Koos Jansen
BullionStar Blogs
Koos Jansen
Posted on 25 Jun 2015 by

Why SGE Withdrawals Equal Chinese Gold Demand And Why Not – The Argument List

This post is part of the Chinese Gold Market essentials series. Click here to go to an overview of all Chinese Gold Market Essentials for a comprehensive understanding the largest physical gold market globally.

The difference between SGE withdrawals and Chinese consumer gold demand as disclosed by the World Gold Council has aggregated to 3,193 tonnes from 2007 until 2014 (the period this article will focus on). Naturally, we’re here to get the finest understanding of Chinese gold demand. To explain how the difference is caused Western consultancy firms have presented several arguments in publications and lectures at conferences, though none of them can explain the difference in full. This post is an overview of all such arguments (supplemented by my own arguments).

Below we’ll examine to what degree the arguments can or cannot have caused the difference. Subsequently, we’ll discuss the details of all metrics that can be applied, to eventually be able to calculate our best estimates of genuine Chinese gold demand 2007 – 2014 within every metric.

This is the argument list so far in chronological order:

  1. industrial demand
  2. stock movement change
  3. round tripping
  4. leasing
  5. official purchases
  6. recycled gold
  7. export
  8. the Shanghai International Gold Exchange
  9. smuggling
  10. ETF demand

In 2014 the World Gold Council (WGC) came out with two special reports about the Chinese gold market that should have shine a light on the difference (China’s Gold Market: Progress And Prospects form April 2015 and Understanding China’s Gold Market from August 2014). However, these reports contain many false statements and the segments on the difference fail miserably, as I’ve pointed out in several posts (one, two, three, four, five, six, seven, eight). Surprisingly, after the reports were published Western consultancy firms came up with new arguments that should explain the difference. I would like to direct your attention on this shift in arguments; when the old ones failed, the firms impudently moved on and came up with new ones. The fact this list of arguments is constantly changing confirms the weakness of all arguments it holds, and the apparent ‘ignorance’ of Western consultancy firms regarding the Chinese gold market.

First let’s go through all the arguments to investigate which ones make any sense, at the end of the post we’ll do some number crunching.

1) INDUSTRIAL DEMAND. The first argument ever presented to me was from the WGC. In August 2013 I’ve asked the Council what their explanation was for the difference between their Chinese gold demand and demand disclosed in the CGA Gold Yearbooks (co-written by the PBOC), which exactly equaled SGE withdrawals. They replied to me by email:

The data that we publish in Gold Demand Trends are collected for us by Thomson Reuters GFMS. Our data represent jewelry and bar & coin demand and do not incorporate any industrial demand or fabrication, which is included in the PBoC figures. As I am sure you will appreciate, data collection of this sort relies on a number of proprietary sources and these will not necessarily be the same for both GFMS and PBOC. It is, therefore, perhaps not surprising that the estimates of demand differ somewhat.

Not very credible the WGC identifies a gap of 3,193 tonnes of gold with industrial demand, but, although being small, industrial demand isn’t captured in WGC Chinese gold demand, for whatever reason, and thus partially explains the difference from a metrics point of view.

The reason I tend to compare SGE withdrawals to Chinese gold demand as disclosed by the WGC, and not by GFMS, Metals Focus or CPM Group, is because the WGC is globally the easily (free) accessible data source for investors. Usually investors and news agencies worldwide consult the WGC for supply and demand statistics, which make these the most important to test for their accuracy. Why the WGC doesn’t include industrial demand in their data is beyond me, but for our own investigation we’ll simply take notice.

Industrial demand is a legitimate argument and its volume will be taken into account for our own calculation of genuine Chinese gold demand at the end of this post.

2) STOCK MOVEMENT CHANGE. When I asked GFMS in August 2013 about net investment – which is how the difference was titled in the CGA Gold Yearbooks – they wrote me by email:

We have checked with our Data Specialist and confirmed that we use a different methodology. Total Chinese demand used by Thomson Reuters GFMS only includes jewelry, physical bullion bars/coins and all industrial demand. Any stock movement change (which is essentially the item 6 net investment) will not be included as underlying demand.


So according to you category six is “stock movement change”? This would be gold added to the stocks from jewelers, the mint, industrial companies, etc? (this is a few hundred tons each year!)


That’s correct based on the resolution provided by our data specialist.

Because SGE withdrawals capture wholesale demand the difference is partially what jewelry companies, refineries, industrial companies and the mint have purchased at the SGE, but not yet sold in retail. And so, stock movement change is a legitimate argument, though the amount of gold in stock can never explain the full difference of 3,354 tonnes.

According to an estimate by the WGC as much as 125 tonnes of gold can have been absorbed as inventory in the Chinese domestic gold market from 2009 until 2013:

… It is, however, indicative that as jewelers expanded, so too did their inventory levels and it is our judgment that across the industry between 75t to 125t may have been absorbed in the supply chain since 2009.

As the period we are investigating with respect to the difference spans wider, from 2007 until 2014, we’ll use an estimate of 200 tonnes for stock movement change.

Stock movement change is a legitimate argument and its volume will be taken into account for our calculation of genuine Chinese gold demand at the end of this post.

3) ROUND TRIPPING. In April 2014 the WGC published a report on China titled China’s Gold Market: Progress and Prospects. It certainly was not the first WGC report on China, in 2010 China Gold Report was released, but it was the first time the Council elaborated on the structure of the Chinese gold market, the Shanghai Gold Exchange and the ‘supply surplus’ in the Chinese gold market. Logically, the Council had some explaining to do, as it was clear China imported substantially more gold than what they disclosed as demand.

For the first time Chinese Commodity Financing Deals (CCFD) were introduced to the Council’s wide reader base. This type of financing is pursued to acquire cheap funds, it can be done trough round tripping or gold leasing. The Council wrote:

These operations fall into two broad categories, although there is some overlap between the two. Firstly, there is the use of gold via loans and through letters of credit (LCs) as a form of financing. Secondly, there is the use of gold for financial arbitrage operations that will also be based upon gold loans or LCs. In most cases the gold is quickly re-exported to Hong Kong, often as very crude jewellery or ornaments to get round tight controls on bullion exports. (This is the practise commonly referred to as ‘round-tripping’. Moreover, because nearly all gold flowing into China goes through the SGE, round-tripping can inflate the SGE delivery figures.) In other cases the metal is stockpiled in vaults in China or Hong Kong.  

In particular the part in bold is not true, as we could read in my previous posts. Basically, round tripping gold flows are completely separated from the Chinese domestic gold market and the SGE system, therefor they can not inflate SGE delivery or withdrawals.

So, round tripping is not a legitimate argument. To my understanding the WGC has abandoned this argument all together, though GFMS still thinks round tripping inflates SGE withdrawals. In their Gold Survey 2015 it’s written (page 78):

…the round tripping flows between Hong Kong and the Chinese mainland, which also inflates the SGE turnover and withdrawal figures…

4) GOLD LEASING. The other CCFD is leasing. In the WGC report from April 2014 it’s stated:

No statistics are available on the outstanding amount of gold tied up in financial operations linked to shadow banking but Precious Metals Insights [PMI] believes it is feasible that by the end of 2013 this could have reached a cumulative 1,000t…

PMI insinuated 1,000 tonnes is tied up in CCFDs, but as I’ve clearly demonstrated in previous posts, this is not true. There is no need to go over this again – if you wish please read my previous posts for a detailed analysis.

5) OFFICIAL PURCHASES. Often it’s being thought in the gold space SGE withdrawals end up in the vaults from the People’s Bank Of China (PBOC). Early 2013 the WGC speculated the difference could be explained by official purchases, later that year the Council changed its mind. From the July 2014 WGC report on China, Understanding China’s Gold Market, we can read:

China’s authorities have a range of options when purchasing gold. They may acquire some of the gold which flows into China; there has been no shortage of that. But there are reasons why they may prefer to buy gold on international markets: gold sold on the SGE is priced in yuan and prospective buyers – for example, the PBoC with large multi-currency reserves – may rather use US dollars than purchasing domestically-priced gold. The international market would have a lot more liquidity too.

In my post PBOC Gold Purchases: Separating Facts from Speculation I’ve analyzed why it doesn’t make sense for the PBOC to purchase gold through the SGE. The firms will agree the PBOC is not likely to buy gold through the SGE and thus official purchases cannot make up the difference we’re after.

6) RECYCLED GOLD. The most obvious argument to explain elevated SGE withdrawals, one would think, is recycled gold through the bourse counted over and over as withdrawn. Though, SGE rules command bars withdrawn are not permitted to re-enter the vaults before being remelted and assayed by an SGE approved refinery. Which is not say it doesn’t happen.

Arguments presented by the firms regarding recycled gold must be divided in subcategories. In Understanding China’s Gold Market the WGC was correct in pointing out there are two sorts of scrap flows going through the SGE; gold-for-cash and gold-for-gold.

Gold can be sold for cash, thereby increasing supply, while gold can also be sold for gold, increasing both supply and demand. Gold-for-gold supply does not affect the supply-demand balance, hence it’s not counted as supply in WGC metrics – nor is the matching demand side. I would say gold-for-gold cycles through the SGE are a legitimate argument that explain SGE withdrawals to inflate and therefor will be taken into account for our calculation of genuine Chinese gold demand at the end of this post.

Let’s have a look at examples of gold-for-gold:

6.1) Process scrap. This argument was first presented by CPM Group. In short, CPM states industrial companies produce 50 – 70 % scrap supply of the gold used in manufacturing. The scrap spillover flows back to the SGE. Process scrap thus inflates SGE supply and demand, because the gold was bought at the SGE (demand), but flows back for a significant part (supply). Although, it’s unknown how much of this gold actually flows back to the SGE or is brought to a refinery for toll refining (a refinery producing bars or wire from the process scrap for the industrial company in return for a fee).

Process scrap, described in detail by Jeffrey Christian at the very end of this post, is a form of gold-for-gold scrap supply.

6.2) Arbitrage refining. This argument was brought forward by GFMS on 17 February 2015 at the Reuters Global Gold Forum when Jan Harvey interviewed Samson Li (GFMS).

Jan Harvey:

Some people see withdrawals on the Shanghai Gold Exchange as a proxy for Chinese demand. Do you think this is valid?

Samson Li:

It depends on the methodology used. For example there are refiners that would, at times, withdraw 9995 gold bars from the SGE, refine it into 9999 bars whenever there is profitable opportunity, and then deposit it back into SGE vault……

Presumably there can be an arbitrage opportunity at the SGE if Au99.95 gold is an X percentage cheaper than Au99.99 gold. Such a spread would be a classic example of one of the contracts being under or overvalued relative to the other.

I’m not a trader, but I can imagine a way to close the arbitrage through gold leasing. This is my theory: if a spread occurs Au99.95 is bought, concurrently Au99.99 (LAu99.99) is borrowed and immediately sold. Then the Au99.95 is withdrawn, refined into Au99.99 and returned to the lender.

If the arbitrage described above can be closed inter alia depends on the speed to which a lease contract can be settled. If a spread occurs and the refiner has to wait 2 days before it can take delivery of Au9999, the arbitrage won’t fly. I’ve asked the ICBC gold lease desk what would be the fastest possibility to sign a lease contract. They told me usually it takes several days or weeks as the lessee’s credit rating must be determined. Though, for regular customers the lease ca be executed in one hour.

It’s hard for me to say if arbitrage refining is really possible according to the aforementioned theory, because it depends on many variables and the established relationship between lessor and lessee. In addition, why would anybody sell Au99.95 if it was undervalued? In my opinion the argument that arbitrage refining inflates SGE withdrawals can be doubted.

6.3) VAT schemes. This argument is legitimate. though it’s unknown to what extent it has been used. Read more here.

On the South China Morning Post there was an article published wherein an illegal VAT scheme is described which can inflate SGE withdrawals.

So how much is gold-for-gold in total (all aggregated process scrap and potential arbitrage refining and VAT schemes)? We can estimate total gold-for-gold supply and I will tell you how. The advantage of calculating total gold-for-gold supply is that it doesn’t require us to know exactly how much all separate gold-for-gold volumes are.

Every year the Chinese publish the composition of total supply in the Chinese wholesale gold market in the CGA Gold Yearbooks. Have a look at the next chart (I’ve added GFMS scrap, which is gold-for-cash supply):

WGC demand vs SGE withdrawals 2
The gap between the green bars (total supply) and the red bars (SGE withdrawals) are caused by additional scrap flows (gold-for-gold) through the SGE.   

By knowing, (i) SGE withdrawals, (ii) Import, (iii), Mine output and (iiii) Gold-for-cash (labeled as GFMS scrap in the chart) we can calculate gold-for-gold scrap, because of the structure of the Chinese gold market:

Gold-for-gold scrap = SGE Withdrawals – Import – Mine production – gold-for-cash

It may be difficult to track process scrap and potential arbitrage refining directly, indirectly the Chinese disclose the volume of both flows as gold-for-gold. By knowing import and we can fill in the equation.

To come to a thorough understanding of Chinese gold supply/demand metrics gold-for-gold scrap flows are very much worth measuring. Especially, because since 2013 this supply category has grown.

7) EXPORT. This argument was brought forward by PMI. On a conference in London Phillip Klapwijk, Managing Director of Precious Metals Insights Limited (PMI), stated China exports about 1,000 tonnes a year (from the domestic gold market). However, at this stage the rules prohibit gold export from the Chinese domestic gold market. I’ve written an extensive analysis on Klapwijk’s presentation (click to read), no need to go over this again here. The export argument is not legitimate.

8) THE SHANGHAI INTERNATIONAL GOLD EXCHANGE. This argument is from myself. As we could have read in The Workings Of The Shanghai International Gold Exchange and SGE Withdrawals In Perspectivethe gold withdrawn from the vaults through the SGEI in the Shanghai Free Trade Zone can be exported abroad and thereby distorting Chinese wholesale gold demand when measured by SGE withdrawals. However, up until now (16 December 2015) it seems SGEI withdrawals are rarely exported abroad, according to several sources. The SGEI argument is not (yet) legitimate.

9) SMUGGLING. Naturally, smuggling can cause SGE withdrawals to be inflated. Although, we have no numbers on smuggling so I can’t take it into account for our calculation of genuine Chinese gold demand.

10) ETF demand read this.

How much is genuine Chinese gold demand?

The difference between SGE withdrawals and WGC Chinese consumer gold demand from 2007 until 2014 accounts for 3,193 tonnes. By subtracting the volume of gold involved in legitimate arguments from the total difference we can calculate genuine Chinese gold demand. Let’s put to work the numbers and see what happens.

1) INDUSTRIAL DEMAND. Data on Chinese industrial demand wildly varies. Thereby, both the CGA and GFMS publish industrial demand including the use of scrap. Because the scrap produced by fabricators flows back to the SGE and must be subtracted from SGE withdrawals as gold-for-gold to compute genuine Chinese gold demand, we need to measure industrial demand excluding the use of scrap. We’ll use an estimate of 200 tonnes for Chinese industrial gold demand (excluding the use of scraps) from 2007 until 2014:

3,193 tonnes minus 200 tonnes = 2,993 tonnes

2) STOCK MOVEMENT CHANGE: For growth in wholesale inventory from 2007 until 2014 we’ll also use an estimate of 200 tonnes:

2,993 tonnes minus 200 tonnes = 2,793 tonnes

6) GOLD-FOR-GOLD. This type of recycled gold supply from 2007 until 2014 accounted for 903 tonnes:

2,793 tonnes minus 903 tonnes = 1,893 tonnes

When we subtract the tonnage from all legitimate arguments from total SGE withdrawals we’re still left with a difference of 1,893 tonnes of gold. Contrasting metrics can only explain the difference for 1,300 tonnes (3,193 – 1,893).

While, the 1,893 tonnes cannot be labeled as anything else than genuine gold demand (and many Chinese gold industry executives have publicly disclosed to wholeheartedly agree with me). To repeat myself, the residual difference can only be caused by direct purchases from individual and institutional customers at the SGE that withdraw their metal from the vaults. 

China SGE gold demand data
Gold-for-gold can be calculated by subtracting ‘GFMS scrap’ from ‘recycled’.

In the chart above you can see the data I’ve used to write this post. Total SGE withdrawals from 2007 until 2014 accounted for 8,822 tonnes, while WGC demand accounted for 5,629 tonnes over this period. We can only correct the difference of 3,193 tonnes by 1,300 tonnes when taken into account contrasting metrics, leaving a difference of 1,893 tonnes.

5,629 tonnes (WGC demand) + 1,893 tonnes = 7,522 tonnes, which is genuine (estimated) Chinese gold demand for 2007 until 2014.

Koos Jansen
E-mail Koos Jansen on:

  • Matthew

    Where do you get GFMS saying 110t of industrial demand 2007 – 2012 – must be much higher than that?

    • JanNieuwenhuijs

      From The Gold Survey 2015, but I see I missed a large segment (electronics!). If I combine electronics and “other industrial uses” I get 530 tonnes. Would that be more accurate in your opinion?

      • Matthew

        Sounds more reasonable. I’d ignore the WGC figures and just use the GFMS total fabrication plus bar and coin.

        • JanNieuwenhuijs

          I’ll change it (though I’m not sure what they mean with INCLUDING THE USE OF SCRAP, that must mean they use their own scrap no?)

          • Matthew

            The difference between fabrication inc scrap and fabrication ex scrap is their scrap estimate for China. So basically if one buys a 1 tonne of gold but scraps 0.8t then the latter will only show 0.2t. You should use ‘inc scrap’ if you are putting scrap on the supply side.

          • JanNieuwenhuijs

            Hmm, well I already subtracted “gold-for-gold” so I would need fabrication ex scrap. But, where to get this figure?

          • Matthew

            I’d have to think about that but the GFMS report has ‘fabrication excluding scrap’ the table following ‘fabrication including scrap’ (or it always used to)

          • JanNieuwenhuijs

            I think 200 t is not a bad estimate for industrial demand ex scrap.

  • wheresthefreemarket

    BlackRock’s Warning: Get Your Money Out Of All Mutual Funds

    BlackRock Inc. is seeking
    government clearance to set up an internal program in which mutual funds
    that get hit with client redemptions could temporarily borrow money
    from sister funds that are flush with cash. – Bloomberg News

    We may have been early on warning about leaving your savings in the
    financial system. It’s okay to be too early getting your money out of
    the system but it’s fatal to be just one second too late. The gates are
    already in place in money market funds just waiting for the signal to
    be lowered

    BlackRock’s filing with the SEC to enable “have cash” funds to lend
    to “heavy redemption” funds should send shivers down the spine of anyone
    with funds invested in any BlackRock fund. In fact, it should horrify anyone invested in any mutual fund.

    Larry Fink, BlackRock’s chief
    executive officer, said in December that U.S. bond funds face increased
    volatility, adding that he expected a “dysfunctional market” lasting days or even weeks within the next two years. – Bloomberg


  • Linda Serena

    The World Gold Council reports a 3% reduction of Chinese gold demand in Q2 2015 versus Q2 2014.

    This figure does not match with Shanghai Gold Exchange withdrawals. Eyeballing the chart, I would estimate average withdrawals of 30 tonnes in Q2 2014, and 40 tonnes in Q2 2015, an increase of about 30%.

    What can be the reason for this and where did that gold go ?

    • Linda Serena

      And the gap is expanding, SGE withdrawals already 400 tonnes higher than last year.

      I think People’s bank of China may well be a buyer. They buy at their national SGE and not overseas and if they wish to sell dollars, they do it in another transaction. Could province banks be buyers as well ?

Copyright Information: BullionStar permits you to copy and publicize blog posts or quotes and charts from blog posts provided that a link to the blog post's URL or to https://www.bullionstar.com is included in your introduction of the blog post together with the name BullionStar. The link must be target="_blank" without rel="nofollow". All other rights are reserved. BullionStar reserves the right to withdraw the permission to copy content for any or all websites at any time.