Central banks have been buying a squillion tonnes of gold, promise
Are the numbers real?
Gold continues to soar because of reasons. One of the most cited reasons of late has been all the central bank buying. Per the ECB in June:
The demand for gold by central banks remained at record highs in 2024, accounting for more than 20% of global demand, in contrast to around one-tenth on average in the 2010s.
But how do we know? You’d think that maybe central bank buying would show up in increases in central bank holdings of gold. And it does, sort of. But the data doesn’t correspond to the headlines.
Central banks’ reported gold holdings have increased by 228 tonnes over the most recent twelve months reported. Is this a lot? It’s the weight of about forty African bush elephants, so objectively yes. But 228 tonnes is actually not that much in central bank purchase terms.
In fact, 228 tonnes puts the twelve months to June 2025 in the bottom quartile of gold purchases over the past five years, 10 years and 15 years.
So why is everyone is talking about unprecedented levels of central bank buying?
It turns out that these reports rely on estimates of ‘unreported’ official gold demand. Add these in and total central bank purchases jump to a whopping 804 tonnes.
And 804 tonnes — as well as being about seven-and-a-half times as heavy as a blue whale — puts the total purchases into the top tercile of the past five and 10 years, and top quartile of the past 15 years.
What even are ‘unreported’ gold purchases? To understand this, we first need to understand the official numbers.
Official central bank gold holding numbers come from the IMF. But reporting to the IMF is purely voluntary. According to the ECB, the “Central Bank of Russia stopped reporting its gold purchases one month before the sanctions were imposed, although it is expected to have been a large purchaser in 2022.” And so the ECB, like the rest of us, turn to the World Gold Council for estimates of ‘unreported’ purchases.
As a reminder, the WGC is the gold mining trade body dedicated to pumping up the gold price serving the gold market and its participants/ making weird videos about Elton John’s kneecaps.
The WGC report total central bank purchases that are far higher than the official numbers. But perhaps conscious of their cheerleading reputation, they have since 2014 outsourced estimates on the demand side to Metals Focus, an independent precious metals consultancy.
Estimating unreported central bank purchases is hard. In fact, this is not actually something they even do — though they do estimate net purchases by central banks and other official sector institutions like sovereign wealth funds.
But with SWF assets under management eclipsing $13tn — around the same size as total central bank reserves — it’s possible that central banks’ gold reserves are pretty much what they report to the IMF, and that gold purchases are built up entirely outside central banks. Maybe we’re splitting hairs.
Furthermore, as the WGC writes in its methodology document:
Given the opacity and complexity of the market, supply and demand statistics are best thought of as estimates, with some estimates being more accurate than others.
Most important statistics, like GDP and the CPI, are estimates. But the process by which they are estimated and aggregated is a matter of public record and subject to scrutiny.
Alphaville doesn’t doubt the level of professionalism and expertise going into Metals Focus numbers. But it’s not possible to interrogate non-public information gathered together from off-the-record conversations with miners, refiners, bullion dealers, etc, about gold flows. And estimates will vary depending on who is doing the estimating.
Back in 1999, and all the way through 2010, the WGC sourced its data from GFMS Thomson Reuters. When it switched data provider to Metals Focus, in order to avoid big jumps in published estimates of demand, they published a hybrid series and smoothed the transition over four years — slowly de-emphasising the GFMS Thomson Reuters data and increasing the weight of Metals Focus data.
We can understand their desire not to spook the market with janky data spikes, but the need to do this highlights quite how much art must go into them.
All that said, putting the gold rally at the door of central bank buying is massively intuitive. Following Russia’s invasion of Ukraine their central bank was sanctioned and its reserves frozen. European leaders are still chatting about whether they can be seized within the current legal architecture. And so it doesn’t take a genius to see that even a shiny pet rock looks a safer bet than electronic IOUs controlled by your enemies/ frenemies.
And it’s in this context that the relationship between the yield on US inflation-linked government bonds — the TIPS yield — and the yellow metal has collapsed:
So we’re not going to rain on the goldbugs’ parade. In fact, given how tinfoil hat goldbugs can get, maybe it’s fitting that at least part of the narrative about the rally is based on statistics that seek to describe opaque backroom chatter rather than official data.