China’s most visible US holdings (the Treasuries held with US custodians) ticked up in the November data, so the TIC release didn’t get much coverage (valuation changes should have pushed the market value of all foreign holdings up in November)
I usually push back against overly dramatized headlines about China selling Treasuries — but in November, the more advanced metrics actually do suggest some (modest) sales. Valuation should have pushed up the reported stock more, the (imperfect) flow measures show sales
Belgian holdings (Euroclear, which historically has been Chin’s main offshore custodian) rose, but not by as much as they should have given the market moves. Their equally wasn’t a big rise in Agencies.
my preferred measure for China (China’s US custodied Treasuries + the euroclear account) is down from June, and Treasury valuations have more or less round tripped — which suggests some true sales
and there aren’t fully offsetting rises (since June) in other important measures (US deposits, US custodied Agencies) – even if November was slightly “up” on valuation changes
So if you want to tell a story of modest diversification, you can — China’s visible holdings of US assets are down a bit (the blue line, which includes euroclear custodied Treasuries) relative to total reserves/ estimated USD reserves (at 58/59%), the dotted green line.
But with the Chinese data, nothing is ever totally clear and I at least think it is important to consider alternative theories. And I am not confident enough to say China is reducing its USD share, for at least 4 reasons
But with the Chinese data, nothing is ever totally clear and I at least think it is important to consider alternative theories. And I am not confident enough to say China is reducing its USD share, for at least 4 reasons
For (1) in the past, during periods when China has disclosed its USD share (just under 60% from 15 to 17), China’s visible US assets have been as far from the predicted dollar share as they are now.
I know enough to know that China has dollar assets that aren’t in the TIC data (deposits in global banks, cross currency swaps with Japanese banks). A shortening of the duration of China’s portfolio to capture high s-term yields could lead to lower security holdings.
Super technical, but China’s reported US holdings would include the US custodied holdings of the state banks. The bulk of the holdings are clearly from SAFE, but there is a bit of CIC and probably a bit of the state banks … and CIC and the state banks could be selling
And most important (4) the ‘Belgium” adjustment isn’t arbitrary — it reflects the evolution of the Belgian account relative to China’s reserves from 2010 to 2016 or 17. But China could change its offshore custodial relationships easily
The Russian reserve freeze made Euroclear very visible (it also showed that Russia’s dollar bonds custodied at Euroclear were small relative to the total, which reinforces the case that China = Belgium)
But it wouldn’t be a total surprise if China was diversifying its offshore custodial relationships, and the quantity of Treasuries held by the UK and the other European custodial centers is WAY up (absolutely and v Belgium)
But I also don’t think it is a slam dunk case, as all methodologies for tracking the offshore assets of big authoritarian countries aren’t perfect — so my best efforts are just that, reasonable (I think) estimates using the best available data, but only estimates.
So I do think that the evidence that China has reduced its US holdings in h2 2023 is much stronger than the evidence that China reduced its US holdings between end 21 and mid 23 …