Key takeaways
- Gold rose 67% in 2025 while bitcoin fell 6%, so one debasement story can't be doing much explanatory work for both assets.
- In the World Gold Council's regression, US fiscal stress carried a coefficient of 0.02 against the dollar's -0.74, over three years of daily data.
- Erb and Harvey found real yields explained 9% of the UK real gold price, while a meaningless time trend fitted better at 0.87.
- US inflation hit 9.1% in the year to June 2022. Bitcoin fell 64.4% over that calendar year and gold gained just 0.4%.
- Central bank net gold buying fell 21% to 863.3 tonnes in 2025, the same year the gold price rose 67%.
Two people, one worry — and if you hold gold or bitcoin, it's probably your worry as well. Governments are borrowing more than they can ever repay, both of them think, and sooner or later they'll inflate their way out of it. So one buys gold. The other buys bitcoin, on the reasoning that a fixed supply schedule is the purer version of exactly the same idea.
Say they each did that on the last trading day of 2024, and put in the same money. Over the following year the LBMA gold price went from US$2,609.10 an ounce to US$4,367.80, a gain of 67%. Bitcoin went from $93,390 to $87,517, a loss of 6%. So the ounce your first friend bought for US$2,609.10 is worth US$4,367.80 by the end of the year, and the identical sum your second friend put into bitcoin has come back smaller than it went in.
Same trade, same twelve months, more than 70 percentage points apart. So which of them was right about debasement?
That's the awkward question sitting underneath the whole thesis. You'll have heard gold and bitcoin sold in the same breath, as the two things governments can't print. Before you treat that as a reason to hold either one, it's fair to ask whether they behave like one idea at all.
The debasement trade needs a testable form
Stated loosely, the debasement trade says fiscal deterioration pushes capital out of government paper and into scarce assets. Deficits widen, debt ratios climb, real yields get suppressed, money keeps growing, and gold and bitcoin rise.
The loose version can't be wrong, and that's its problem. If gold rises, that's debasement. If gold falls, the market hasn't woken up yet. If bitcoin falls while gold rises, bitcoin is early or gold is the safer expression. Every outcome confirms it, which means no outcome tests it.
A testable version has to name variables and a sign. Gold and bitcoin returns should rise with deficit ratios, debt-to-GDP and money growth, after controlling for the dollar and real yields. That version is checkable. Most of what you'll read about the debasement trade isn't.
The fiscal facts themselves aren't in dispute. The US federal deficit ran at 5.77% of GDP in fiscal 2025, on the Federal Reserve Bank of St Louis series. Gross federal debt reached 122.6% of GDP in the first quarter of 2026. M2 grew 5.5% in the year to June 2026, and 56.5% since June 2019. The same pressure shows up worldwide, and it's the subject of our piece on global public debt at 93.9% of GDP and the 117% tail scenario for 2028. What's disputed is whether any of it explains asset prices.
Fiscal stress shows up, but faintly
The World Gold Council ran the direct test in June 2025. It regressed log daily gold returns on three variables: a fiscal stress proxy, the DXY dollar index, and the US 10-year real yield. The sample ran from 15 June 2022 to 15 June 2025. Three variables, three results:
| Variable | Coefficient | t-statistic |
|---|---|---|
| Fiscal stress proxy | 0.02 | 2.1 |
| DXY dollar index | -0.74 | -11.5 |
| US 10-year real yield | -0.01 | -1.8 |
Fiscal stress came out significant. Its coefficient was 0.02, with a t-statistic of 2.1 and a p-value of 0.025. Stop reading there and the thesis looks vindicated. So does that make fiscal stress the reason gold moved?
Look at the company it keeps. The dollar index carried a coefficient of -0.74 with a t-statistic of -11.5. The real yield came in at -0.01 with a t-statistic of -1.8. The dollar term is thirty-seven times the size of the fiscal term. Statistical significance and economic significance are different things, and here the gap between them is the whole finding.
There's a second problem with the fiscal variable. The proxy is the US 10-year Treasury swap spread. The Bank for International Settlements looked at why swap spreads turned negative and attributed it largely to plumbing: dealer balance sheet costs, repo funding, quantitative tightening and the mechanical effect of moving from Libor to SOFR. Some of that spread is fiscal. A lot of it is intermediation capacity. The Council says as much itself, noting that a fuller analysis would be needed to judge the proxy.
The inflation link is weaker than the story needs
Debasement is meant to arrive as inflation, so gold's inflation-hedging record is the natural place to look. Erb and Harvey examined it across 1975 to 2011 and found effectively no correlation between gold and unexpected inflation. What positive relationship exists is driven by a single year, 1980.
Their 2024 follow-up is blunter. The correlation between ten-year nominal gold returns and ten-year inflation rates is essentially zero. The correlation between ten-year nominal gold returns and ten-year real gold returns is about 0.98. Gold's returns are the real price of gold moving. Inflation is close to irrelevant over the horizons you'd actually invest across. Tested window by window, the gold inflation hedge held in 53.1% of one-year periods since 1975 and 53.4% of ten-year periods.
So what are you holding, if not a hedge? You're holding an asset whose value moves with the real price of gold. It isn't a claim indexed to your cost of living, however it was sold to you.
The currency version fares no better. Across seven currency pairs gold's betas ran from -0.09 to -0.24, with R-squared values of 6.3% to 15.0%. Gold isn't a clean hedge of any single currency's decline, including the one your bills arrive in.
Erb and Harvey also give the valuation warning. The ratio of the gold price to the US CPI index averaged about 3.2 over their sample and stood at 7.31 when they wrote. High readings have been followed by weak real gold returns. That's mean reversion, not debasement, and it points the opposite way.
The money supply version is an identity, not a forecast
The most concrete form of the argument is the one you've probably been shown on a chart: divide the money supply by official gold holdings and read off a price. Erb and Harvey walk through it. With a US monetary base of $2.7 trillion and official holdings of 8,300 tonnes, the arithmetic gave a shadow gold price of about $10,000 an ounce.
That looks like a valuation. It's an accounting identity with a free parameter. The Federal Reserve publishes a monetary base, an M1 and an M2, and each one produces a different answer. Nothing in the argument tells you which to use. Pick the aggregate that gives the number you already believe and the identity will confirm it, which is why the useful thing to ask of one of these charts is which aggregate sits on top of the fraction.
It's also the inflation-hedge claim in different clothing, which is how Erb and Harvey read it. If gold doesn't track realised inflation over investable horizons, there's no reason a money-supply ratio should do better.
The strongest objection to the whole thesis
The best counter-argument doesn't come from gold sceptics. It comes from people who think gold is perfectly explicable without any fiscal story at all. Their claim is that real yields do the work. Gold pays you no coupon, so when the real return on inflation-linked bonds falls, holding gold costs you less and the price rises. No debasement required. The 2021 to 2025 record of each inflation sleeve fits that reading better than it fits the fiscal one, because gold's gain arrived after the price shock rather than during it.
The evidence for that looks strong. From the launch of US TIPS in 1997 to 2012, the correlation between the ten-year TIPS real yield and the real price of gold was -0.82. That's a tight fit for a single variable. So why isn't that the end of the argument?
Because Erb and Harvey stress-tested it, and this is the part that matters. Using UK index-linked data, which starts in the early 1980s and covers a longer window, the same correlation drops to -0.31. Real yields explain roughly 9% of the variation in the UK real gold price. The relationship isn't stable across samples.
Worse, over that US window real yields correlated -0.90 with a simple time trend, and the real gold price correlated 0.87 with a time trend. A variable with no economic content fits gold better than real yields do. That's the signature of a spurious relationship, and it's why the -0.82 figure shouldn't be treated as a mechanism.
Recent attribution work says the same thing from the other direction. The World Gold Council's own return attribution model, covering the first half of 2026, assigns 24% of gold's price variability to momentum, 17% to risk and uncertainty, 14% to currency and 12% to economic expansion. Rates get 3%. Thirty per cent is unexplained. For the whole of 2025 to 28 November, gold returned 61% and the model's named drivers accounted for 41 percentage points, leaving 20 points in a residual bucket the Council attributes partly to central bank buying.
So the debasement story and the real-yield story fail in the same way. Both are fitted after the fact, both work in one window and not another, and neither has a coefficient stable enough to trade on. Our piece on why real yields at 2.41% reprice long-duration assets covers what real rates do explain well. Gold isn't on that list.
Where bitcoin breaks the debasement trade story
Bitcoin's supply schedule is fixed, which makes it the purest version of the debasement argument. It also gives the thesis its cleanest failure.
Picture the experiment you'd design if you wanted to prove bitcoin hedges inflation. You'd want a huge inflation surprise, and you'd want to watch bitcoin hold its value through it. That experiment has already been run for you. US consumer prices rose 9.1% in the year to June 2022, a forty-year high. Bitcoin fell 64.4% over that calendar year. Gold gained 0.4%. The single largest inflation surprise of the modern era produced the worst annual result in bitcoin's institutional history.
Academic work finds a hedge, but a fragile one. Rodriguez and Colombo ran a VAR on monthly data from August 2010 to January 2023 and found bitcoin returns rise after a positive CPI surprise. Swap CPI for core PCE and the response turns negative. Drop the early sample, before institutional adoption, and the hedging property largely disappears. They conclude the effect is context-specific and diminishing.
A 2026 study from University College London and Deutsche Bank pushes further. Using a monetary policy expectations index built from over 118,000 classified market messages, the authors find hawkish central bank narratives trigger negative bitcoin price responses independently of what the Federal Reserve actually does to rates. Their conclusion is that bitcoin is a procyclical exposure to global liquidity conditions rather than a structural hedge. On that reading, what you own is a leveraged bet on easy money, not a hedge against loose money.
The recent record agrees. M2 kept growing through 2026, yet bitcoin fell 26.7% between the end of 2025 and 4 August 2026. Money creation didn't stop; the asset that's meant to price it fell by a quarter. If you want to see how the flow data behind these moves gets constructed, our explainer on what CoinShares crypto fund flow data really measures sets out the limits of reading intent from flows.
Central banks are where the fiscal link actually holds
There's one place the debasement thesis finds real support, and it isn't in price data. It's in who buys gold and why.
The IMF studied the determinants of gold's share in official reserves across 144 economies from 1980 to 2021. The gold share is negatively related to the fiscal balance, and the effect is larger for advanced economies than emerging markets. Weaker public finances, more gold. That's the debasement thesis' best evidence, and it's genuine.
The same paper undercuts it. The level of public debt to GDP is negatively related to the gold share in emerging markets, and shows no relationship at all in advanced economies. If debt levels drove gold accumulation, that sign would run the other way. Two fiscal variables, two different answers.
What the paper does find cleanly is seizure risk. Financial sanctions from the US, UK, EU or Japan raise a country's gold share by around 2 percentage points, and multilateral sanctions matter more than unilateral ones. All 14 of the identified active diversifiers into gold are emerging markets. That's a story about assets nobody can freeze, not about assets nobody can print.
Then there's the timing. If official buying was the engine under the 2025 price, what happened when the engine slowed? Central bank net purchases fell 21% in 2025, to 863.3 tonnes from 1,092.4 tonnes in 2024. That was the year gold rose 67%. The buyer credited with the move stepped back while the price ran, which our coverage of gold demand reaching 5,002 tonnes in 2025 sets out in detail.
2026 so far makes the point again. Go back to the two friends from the opening and leave both positions alone. Between the end of 2025 and 4 August 2026, gold fell 6.5% and bitcoin fell 26.7%. Deficits didn't shrink over that stretch and debt ratios didn't fall. The fiscal picture that's meant to drive both assets got no better, and both of them went down anyway.
What would change the conclusion
Several things would move this. A fiscal variable with a large coefficient that holds out of sample, across countries and across decades, would do it. So far nobody has produced one.
Gold and bitcoin moving together on fiscal news would help, rather than 67% apart in a single year. So would bitcoin holding its value through a year when inflation surprises high and equities fall. It hasn't managed that yet.
A formal attribution model that adds a fiscal driver and finds it explains a meaningful share of gold's return would be strong evidence. The Council's model doesn't contain one.
On the other side, Erb and Harvey's mean reversion could simply fail. If the real gold price stays at roughly twice its historical average for another decade, their framework loses its predictive claim, and the case for a structural regime change gets stronger.
If you want to check any of this yourself, here's what it rests on, weak points included. The Council's regression is three years of daily data with one imperfect proxy. Erb and Harvey's samples end in 2012 and 2024, and their valuation call has been wrong for a while. The IMF panel stops in 2021, before the largest official buying wave on record. Bitcoin has lived through roughly one and a half inflation cycles, which is not enough to settle anything.
What survives all of it is modest. Fiscal stress is detectable in gold's price, at about a thirty-seventh of the dollar's weight. Central banks with weaker fiscal positions do hold more gold, though sanctions risk explains their behaviour better than debasement does. And bitcoin's response to inflation depends on which price index you pick and which years you keep, which is another way of saying it isn't a finding yet.
None of that makes gold a bad asset. It makes the debasement trade a bad reason to own it. Which leaves you with a narrower and more honest question than the one the two friends started with: if not that, what is the position in your portfolio actually for?