Things That Cannot All Be True: China RMB Flow Edition Part I

I have a couple of guiding principles when it comes to how I approach studying Chinese data. First, details matter.  Broad blunt measures like debt to GDP might provide a good headline but do little to advance our understanding of what is really happening. Second, numbers must reconcile relatively closely.  There is enough data throughout the Chinese economy that we should be able to match, within some reasonable error or noise level, a wide variety of data. Third, a long and broad memory is important to best utilize points #1 and 2.

Let us start with a rough estimate of how much RMB is leaving China.  This is not FX transactions conducted inside China, but rather international transactions that are denominated in RMB.  As a final caveat, it is worth noting that about 75% of international RMB transactions are conducted between China-China or China-Hong Kong.

There is a very close relationship between RMB outflows, the net balance of international RMB transactions and RMB denominated balances in Hong Kong the primary offshore center for the RMB.  Since we have international RMB transaction data back to 2010, there has been a pretty close relationship between net RMB outflows and RMB balances in Hong Kong.

This is intuitive and straight forward.  Hong Kong is the counterparty in never less than 70% of international RMB transactions and remains the dominant source of offshore RMB in the world.  As I frequently stress, we are not looking for exact matches or reconciliation between numbers but rather numbers that are so grossly out of place to cause concern.  For most of the period we have data for, the relationship between RMB outflows from China and Hong Kong RMB deposits is relatively stable.

There are a number of ways that we can conclude that the relationship between RMB ouflows and Hong Kong RMB balances is pretty stable.  I will just give you a few data points.  First, in August 2015 the difference between the aggregate outflow of RMB from China since January 2010 to RMB deposits excluding the starting balance as of January 2010 was less than 38 billion RMB.  By comparison, total RMB deposits in Hong Kong in August 2015 was 979 billion RMB, so the discrepancy was equal to 3.9%.  Given the total size of flows and number of offshore RMB centers, this is a relatively small difference.

Second, if we compare the difference between the September 2015 and November 2013 aggregate outflows and Hong Kong RMB balances, we see how closely related they are.  During a time when aggregate outflows went from 940 billion RMB to a peak of 1.72 trillion before falling back to 940 941 billion, Hong Kong RMB deposits witnessed nearly an identical pattern with an important caveat.  While Hong Kong RMB deposits did go up during the same time frame, they increased much less than the total amount of outflows.  While aggregate outflows during this period peaked at 779 billion, RMB deposits in Hong Kong never rose by more than 176 billion.  It was during this time that many other offshore centers were gaining large inflows of RMB.  However, by September 2015 RMB had moved back to Hong Kong so that even as aggregate RMB outflows were up only 628 million, RMB bank deposits were up 68 billion.  By November this gap between aggregate outflows and RMB deposits had shrunk from a 69 billion to an insignificant 15 billion.

Third, Hong Kong RMB deposits as a percentage of RMB outflows have averaged about 65-85% depending on some various measures.  Given that more than 70% of international RMB transactions involve Hong Kong, this number again tells us that as RMB leaves China a pretty stable amount of it ends up in Hong Kong.

However, since August 11, 2015 these numbers have changed dramatically.  In November 2015, the Hong Kong RMB deposit to aggregate outflow ratio stood at 70%.  This matches the transaction volume and other metrics of where RMB was going and how much was leaving China.  However, since November 2015 this ratio has fallen to 19%.  In other words, Hong Kong deposits of RMB are equal to only 19% of the aggregate outflow.

What is notable is that both numbers have changed dramatically in the wrong direction. Since October 2015, aggregate RMB outflows, as measured by net receipts from banks, grew from 1.02 trillion RMB to 3.17 trillion RMB.  In other words, in one year there were outflows from bank receipts less payments totaling 2.15 trillion RMB.

All this outflow should have shown up in higher RMB denominated bank balances right? Wrong.  In that same period, RMB denominated bank balances shrunk from 854 billion RMB to 663 billion RMB or by 192 billion RMB.  Put another way, during a one year period when RMB was flooding out of China by more than tripling the aggregate net outflow level, RMB deposits rather than growing roughly in line with a historical trend fell by 22%.

If we take just the fall in Hong Kong RMB deposits, this implies that there is approximately 2.34 trillion RMB or $339 billion USD, using current exchange rates, that we should be able to see somewhere in the offshore market that simply isn’t there.  It is worth noting that RMB deposits in other offshore centers have fallen by similar relative levels.

We are now left with a simple conundrum: if RMB denominated outflows from China exploded and RMB bank balances dropped sharply within the past year where did that RMB go? Even in China, this is simply an unexplainable amount of money.  From January to October this year, the last month for which we have banking flow statistics, the combined amount of outflows and drop in RMB deposits equaled 1.56 trillion RMB or $228 billion USD.  However, FX reserves in China had only dropped $110 billion.

While the PBOC would have been, by its own numbers, unable to soak up all the new RMB in offshore centers, this leaves us with two specific alternatives. First, the PBOC numbers are unreliable.  While we cannot rule that out, I think it is pretty unlikely that the PBOC is releasing fraudulent data for many reasons.  Second, the more likely explanation is that there are unofficial official actions being taken to drain the RMB liquidity leaving China from settling in offshore centers and pushing the wedge between the CNY/CNH.

I want to stop now because there is so much more to write on this topic, as we piece together how the money is flowing and why these numbers are simply inconsistent.  However, we can say now that the amount of RMB that is leaving China on a net basis simply cannot be reconciled with the amount of RMB we see showing up in offshore centers.  That leaves us the question for the next time: if the money is leaving China but isn’t showing up in offshore centers and offshore center RMB deposits are falling, where is this enormous amount of RMB going and who is doing it?

 

Some Brief Thoughts on Outflows

So I have been travelling to much and am currently enjoy a strenuous regimen of two a day umbrella drinks and naps on a south-east Asian beach.  The battery is getting recharged and looking forward to writing more.

I wanted to put out something someone sent me about the rapidly shrinking payments gap. As you can see below, the difference between bank payments for imports and the customs reported imports has shrunk rapidly and dramatically.

Since the recent peak discrepancy number, of $58 billion in January and writing about it here in February when the discrepancy dropped slightly to $47 billion, the difference between bank payment for imports and customs reported imports have fallen dramatically.  In August, this discrepancy was just above $10 billion USD.

The fall off in this discrepancy has been nothing short of stunning.  The last time there was a single month this small was in September 2013 with periods of 2012 and 2013 matching some type of moving average.  SAFE is clearly cracking down on moving money out of China this way.

Placed in larger context it gets even more interesting. First, this drop is responsible for essentially all of the supposed slow down in outflows from China.  If this number returns to the pre-crack down average, outflows from China would be approaching $100 billion per month.  Second, there is a game going on here which we can call whack a mole.  Shutting this avenue down will only drive the money out other channels which we already see evidence of as other channels become more prominent.  Third, this movement represents a structural outflow of capital.  As I have noted before, this is not due to 25 bps in New York but rather a structural and likely quasi-permanent shift in the demand for foreign assets by Chinese citizens.

Interesting stuff now back to my pina colada.

On the Recent RMB Strengthening

There have been questions raised in the past few weeks about the state of the RMB.  Questions have focused on why the market is not reacting more strongly to continued depreciation, whether the PBOC is engaging in active price manipulation, and the direction of the RMB.

These questions at their heart revolve around why the RMB depreciation path seems to have halted and even reversed in the past 1-2 weeks.  In fact, the RMB has strengthened recently which seems to have caught many off guard.  We believe there are clear and straight forward answers for what we are seeing the RMB FX market.

First, according to my esimates, the RMB against the CFETS basked has been relatively stable over the past month with small strengthening over the past 1-2 weeks.  My model shows slight strengthening of the RMB against the CFETS basket whether measured in 1 or 2 week increments even over the last month.  In other words, if the RMB is generally following the CFETS basket, the RMB should have strengthened which is what we see.  This is the spot rate and the Wind estimate of the CFETS but mine and other replications of the CFETS show similar strengthening.

As many have noted previously, there is an asymmetric pattern for when the USD weakens.  The RMB is stable against the basket when the USD strengthens, but when the dollar is weak, the RMB maintains stability against the USD.  Consequently, when the basket is generally strengthening against the USD, the RMB will see mild strengthening which is what we have seen.  The past few weeks therefore, should not come as any type of significant surprise.

Second, the fixed nature of the RMB makes the RMB much more prone to exogenous shocks.  Given a relatively rules based regime, whether moving directly inline with the CFETS basket or with some flexibility to the USD, the RMB tracks other global currency movements rather than building its own internal market that others respond to.  As global currencies have stabilized over the past few weeks and months, it does not come as a surprise that the RMB has stabilized.

Third, there remains overwhelming evidence that the PBOC either directly or via proxies is heavily involved in the market ensuring pricing it wants.  For instance, spreads after factoring in all costs continue to predict a strengthening of the RMB over the next 1-12 months.  Looking at the swaps market, even as the spot price has depreciated, the swaps price post August 11 has tightened considerably.

This is fundamentally counter intuitive.  Before August 11, when there was no expectation of future weakening, the spread was large.  Post August 11, when the market almost uniformly expects depreciation, the swaps price has narrowed so much it actually predicts RMB strengthening.  Spreads on various futures products remain tight even as markets continue to expect longer term depreciation.  Traders continue to report difficulty executing trades at posted prices for various products.  Liquidity appears to remain tight or potentially worse indicating less than normally functioning market.

Fourth, the long term trend remains for continued depreciation.  Capital continues to move out of China at a relatively steady rate over the past 3-6 months and slower than its late 2015 rate.  As previously noted, there is strong evidence that the PBOC is enlisting other parties to prop up FX reserves and slow their depletion, but given the ongoing outflow of capital out of China it seems clear the trend remains to expect further depreciation.  It is worth noting that the RMB outflows have slowed, but still continue.  Foreign inflows are down significantly and net bank payment and receipt surplus is only slightly behind the total for all of 2015. There is pressure within China to allow further depreciation and the continued net outflows necessitate further depreciation.

As the markets have become distracted with Brexit, US elections, and Japanese easing, focus on the RMB has eased as expectations have changed.  However, all factors seem at play to expect ongoing steady depreciation barring some large exogenous shock.  The PBOC has learned how to better manage market expectations and we believe ongoing depreciation should be expected.

How PBOC Making FX Reserves Look Better Than They Really Are

The PBOC has released foreign exchange reserves data and the results are puzzling.  Even major investment banks releasing their notes on post-FX reserve analysis have expressed various degrees of bewilderment at the results.  Fundamentally, it is becoming increasingly difficult to reconcile the stock value of FX reserves and the flow changes we witness every month.

There are numerous pieces of data that form our picture of the whole as to why we say this. Let’s break this down piece by piece show why there is increasingly contradictory evidence.

  1. According to our model, which is similar to other estimates of PBOC reserve composition, and general FX reserve holdings, the PBOC USD value of foreign exchange reserves should have remained essentially unchanged between May and June 2016. The rapid rise in the JPY in June should have largely been offset by the rapid fall in GBP.  While we cannot know the exact weighting of the three primary non-USD currencies, given a range of reasonable parameters would leave this portion of the basket fluctuating around no valuation change. The only plausible method to arrive at a material USD valuation change between May and June in the non-USD portfolio is to assume extreme parameters in EUR, GBP, and JPY assets.
  2. Even if we extend this basic valuation change back to the beginning of the year, there should be a relatively minimal change in the USD value of the non-USD asset portfolio of PBOC FX reserves. We estimate the non-USD portfolio, absent non-USD depletion, to have benefited from an approximately $30 billion valuation increase.  Foreign exchange reserves however through the first six months of 2016 have only declined $26 billion.  Absent other valuation or unrecorded inflows changes, this would imply total net outflows between $55-60 billion.
  3. However, just according to official SAFE data, the YTD bank receipt less bank payment for international transaction reveals a net outflow of $145 billion USD through May. If we add in the expected value for June, this would give us a forecast net outflow from bank transactions of $170-185 billion USD nearly on par with all of 2015.  Given the estimated valuation increase and the official decline in PBOC reserves, this would leave an approximately $115-130 billion USD that we cannot account for in our calculations.
  4. Even if we look at the net flows by currency type, the numbers tell a story of similar outflows. Looking at just the top two currencies, we see that USD net flows were in surplus by $52 billion while RMB net outflows totaled $106 billion in USD terms.  HKD, JPY, EUR, and all other currencies summed to the previously noted $145 billion net outflow.
  5. Breaking it down by currency however actually gives us a clue as to what is likely happening. The $106 billion RMB outflow in USD terms is leaving China for international transactions.  Theoretically, this should result in ever expanding offshore liquidity.  Conversely, we actually see quite the opposite happening in offshore centers with RMB trading and deposits.  Where RMB deposits have been shrinking, specifically in the primary offshore center Hong Kong relatively rapidly.
  6. Bank buying of FX from non-bank customer through May totaled $661 billion USD while sales of FX totaled $541 billion USD for net purchases by banks of $120 billion USD. Given the previously mentioned net outflows from bank payments of $145 billion and the approximately $25 billion in revaluation over the same period, we are able to reconstruct the numbers through May relatively closely.
  7. This conclusion though has a very important implication. This means that commercial SOE banks are essentially acting as a central bank purchasing surplus RMB either on the Mainland or in Hong Kong to prop up the RMB.  It is worth noting that the Bank of China acts as the primary settling bank or cross border RMB and takes a small fee for all offshore RMB remitted to the mainland.  Given that spreads between the bid and ask is less than the fee BoC takes for remitting offshore RMB back to the Mainland, it is likely they are essentially operating a large churning operation propping up the RMB.
  8. We actually see evidence of this in the Bank of China Q1 2016 report. They list a 31% drop in “Net Trading Gains” which they attribute to “decrease in net gains from foreign exchange and foreign exchange products.”  What makes this so interesting is that even though BoC is the primary settlement bank for the PBOC of international RMB transactions, FX market turnover was up 20%.  It seems difficult to understand how with a market up 20% the near monopolist firm see revenue drop 30%.  The most likely explanation is that they are essentially acting as a central banker, soaking up the liquidity at the spread, profiting from the repatriation fee, and churning.  Though much of their purchases are offshore, forcing them to incur a loss, the repatriation fee compensates them harming their margin but upholding the national interest.

We need to keep an eye on this especially as we move forward and BoC trading revenue and matching up the outflows to the SOE/PBOC churn.

April Trade Data and Foreign Exchange Reserves

A lot of how you decide to view the Chinese April trade and foreign exchange report, depends on what exactly you measured.  April exports were higher than March exports but were down YoY and YTD YoY if measured in USD.  However, if measured in RMB exports YoY was actually up 4% but remains down YTD 2.3%.  In some ways, this data can be viewed positively or negatively, but I am going to try and help provide some personal perspective.

  1. While the month to month and year over year snapshots are important, I firmly believe that the YTD are much more important. MoM and YoY can induce a sense of noise or bias into analysis that skews our understanding.  YTD exports are down 8% from 2015 and imports YTD are down another 13%.  What makes the import growth some amazing is that full year import growth was down strongly in 2015 and flat in 2014.  It is difficult to see how these are positive signals for an economy as you stretch the time horizon out.
  2. While the trade surplus again remains strong this is a very deceptive measure for a couple of reasons. The trade surplus remains strong not because trade is increasing but because imports are shrinking much faster than exports.  Whether you look at it on a YoY or YTD YoY trend, it is clear that imports are shrinking faster than exports.  While some of this can be attributed to factors like commodity price drops, it is also clear that some of this needs to be attributed to weak Chinese demand.
  3. The other reason that the trade surplus is incredibly deceptive is that the actual surplus if measured by cash, which is really what matters, is much much smaller. Through March, Chinese Customs reported a surplus of $126 billion USD while banks reported a surplus in goods trade receipts of $23 billion.  This means there is a $103 billion discrepancy between the official trade surplus number and what cash is actually flowing into China.  Given the $46 billion surplus reported for April, we can probably expect that this resulted in a bank receipt surplus of $10-12 billion USD.
  4. Extrapolating this into the official amount of FX reserves is where things start to get a little debatable. To date, the only category in surplus on a cash basis in Chinese banks in goods trade and it is small at only $23 billion.  All others are in significant monthly and year to date deficit.  For instance, through Q1, YTD outflows are almost equal to Chinese net outflows through November in 2015 YTD.  Capital account receipts are plunging and outflows are up almost 40%.  This is a very consistent pattern in each month and summing across Q1.  If this patterns holds in April, this would imply a net outflow of at least $30 billion through official bank payment channels.    Despite talk of how USD valuation drove FX reserves up, the EUR was essentially unchanged against the USD in April.  The JPY which was up almost 5% against the USD but by most estimates comprises no more than 15% of PBOC reserves should not swing the portfolio that much.  If we assume the JPY has a 15% portfolio weighting and moved 5% in the PBOC’s favor, this should result in no more than a $24 billion boost.  This at least gets us closer to explaining the PBOC official data that reserves rose but as many have noted is an increasingly difficult number to reconcile to other data.  This would have to imply a much small outflow.
  5. The reason for the skepticism is that it is increasingly difficult to reconcile the ongoing outflows, even after accounting for valuation, with the stabilizing and actually increasing reserves. For example, in the past three months when FX reserves were stabilizing and then slightly increasing net outflows have actually gone up by most measures.  This is simply difficult to reconcile though I think it is fair to say that while there is suspicion and concern, there is as of yet no smoking gun or hard evidence of how they are making this number appear so rosy.
  6. Too many people focus on the level of FX reserves rather than the net outflow number. If you run a fixed exchange rate regime, you cannot sustain net outflows for an extended period of time.  Despite the rosy official trade surplus, underlying cash flows have if anything accelerated this year, though there may be some evidence that capital controls are starting to bite though it is too soon to tell if that is just Chinese New Year seasonal fluctuations.  Even if the FX numbers are perfectly accurate, the ongoing level of sustained outflows should absolutely be the bigger topic of discussion.