There has been a quiet growing discussion about the accuracy of official PBOC FX reserve numbers. The internal data discrepancies are becoming simply too large to ignore. Let’s break this down.
Between November 2016 and the June 2017, official PBOC FX assets are effectively flat. $3.052 trillion in November, they stand currently at $3.056 trillion for an official increase of $5.2 billion. This has largely been greeted with a sigh of relief in international financial markets but there are good reasons to look closer at these numbers.
Let’s start with the change in bond yields. On October 31, 2016, 10 year Treasury yields stood at 1.84% but had jumped to 2.37% on November 30, 2016. From October to November, FX reserves fell by $69 billion. Given that the increase in interest rates would have resulted in an estimated mark to market loss on band value of $100 billion if we estimate that roughly two thirds of the PBOC FX reserves are in USD fixed income primary government securities, this does not match up perfectly but at least we are in the neighborhood. Since the end of November, 10 year Treasury yields have traded in pretty tight range, so for our purposes, let us posit that there is no USD bond valuation discrepancy but hold on to that $31 billion difference for later.
Now let us assume that the remaining $1 trillion in PBOC FX reserves is in EUR denominated government debt. From October 31, 2016 to November 30, 2016, German and French 10 year government rates went from 0.13% to 0.47% to 0.2% and 0.76% respectively. If we split that difference, that results in a roughly $20 billion bond valuation loss. Additionally during November, the EUR lost roughly 4% against the USD, as did the other primary reserve currencies. This should have imposed, using our albeit rough estimate, of $1 trillion in EUR denominated holdings, additional losses of roughly $40 billion.
There is one additional point to make about the month of November. The PBOC reports being net seller of FX to the tune of roughly $33 billion in November. Taken together we have bond, currency, and net sales (we will return to the net sales issue in a moment), of $193 billion while the PBOC registered a decline of officially declared FX reserves of only $69 billion. This is only for one month.
If we carry this general framework forward, what is notable is how stable the FX reserve portfolio should be. On November 30, 2016, 10 year Treasury yields were 2.37% and on June 30, 2017 they were 2.31% touching 2.37% just a few days later. In short US Treasuries at intermediate durations have traded within a pretty tight range. For our purposes, let us assume that there has been no valuation change to the US bond portfolio.
Most Euro denominated government yields continued to climb moderately during this time. German 10 year bunds rose 0.3% and Italian 10 year yields rose about 0.35%. This would result in a bond valuation loss of about $25-30 billion USD. However, during the same time the EUR rose against the USD by about 7.8% which turns that $25-30 billion loss into about a $45 billion gain in USD terms for the Euro denominated portion of the portfolio.
While the asset value of the PBOC FX reserves, registered roughly a $45 billion gain, the PBOC was also net FX sellers to the amount of $120 billion. In other words, there is an unexplained $75 billion in the PBOC FX reserves. If we add in November, this really raises the discrepancy. Since the end of October 2016, the PBOC incurred FX reserve decline of $64 billion, but also spent $153 billion of FX reserves and incurred estimated valuation losses due to interest rates and currency movements of $110 billion. In short this means, that even though there are verified and estimated PBOC losses totaling approximately $263 billion, FX reserves only declined by roughly one quarter this amount or $64 billion. This is a discrepancy of approximately $200 billion.
This raises two specific possibilities. Either the PBOC is engaging in some unique accounting or China is drawing on other sources of FX to prop up the RMB. Based upon available evidence, it seems most likely that China is drawing on unofficial sources to prop up the RMB.
It is worth noting that in December 2015, China stopped publishing data that accounted for bank capital available for FX purchase. Consequently, we have to draw from other variables that might reveal evidence of propping up the RMB. Fortunately, they are sources available which give us a solid basis for comparison.
First, we have a data from the PBOC called the Net Foreign Assets from the Overview of Depository Corporations. What is interesting with this data point, as we can see in Figure 1, is how closely it matches up with the previously ceased data Bank position Available for FX purchase.
The Net Foreign Assets data shows a continual and significant drop. From October 2016 through May 2017, the total drop in net foreign assets was 1.1 trillion RMB or at current exchange rates $159 billion USD bringing us much closer to this estimated $200 billion discrepancy noted above.
What makes this 1.1 trillion decline in net foreign assets is the primary source of decline. Another data point we can draw from is something denominated in RMB but which has been published since January 2016, notice break from December 2015, which is drawn from the Sources and Uses of Credit Funds of Financial Institutions dataset and is categorized as “Funds Uses: Foreign Exchange”. From October 2016 to May 2017, the RMB balance here declined by 993 billion RMB or $146 billion.
While there are numerous other ways we could make this case, I will leave you with one more. The PBOC maintains another dataset on the External Balance Sheet of the Banking Industry and is reported quarterly. The data point Net Foreign Currency Assets of the Banking sector from September 2016 (remember reported quarterly not monthly) to March 2017 (June data for this dataset will not be available for a few months) declined by $43 billion.
Consequently, if we average out the Net Foreign Assets number and Funds Available for Foreign exchange number, a reasonable assumption extrapolating backwards, this leaves us $3.47 trillion at current exchange rates in capital to defend the RMB. After removing the PBOC official FX assets, this would imply there is roughly $416 billion in unofficial FX assets available. This would imply based upon current rates of net sales, which appear to come primarily from banks, that within the next 9-18 months, the RMB will have to start drawing primarily from PBOC reserves rather than bank positions.
Though I have tried to be faithful, let’s assume I have mildly overstated the FX discrepancy and there is some statistical noise here, this would match the decline in foreign assets quite closely. What is important to note here is this: China appears to be using third parties to prop up the value of the RMB. What is interesting is that this decline in net foreign assets does not appear to be driven by the usual suspects of the major SOE banks like Bank of China and ICBC. The net foreign asset position by large commercial has actually grown significantly implying there is some large sector of financial institutions propping up the RMB by depleting their foreign exchange reserves by a very large amount. It is not entirely clear where this decline in net foreign assets is taking place because it is not taking place at the PBOC or large state owned banks.
What is most important is that the PBOC appears to be shielding itself from the worst of capital outflows by enlisting quasi-public entities to prop up the RMB though at the current rate of decline, this has a limited shelf life before the PBOC will need to be the primary institution.