Executive Summary

• What if Gold Replaces US Treasuries?

  One of the most frequent questions of the past few years in economics and finance has been “will the US dollar (USD) maintain its reserve status, and what does it mean if it loses this status?”

• Central Banks Care and are Reacting

  The question became serious after the Global Financial Crisis (GFC) and acute after US sanctions on the Central Bank of Russia’s US treasury reserves.

• Developed Markets (DMs) Always Stimulate in Response to Adversity

  DMs are arguably subject to “fiscal dominance”, a common situation in which central banks lose traction due to high government debt. This refocuses attention on the central bank balance sheet.

• The Definition of Money Keeps Expanding From M-0 to “M-infinity”

  A quick reminder that the M0, M1, M2, M3 aggregates came about due to runs. The Fed and Treasury backstopped the global financial system during the GFC, and that modality was repeated in the 2020 Lockdowns.

• We Are Just Dividing Two Numbers – Money Liabilities by Gold Reserves

  All we do is divide money liabilities (M0, M2) by gold reserves.

• Central Bank M0 Money Liabilities Divided by Gold Reserves = $39,000 Per Ounce (weighted by daily FX turnover)

  Using a sample of major global central banks’ M0 money liabilities by their gold reserves and weighted by their share of global daily FX turnover, we calculate the price of gold that equalizes M0 is $39,210.

• Central Bank M2 Money Liabilities Divided by Gold Reserves = $184,000 Per Ounce (weighted by daily FX turnover)

  Using global M2 (where we have a global series unlike for M0), we calculate the global price of gold that equalizes M2 is around $100,000 per ounce. Weighted by daily FX turnover (for our sample), the gold price that equalizes M2 is $184,211

• Our View on US Dollar Status

  Our view remains that the USD will not lose its reserve status, rather it will gradually share that status with other deserving currencies, including gold but also government bonds of fiscally sustainable emerging markets (EM) countries.

• Fun Observations

  The poster children for DM fiscal dominance - the UK and Japan - look incredibly levered on these metrics and Gilts and Japanese Government Bond (JGBs) trade like it. Kazakhstan and Russia have enough gold and so few money liabilities that they could peg their currencies to gold. China would have to purchase around 325 million troy ounces in order to have a peg to gold (they currently have 74 million ounces excluding state banks). South Africa’s gold backs roughly 60% of Rand M0 and their 10-year bonds pay over 8%, while Japan’s gold backs only around 3% of Yen M0 and their 10-year bonds pay just under 2%.

What if Gold Replaces Treasuries?

One of this era’s key questions in economics and finance has been “will the USD maintain its reserve status, and what does it mean if it loses this status?” One can say the question is ridiculous, very unlikely, or even that it’s your central case. We are explicitly not discussing the likelihood of this scenario other than by restating our longstanding view that the USD will not lose its status but will likely gradually share this status with other deserving currencies. But again, the point of this paper is not to make that case one way or another. The point of this paper is simply to answer the question asked – what would it mean if the USD lost its reserve status, regardless of what one thinks of the question. Too often, the answer to the question is prose or maybe technical analysis normalizing the gold price move relative to an economic variable (like inflation, fiscal policy, interest rates, money supply, etc.).

We attempt here to have a precise answer on the gold price, across all the major currencies/central banks, using consistent econometrics and a simple framework. Each gold price in our exhibits will be that central banks’ “answer” to the question on what the impact of a loss of USD status would be on the gold price (for that central bank). Presumably any such development would affect all balance sheets, so the output could be especially useful in their relative effects (between the central banks).

Central Banks Care and are Reacting

The question about USD reserve status became a legitimate concern after the further rounds of monetary and fiscal forbearance in the GFC but became acute after US sanctions on the Central Bank of Russia’s US treasury reserves. We wrote our initial paper in August 2012, (before sanctions) following the forensic understanding of the GFC (remembering that the contemporaneous understanding of the GFC being a “one in a thousand years storm” differed significantly from the ex-post analysis, the heroes of which are Laurence Kotlikoff of Boston University and Mark Pittman and Bob Ivry of Bloomberg News, which sued the Fed for documents and won). Our impulse was simple – something that wasn’t supposed to happen to US markets happened. Our framing was also simple – the global financial system was essentially backstopped by the Fed and Treasury in the GFC, and the re-emergence of this forbearance in the 2020 lockdowns means they are permanent.

Sanctions, though, added urgency to this general concern over “fiscal dominance” and the examples of endless monetary forbearance above. Sanctions injected actual concerns of total loss of your asset due to politics, which happened to the reserves of the Central Bank of Russia (CBR), and prominent economists have described it as a form of default. This is clearly the opposite feature one wants in a reserve asset, so the permanent rise in concern on the part of reserve managers was inevitable. Moreover, sanctions were imposed by countries with large domestic and external financing requirements – it is a clear sign of a deep commitment to sanctions that DMs are willing to sanction their lenders despite their large offshore borrowing requirements. Sanctions triggered the latest move higher in gold prices.

Exhibit 1 – Central Bank Gold Purchases Surge After Sanctions

Source: Bloomberg, IMF. As of September 30, 2025

DMs Always Stimulate in Response to Adversity

DMs employ monetary and fiscal forbearance as a policy tool with little reluctance, viewing higher risky asset prices as a means of boosting demand (the “portfolio balance channel”), whereas EM authorities are much more cautious about taking on liabilities and more focused on inflation and financial stability. We like to remind that Thailand let its largest bank, Finance One, default in the 1997 Asia crisis, whereas the US changed the definition of “state-owned” (as the IMF’s largest shareholder) when the IMF told them to clarify the status of Fannie and Freddie during the GFC. EMs cannot afford these expansions of government financial liabilities and therefore avoid them. The US jumped in with forbearance in its GFC, though. As EM economists, this reminded us of the “fiscal dominance” and resultant monetary forbearance that used to characterize EMs before their crises in the late 1990s. We’ve written about this extensively in other reports, but we should highlight that the problem in the GFC was not the “save” by monetary experimentation (QE, etc.) which was likely necessary. The problem was the absence of moral hazard – the “save” became permanent. For example, US Treasury borrowing from offshore hedge fund financing using near-zero-cost repo is the latest obvious example.

The Definition of Money Keeps Expanding Due to Runs – The Story of “M”s

A quick reminder on economics and history. You might remember your monetary policy class in which you are taught the central bank’s money liabilities. M0, M1, M2, M3, etc. It is worth reminding how those aggregates became money liabilities as a result of runs. The “story” is: once runs and crisis interventions made clear which promises the state would effectively stand behind as cash‑equivalents, central banks recast M0, M1, and M2 as structured stacks of specific short‑term liabilities—central bank reserves and notes, then bank deposits, then near‑money claims like money market mutual funds (MMFs)—rather than as a loose conceptual spectrum of liquidity. MMF runs exposed which instruments the public really treated as payable‑on‑demand cash. The MMF crises in 2008 and 2020 then forced regulators to pin down whether MMF shares were effectively part of broad money (and run‑prone bank‑like liabilities), tightening the perimeter of what is regulated as money and what remains a portfolio. Exhibit 2 summarizes how they were and are defined in the US.

Exhibit 2 – US “M”s Today and Yesterday

Source: Federal Reserve, Richmond Fed, Investopedia, Wikipedia

Let’s quickly describe how these “M”s came about due to runs. M0, M1, M2, and M3 emerged as increasingly broad “run maps” of what the public actually treats as money once crises force governments to decide which promises they will back.

• Early era – vague “money in circulation”: Initially, official statistics focused on currency and bank reserves (what is now M0/monetary base), because notes and specie were the things people ran to and banks ran out of in panics.​
• Deposit runs → M1: After repeated banking panics and then the Great Depression, policymakers realized sight deposits behave like money in a run. “Narrow money” (M1) was defined as currency plus checkable/demand deposits, i.e., exactly the instruments that can be withdrawn immediately at par and cause classic retail bank runs.
• Savings/time deposits, thrifts, MMFs → M2/M3: Post‑war financial innovation and the 1970s–80s disintermediation waves showed that savings accounts, small time deposits, and then money‑market funds are also runnable when rates move or confidence cracks. Broad aggregates (M2, later M3) were built to add those “near‑money” claims that had shown, in crises, that the state would likely protect or support them.​
• Why the layers matter: Each step up the ladder reflects a political recognition: once a category has actually run and forced intervention (thrifts, MMFs, repo), it tends to get pulled into the conceptual money/credit perimeter, even if labels and publication practices change (e.g., the Fed dropping public M3 in 2006 while still tracking broad liquidity).​

We Are Just Dividing Two Numbers: Money Liabilities and Gold Reserves

All we do in the output below is answer the question: what if the USD loses its reserve status and is replaced by gold as the sole global reserve currency? Now, even in such a scenario, gold is extremely unlikely to be the sole reserve. Capital controls would also likely be part of such a mix, but that is part of the usual blah blah blah prose you get, and muddies the calculation aimed to answer a simple question. So, that is a layer (of game theory, basically) we won’t conduct, we assume no capital controls only for this exercise – we use the gold price only to measure strains central bank balance sheet by central bank balance sheet because it is the most complete and econometrically reasonable exercise. And central banks have actually been buying gold and have always treated it as a reserve currency. If one wants to divide the strain to be spread onto more than just gold prices, please go ahead, that’s a commendable exercise, but not one we are conducting. One price, in gold, for every central bank, assuming USD is replaced by gold, is what we are calculating (and then we “add” them up into headline “global” numbers for you, of course). All that is involved in the exercise is dividing gold reserves (on the asset side) by money liabilities (we calculate using only M0 and M2).

Although we noted above that the Fed/Treasury backstopped the global financial system in the GFC, we won’t calculate “M-infinity” itself. Our only point was that we assume this backstop and forbearance is permanent. (We do have an initial internal paper on this “M-infinity” framework and calculation, reach out if you’d like to be part of that discussion.) In any case we already know that dividing infinity by any gold reserves divided by “infinity” money is an extremely high number. We just wanted to remind of that key context. In the following exercise, every central bank has a different amount of formal M0 or M2 liabilities that are known and a different amount of gold in reserves which is known, with all data from the same source/econometrics. We use only central bank balance sheets, not state banks, for econometric consistency, but we would normally include state banks in a country-specific report (but doing that for all sovereigns is econometrically challenging and would require a lot of assumptions). Note that we translate money liabilities into US dollars at mid-December exchange rates; we translate gold reserves at market prices too, of course (the underlying data is number of troy ounces). Whenever we say “ounce” we mean troy ounce.

No individual academic is “credited” with this framework as it is essentially dual-entry accounting at its core, but these measures were called things like the “gold reserve ratio” and they were codified in Britain’s Bank Charter Act of 1844 and the US’ Gold Standard Act of 1900. In EMs, “currency boards” pegging a home currency to a different offshore currency by holding government bonds in/of the offshore currency was one of the standard policy options and your authors have been economists and traders of such countries over the decades (Argentina and Bulgaria are examples). One could also point out that certain methodologies like the “balance sheet approach to money” are similar to what we are conducting. Today, most modern central‑bank and BIS/IMF documents talk about the “central bank balance sheet,” “quality of assets,” “capital and risk buffers,” and “credibility of the monetary authority,” but do not canonize a specific phrase like “asset‑quality valuation model of currency.” Our first paper was authored by Eric Fine and David Austerweil in August 2012, and our follow-up papers have been written by Eric Fine, Natalia Gurushina, and David Austerweil. See our latest IMF recap below. 

Central Bank M0 Divided by Gold Reserves is $39,000 Per Ounce (weighted by global daily FX turnover)

We’ll start by using the M0 liability of central banks, so M0 divided by reserve gold holdings. Already we have an econometric problem because we don’t have a good data series on “global” M0 (we do on M2, more on that later). So, we use M0 for the selected central banks in Exhibit x below and calculated the mean and median outcome for that set instead - $31,612 for the mean, $18,205 for the median. Exhibit x shows the answer on a central-bank-by-central-bank basis, using the existing exchange rates (as of mid-December 2025). Now, (according to the original data from the BIS’s 2025 triennial survey) the USD is 50% of global daily FX turnover in our sample (EUR is 16%, JPY is 9%, etc.), so we applied these weights to the respective country’s money-equalizing gold price (US counts 50%, EUR counts 16%, etc.) That FX-turnover weighted average is $39,210. A separate idea we’re filling in, of course, is that we are also trying to measure the strains on individual central banks relative to each other. All central banks would be under strain in such a situation, but some far more than others, and the relative magnitude of the differences between central banks could be viewed as a measure of the strain. (Again, capital controls would be imposed long before such a scenario materializes, but that kills any ability to make these calculations which are basically measures of the strain itself.) Those central banks whose gold price in these calculations varies the most from current prices would be the most strained. Also note that money liabilities and gold reserves vary considerably, that’s why we’re doing these calculations – it is not the amount of reserve, it is the amount of reserve relative to the liability. And then, those central banks’ ratios relative to each other.

Exhibit 3 – Price of Gold (in USD) That Equalizes Gold/M0 Ratio

Source: VanEck Research, IMF, Bloomberg LP; exchange rates as of December 18, 2025

This framework sure seems to have predicted the core asset price paths (yields higher, currency weaker) of the UK and Japan! These two countries are poster-children for “fiscal dominance” in the DM. Why do we say this here? The numerator in the calculation above is the M0 money liability, and fiscal dominance says central banks are essentially co-opted when government debt is too high. Seems to be the perfect framing, and it’s a reminder that it is not so much the low amount of gold on the asset side, but more the high amount of money liability. We won’t analyze every central bank here; we mainly wanted to riff off of these results to give you a sense of our interpretation. Let’s look at the opposite extreme – Kazakhstan and Russia (countries your authors have long and deep experience in). According to the output above, both countries could have a credible currency board against gold, literally having enough gold to back M0, at below current market prices for gold and for their respective currencies. Remember you get paid a yield in bonds denominated in those currencies, too. Their asset sides (gold) are actually fairly high, but again it is the low money liability that is arguably much more important. (We are not making a policy prescription out of a ratio, just making observations.) These are extreme versions of EM countries that abhor leverage and have to go to great lengths to establish stability is our point. The table above says some have already.

Gold M2/Gold Reserves = $184,000 Per Ounce (weighted by FX turnover)

Now, we do have a good data series on total global M2 and the respective gold reserves, which you see in Exhibit x below. The gold price which equalizes global M2 to gold reserves is circa $100,000 per ounce. You can eyeball the steep climbs during the 2008 GFC as well as following the 2020 lockdowns in the chart. We also did the same calculation as we did above with M0 – weight the M2-equalizing gold price by the major trading FXs (USD, EUR, etc.) and that weighted average is $184,211. As we stated at the beginning, we are not addressing the odds of gold becoming the sole reserve, nor are we addressing whether strains will be from M0, M2, in fact we alluded above that the real strain will be from M-infinity. This context must be kept in mind because the notional of derivatives is incomprehensibly large and many in the market have glossed over the reality of the GFC.

Exhibit 4 – Global Gold Reserves/M2 Nearing $100,000 Per Troy Ounce

Source: VanEck Research; IMF - World Reserve Gold Holdings (001.046 index); Bloomberg - Global M2 CIX index (.glmosupp index)

A Table Summarizing the Calculations

So, we calculated the gold price that equalizes the monetary base for:

The exhibit below lays it out in detail for convenience. Note that we didn’t publish the graph on M2 by every central bank as we did with M0. The ranking of central banks using M0 and M2 was in the same general order.

Exhibit 5 – How Gold Price to Equalize Monetary Base and M2 Weighted by FX Daily Turnover Works

Weighted Averages Based on Daily FX Turnovers (BIS Triennial Survey, 2025)
(weights are rebased from BIS original numbers for the current sample)

Source: VanEck Research, BIS, Bloomberg, IMF

How Did This Output Compare to Our First Calculations in 2012?

Because we’ve been following this issue (largely through the lens of “fiscal dominance”, which forces you to look at the central bank’s balance sheet with greater attention), it’s reasonable to ask what changed since 2012 that is noteworthy. Keep in mind that all our calculations use contemporaneous exchange rates, which have obviously moved, but also official ounces of gold holdings, which also move. We present the results of the intertemporal comparison between our first exercise in 2012 and the latest data in the table below. One thing that pops out immediately is that most emerging markets posted significant improvements over this period (i.e. the “equalizing” price of gold went down), whereas advanced economies found themselves on the opposite end (the “equalizing” price of gold went up…by a lot). This divergence reflected a slower pace of monetary expansion in EM compared to advanced peers, consistent with our framing that “fiscal dominance” characterizes some DMs, while many EMs are not characterizable that way. EMs are more closely scrutinized by the market and rating agencies, and transgressions are swiftly punished. EMs have to work hard for financing, generally speaking, compared to DMs. Thus, EM orthodoxy – we’ve written about this history, and the best initial scrutiny came from the IMF itself with its lending programs and conditionality in the 1997 Asia crisis. And, of course, there was a major pickup in the reserve gold purchases by EM central banks after the GFC in 2008. China’s “no change” status also stands out – the authorities were routinely criticized for their leverage-based growth model, and China’s money supply went ballistic after the pandemic. However, China leads EM (and the world) in the gold reserve accumulation, and the official numbers that we used in this exercise are most likely very conservative because state banks are excluded from calculations as explained earlier.

Exhibit 6 – How These Measurements Changed Since 2012

Source: VanEck Research, IMF, Bloomberg

Note: Hungary is an obvious outlier. It started the series with very low gold reserves (it was in a unique situation for many reasons, two of your authors were sell-side Hungary economists since the 1990s if you want to discuss), and the authorities subsequently embarked on a publicly-discussed gold purchasing program (you can maybe be public if you are small). This is behind its outlier performance, though we aren’t doing country-by-country analysis in this piece; we just wanted to comment on this outcome.

What Are Central Bank Gold Holdings Across the Number of Central Banks?

Some may be interested in the distribution of results by number of central banks, which we show in the Exhibit below. It seems a straightforward conclusion that there are a large number of central banks with very low gold holdings. Also noteworthy is that 50% of central banks hold zero gold according to IMF data, generally poor countries with low overall reserves (i.e., including treasuries, etc.). This is food for further thought, like ‘what might they have to buy’ (putting on our trading hats), with CNY being an appealing option for many central banks, perhaps.

Exhibit 7 – Central Bank Gold Holdings by Number of Central Banks

Source: VanEck Research; IMF; Bloomberg LP

Major Implications for Relative Exchange Rates

In the scenario when/if the dollar loses its reserve status, it would lead to gold prices being higher like suggested by our estimates. But this is not the only implication – this scenario would also lead to a massive relative price adjustment of individual exchange rates vs. the dollar. Using the prices of gold in dollars for each country might give us at least some idea about a new exchange rate for those countries relative to the dollar. Our estimates suggest that some currencies might experience significant appreciations (assuming no intervention from central banks), while others will move in the opposite direction. There will be big shifts both in FX adjustments and gold price increases until they reach a new equilibrium. But this is the topic for the new research paper. Stay tuned!

To receive more Emerging Markets Bonds insights, sign up in our subscription center.