Ammous: Does Bitcoin’s Fixed Supply Make It Too Volatile to Be Money? A Response to Professor White; and Ammous vs. Selgin

Below I post Saifedean Ammous‘s article “Does Bitcoin’s Fixed Supply Make It Too Volatile to Be Money? A Response to Professor White” (Oct. 13, 2026) and some some related recent exchanges between Ammous and freebanker George Selgin.

Related

• Stephan Kinsella, The Great Fractional Reserve/Freebanking Debate (Jan. 29, 2016)
• ——, “Comments on Block and Barnett on the Optimum Quantity of Money,” StephanKinsella.com (March 26, 2021)
• ——, “On Gold, the Optimal Amount of Money, Hypersubjectivism, and Demonstrated Preference,” StephanKinsella.com (May 10, 2025)
• ——, On Coinbase, Bitcoin, Fractional-Reserve Banking, and Irregular Deposits (July 12, 2022)
• ——, “Musings on Fractional-Reserve Banking in a Bitcoin Age; Physicalist Metaphors” (May 8, 2023)
• Stephan Livera “In Defense of Bitcoin Full Reserve: Not Anti-Credit, but Anti-Fiduciary Media,” Bitcoin Magazine (July 28, 2022)

I’ve never come across an economist willing to explain clearly why Bitcoin’s fixed supply is a problem and why and how this compromises its operation.

It would be enormously valuable if someone literate could present that argument in a form more coherent than snarky eyerolls.

— Saifedean Ammous (@saifedean) October 19, 2019

🧵The literature concerning why a constant money stock would pose problems is vast, with contributions from economists from numerous schools, including Hayek among other Austrians and sympathetic fellow travelers like the great Leland Yeager. https://t.co/ZNXSQX2TDu

— George Selgin (@GeorgeSelgin) October 13, 2026

Statements in Saifedean’s next paragraph suggest that he understands neither the meaning of monetary equilibrium nor that of the “demand for money.” The last means demand to _hold on to_ (HODL, if you will) money balances, not simply willingness to receive money! 1/2 https://t.co/XDL6CHkjFd

— George Selgin (@GeorgeSelgin) October 14, 2026

If Saifdean hasn’t “come across” these arguments, which pertain to any sort of fixed money stock regime (so it hardly matters that they don’t specifically refer to Bitcoin), it can only mean that he is woefully unfamiliar with the literature on monetary economics.

— George Selgin (@GeorgeSelgin) October 13, 2026

I have wasted enough time on Selgin’s snarky eyerolling. I flew to NYC to debate him on this and he got on stage and said he thinks bitcoin’s monetary policy is good and has no problems, and instead he repeated silly bcash shitcoiner talking points about bitcoin that have nothing to do with monetary policy. Then, behind the safety of his keyboard, he went back to snarky content-free eyerolling about bitcoin’s monetary policy, failing to make a single substantive point. You can search for our discussions and see the video below. I have a strict policy of never letting anyone waste my time twice in life, and Selgin blew his chance. I saw a comment from Larry White referencing his book, I will check that out and answer it.
https://t.co/sNqxAyITG1

— Saifedean Ammous (@saifedean) October 12, 2026

https://t.co/DtJw78w8iQ

— Saifedean Ammous (@saifedean) October 13, 2026

Does Bitcoin’s Fixed Supply Make It Too Volatile to Be Money? A Response to Professor White

Oct. 13, 2026
Saifedean Ammous

In response to my tweet [11] on the lack of a coherent argument against bitcoin’s fixed supply, Professor White has suggested to me that his book Better Money: Gold, Fiat, or Bitcoin?[1] offers an explanation for why Bitcoin’s fixed supply makes it too volatile to function as a medium of exchange. I am grateful to Professor White for responding and for writing this book, as it represents the first coherent attempt I have encountered to present this argument. Yet I remain unpersuaded. Below is my response to him:

My disagreement begins with your explanation of gold’s monetary success. You treat its supply response as an advantage over Bitcoin, but resistance to supply expansion is precisely what distinguishes gold from inferior monetary commodities.

If a supply response were desirable in money, why did monetary competition favor the commodity whose existing stock was hardest to expand? Why gold rather than copper, iron, or tin? Divisibility, durability, and recognizability cannot explain this by themselves: other metals possess these properties. If responsiveness of supply to demand changes were desirable in money, then other metals would have been superior as money to gold and would have triumphed in the market competition. Yet the opposite happened; even though iron, copper, and silver were more commonly used as money in the distant past, they all eventually got demonetized and replaced by gold. My explanation is that gold’s extraordinarily low supply growth rate makes it the best physical money, precisely because it responds least to changes in demand. Your argument requires explaining why the characteristic that made gold superior to other commodities becomes a disadvantage when Bitcoin improves upon it.

More fundamentally, why must an increase in monetary demand be offset by additional production? With a fixed supply, greater demand raises money’s purchasing power. The existing stock can then accommodate larger real cash balances. Holders are rewarded for having saved. Nothing requires restoring the previous purchasing power by producing additional units.

I also reject the idea that there is any value from increasing the supply of money in principle. Of course demand for money is larger than the existing supply. Of course everyone would like more money. But the way to have more money is to produce things others value. It is an imperfection in money that it can be produced without providing value to others. So responsiveness to demand is truly undesirable, and gold emerged as money precisely because it is the least responsive to it. A supply response simply undermines the ability of the money to hold its value to the future. A revaluation response enhances it. Subjectively, I would prefer the revaluation response, but it does not matter what I prefer. The money that responds by revaluing upwards will hold and increase its value over time far better than the money that responds by increasing its supply, and this means, in the long run, the revaluing money will have a higher valuation, larger cash balances, and more moneyness.

The same logic applies when demand declines. Then, money becomes cheaper for people who wish to accumulate it. It is clear to me that the significant price response and limited supply response enhance gold’s monetary function and its operation as money: they encourage selling money when the demand rises, and they encourage buying money when the demand declines. Supply and demand adjust through price without the quantity needing to adjust. Neither adjustment requires industrial consumers or miners to rescue the monetary system. If the adjustment were desirable, gold would have been replaced by copper, silver, iron, or another metal.

I am not merely arguing that nonmonetary demand does not help gold’s monetary role; I am saying it undermines that role by making the value of money depend on factors unrelated to monetary demand, which is undesirable. Nonmonetary demand introduces its own uncertainties. A monetary asset exposed to industrial demand must respond to changing technologies, substitutes, and business conditions unrelated to people’s demand to hold money. Why should adding these disturbances produce greater monetary certainty? I believe ideal money is pure money, with a predictable supply and no industrial uses competing with its monetary function.

Mises seems to agree. In Chapter 13 of The Theory of Money and Credit, he described as “considerable disadvantages” the influence on money’s value of “fluctuations in the conditions of production of the metal and variations in the industrial demand for it.” [4] He observed that these effects were relatively small for gold and silver, and argued that these metals are desirable as money precisely because these fluctuations are so small compared to the fluctuations government control of money would produce. In other words, the industrial and mining variations in gold and silver are imperfections, not desirable features. Gold and silver are money despite these factors, because all other moneys (until then) had these imperfections to a larger degree. Bitcoin advances that monetary ideal: predictable issuance, no industrial demand, and no discretionary issuer.

I think the gold-mining numbers support my argument. The World Gold Council estimates average stock growth of only 1.7% annually over the past 25 years. At that rate, even a 10% increase in annual production adds only 0.17 percentage points to stock growth. [2] I do not see how these tiny variations explain monetary stability against the enormous demand disturbances caused by government monetary manipulation. This is also a self-defeating dynamic: every additional ounce enlarges the stock, making it necessary to produce still more just to maintain the same percentage growth rate. Because of gold’s indestructibility, the growth rate stays bound in the same range and seems essentially unrelated to increases in price, which means the supply response to increased demand does not really exist. It’s held to the 1-2% range throughout the entire last 125 years, with two world wars, demonetization of gold, great depressions, stagflation, and much more. Gold production in 1980 was lower than it was in 1970, even though the price had increased ~20x in the period. The annual growth rate in gold supply during the 1970s, when gold rose from $35 to $800, was around 1.34%, but in the 1980s, when the price declined from $800 to $380, it was 1.39%. Both of these numbers are practically rounding errors next to the enormous variations in demand caused by monetary factors. If a ~20-fold increase in price over a decade elicited roughly the same supply response as a 50% decline over another decade, then the supply response is practically nonexistent. The literature on gold supply elasticity finds no effect, a very small effect, or a very small effect after several years or decades [10], suggesting the effect is largely irrelevant to responsiveness to market conditions. If supply takes a decade or two to respond, monetary demand and price will have responded, and monetary conditions will likely have reversed before any supply response makes a difference. It is difficult to see how this supply response works, which makes it even harder to imagine that it is what makes gold a superior money. Gold’s supply increases by about 1.5 to 2% every year, regardless of market demand or monetary policy.

The short-run argument about converting jewelry into monetary gold is also unpersuasive. People hold jewelry to preserve wealth, so it functions as a monetary asset as well. Melting a necklace into a coin changes its form without creating additional gold or adding to monetary wealth. Your chapter itself notes the use of gold jewelry as collateral for loans. Further, industrial uses of gold are historically an insignificant source of demand for gold. Most of these uses emerged in the twentieth century with modern technology, and with the demonetization of gold, making it doubly unlikely they reinforce its monetary role. Gold served as money for thousands of years before modern industrial applications emerged. Industrial demand neither created gold’s monetary role nor explains its enormous liquidity and relative stability. These emerged from the accumulation of monetary cash balances, made possible by gold’s resistance to supply expansion. [3]

My explanation of monetary stability therefore runs in the opposite direction from yours. Resistance to dilution makes an asset attractive for saving. Growing demand raises its value, allowing it to hold larger cash balances. Those balances support deeper liquidity, narrower bid-ask spreads, and greater salability. Stability emerges as this monetary market grows. Your diagrams do not explain how resistance to dilution affects the demand to hold money. The size and persistence of cash balances depend on the monetary properties that make people willing to accumulate them over the long term, and that allow the money to hold and increase its value over time.

Gold earned its monetary position through its exceptional hardness; attributing that position to its tiny, almost imperceptible supply elasticity is like saying Usain Bolt won the 100m dash because he slowed down near the finish line. Yes, he slowed down, but he won because he is the fastest. If you invented a synthetic Usain Bolt that didn’t slow down near the finish line, he would beat the real Usain Bolt.

Your chapter rejects the claim that rising capitalization reduces Bitcoin’s volatility, stating: “In fact, as its market capitalization has grown over the years, Bitcoin’s volatility has not declined.” This was likely wrong when you published it if you had zoomed out, but by now, it is extremely difficult to substantiate.

Fidelity documents a long-term decline in realized volatility as you can see in this chart. [5] @JPMayall has shown that bitcoin’s 252-day realized volatility has now declined to only 1.52 times gold’s. [6] Kim-Oanh Do recently published an important paper showing that bitcoin’s mean daily realized volatility is declining significantly, while bid-ask spreads are narrowing, price impact and illiquidity have decreased, and volatility shocks are having smaller, less persistent effects on liquidity. [7] Charles Schwab also finds a significant reduction in annualized historical volatility. [8] Huda Aldhahi finds similar results from hourly observations. [9] More such studies can be found.

Bitcoin’s market capitalization and liquidity have grown while its volatility has declined markedly. This is exactly what I would expect. Gold did not become the least volatile monetary asset because it had the best supply response, but because it had the worst supply response. Since bitcoin has invented an even worse supply response, I would expect it to become even less volatile when it matures and becomes larger in market capitalization than gold.

Holding bitcoin is using bitcoin as money, and its lack of widespread use as a method of payment is no evidence against its monetization unfolding. Someone holding Bitcoin to preserve purchasing power for future expenditure is holding a monetary balance. Calling that person an investor does not remove the monetary purpose of the holding. Accumulating savings is central to how an emerging money develops the liquidity needed for widespread exchange. Bitcoin is still around 1% of global cash balances, but it is growing fast. Expecting it to have widespread payment adoption before it develops significant cash balances puts the cart before the horse.

This also answers your Tesla comparison: you argue that the rise in Tesla stock will not turn it into money. A share’s value depends on a company’s profits, costs, management, competitors, and issuance decisions. These introduce countless nonmonetary influences. A company facing enormous demand for its shares as money can issue more shares to finance its operations, compromising the scarcity that attracted monetary holders. Bitcoin has neither that issuer nor that exposure to a particular business. Comparing their capitalization without considering these differences misses the argument.

I expect continued Bitcoin monetization to produce larger cash balances, deeper liquidity, and lower volatility. Its transition from a newly invented asset to global money necessarily involves substantial appreciation and price discovery. In a mature Bitcoin standard, I expect appreciation to become slow and steady, reflecting economic growth rather than the rapid acquisition of monetary demand. There is no reason to assume the transition’s volatility describes the destination.

Finally, dollar stablecoins are economically a new form of fiat checking balance. Their transaction volumes belong to the already dominant fiat monetary network, which is of course far larger than that of bitcoin. Their popularity tells us that people use dollars and appreciate better ways to transfer them. It does not establish that Bitcoin needs an elastic supply, any more than the popularity of ordinary checking accounts does.

The question remains: gold’s outstanding monetary characteristic was how difficult its stock was to expand. Why should eliminating the remaining possibility of dilution make a better version of that monetary property into a defect?

References

1. Lawrence H. White (2023), Better Money: Gold, Fiat, or Bitcoin?, Chapter 6, “Comparing and Contrasting Gold and Bitcoin Standards,” Cambridge University Press.
2. World Gold Council (2026), Gold’s Key Attributes: Return. Documents average annual above-ground gold stock growth of approximately 1.7% over the preceding 25 years. https://www.gold.org/goldhub/research/gold-strategic-asset-2026-edition/20447
3. U.S. Geological Survey, Gold Statistics and Information. Discusses gold’s historical monetary role and its emergence as an important industrial metal in the late twentieth century. https://www.usgs.gov/centers/national-minerals-information-center/gold-statistics-and-information
4. Ludwig von Mises (1912), The Theory of Money and Credit, Part II, Chapter 13, “Monetary Policy,” Section 5. English translation by H. E. Batson, Liberty Fund edition. Full Mises quote:
   

   “The significance of adherence to a metallic-money system lies in the freedom of the value of money from state influence that such a system guarantees. Beyond doubt, considerable disadvantages are involved in the fact that not only fluctuations in the ratio of the supply of money and the demand for it, but also fluctuations in the conditions of production of the metal and variations in the industrial demand for it, exert an influence on the determination of the value of money. It is true that these effects, in the case of gold (and even in the case of silver), are not immoderately great, and these are the only two monetary metals that need be considered in modern times. But even if the effects were greater, such a money would still deserve preference over one subject to state intervention, since the latter sort of money would be subject to still greater fluctuations.”

   https://www.econlib.org/library/Mises/msT.html?chapter_num=17

5. See: https://institutional.fidelity.com/app/proxy/content?literatureURL=%2F9922887.PDF
   Also: Zack Wainwright (May 1, 2024), A Closer Look at Bitcoin’s Volatility, Fidelity Digital Assets. Documents Bitcoin’s declining realized volatility over time, including comparisons at similar price levels in different market cycles. https://www.fidelitydigitalassets.com/research-and-insights/closer-look-bitcoins-volatility
6. https://x.com/jpmayall/status/2107851442031566871
7. Kim-Oanh Do (2026), Maturation of Bitcoin Market Microstructure: Evidence of Institutional-Grade Liquidity and Efficiency (2012–2025), Borsa Istanbul Review, 26(5), 100869. https://www.sciencedirect.com/science/article/pii/S221484502600089X
8. https://www.schwab.com/learn/story/bitcoin-volatility-shrinks-to-magnificent-7-levels
9. https://www.mdpi.com/1911-8074/19/9/707
10. Here is an AI summary of the literature on gold supply responsiveness:
    The empirical literature finds that gold supply responds to its real price late and weakly. For the classical gold standard, Rockoff (1984), the usual source for an induced response, presents no regression. He argues that the Klondike discoveries and the cyanide process were price-induced, and concludes that “the long and uncertain lags that marked the supply response hardly seem acceptable in an impatient age” (p. 640). The one regression from the period is an unpublished estimate by Rush, reported by Barro in his comment on Rockoff. It uses seventeen annual observations (1897–1913) and gives an output elasticity of 0.3 in the short run and 1.7 in the long run, with simultaneity unaddressed. Bordo (1981) correlates the detrended world monetary gold stock with the purchasing power of gold over 1821–1914. He finds that purchasing power leads the stock by 25 years (r = 0.44), or by 16 years for 1879–1914 (r = 0.86); his stronger result runs the other way, with the stock leading purchasing power by two years (r = −0.64). Cagan (1965) declined to compute a correlation and reported only that the relation was positive with a long lag: prices fell from the early 1870s, but output did not expand appreciably until after 1890. Eichengreen and McLean (1994) conclude that the influence of real gold prices on production “is not dominant” and that conditions such as prior settlement mattered.
    Modern studies point the same way. In the short run the response is zero or negative. Marsh (1983) confirms Keynes’s observation that a higher price leads mines to work lower-grade ore and produce less. Cooper (1982) notes that world output declined through the 1970s while the real price rose four- to fivefold. Schwartz (1987, p. 386) writes that production “responds sluggishly to changes in market price and since the 1960s has responded perversely.” A positive response appears only after several years. Selvanathan and Selvanathan (1999) find none within a year and roughly unit elasticity after five in Western Australia (1948–94). Rockerbie (1999) finds about eight years for South Africa (1970–95), and the World Gold Council (2026) puts the lag at “at least six years.” Nearly all of this work measures mine output, not the stock. Of these studies, only Bordo searches across lags, and none estimates how much of a price reversal the added supply accounts for.
11. https://x.com/saifedean/status/2109259766895988815