This week for the first time - in the name of "transparency" - the London Bullion Market Association unveiled that as at 31 March, 2017 there were 7,449 tonnes of gold, or 596,000 gold bars, valued at $298 billion sotred in the vaults around London as well as $19 billion in silver. Only the gold hoard at Fort Knox and among Indian households said to account for more than the LBMA's inventory, which clears just over $18 billion in gold daily. Most London gold is stored in the Bank of England, with the rest in private vaults, including those operated by HSBC and JPMorgan, both profiled previously (here and here).
The Bank of England holds most of the gold and silver in London, or over 60% of the total gold, and already publishes some details of its holdings. The new LBMA data supposedly also reveals how much private custodians, HSBC, JP Morgan, and ICBC Standard Bank among them, keep in their vaults.
The publication of vaulting statistics marks the first step toward the LBMA's promise of greater transparency, which will eventually be enhanced further by trade reporting that is set to also be published later on. respectively.
As UBS strategist Joni Teves writes, this addition to the available pool of information helps various market participants, as well as regulators, have a better assessment of market activity in precious metals. For instance, comparing the newly released vaulting statistics with trade flows allows for a better understanding of the movement of metal and the dynamics between investment and physical demand. Combined with other indicators of market length, such as Comex net longs and ETF holdings, it also enables market watchers to have a better gauge of market positioning. The data released so far confirms relatively limited investor participation this year – perhaps as a result of a broad shift toward cryptocurrencies as the new "safe" currency, at least among some age groups. Bank of England (BoE) data, which extends further back to 2011, confirms that investor positions have declined considerably from the peak during the height of the bull-run and that there's room for gold exposures to grow from here.
As LBMA data series extends up ahead, we should get an even better picture. All in all, changes in gold holdings as illustrated by LBMA and BoE reinforce a positive stance on gold.
Courtesy of UBS, below we present extended observations on what the LBMA data reveal, and the implications for the broader gold market in general, as well as investing in the precious metal in particular.
Never 'too much of a good thing' with market data
The London Bullion Market Association (LBMA) and the London Precious Metals Clearing Limited (LPMCL) have released data on gold and silver inventories sitting in London vaults (loco London gold and silver), a move that is consistent with wider developments in the industry towards greater transparency. This addition to the available pool of information helps various market participants, as well as regulators, have a better assessment of market activity. Comparing the newly released vaulting statistics with trade flows allows for a better understanding of the movement of metal and the dynamics between investment and physical demand. Combined with other indicators of market length, such as Comex net longs and ETF holdings, it also enables market watchers to have a better gauge of market positioning. The publication of vaulting statistics marks the first step towards more transparency in the industry; trade reporting that is set to also be published later on should further enhance this. Having data on loco London gold and silver trading volumes would allow for better comparisons of trading activity and interest across various regions. As far as market analysis is concerned, there's no such thing as 'too much of a good thing' when it comes to information that provides additional insights.
The LBMA and the LPMCL first announced the plan to publish gold and silver vault holdings in May. The data published today represents aggregated data collected from the Bank of England (BoE) and seven other custodians that offer vaulting services, and covers end-period holdings of gold and silver from July 2016 to March 2017. Going forward, this data will be published each month, with a three-month lag. The data includes gold and silver in the form of large bars, kilo bars or coins, but excludes jewellery and other private holdings held in vaults that are not part of the London clearing system. These statistics only include gold and silver that is held physically in London, defined by the LBMA as “held within the environs of the M25”. This metal underpins the daily trading and clearing activities in the London bullion market.
London continues to be a key gold trading centre globally, with an average of $18.1bn worth of gold cleared each day in March 2017, according to LBMA net daily clearing statistics. Clearing data is different from turnover data, so until we get LBMA trade reporting statistics, comparing London market activity against that of other regions remains challenging. But to somehow put things into context for now, it probably still helps to note that the average daily turnover of physical gold spot contracts on the Shanghai Gold Exchange is over $1bn, while an average of about $32bn worth of gold futures trade on Comex each day.
Estimating investor gold holdings in London
One of the recurring questions that gets asked by clients and various market participants is on the level of gold or silver holdings or the level of investor positioning. This has always been very difficult to answer. The market has had to rely on third-party estimates of above-ground holdings, such as those provided by Thomson Reuters GFMS, the World Gold Council or Metals Focus. Other ways that we've tried to address this type of question is to look at Comex net long positions and ETF holdings. Now, with the availability of vaulting statistics, we can add another dimension to the analysis. We can start by saying that 7,449 tonnes or 239.49 moz of gold and 32,078 tonnes or 1,031.32 moz of silver were sitting in London vaults as of end-March.
For gold, we can take the analysis further: as the BoE also publishes gold vault holdings data separately, we are able to split out – at least in aggregate – what is held in other vaults. This is not possible for silver as the BoE does not provide silver custodial services. The BoE provides gold custody services primarily to central bank customers to help them get access to London gold market liquidity, although it also holds some gold custody accounts on behalf of certain commercial banks that support central bank access to the liquidity of the London gold market. On the assumption that the BoE custody service primarily caters to central bank customers, it’s probably fair to say that 1) the bulk of BoE gold represents official sector holdings and 2) gold held in other vaults is likely to be more investment-related. To be sure, this is somewhat simplistic in that, as mentioned earlier, the BoE also offers custody accounts to certain commercial banks, while at the same time it is also possible for central banks to hold their gold in non-BoE vaults.
To put some context around this, let us consider for illustration's sake that about 80% to 90% of BoE gold holdings are accounted for by the official sector. Let us also consider for simplicity that the vast majority of gold held in commercial vaults represents investor holdings and only a negligible amount comprises official sector gold accounts. These simplistic assumptions would therefore imply that over the past year, an average of about 96.68 to 110.92 moz ($119-139 bn) of investment-related gold was held in London. These inventories would include metal held on behalf of ETFs. Based on our database of gold ETFs, we estimate an average of around 48.78 moz or about $60bn worth of gold was held in London vaults to back ETF shares during the same period. Taking these ETF-related holdings into account would then leave roughly around 47.90 to 62.14 moz or about $60bn to $78bn worth of gold in unallocated and allocated accounts as available pool of liquidity for OTC trading activities.
While we now have a starting point with the vaulting statistics, a breakdown of the type of gold holdings remains unavailable. As such, the discussion above and the charts and tables below are presented here to provide some context. Nevertheless, despite these limitations, the data still somehow supports the broad assumption on how loco London gold stocks may be split between official sector and investor holdings. For instance, the data shows that gold held at the BoE appears to be more stable relative to gold held in other London vaults, such that changes in total loco London gold over the past nine months have been driven more by changes in non-BoE gold. Intuitively, it would make sense for investment-related gold holdings to be more volatile relative to official sector holdings. Central banks hold gold as part of their official reserves and also serve monetary purposes, typically having much longer.
While we now have a starting point with the vaulting statistics, a breakdown of the type of gold holdings remains unavailable. As such, the discussion above and the charts and tables below are obviously only presented here to provide some context. Nevertheless, despite these limitations, the data still somehow supports the broad assumption on how loco London gold stocks may be split between official sector and investor holdings. For instance, the data shows that gold held at the BoE appears to be more stable relative to gold held in other London vaults, such that changes in total loco London gold over the past nine months have been driven more by changes in non-BoE gold. Intuitively, it would make sense for investment-related gold holdings to be more volatile relative to official sector holdings. Central banks hold gold as part of their official reserves and also serve monetary purposes, typically having much longer time horizons.
Matching ETF data against total loco London inventories
Gold and silver held in London vaults that back shares in exchange-traded funds (ETFs) are also included in the statistics, regardless of where the ETF is listed. In general, this means that although gold or silver ETF shares may trade on exchanges elsewhere, as long as the metal is held physically in London, these would be considered loco London and form part of the LBMA’s monthly statistics. For instance, the largest gold ETF is listed and trades on the New York Stock Exchange (NYSE), but its designated custodian ultimately holds the underlying metal in its London vaulting facilities. Similarly for silver, the ETF with the largest holdings trades on the NYSE, but some of the metal underpinning the shares is allocated in London.
We estimate that as of last week there are around 68.65 moz of gold and about 687.57 moz of silver held in ETFs globally. Our database of gold and silver ETFs suggests that, using average gold and silver prices over the past year, about 48.78 moz of gold worth about $60bn and about 427.67 moz of silver worth about $8bn are likely to be held in London to back ETF shares. This is obviously a rough estimate that’s provided here only for indicative purposes.
Looking at changes in ETF holdings vs changes in total vault holdings between July 2016 and the end of the first quarter this year shows that the trend is broadly similar for gold, yet divergent for silver. Considering the period as a whole, the divergence continues for silver with total holdings up by 8% or 79.89moz, but ETFs down 3% or 12.06 moz. For gold, the change in total stocks vs the change in ETFs also diverges when the entire period from July 2016 to March 2017 is taken into account – total loco London gold inventories increased by a modest 2% or 5.34 moz but ETF holdings were down 8% or 3.87 moz.
Given the data series is quite short, it is challenging to form any strong conclusions at this juncture. We expect that as the series extends in time we will be able to gather more insights from these flows. Intuitively, divergences in total London gold stocks and ETF holdings could reinforce other indicators that suggest declines in ETF shares do not always reflect net exit from gold or silver exposures. Instead, outflows from ETFs could represent a switch to different forms of exposure such as allocated or unallocated accounts.
Tracking gold trade flows vs changes in London stocks
Comparing total gold holdings data with UK gold import and export flows confirms a strong link between the two, regardless of whether one is using UK trade data or what may be implied by Swiss Customs statistics. This is intuitive in that net inflows of gold into the UK should understandably translate into a build in holdings, while net outflows would suggest a reduction in overall levels. In addition to UK trade statistics, Swiss data on gold flows from and to the UK is also a useful indicator of how much gold is leaving or entering London vaults. Given Switzerland’s key role in global gold refining, it is an important hub for the movement of metal and essentially acts as an interface between investment demand in Western markets and consumer demand in Asia. Combining Swiss and UK trade data with London vault holdings allows for a more complete picture.
Periods of strong physical demand out of key markets such as India and China could attract the movement of large investment bars sitting in London vaults to Switzerland for conversion into kilo bars. These gold kilo bars would subsequently be shipped to physical buyers in Asia and ultimately be converted into jewellery or other investment products to meet consumer demand. Net outflows of gold from the UK should coincide with declines in vault holdings in this case. In fact, this scenario occurred back in 2013: as the Fed's taper tantrum led to a sharp rise in US real rates, gold collapsed by 29% as investors exited gold exposures. In turn, the sharp drop in the gold price made it very attractive for physical buyers, especially in the context of a decade-long bull market. Investor gold positions held in London vaults – whether OTC or ETF-related – were being heavily liquidated. This metal then made its way to Switzerland, was converted into kilo bars, and shipped to Asia to satisfy the surge in demand in China, India as well as other parts of the region.
Interestingly, LBMA data shows that in Q3 last year, the opposite occurred – the UK was a net importer of gold and vault holdings increased. This coincided with weak physical demand for most of last year, which was more than offset by a pickup in investment interest. Given strong demand from the investor community, gold still managed to rally 8% in 2016 (and as much as 30% from trough to peak) despite the weak fundamental backdrop. Towards the end of the year, market forces became conducive for gold to once again leave London vaults and be shipped out of the UK. Investors unwound positions in anticipation of the first Fed rate hike, while at the same time, seasonal and regulatory factors in key markets boosted physical demand, allowing buyers in Asia to absorb the liquidation from Western investors. Demand out of China kicked in heading into year-end, in anticipation of the Lunar New Year holidays, while the demonetisation of highvalue banknotes in India led to a knee-jerk spike in demand. So far this year, physical markets have been more stable, while investment demand has been positive yet broadly more subdued. As mentioned earlier, the data tends to reflect this, with changes in loco London gold holdings relatively limited during the first quarter of 2017.
A marginal benefit of this link between vault holdings and trade flows is that there is a shorter lag in trade data reporting. This means that trade statistics should help market watchers anticipate changes in loco-London inventories before getting confirmation a couple months later from the LBMA report.
Insights on investment activity
Looking at the entire data series from July last year to March this year, an observation that's worth noting is the modest 2% increase in total gold holdings equivalent to around 166 tonnes. This change is likely a function of investor liquidations in Q4 2016, as markets anticipated a Fed rate hike, which has since reversed but at a subdued pace. This also coincides with ETF flows – heavy liquidations towards the end of last year have been succeeded by net inflows this year, but volumes lag considerably compared to the same period in 2016.
More importantly, according to LBMA data loco London gold holdings during Q1 this year were relatively flat. This coincides well with the broadly limited investor participation in the gold market that we have been highlighting. Many investors are keeping an eye on the gold market and continue to appreciate gold's value as a diversifier in a portfolio, yet few have been actively involved in putting on meaningful, strategic positions. Subdued investor participation has been an important factor holding gold back, in our view, in spite of supportive macro factors.
Subdued investor activity has been apparent in speculative positioning data indicated by the CFTC Commitment of Traders Report as well as gold ETF flows. In June, sentiment towards gold understandably came under pressure given the rise in real rates – not just in the US but also in Europe – and the seemingly hawkish shift in tone among central banks (see Gold falls as real rates rise). As of early July, CFTC data showed that gold net longs on Comex had fallen to the lowest levels since January last year. Despite the rebound in gold prices in recent weeks, net speculative positions on Comex have remained very lean. Meanwhile, gold ETFs marked the first month of net outflows in July. Although gold ETF holdings are up on a net basis year-to-date, the increase is much more subdued at only 107 tonnes compared to over 598 tonnes during the same period in 2016. Changes in gold positioning on Comex and ETF holdings coincide with the subdued changes in loco London gold inventories.
As mentioned earlier, the data shows a larger increase in silver vault holdings vs gold. Relative speculative positioning between gold and silver on Comex similarly shows a stronger build in investor positions in silver than gold, at least during the first half of the year. Given silver's large industrial demand component, it initially attracted some attention on the back of growth and risk optimism amid the reflation theme. This translated into persistent gains in Comex investor positioning, which reached a fresh all-time high in May. Yet as fiscal stimulus hopes faded and reflation trades were unwound, so did the interest in silver. Comex positioning has since given back about 81% from the highs, mainly driven by a strong increase in gross short positions, which have been reaching record highs in recent weeks. It would be interesting to see Q2 and Q3 data on loco London silver holdings to find out whether OTC positions followed a similar trend. Loco London silver holdings could potentially be relatively more robust than what Comex speculative positions imply. Silver ETF holdings have generally been resilient over the past few years, despite heavy liquidations in gold ETFs, which could mean that silver OTC positions could also show a similar sense of stability.
Finding clues on positioning from BoE gold data, for now
Given that only limited historical data is available, it is difficult to put total London gold inventories fully into perspective. For instance, we cannot compare current levels versus the amount of gold and silver held during the peak of the bull run nor during the trough reached at the end 2015. However, for gold we can look to BoE vault holdings, which go back to 2011, for some hints. Based on LBMA data, the BoE accounts for the bulk of loco London gold inventories, representing about 68% of the total on average over the period covered. As the LBMA data series extends up ahead, we should be able to have a better idea of how closely linked trends in BoE and non-BoE gold are, and in turn get a better perspective on investor positions.
The amount of gold held at the BoE was at a high of 6,250 tonnes or 200.50 moz in February 2013 and fell to a low of 4,693 tonnes or 150.87 moz in March last year. As of end-Q1, levels are about 8% or 12.49 moz higher at 163.36 moz, but still considerably below the highs. As we argued earlier, the BoE's gold holdings likely mainly reflect official sector positions. Some central banks manage their gold reserves more actively than others while there have been a few such as the Bundesbank which have repatriated gold held in various foreign locations over the past few years. These flows are likely also reflected in the BoE's data. Nevertheless, the BoE does hold some gold custody accounts for certain commercial banks. To the extent that non-official sector gold holdings influences these flows, the trends should be broadly similar to changes in loco London gold held in other vaults. The trend in gold holdings would in this case be consistent with our view that exposure to gold has been reduced considerably from the highs during the peak of the bull run. And although investment interest has been revived over the past year and a half, levels of exposure likely remain limited by historical standards. Much leaner positioning is one of the key factors supporting gold this year and suggests that there is ample room for positions to be rebuilt against a backdrop of benign rates, soft dollar and lingering uncertainties.