Follow Up to Whether China has a Trade Surplus

I wanted to write a follow up to my previous post about whether China has a trade surplus because I received a number of very good questions and I felt there were some confusion about the implications of what I was trying to say.  Let me emphasize that in some areas here my thinking is still fluid and there are good arguments are both sides, so I won’t say that my thinking is final by any means.

  1. This is not misinvoicing. This is related to misinvoicing in that it achieves the same ultimate objective of disguising capital flows but it is not misinvoicing.  Misinvoicing is the process where a $100 export from Hong Kong to China turns into a customs reported $200 Chinese import from Hong Kong.  This is a mis-payment scheme.  In this case, Hong Kong reports a $100 export to China, China reports a $100 import from Hong Kong.  However, now when the Chinese importer goes to the bank to pay the Hong Kong exporter, the importer tells the bank to pay $200 rather than the $100 reported to Customs.  This may happen by the importer presenting forged Customs documentation to the bank or by bribing bank officials.  This may seem like a semantical difference but it is important to understand the difference and I will explain why the difference matters a little later.
  2. There has been some confusion about whether I am arguing whether this means China has a real trade surplus or whether imports are actually higher than reported. I think it is highly unlikely that that other countries are exporting to China $2.7 trillion worth of goods but Chinese customs is only reporting $1.7.  The reason is simple: that discrepancy would easily be caught by comparing Hong Kong/Australian/German/Japanese/US exports to China and Chinese imports of Hong Kong/Australian/German/Japanese/US products.  We do not see that discrepancy.  This means that that the value of physical goods imported in China is most likely $1.5-1.7 trillion.  I have dealt with Chinese data and nuances long enough to never rule anything out, however, it seems unlikely that China is actually importing $2.7 trillion worth of goods.
  3. However, we need to be careful about immediately jumping to the conclusion then that China has a real trade surplus. Though it may be fair that the market value of the goods trade would have resulted in a surplus, net exports for national accounting purposes are not based upon physical output or a market value but essentially on a cash basis.  For instance, if physical oil exports rise 10% from 100 to 110 but the price of oil drops from $100 to $50 then the cash value of exports drops from $10,000 to $5,500 for a 45% drop. The “real” increase in export shipments is irrelevant in some ways.  It helps inform our understanding of the whole but for national income accounting with regards to net exports (X-M) is irrelevant.  In real terms yes I think it is fair to say continues to have a trade surplus, but in very real ways this does not mean anything as cash is how trade balances are calculated and pass through to GDP growth.  In other words, people no longer trade seashells but cash and on a cash basis, China does not have a trade surplus.
  4. My thinking on what this means for GDP growth is more fluid and I would appreciate comments from someone who is a true national income accounting expert. I lean towards the argument that this has direct and negative pass through effect on the GDP growth and let me explain how.  In nominal terms, Chinese official data says GDP was 67.67 trillion RMB at the end of 2015 compared to 63.59 trillion RMB in 2014 for total nominal growth 4.08 trillion RMB.  Now the nominal RMB Customs reported trade surplus was 3.69 trillion RMB.  Now if reverse the Customs reported trade surplus and account for the trade balance on a cash basis including the service deficit, this turns into a trade in goods and services (emphasis not current account) deficit of 226 billion RMB.  This means that nominal growth when using the cash basis net export trade deficit turns into nominal growth of only 168 billion RMB or less than 0.3% in 2015.  I should strongly emphasize that I am implicitly assuming that China is counting net exports as equal to the trade balance reported at customs as part of the old equation Y = C + I + G + NX. Now here is the primary reason I believe this has a negative impact on growth and that growth should be reported, and again my thinking is fluid here.  Running a trade surplus implies that there is surplus cash flowing into the economy.  This cash is creating additional savings, investment, jobs, and demand throughout the economy.  However, because there is no surplus cash there is no additional savings, investment, jobs, and demand.  In fact, we see the impact pretty clearly.  When trade surpluses were high and the PBOC was sterilizing, China was awash with cash and investment was booming.  Now Customs reported trade surpluses are at record highs, but liquidity is tight and rather than sterilizing the PBOC is propping up the RMB.  Furthermore, as previous noted, because NX are always thought of on a cash/nominal basis, it seems that the impact on growth needs to be considered on a cash/nominal basis rather than a real basis.  Before I published my original post, I consulted a couple of very knowledgeable people about some of the implications like this and they all came back to me and said they really didn’t know and would have to think about it.  I say this definitely not to blame anyone I consulted as all errors are mine, but rather to note that this is a thorny issue about the implications and that this is what I currently think about the implications for growth but I am open to changing my mind and these are not deeply held convictions.
  5. This is also why capital controls would not work. For all the chatter about China imposing capital controls, it is obvious they simply wouldn’t work.  Let me explain.  Impose the most draconian capital controls you could imagine and money would still flood out of China via the free currency transactions via the current account.  I can already hear people saying, just tighten capital flows on the current account side.  The only way that could be achieved is by bringing goods and services flows into and out of China to a near standstill.  Think I am exaggerating?  Imagine customs inspectors having to count every physical unit into and out of China; establish an independent market price for every good and service into and out of China; verify the agreement between importer and exporter is at the independent market price; follow the importer exporter to the bank to verify that the payment or receipt perfectly matches the customs approved value.  In other words, at every possible step for every unit, everything would have to be verified and approved by the government.  This would necessitate bringing goods and services trade into and out of China to a complete standstill.
  6. Finally, I think there are a couple of things driving this shift in flows. First, China opened up the current account in 2012 to relatively free currency transactions and the data shows this discrepancy began growing rapidly in 2012.  That should not come as any surprise.  Second, in late 2012 and early 2013, China had a power hand over and it is likely that a not insignificant part of this outflow is driven by security concerns.  Chinese brokerages who survey their clients regularly find that high net worth individuals are concerned about a variety of issues with large numbers of them looking to leave China.  Third, investment returns in China have been declining for about 18 months.  For instance, 10 year local government Chinese debt is lower yielding than good quality US corporate debt.  Given the junk/near junk status of local government offerings, the risk return there is easily worth examining closer shall we say.  Fourth, I don’t think this is temporary in the least or about the stop or reverse.  This has been picking up speed since 2012 and shows no sign or reason to stop.  Outward FDI or official capital flows have no reason to use this channel.  This should be thought of as long term or permanent removal of capital from China.  Whether someone is moving from China to Canada or a small to medium size business is expanding elsewhere, most companies have no reason to use this channel.  This is not short term opportunism but rather long term to permanent capital expatriation.  Yet another reason, capital controls are very unlikely to work.
  7. I apologize if I missed any issues as I tried to cover them.

18 thoughts on “Follow Up to Whether China has a Trade Surplus

  1. First off, much kudos to you for identifying a major discrepancy in the data. It’s a critical piece in the puzzle around China’s capital flows and, by extension, pressure on the CNY.

    At the same time, however, I don’t think some of your conclusions flow logically from the known facts, and unfortunately this follow-up post is not as clear as you could have made it. In particular, it doesn’t appear to benefit from the very helpful comments discussion with Simon Cox [whom we can all thank for clarifying several key issues]. My take on your discussion runs as follows:

    1. You have identified a very large discrepancy between China’s large trade surplus as reported by customs data and the small deficit as suggested by bank payments data. This relates almost entirely to imports of goods and services being substantially larger using the bank payments data than the customs data, i.e. it’s an issue with the accounting for imports.

    2. As clarified by Simon, there are two rival/exclusive explanations for this discrepancy:

    (a) imports are understated in the physical/customs data, therefore China’s trade surplus is massively overstated (i.e. in reality it’s a deficit); OR

    (b) Chinese corporates are over-invoicing/over-paying for those imports, thereby inflating the payments for imports and disguising capital outflow.

    3. As Simon notes, these can’t be equally true, in that imports can’t be understated and over-invoiced at the same time – to the extent that (a) is true, (b) is less true.

    4. As both you and Simon conclude, (b) appears far more likely than (a). In particular, as you point out above, it’s hard (impossible?) to reconcile China’s imports data discrepancy with the exports data of its trade partners.

    5. The implications are that:

    (i) China’s reported trade surplus is real; but also
    (ii) there is massive disguised capital outflow taking place through the current account via over-invoicing for imports.

    On (ii) Simon notes that this might explain the substantial increase in China’s balance of payments “errors & omissions”, which is the larger part of the increased pressure on FX reserves. This is essentially the “smoking gun” of capital flight from China.

    6. Turning to GDP effects, it’s not clear that GDP growth would be as low as you first argued (and still argue, it seems) EVEN IF the trade surplus were massively lower. As Simon pointed out, you are confusing flows with growth in those flows. If we follow the neoclassical model,

    Y = C + I + G + NX

    (where Y = GDP, C = private consumption, I = investment, G = govt consumption and NX = net exports),

    the CHANGE in GDP, dY, is the sum of the CHANGE in its components. Thus dY should not be compared to the level of NX, but the CHANGE in NX, i.e. dNX.

    If NX is overstated, then to determine the “true” level of dY requires rebasing both the beginning and ending Y for the correct levels (beginning and end) of NX and measuring the effect of the change in (adjusted) NX on the change in (adjusted) Y. The mistake you made was to subtract the LEVEL of the discrepancy in NX from the CHANGE in Y, and this isn’t accurate.

    7. However, that point is moot if you agree – as you do – that China’s reported trade surplus probably is real.

    8. Finally, it seems to me that the key conclusion here is that over-invoicing of imports appears to be a (the?) major channel for underlying capital outflow from China. We’ve known the over/under invoicing of China’s trade flows to be a source of data discrepancy for several years now because of the Hong Kong data, but it’s only in the past year that such flows have turned sufficiently materially negative (i.e. to net outflow) to threaten, potentially, PBOC control of the exchange rate. I’m more confident than you that Chinese authorities can curtail or choke off these flows as you suggest. As the discussion above reinforces the point that it is a payments issue, not one related to physical trade, it’s possible that bank-level controls could be successful in at least dampening and/or slowing the pace of outflow. To be sure there would be an economic cost, but from the regulatory perspective there is a huge and growing opportunity cost attached to the alternative of doing nothing. Knowing China’s whack-a-mole regulatory environment this might only result in a shift to other forms of capital export, but I’d expect the government to attempt it. More broadly, I think China will have to raise capital controls much more aggressively to suppress the natural inclination of Chinese investors to escape the deflationary consequences of a massive debt bust. I suppose we’ll find out the results within a few years…

    • A. 1-5 I think we are in agreement.
      B. I need to double check my national income accounting but I am pretty sure that it isn’t dNX but the level of NX. This is actually a confusing point for many people. dY is clearly noted in the post and if NX drops to zero this has a direct pass through effect on GDP. I do agree that doing it correctly would require rebasing every previous year to account for this discrepancy. Think of the problem like this: it isn’t a question if the trade surplus increases 10% from $100 to $110 but rather what percentage that $110 contributes to GDP. That $110 contributes directly to GDP.
      C. As I noted, I would advise caution in saying “real” trade surplus. There really is no such thing as a real trade surplus because the entire concept of a trade surplus is based upon a nominal or cash surplus. The market value of goods exported by China did exceed the market value of goods imported, which is important to remember but on the more important cash metric which is how trade surpluses matter and are measured it did not. So I guess on this question I say yes and no.
      D. I would also caution first of all against just describing this is misinvoicing. This isn’t misinvoicing. In fact, the discrepancy between China and other countries on trade invoices for the capital outflow channel is almost within the normal discrepancy range. Hong Kong by itself is an outlier and not big enough to move the total needle. Misinvoicing on the imports into China side has actually been bouncing around close to zero for a number of years. This is mispayment which is a different step in the transaction. I am not saying they didn’t know about this but at the same time I’m not saying they did. I’m leaving open both possibilities. This has continued to grow pretty rapidly since 2012 and for your confidence that they can handle this, they show no signs of handling this. As you said, in this whack a mole environment we’ll see if they crack down on this. Watch this: if they start cracking down on this in the next few months, we’ll know they didn’t know about this.

  2. B. OK, let me try again. Reconsider your example above:

    Y (2014) = 63.59tn
    Y (2015) = 67.67tn

    dY = 4.08tn, or growth of 6.4% (i.e. 4.08 / 63.59).

    Now, suppose that NX (2015) has been overstated by 3.916tn (i.e. reported NX surplus of 3.69tn should really be a deficit of 0.226tn). Assume for the moment that the level of ACTUAL and REPORTED NX were the same in 2014 and 2015, i.e. the discrepancy was roughly static. In that event, Y would be 3.916tn lower in each year, right?

    That is,

    Y (2014)* = 59.674tn (63.59tn less the 3.916tn NX adjustment)
    Y (2015)* = 63.754tn (67.67tn less the same 3.916tn NX adjustment)

    In that case, dY* is now…4.08tn. The growth rate actually increases, to 6.8%, but only because the denominator has been adjusted lower.

    Rationally, using the same national income accounting you ONLY get a material fall in adjusted dY (i.e. “dY*”) if ADJUSTED NX falls materially. Do you see the difference now?

    Now, it’s possible that the size of the adjustment/discrepancy itself has changed, artificially inflating Y, but it doesn’t look like the discrepancy moved that much from 2014 to 2015, and certainly not by as much as 5-6% of GDP.

    Again, this all assumes that the trade surplus has been overstated massively, and I don’t think it has.

    C & D. There is an important issue around accounting for value here, but if the over-payments (let’s skip the semantics on “invoicing”) for imports implicit in the bank payments data are disguising capital flow, then you are moving out of the “real” economy into financial flows. That is, it’s a transfer of value (capital/financial flow), not the creation/addition of value (GDP).

    On the growth since 2012, I absolutely agree that the discrepancy has grown substantially and there’s little evidence of it being “handled”. However, as is often the case in China, the mole often has to pop a long way out of the hole before he gets whacked! As you would have seen from Alphaville, Deutsche Bank has just published a report looking at precisely this issue that you raised a week ago. If DB’s onto it, the PBOC – and presumably its Chief Economist (a DB alumnus) – won’t be far behind.

    • 1. Completely willing to agree that to get a true idea of the impact you would have to build a time series factor out these things and that would impact the 2015. Agreed. I was just taking a one year static calculation.
      2. I agree with you about the financial flow issue disagree about what the GDP implications are and disagree is probably the wrong word. Rather I am hoping to find a national income accounting specialist to help disentangle this issue. As I said in the post, a trade surplus boosts growth by essentially providing surplus cash which boost wages, investment etc etc. However, here even if “real” trade is in surplus nominal/cash trade (or at least the things that should pass through from a trade surplus to growth) isn’t, what does that mean for growth? Trade here is linked to a financial flow. It’s a very tricky issue and my thinking is not set in stone but I don’t think you can automatically say without examining the issue closer, that this provides a boost to GDP because the positive spillovers essentially don’t appear.
      3. Agreed also. If they didn’t know before, they know now.

      • For GDP purposes, it’s the real value of imports, exclusive of the additional financial (out)flows, that should be used. It goes back to the basic fact of GDP being Gross DOMESTIC Product: i.e. an import is deducted from DOMESTIC economic output (thru the Net Export account) because its value isn’t domestically generated. But the “value” we’re concerned with here is exclusively the “real” value, not including any inflated financial flow. If a country pays $100 to import an item that actually sells for $75 on the international market, the extra $25 of financial flow doesn’t enter the domestic economy via consumption or inventory, and hence only the $75 real value should be deducted from domestic economic output.

        So yes, China is losing money, but it’s precisely because a large REAL trade surplus of nearly $600 billion is overwhelmed by even larger capital outflows of about $1 trillion – about half of which, according to Deutsche Bank, is attributable to import overpayment (from FT Alphaville,

        • There is no such thing as a “real” trade surplus. Trade surplus is counted as the nominal/cash(financial flow) value of trade. Consequently, Net Exports are counted not on a physical output basis but on a cash/financial flow basis. A (nominal/cash) trade surplus should generate surplus inflows into China. This is not happening. There is no such thing as a “real” trade surplus.

          • 1. I believe that phu618 is using the term “real” to mean “actual” or “true” rather than “volume” or “price-adjusted”.

            2. In the calculation of real GDP, nominal net exports are converted to real net exports using export and import price indices.

            3. I think that phu618 is basically making the same point as I did in my post.

          • 1. That is a fair and accurate description. I’m okay with that.
            2. Agreed but here is the problem as I see it. When converting into real GDP because the nominal is lower you would need to convert into real with a much lower nominal. If nominal is essentially the value(cash/financial flow) number of net exports you would need to use a very different nominal value.
            3. Agreed.

          • As steve pointed out, I was actually referring to “actual”, not “real” in the technical sense.

            It would appear either way that China no longer has a combined positive balance of payments, but as the above detailed breakdowns show, this doesn’t mean GDP growth is overstated by 5-6 percentage points, even if it means GDP itself is overstated by the overestimate in Net Exports (still open to interpretation, IMHO).

            Capital clearly wants to leave China – it has already done so in ever-increasing volumes since 2011. But it seems like the dam really broke only last year, specifically the second half starting with the July stock market crash and August devaluation. What really turned the broader sentiment was PBOC itself apparently conceding that a weaker exchange rate was needed. After two such waves of yuan turmoil, however, PBOC now realizes the RMB can’t be loosened too quickly, and that global markets demand far more than what China’s willing to concede at this juncture. PBOC will thus try to maintain the “impossible trinity” as long as possible, but at least it’s easier to convince Xi Jinping’s leadership team that this can only provide some breathing space for the supply-side structural reforms to get off the ground and the central leadership to tackle the vested SOE and bureaucratic interests at the provincial and local levels.

            If such reforms fail, a day of reckoning won’t be too far off. Even then, however, depending on how confident Xi is in his consolidation of power, he will likely sacrifice capital openness for currency and reserve stability, at least as a temporary provisional measure. And chances are he’s already preparing for the next financial blowup as an opportunity to downsize the state sector – for political more than economic reasons. The way he probably sees it, if he can’t force the state economy to fix itself, the only alternative is to wait for it to crumble of its own distortions and be there to rebuild it from the top down. He’s already refocusing on the 19th party congress in fall 2017 – whatever happens economically between now and then, he wants a much more streamlined and loyal bureaucratic apparatus in his second term (2017-2022) than he’s had in his first.

          • Generally agreed. I would argue and will in a post that we can definitely grapple with the issues but the positive effects of a trade surplus are clearly and obviously not present. Also, people noticed the capital outflows last year. It really started in 2012 and grew through last year. Not sure if it will continue rising but definitely expect the levels to remain high

  3. This post just relates to the import side anomaly. Obviously, I have not used the actual numbers.

    Recorded transaction:

    Debit of100 (local currency units) on the balance of trade (and hence current account)

    Actual (suspected) transaction:

    Debit of 200 on the balance of payments

    Interpretation of this scenario:

    Debit of 100 on the balance of trade (recorded)
    Debit of 100 on the secondary income part of the current account (unrecorded)
    …if the “correct” price of the imported goods is judged to be the debit of 100 that is actually recorded.

    Given this interpretation:
    The recorded balance of trade reflects the actual balance of trade
    The recorded balance on the current account omits the illicit transfer of 100.

    Impact on GDP:
    On this interpretation, there would be no change to recorded nominal GDP, since the unrecorded transaction did not affect nominal net exports.
    However, actual gross national disposable income would, in principle, be lower than recorded (if it was recorded). Savings have been illicitly transferred out of the territory. If these had not been transferred out, they could have been used in the territory in many different ways, each of which would have produced a range of different impacts on the real and monetary sectors of the economy. Thus, the recorded level of nominal GDP at the end of the year might have been different to the one that was actually recorded. And that is about as much as you can say.

  4. Couldn’t stay away…

    Let me try and be ploddingly concrete. Suppose an overseas firm charges $100 for its goods. A Chinese resident buys one. He declares to customs that he has imported a $100 good. The $100 import is duly recorded in the balance of payments. Then comes the twist: the Chinese resident instructs his bank to pay an overseas bank account $120 for the import. The bank reports a $120 “import” payment to SAFE.

    I think we’re all agreed this sort of thing could be happening. It results in a discrepancy between the customs/balance-of-payments data, on the one hand, and SAFE’s bank-payments data, on the other. Spotting that discrepancy was a helpful contribution to the debate.

    The question is how should we characterise and label the extra $20? Chris insists on calling it an import (or payment for an import). He concedes that 20% of extra stuff hasn’t crossed the border. But he argues that this transaction is analogous to the case of an import rising in price by 20% from one year to the next.

    But we can’t call the extra $20 an import (or a payment for an import or an import “on a cash basis”).

    Who do we imagine keeps that extra $20? Presumably it isn’t the overseas firm. (If it were, then the $20 should be characterised as a $20 gift bundled together with a $100 import. Gifts should be thought of separately in the balance of payments.)

    Presumably, the extra $20 is kept, in some shape or form, by the Chinese resident (minus a cut for middlemen). The Chinese bank transfers money to the overseas account. $100 ends up in the hands of the overseas supplier. Most of the rest stays in the hands of the Chinese resident.

    So the Chinese resident has acquired two things not one. He has acquired a foreign good (the $100 import) and in addition he has acquired a foreign asset (the extra $20 or so he now holds in an overseas bank). The acquisition of foreign assets by residents is a capital outflow, by definition. It is not an import. The word import, in this context, is reserved for the purchase of foreign goods and services.

    A meta-comment, if I may. I agree with Fyodor that it was well worth highlighting the discrepancy between SAFE’s balance-of-payments data and its bank data. But I can’t help feeling that insisting that this discrepancy represents additional imports (“on a cash basis”) is obscuring that contribution not illuminating it.

  5. Hi Chris, Deutsche Bank did a similar piece on 29/02 title “China:Unveiling the Capital Flows”, this pegged the difference for 2015 at $500bn. As there are no links to your source of banking payment data, it’s impossible to verify the discrepancy. Can you please provide a link or send the data set you are using?


    • The data I received from Deutsche Bank (which is in Chinese, so I can not investigate), highlight the payments for imports at $2.2 trillion, consistent with their report.


  6. Pingback: Collection of Thoughts on Chinese Economy | Balding's World

Comments are closed.