Libra's Unresolved Puzzles

Libra, mutual fund, cryptocurrencies, currency board
Libra, mutual fund, cryptocurrencies, currency board, private money
Libra Association website via TechCrunch

By now everyone has heard of Libra, the new private digital money system announced last week, sponsored by Facebook and a consortium of other firms, to be rolled out in 2020. The project’s official white paper, “An Introduction to Libra,” calls it “a new decentralized blockchain, a low-volatility cryptocurrency, and a smart contract platform.” A second official document on “The Libra Reserve” goes into more detail. An overview of the system’s potential by Diego Zuluaga is available here, a more critical view by Stephen Williamson here. Alarmed denunciations of Libra for being too corporation-y can be found here and here.

Let’s put aside the questions of how decentralized Libra will really be, what kind of blockchain it will employ, and its use in smart contracting, to focus on the question: Exactly what kind of money will it be? I find that the white papers under-specify the core mechanism that is supposed to control the value and quantity of Libra. My best guess is that Libra is meant to be like a mutual funds share, redeemable (by authorized resellers, not by the public directly) at the net asset value of a low-risk mutual fund diversified across several fiat currencies.

First, a verbal and symbolic aside: The “libra” historically was a medieval monetary unit. In Latin, libra is singular noun meaning “pound” or “scale” (plural librae). Historically this is why the sign of the British Pound is £, a cursive capital L. In the Libra white papers, “Libra” is variously used as a singular noun, a plural noun (the word “Libras” is never used), and as an adjective. Herein I adopt as the Libra currency sign the zodiac emoji for Libra, so for “10 Libra” I write ♎10. Apparently, one of the more mundane aspects of creating a new currency is waiting for HTML to catch up with your preferred symbol.

The structure of the Libra system

First, the structure of the system. The public can buy and sell Libra “coins” in exchange for existing fiat money. Unlike Bitcoins, the exchanges (at least initially) will not be independent of Libra and competitive with open entry, but will be a set of Authorized Resellers. They sell coins to the public at a price reflecting the wholesale value of the Libra plus a bid-ask spread and a transaction fee. They buy coins back at the wholesale value minus bid-ask and transaction charges. The Resellers manage their inventories by interacting with the Libra Reserve, swapping fiat for Libra and vice-versa. But here’s the key unanswered question: on what terms do they swap?

If we take the white papers’ talk of “backing” seriously, it suggests that the value of Libra coins in circulation is matched by the value of assets held in the Reserve, ready to buy back or redeem the coins. The papers say that Libra will be backed by a portfolio of $-denominated, €-denominated, and other fiat-denominated securities. But is a coin in the hands of a Reseller a debt claim or an equity claim on the Reserve? In particular, when a Reseller brings ♎1 to the Reserve, she might either have an IOU, entitling her to a specified medium of redemption, like a Paypal account balance or a Hong Kong Dollar note redeemable in US Dollars. Or she might have a share claim on the Libra Reserve portfolio, like a mutual fund share. For the Reserve portfolio to provide full backing, the share claim will have to be redeemable in a bundle of currencies whose composition mirrors the composition of the portfolio.

The official papers ambiguously suggest both debt and equity characteristics. In places, they liken the Libra Reserve to a currency board. An orthodox currency board note issues debt claims (local currency notes), each redeemable for a fixed amount of the anchor currency (HK$7.8 = US$1), and holds at least 100 percent reserves in the anchor currency. If that is the Libra arrangement, then there is a fixed exchange rate between Libra and a pre-specified fiat currency basket. The proportions of fiat currencies in the medium-of-redemption basket would be pre-specified. To provide full backing the proportions would have to correspond exactly to the proportions of currency-denominated assets in the Reserve’s portfolio. Otherwise, adverse exchange rate movements could reduce the portfolio value below 100 percent of the par value of Libra in circulation.

On the other hand, the Resellers are not described as redeeming Libra at the Reserve. A different backing arrangement would provide that returning ♎1 always gives the Reseller a fixed-proportions fiat currency basket equal in value to 1/N, where the portfolio’s market value is ♎N. The Reserve is then a kind of mutual fund, and ♎1 in the hands of a Reseller is a mutual fund share (a possibility Williamson identifies). This would be a novel arrangement – a mutual fund redeemable in a multi-fiat medium of redemption, with shares used as a medium of exchange. The value of the ♎1 share would not be perfectly steady in terms of the defined currency basket, but would be as steady as the nominal net asset value of the portfolio in currency baskets.

The “Introduction” frames the case for the Libra essentially by contrast to Bitcoin: “Mass-market usage of existing blockchains and cryptocurrencies has been hindered by their volatility and lack of scalability, which have, so far, made them poor stores of value and mediums of exchange.” I agree, although I would have said “media of exchange.”

This framing of the case suggests that we are to view Libra as a “stablecoin,” that is, a cryptocurrency whose value is somehow linked to a bundle of existing government fiat currencies. But this does not tell us the backing arrangement, because stablecoins have tried various methods for linking to single fiat currencies. One simple method is redemption at a fixed rate. The most prominent stablecoin today is Tether USD, which promises on its website that Tether is “1-to-1 pegged to the dollar” and “100% backed by our reserves,” though it avoids speaking of “redemption.” A more transparent model is PAX USD, which promises that the parent company will be “issuing and redeeming each token in exchange for $1.00” while “backing every token with a US dollar held in segregated accounts at FDIC-insured, U.S. banks.” Other existing stablecoins do not promise redeemability to insure that they trade at or close to par, but instead promise to correct deviations from par by mechanisms that are supposed to vary the quantity of coins in circulation as appropriate to match demand at the par exchange rate

What will Libra be stable in terms of? “The Libra Reserve” paper in some places suggests stability in terms of purchasing power over goods and services. Like holders of euros who can count on stability in the euro price of coffee, it says, “holders of Libra, too, can be confident the value of their coins today will be relatively stable over time.” By contrast to Bitcoin, the paper asserts that this stability will remove the volatility of value that obstructs widespread adoption. No mechanism for purchasing power stability is specified, however.

The goal that the Libra system actually seems to aim for is stability of Libra’s value in terms of a basket of existing fiat currencies that lose their purchasing power slowly, that is, have low and stable inflation rates. That goal points to another possible barrier to adoption: inertia. Why would anyone take the trouble to switch to Libra if it holds its purchasing power no better than the average of the least-bad fiat currencies? The positive case for switching must lie in other features of the Libra payment system, like low transaction fees for intra-system transfers. But lower than the fees of low-cost fiat transfer systems like Paypal and Venmo? FX fees will be a burden until the Libra community achieves critical mass. The killer use that will make Libra gain critical mass remains unclear.

How will Libra achieve stability of value in terms of a basket of fiat currencies? The second paper states: “The reserve is the key mechanism for achieving value preservation. By fully backing each coin … users can have confidence that they will be able to sell any Libra coin at or close to the value of the reserve at any time.” This reference to backing by “the value of the reserve” is what persuades me that the authors have a mutual fund model in mind. Resellers redeem Libra at the Reserve portfolio’s NAV (net asset value) per Libra.

Is the Libra Reserve really like a currency board?

In crucial respects, Libra is very different from a currency board. Libra does not have a single anchor currency. In fact, it seems to lack any specific anchor. Libra coins are never referred to as liabilities.

We can imagine an anchor currency basket, for example, such that the fixed par rate is ♎10 = {$5 + €2 + £1 + ¥100 + CHF1}. In that case, the first white paper would be justified is its claim that “the value of Libra will be effectively linked to a basket of fiat currencies,” even if not to any single fiat currency.[1] In such a system, a Reseller who turns in ♎10 would always be paid {$5 + €2 + £1 + ¥100 + CHF1}, with the same exchange rate in the other direction for a Reseller turning in fiat currency baskets. Or, if the Greenfield and Yeager concept of “indirect convertibility” could be successfully implemented, separating the medium of redemption from the unit of account, the Reseller turning in ♎10 would be paid ♎10 worth of dollars (or euros, etc.) at up-to-the-second wholesale foreign exchange mid-market rates or asking prices. But no specific basket is mentioned in the white papers, and there is no talk of a fixed exchange rate.

The only thing like a “basket of fiat currencies” is the Libra Reserve’s portfolio, which will hold “a collection of low-volatility assets, including bank deposits and government securities in currencies from stable and reputable central banks.” The reserve portfolio is explicitly to be diversified across “multiple governments” and currencies. Saying so does nothing to specify a par redemption rate, however.

Is the Libra Reserve like a mutual fund?

As the papers describe it, the Reserve portfolio sounds like a government-bond money-market mutual fund. It is to hold only short-term government bonds and cash equivalents. The bonds are all to be low in default risk, highly liquid, and low in duration risk. Via new issues and buy-backs of Libra, the Reserve fund is open-ended. The liquidity of its assets “allows the size of the reserve to be easily adjusted as the number of Libra in circulation expands or contracts.” The choice of the passive voice here (“be adjusted”) is odd, in that the size of the reserve is not supposed to be deliberately adjusted by anyone, but is rather supposed to expand automatically when the public buys Libra from Resellers and contract automatically when it sells them back.

Also, unlike an ordinary mutual fund, which allows members of the public to buy shares directly and sell them directly back to the fund at their current net asset value, Libra only allows Resellers to interact directly with the Libra Reserve. To be sure, it is nowhere directly stated that Libra in the hands of Resellers constitute a claim with a redemption value defined by a share of the portfolio in the manner of mutual fund shares. Yet some such arrangement would have to exist to make sense of the statement that Libra “will remain fully backed across time.” Full backing means that the market value of reserves assets equals or exceeds the total value of redeemable claims. With claims defined as shares, and assets continuously marked to market, there is by definition no opportunity for the value of all claims to exceed the value of the portfolio.

The first white paper elaborates in the next sentence: “This means that today, tomorrow, and in the future, users have confidence that any Libra coin they hold can be sold for fiat currency at a narrow spread above or below the value of the underlying reserve, when a competitive market for exchanges is present.” This would be true if two conditions are satisfied: (1) “the value of the underlying reserve” means the value of ♎1 as a 1/Nth share of the reserve portfolio; and (2) the Reserve always pays a Reseller a proportionate share of the Reserve portfolio for each ♎1 turned in. A “competitive market for exchanges” implies that Resellers add only a small (competitive) bid-ask spread to cover their transaction costs and adverse information risk.[2]

What is missing from the white papers is a clear account of whether these two conditions hold. Of course they are jointly sufficiently conditions, not necessary conditions, so the authors may have some other mechanism in mind for linking the Libra’s market price in dollars or euros to the Reserve portfolio’s value in dollars or euros. It would be helpful if the architects of Libra would tell us explicitly whether they have in mind such a mutual-fund-type purchase and redemption arrangement between Resellers and the Libra Reserve. Otherwise, potential users are being asked to rely, not on a specified pre-commitment, but on mere promises.

A mutual fund structure has advantages and disadvantages

Advantage: a mutual fund is not run-prone. There is no payoff in running to the head of the line to redeem, because you don’t get any more than people at the end of the line. Even if you hear that the Libra Reserve portfolio has taken a hit (say it held Japanese bonds and Japan’s credit rating suddenly plunged), there is no point in racing to redeem, because the value of your ♎1 claim has already dropped so that the total of all claims still sums to the value of the portfolio. If a Libra is at bottom a mutual fund share, not a deposit or other kind of debt claim, there is no rationale for subjecting the Libra system to government deposit guarantee scheme with its taxes (premiums) and portfolio restrictions. It should be noted that the Libra Association is incorporated in Switzerland, not in the United States.

Disadvantage: the value of a mutual fund share cannot be fixed in nominal terms when the portfolio value can sometimes decline. In a world of near-zero and sometimes negative interest rates on government bonds, a small asset loss can wipe out a day’s interest earnings. Holders of Libra will have to live with some degree of nominal volatility, even if it is small, in terms of the defining currency basket. Market exchange rate changes among the constituent currencies mean that there will be slightly more volatility in terms of the dollar or any other single currency.


[1]Such a setup would be reminiscent of Warren Coats’ proposal for the IMF to promote use the SDR basket, a bundle of fiat currencies in defined weights, as a unit of account and medium of redemption for private bonds and payment media.

[2] On the microeconomics of bid-ask spreads see Malte Krueger (2012).