China:SAFE data suggest FX outflow moderated in October to US$40bn

Bottom line:Our preferred gauge of FX flow (based on SAFE data) shows that FX outflowsmoderated to US$40bn in October (from US$78bn in September). Our gaugeincorporates information on both onshore net FX demand by non-banks and crossborderRMB movements.

Main points:As in the last few months, we focus on two separate sets of SAFE data to gauge theunderlying FX flow situation: one on onshore FX settlement, and the other on thecross-border movement of RMB.

According to the first SAFE dataset on onshore FX settlement, n net FX demand bynon-banks onshore in October was US$11.4bn (vs. US$33.3bn in September).

This is composed of US$7.4bn via net outright spot transactions, and US$3.9bnvia net freshly-entered forward transactions.

The second SAFE dataset on cross-border RMB flows shows that net flow ofRMB from onshore to offshore was US$29.0bn (vs. $44.7bn in September). Weincorporate this data in our measure of net FX flow, for reasons we havediscussed previously (see here).

Our preferred gauge of underlying flow therefore indicates a total net FX outflow ofUS$40bn in October (US$11.4bn from net FX demand onshore plus US$29bn in FXoutflow routed through the CNH market). We note that the onshore FX marketswere shut for a whole week in early October given the National Day holidays,although historically there is no strong evidence pointing to special Octoberseasonality on FX flow.

Separately, the PBOC has earlier released two related data points on its FX assetholdings for October. One is the widely-followed FX reserves data (out on Nov 7),which fell US$8bn, after adjusted for our estimate of the currency valuation effect.

But valuation effects are highly uncertain and our estimates on those can be noisy,hence we rely on another dataset called “PBOC’s FX position” (which shows theamount of PBOC’s FX assets recorded at book value, out on Nov 14) as a crosscheckon PBOC’s FX sales net of valuation effects. According to that, PBOC soldUS$40bn in FX in October (which is quantitatively the same as what is indicated byour FX flow measure). Similar to the case in September, this data suggests