Quantifying the shift in China's FX policy

Following on from Wednesday's post, it seems clear that at some point over the last year there's been a shift in China's FX policy.  Frustrated by the implications of dollar strength on the RMB nominal effective exchange rate (a natural outcome of pegging to the dollar), the authorities not only moved to a basket peg but also changed how the RMB's parity to the basket changed vis-a-vis changes in the dollar's value.

Macro Man ran a series of correlations over several time frames after adjusting for issues like lags in the data (RMB fixings for day t reflect  FX changes on day t-1), Chinese holidays, etc.  Because some of the FX fixings are not available before the beginning of last year, he calculated a synthetic CFETS basket using tradeable currencies (including the CNH instead of a CNY fixing.)  As you can see, for the most part the two series line up pretty well, especially before last August.

Obviously, the CNH appreciated markedly versus the reference basket from October 2010 though last summer.  Not only was the USD appreciating against other basket constituents at the time, but the CNH was appreciating against the dollar!

If we look at a full sample of the tradeable basket versus the DXY, what do we find?  Even during (or perhaps because of) the persistent rise in the RMB's NEER, the currency was generally more responsive to following along with dollar down days than up days.  Not only is the R-squared higher, but the beta coefficient is as well.

OK, so what if we break that down into different time buckets?  Let's first look at October 2010 (when Macro Man's data begins) through the end of 2013, when things were ostensibly still hunky dory in China:

Although the R-squareds are much closer, the beta was still notably higher on down days than up days.   Let's now look at 2014-15, when the rumblings of trouble first emerged and the pressure on the external balance exploded into the public consciousness last August:

It's interesting to note that even as correlations fell, the betas increased.  This is consistent with the notion of China gradually introducing more volatility into FX regime.  Of course, it could also merely reflect those few days last August when the RMB had a very sizeable beta indeed during the mini devaluation.   What does 2016 look like?

OK, so this is very, very interesting.   While the betas have dropped back towards historical norms (perhaps reflecting the lack of step-deval?), the correlation of the tradeable basket to the DXY collapsed on dollar up days, while the correlation on dollar down days remains as it was.  The upshot appears to be that when the USD sells off, the RMB tags along for the ride (rising slightly against the dollar but falling more dramatically against non-dollar currencies), but when the dollar rallies, the CNH weakens enough to almost entirely wipe out its gains against other currencies.  In other words, USD/CNH realized vol is a lot higher to the upside than the downside, which of course you knew and saw reflected in risk reversal prices, but perhaps did not consider the consequences of.

Of course, one could argue that the CNH is only a dirty peg that the authorities may only partially control, and perhaps does not reflect the actual policy intentions of the PBOC.  Will we find anything different if we look at the official fixings in 2015-16? Consider last year first:

It is interesting but unsurprising to see the betas lower than for the tradeable basket; after all, there is no intraday movement or volatility with an FX fixing.  It's notable, however, that we already see a large discrepancy in the R-squareds.  Let's look at this year:

Hoo boy, there you go.   The CNY is more responsive versus its basket than it's ever been (on an R-squared basis) when the dollar goes down, and less responsive (based on beta and nearly on R-squared) when the dollar goes up.  The difference in each measure when the dollar goes down and when it goes up is striking.

And this is how the Chinese have managed to fool the world (or at least lull some parts of it into a false sense of security); buy allowing the RMB to "strengthen" (versus the USD) only modestly when the dollar goes down, the authorities manage to ensure that the RMB weakens considerably versus its reference basket.   When the dollar strengthens, however, the CNY takes enough of the brunt that the basket barely moves, as we saw to some degree on Wednesday.

It's all well and good as long as the dollar trades sideways to lower; if trend appreciation ever reasserts itself, however, the Chinese FX mechanism will find itself in a spot of bother.