly as complements rather than substitutes. A purchase of GLO is not a charitable donation, can’t be deducted, etc. Our aim is to invest in only very safe options—cash and equivalents—and we don’t foresee major fluctuations therein.
If KYC fails, we’ll have to evaluate our options based on the specific failure, which could take a lot of potential forms. But this is one reason why we’re taking care to work with reputable partners.
RE: algorithmic stablecoins—we think that the GLO project is fundamentally different than those projects so there’s not much obvious relevance, to my eyes. I would say general lessons like “don’t say anything on Twitter that will look really bad if something goes wrong” is a good one though 😃
RE: the security/usability tradeoff—we have some nascent thoughts about this but I’m afraid that I personally don’t have much to say about it. But you’re right that in the long run, this is going to be important, for GLO but also for crypto generally.
OK, so I used “security” in a couple different senses in my prior comment, thank you for putting up with that.
There could be fluctuations in returns on treasuries, given that there have been before, how does it work if they dip into negative territory unexpectedly? Do you pull GLO out of circulation?
So, it looks like going back to 2017, there’s only been one week in which returns were negative , which was late March 2020: https://ycharts.com/indicators/3_month_t_bill . And that was (clearly) a very unusual moment so I’m not worried per se about rates truly going below zero. I am more worried about rates going to zero, in which case, yeah, we’d have to figure out some other low-risk investment opportunities for the model to keep working. This is on my research agenda—get a good sense of the ROI of, say, short-term bonds from all the OECD countries, rank them by yield and perceived risk, and then rank things that fall near that on the risk/reward scale but that still definitely qualify as conventionally safe. I’m afraid I haven’t gotten to it yet, this week I’ve been focused on writing something up about why we are donating to GiveDirectly and not some other great charity (e.g. GiveWell’s top ranked charities, which as of this week no longer include GiveDirectly; GD’s response is interesting https://www.givedirectly.org/giving-directly-still-means-giving-well/).
Yeah, it looks like, the rates stayed close to zero between 2008 and 2015, and more recently as well, March of 2020 to January of 2022.
I took a look at GiveDirectly’s response. They mention the multiplier effect of cash, and the possibility that Givewell might lower its cost-effectiveness requirements from 10X. Now I am confused about GiveWell’s estimates of the cost-effectiveness of cash and in what circumstances GiveWell actually makes the comparison.
GiveDirectly’s point about meeting the needs of its charity recipients while respecting the choices of the recipients is an interesting point. Pro and con arguments might track arguments about UBI in general. Your efforts here could reflect your philosophy about UBI, I think.
Hi Noah,
Thanks for your thoughtful summary and questions. I would like to give you solid answers but first, two comments and a few requests for clarification.
Comment 1: converting money to GLO is not a donation, but a currency transfer.
Comment 2: the future stuff you describe is broadly possible and, IMO, desirable—but I just want to be clear that GLO the company is not making any specific commitments to this as a roadmap; rather, our post serves to state our big picture intent at this juncture, and our intended first step on that path.
Requests for clarifications
do you mean ‘keep as dollars rather than treasury bills’?
I’m pretty sure whatever numbers we put in, a plan that distributes money to the truly needy (GiveDirectly) and that has a much lower probability of the currency reaching $0.00 (as might happen with any uncollateralized crypto) is going to be higher EV for now. But did you have something more formal in mind? Frankly, I’m not totally sure what parameters we should choose, nor how to estimate their values. I will ponder this but I think that something fuzzy/directional is probably as good a map of the territory as something more structural.
Do you mean: if some large accounts got frozen — let’s say that a whale who holds 50% of all extant GLO turned out to be Someone Who Is Very Bad — would we need backup, uninvested reserves to be able to keep liquidity?
I’m not sure what you mean by ‘downside across the inflection point’—like, as we’re converting, risks of that conversion? There are many possible paths here, including introducing a new currency with a different name and a different model, converting one to the other, or anything in between. I personally have some thoughts about pros and cons of different strategies there, but I’m afraid they’re going to be so speculative as to be uninformative.
Hope your Friday is going well,
To clarify:
I think the reserves let you provide dollars to anyone wanting their GLO back as dollars, and you structure your investments and nonprofit expenses so that you can access those dollars on demand. Whether that means you keep cash on hand, or invest it all, I don’t know.
I thought your long-term plan was to use GLO as UBI, but you have a very different plan for now, converting GLO to dollars and giving dollars to Give Directly, right? You are using expected value to form your goals, so I thought it was a natural question.
Yes, another possibility is that you plan safeguards to protect you from money launderers or fraud in the first place, and will not structure your reserves to account for regulatory actions that could reduce GLO liquidity, because you feel you are protected from the worst-case scenarios by your safeguards.
Right, the risks, so rather than the graph shooting up after the inflection, it shoots down. It’s good you have ideas to handle those risks, or are thinking about them.
I am still curious about your nonprofit’s take on the cryptocurrency industry in general, and how to be good stewards of your currency, if you get it into common use.
Thanks! I’l take a stab at answering these here:
Our general plan is to only invest in highly liquid assets, and to fund the operations of our non-profit separately (re: through donations). So while I I hope that we will in time build dashboards that provide complete transparency into how our treasury is distributed/stored, in the meantime, we think that treating our treasury investments as “cash-like” is a fair assumption.
Makes sense, thanks. So—ballpark estimate here—we expect the yield from treasury bonds to be about 2%, and we were aiming for an 8% yield in the seigniorage model. But I personally think that the stablecoin model has at least a 5X greater chance of going somewhere rather than nowhere—of not being driven into the ground by a speculative attack, for instance—so in the short-term, I think EV has gone up by at least 20%. In the long run, we expect to get to that 8% once we meet the conditions described in the post. Comparing the EVs of getting there at Time 1 vs Time 2 would require pinning those dates down, and frankly, that would be straight guesswork.
We will work with exchanges that perform KYC to obviate money laundering, and we’ll build our own infrastructure to support this if it’s absolutely necessary. But in general, we prefer to ‘orchestrate rather than create’, so we’ll hope to build on existing infrastructure, or incentivize its creation.
It is definitely possible that our shift over to a new currency will lead to a devaluation, but it’s also possible that we’ll launch a separate currency to supplement our existing stablecoin. Another potential idea is to peg one coin to the other; perhaps that will be standard protocol by that point rather than in ill repute. This is far enough in the future that it’s hard to forecast. But we would only make the shift when we feel confident that it’s reasonably likely to work.
We’re definitely thinking about this! I don’t have anything definitive to say yet on behalf of the company, but personally, I think a standardized, regular framework for crypto companies to operate in would be welcome. Tyler Cowen has an interesting recent column touching on this.
Thank you, so:
all reserves will be in treasuries. That was what I guessed, but I wanted to check. My confusion about stablecoins is whether there are in fact reserves equivalent in dollars to the full amount of stablecoins available. In the case of GLO, there will be less than a dollar of reserves for every GLO stablecoin, is that right?
I actually am wondering about the expected value to Give Directly UBI recipients of offering the interest yields vs allowing them to receive direct $$ donations. Your action to incentivize currency exchange rather than donations could reduce total UBI donation amounts to Give Directly. After all, you plan to funnel people with charitable interests into buying your currency rather than giving directly to their preferred causes. As the value of your treasury investments fluctuate, there will be fluctuations that could include actual reserve losses. Three month treasuries dipped into negative territory in the last couple years, for example.
There is also the matter of the transaction fees, which could be meaningful in some calculations of expected value.
OK, so in the situation that the KYC fails, will you have contingency plans involving use of your reserves to handle any fallout?
Hm, I am curious about how you intend to learn from other’s mistakes. The trouble with algorithmic stablecoin (based on my very limited understanding) is their vulnerability to a spiral of death: sell-offs, reserve value losses as sell-offs continue, and never enough reserves to recover. For example, with the arrangement of Terra’s stablecoin and its Luna counterpart there was vulnerability to:
attacks (because of Terra’s choice of bitcoin as reserves, shorts took advantage of bitcoin sell-offs as Terra’s stablecoin lost its peg)
bank runs once fears rose about available reserves or loss of the stablecoin peg.
unsustainable interest rate offerings to entice currency buyers.
(let me know if I got anything wrong or missed anything)
I read the article you referenced. I didn’t find any suggestions about specific regulations, however I do see a framing of the issue as how to better regulate cryptofinance when costs and friction involved in issuing securities “on the blockchain” are less than with traditional securities.
Based on what little I know about this area, the biggest claims of web3 benefits, decentralization and blockchain-ensured security, don’t show up in practice for users who keep private wallets on their computers and who rely on a few companies to handle their web3 transactions. The underlying technology that handles their transactions is mostly centralized web2. This makes me doubt the difference in friction and costs that crypto companies can provide versus more traditional securities vendors who can also use web2.
I believe that most in finance have a security in mind for folks to trade that does not involve the blockchain or cryptocurrency, no matter who those folks are, and the only question is if Wall Street can get past regulators and develop that market (for example, put an app in somebody’s hands, and get them to commit dollars in a marketplace for the security). That’s just business as usual. I wonder how you would improve on it in case of web3 and cryptocurrency. I guess that’s a question you’ll have to answer later.
Thanks, Seth, for your time. It’s refreshing to talk to someone who is sincere in wanting to do good in what seems to be a wild west full of speculators and information asymmetries. There’s so much money moving through cryptocurrencies, I see the rationale for wanting to do good through the technology.
Hi Noah, a pleasure to talk to you as well! My apologies for the delay, I have been traveling.
GLO will be fully backed! This means 1-to-1 between cash and equivalents and GLO in circulation.
RE: reducing donations to GiveDirectly—we see a purchase of GLO and a donation to GiveDirectly as complements rather than substitutes. A purchase of GLO is not a charitable donation, can’t be deducted, etc. Our aim is to invest in only very safe options—cash and equivalents—and we don’t foresee major fluctuations therein.
If KYC fails, we’ll have to evaluate our options based on the specific failure, which could take a lot of potential forms. But this is one reason why we’re taking care to work with reputable partners.
RE: algorithmic stablecoins—we think that the GLO project is fundamentally different than those projects so there’s not much obvious relevance, to my eyes. I would say general lessons like “don’t say anything on Twitter that will look really bad if something goes wrong” is a good one though 😃
RE: the security/usability tradeoff—we have some nascent thoughts about this but I’m afraid that I personally don’t have much to say about it. But you’re right that in the long run, this is going to be important, for GLO but also for crypto generally.
Oh, nice surprise to hear back from you, thanks!
OK, so I used “security” in a couple different senses in my prior comment, thank you for putting up with that.
There could be fluctuations in returns on treasuries, given that there have been before, how does it work if they dip into negative territory unexpectedly? Do you pull GLO out of circulation?
So, it looks like going back to 2017, there’s only been one week in which returns were negative , which was late March 2020: https://ycharts.com/indicators/3_month_t_bill . And that was (clearly) a very unusual moment so I’m not worried per se about rates truly going below zero. I am more worried about rates going to zero, in which case, yeah, we’d have to figure out some other low-risk investment opportunities for the model to keep working. This is on my research agenda—get a good sense of the ROI of, say, short-term bonds from all the OECD countries, rank them by yield and perceived risk, and then rank things that fall near that on the risk/reward scale but that still definitely qualify as conventionally safe. I’m afraid I haven’t gotten to it yet, this week I’ve been focused on writing something up about why we are donating to GiveDirectly and not some other great charity (e.g. GiveWell’s top ranked charities, which as of this week no longer include GiveDirectly; GD’s response is interesting https://www.givedirectly.org/giving-directly-still-means-giving-well/).
Hope that helps!
Yeah, it looks like, the rates stayed close to zero between 2008 and 2015, and more recently as well, March of 2020 to January of 2022.
I took a look at GiveDirectly’s response. They mention the multiplier effect of cash, and the possibility that Givewell might lower its cost-effectiveness requirements from 10X. Now I am confused about GiveWell’s estimates of the cost-effectiveness of cash and in what circumstances GiveWell actually makes the comparison.
GiveDirectly’s point about meeting the needs of its charity recipients while respecting the choices of the recipients is an interesting point. Pro and con arguments might track arguments about UBI in general. Your efforts here could reflect your philosophy about UBI, I think.
I wish you the best of success with your efforts!