RESEARCH AND ANALYSIS
This Institutional Investor article How ETFs Alter the Dynamics of Gold is typical of mainstream financial market commentary on gold ETFs. The idea is that gold ETFs were a “game changer for the gold industry”, making it easier for investors to buy gold and having a positive impact on the gold price. It is intuitive and fits in with mainstream narratives around ETFs in general, which have become a significant feature of financial markets, and no doubt promoted by the World Gold Council (WGC), who sponsored the major gold ETFs including GLD.
The reality, however, is a bit more complex. Let’s take the WGC’s own figures from 2004 (when GLD first started trading) to 2012 (ignoring 2013 when gold ETFs divested 880t), a time period heavily favouring the ETF game changer narrative. During that nine year period, 35,034 tonnes was supplied to the market in the form of newly mined gold and scrap. The chart below shows where that gold went (net nine year demand; miners = mine hedge book reductions).
The dominance of jewellery is probably of no surprise, but I doubt many would be aware that physical bar and coin investment consumed 2.7 times more gold than ETFs. The WGC figures are also skewed against the bar and coin investment category given that the WGC includes Asian/Indian 24ct and low premium jewellery as jewellery when one could argue that sort of buying is more of an investment purchase rather than for adornment purposes like Western jewellery buyers.
If we re classify only 50% of India and China consumer jewellery as bar/coin investment (completely ignoring all other Asian countries), then bar/coin investment style investment then accounts for over 11,400 tonnes of physical gold compared to 2,600 tonnes of ETF investment.
A counter argument to the above might be that price is formed at the margin, so it is not necessarily the total size of the categories that matters, but which of those at the margin tips the supply/demand balance. Certainly this was one of the reasons the WGC shifted its budget from jewellery to ETFs as it saw investment demand as the swing factor in the gold market. If gold ETFs have been behind gold price movements, we should expect to see some correlation between their flows and the gold price, that is, increasing inflows as the price rises and outflow as the price falls.
The table below is a simple correlation of the WGC’s quarterly figures for the period 2004 to 2013 (including the year ETFs showed outflows while the gold price decline, again favouring the dominant narrative) against the average quarterly gold price.
This shows that ETF demand is not correlated to the gold price, with a slight skew to the negative. Physical bar and coin (and central bank activity) however has a very high correlation to the gold price. So if you wanted to forecast the gold price, correctly estimating these two demand figures will give you a far better result than looking at ETF demand. The chart below demonstrates this more clearly.
Here you can see that ETF demand was relatively constant during the gold bull market, averaging 50 tonnes a quarter. In the early phase of the bull market it was similar to coin and bar investment which was running at 100 tonnes a quarter. However the two diverge in 2008 at the time of the global financial crisis, where you can see physical coin and bar buyers moving up to around 300 tonnes a quarter while ETF demand showed little change. It is clear that physical coin and bar investment was more of a driver of the gold market post-2008 than ETFs, with ETFs only having a major impact in 2013.
So why do mainstream financial commentators focus on gold ETFs? Because they are highly visible, giving daily trading volume and holding figures. The less visible/frequent physical coin and bar market figures can’t compete for the attention of journalists needing to write something every day. That’s OK – as Ben Hunt says, it is “the business model imperative of financial media” – but just don’t make an investment decision based on it.