Gold ETF Flows: September 2021

Gold ETFs Continue to Slide in September Driving Net Outflows in Q3


September highlights

Gold-backed ETFs (gold ETFs)1 experienced net outflows of 15.2 tonnes (t) (-US$830mn, -0.4% AUM) in September. Outflows in Europe and North America were only partially offset by inflows in Asia. Global gold ETF holdings fell to 3,592t (US$201bn) during the month2  – the lowest tonnage level since April – as the gold price fell on the back of rising yields, a stronger dollar, and a reduction in COMEX managed money net long positions.3


ETF flows chart

Data as of

Sources: Bloomberg, Company Filings, ICE Benchmark Administration, World Gold Council; Disclaimer

September regional overview

European gold ETFs were the primary driver of September’s outflows, followed by North American funds. Larger funds in the UK and Germany led outflows within Europe, which as a whole lost 11.5t (-US$640mn). North America had collective outflows of 6.6t (-US$349mn) attributed mostly to losses from large US funds. Outflows from both regions stemmed from central banks announcing policy tightening going forward – the European Central Bank from its pandemic emergency purchase programme only, while the US Federal Reserve signalled asset tapering in Q4 with higher expectations of interest rate increases next year as well. In contrast, Asian-listed funds ended the quarter positive with inflows of 2.4t (US$135mn), supported again by weakness in Chinese equities in late September, rocked by the Evergrande liquidity crisis. In addition, heightened stock market volatility and a correction in the domestic gold price drove holdings in Indian gold ETFs to their highest level since September 2013.4 Other regions also contributed to global gold ETFs with inflows of 0.4t (US$25mn).5 

Price performance and trading volumes

Gold ended the month approximately 4% lower at US$1,743/oz.6  Our short-term price performance model suggests that gold’s sell-off in September was driven by changes in interest rates, a stronger US dollar, and momentum in futures positioning. Yields rose to quarterly highs in response to central banks’ comments surrounding a likely moderation in easy money policies. In turn this supported dollar strength as real yields mostly increased in lockstep with nominal yields given anchored US inflation expectations.7  This sent gold negative for the quarter, and more than 8% lower on the year after having its worst month since June. Gold daily trading averages increased to US$146bn in September off lows of US$141bn in August, but still below the year-to-date average of US$160bn. COMEX volumes only marginally increased from year-to-date lows in August, while net long futures positioning reflected a general apathy towards gold as of late, reducing to 537t (US$30bn) in the second half of the month. However, this remains just above the historical weekly average level of around 500t (US$28bn).8  For more details see Gold Market Commentary, September 2021

Q3 2021 highlights 9

Gold ETF flows generally tracked gold price fluctuations during the third quarter, led by outflows from North American funds which lost 46.3t (-US$2.6bn, -2.4%) overall. North American outflows were dominated by larger US funds, with a majority of losses coming in August. European funds proved more resilient to price declines over the period as inflation expectations continued to notch higher, resulting in net inflows of 15.2t (US$909mn, 1.0%). This was driven by German funds which, similar to the second quarter, represented more than half of all flows into Europe (8.9t, US$516mn). Low-cost ETFs10  across these regions continued to exhibit strong growth and helped mitigate outflows, adding a total of US$920mn (15.9t). 

Funds in Asia had inflows of US$228mn (3.0%) in the quarter as investment demand remained strong amid volatile local equity markets, particularly in China in the face of stalling economic growth. Moreover, falling gold prices spurred some investors to tactically build gold positions at the dip. Other regions saw inflows of US$13mn (0.4%) during the third quarter, supported by weakness in emerging market equities. 

Regional flows11 

Outflows from European and North American funds in September outweighed inflows into Asian funds

  • European funds had outflows of 11.5t (US$640mn, 0.7%)
  • Holdings in North American funds had outflows of 6.6t (-US$349mn, -0.3%)
  • Funds listed in Asia had net inflows of 2.4t (US$135mn, 1.7%)
  • Other regions had inflows of 0.4t (US$25mn, 0.7%).

Individual flows

SPDR® Gold Shares in the US and iShares Physical Gold in the UK drove global outflows in September, partially offset by inflows into Asian funds and iShares Gold Trust in the US

  • In North America, SPDR® Gold Shares lost 10.2t (-US$561mn, -1.0%), while iShares Gold Trust had inflows of 2.5t (US$145mn, 0.5%). Low-cost ETFs iShares Gold Trust Micro gained 0.6t (US$37mn, 6.2%) and Goldman Sachs Physical Gold added 0.5t (US$29mn, 7.6%) 
  • In Europe, iShares Physical Gold had outflows of 6.2t (-US$351mn, -2.7%) and Invesco Physical Gold lost 1.5t (-US$81mn, -0.6%). On the other hand, Xtrackers IE Physical Gold had inflows of 1.7t (US$93mn, 4.7%), while Xetra Gold added 0.8t (US$48mn, 0.3%)
  • In Asia, Huaan Yifu Gold ETF in China had inflows of 1.1t (US$58mn, 3.4%), while in India Nippon India Gold added 0.5t (US$31mn, 3.9%).

Long-term trends

After a partial recovery in Q2, gold ETF flows across developed markets are again flat to negative in Q3 following continued weakness in gold prices 

  • Year-to-date, gold ETFs have seen global outflows of US$8.3bn (-156t) as large North American and some European funds have lost assets in accordance with fluctuating gold prices, while low-cost and Asian funds have remained steadily positive
  • Bouncing back from declining inflows in Q2, Asian gold ETFs are once again the primary growth driver among global ETFs, having added nearly US$1.2bn (17%) y-t-d on the heels of higher regional economic uncertainty
  • Low-cost ETFs continue to post inflows despite varying price conditions, growing by almost 43% y-t-d (60.1t), and constitute approximately 6% of the total global gold ETF market.


  1. We regularly review the global gold-backed ETF universe and adjust the list of funds and holdings based on newly available data and information.

  2. Based on the LBMA Gold Price PM as of 30 September 2021.

  3. Based on the LBMA Gold Price PM as of 28 September 2021, the last date of available weekly COMEX positioning data

  4. Indian stock market volatility based on the one-month volatility of BSE SENSEX Index for the month ending 30 September 2021. Holdings based on the LBMA Gold Price PM as of 30 September 2021.

  5. Other’ regions include Australia, South Africa, Turkey, Saudi Arabia, and the United Arab Emirates.

  6. Based on the LBMA Gold Price PM as of 30 September 2021.

  7. Based on the Bloomberg US Government Generic 10-year Yield and US 10-year Inflation-Indexed Yield Indices and the US 10-year Breakeven Inflation Index.

  8. From 4 December 2012 to 28 September 2021, based on available data.

  9. A full review of Q3 ETF flows will be available in Gold Demand Trends Q3 2021 at the end of this month

  10. Low-cost US-based gold-backed ETFs are defined by the World Gold Council as exchange-traded open-ended funds listed in the US and Europe, backed by physical gold, with annual management fees and other expenses like FX costs of 20bps or less. At present, these include Aberdeen Physical Swiss Gold Shares, SPDR® Gold MiniShares, Graniteshares Gold Trust, Goldman Sachs Physical Gold ETF, iShares Gold Trust Micro, CI Gold Bullion Fund, WisdomTree Core Physical Gold, and Xtrackers IE Physical Gold ETC.

  11. We calculate gold-backed ETF flows both in ounces/tonnes of gold and in US dollars because these two metrics are relevant in understanding funds’ performance. The change in tonnes gives a direct measure of how holdings evolve, while the dollar value of flows is a finance-industry standard that gives a perspective on how much investment reaches the funds. We have made a few adjustments and improvements to our calculation methodology as of 1 July 2021 that will impact historical and future data. Specifically, we revised the methodology used to estimate changes in gold holdings as described below: • Previously, changes in tonnes were calculated by converting a fund’s AUM (in USD) into gold holdings (in tonnes) and computing the difference over periods. However, currency movements and large daily and weekly gold price movements could distort the difference between tonnage change and US-dollar fund flows during short time horizons. We therefore adjusted tonnage change as a function of fund flows versus AUM and replaced the tonnage change field with fund flows (tonnes). • Now, for most funds, we estimate US-dollar fund flows, as described in section 2.3.2 below, and then convert those flows to fund flows (tonnes). • Fund flows (tonnes) and US-dollar fund flows will now represent a more aligned explanation of investment demand for gold ETFs, while the true holdings of a fund, in US dollars and tonnage, will remain a close estimate, impacted by the currency and price volatility described above. • Based on our initial analysis, the changes are not likely to have a material long-term effect on historical information, particularly on a global or regional aggregate basis, but will adjust short-term fluctuations that can sometimes occur due to input data and timing variations.