No metal exited the trust for the SPDR Gold Shares ETF (GLD) on Monday, but that was due entirely to markets being closed for the Memorial Day holiday. Expect outflows to resume today as the trust has shed gold bars on 31 of the last 34 trading days beginning early last month, just after Goldman Sachs recommended that clients not just sell - but short - the metal.
Combined with rumors that officials in Europe might begin seizing gold from the central banks of wayward eurozone members (and aided by some unusually large futures market sell orders) the gold price began tumbling and owners of GLD shares began selling in earnest.
Despite being only moderately correlated with the gold price in recent years as detailed in "Do More Gold ETF Outflows Mean Even Lower Gold Prices Ahead?," ETF flows have dominated the news coverage of the gold market in recent months and it's worth taking another look at the relationship between these two.
In an update to the graphic from six weeks ago as shown below, one thing is clear - ETF flows and the gold price are once again highly correlated as indicated in red. Only in 2007 was there a closer relationship between the two, that is, back when the price of just about every asset class was rising and global markets were readying for their 2008 crash.
In the five years that followed, there were multiple occasions when GLD holdings increased while the gold price was flat and even more periods of time when GLD holdings were steady while the gold price rose.
Notably, the period from mid-2010 to late-2011 saw gold ETF holdings steady or falling at around 1,300 tonnes while the gold price rose by over 50 percent. That was part of the basis for previously dismissing ETF flows as a major factor in determining the gold price, but, with the two now highly correlated again, a closer look is warranted.
A continuous three-month correlation between GLD ETF flows and the gold price as shown below provides an interesting new perspective on this relationship.
Note that the period from 2005 to 2007 has been excluded from the chart since, for the most part, this is fairly uninteresting data - ETF holdings rose steadily, as did the gold price except during the 2006 correction. Also note that the time period of three months was chosen since anything shorter results in far too much noise in the data and, as it is, the chart is still pretty noisy. So, can anything be gleaned from this?
One thing that struck me when looking at the data is that the last few months are one of only a few periods where ETF flows and the gold price have been highly correlated for extended periods of time as indicated in black. In early-2009, late-2009, and again in early-2010, the three-month rolling correlation remained above 0.8 for two to three months at a time, a mark that we are now closing in on again.
In each of the prior cases, the correlation then plunged (which, in itself is not unusual), but each time the gold price made a steady advance not long afterwards. Back then, it was very much a case of gold ETF holdings changing little as the gold price rose, so, there is no simple corollary to the events we've seen recently where ETF holdings are falling.
One possibility is that we might see gold ETF holdings continue to fall as the gold price stabilizes. This has already occurred (to some degree) in recent weeks as gold establishes new (mostly lower) levels of support. It seems unlikely that ETF holdings will increase anytime soon as sentiment amongst institutional investors and hedge funds is about as bad as it's ever been and, if that were to occur, that change in outlook toward the metal would certainly boost the gold price and keep the correlation high.
Similarly, it's unlikely that we'll see ETF holdings stabilize as the gold price continues to fall. The end of outflows will likely be seen as a bullish event in the gold market, particularly after all the attention this subject has garnered in recent months.
One thing is certain, history shows that the correlation between the gold price and gold ETF flows does not remain this high for longer than two or three months, so, we are probably closing in on a big change of some kind. But, what kind of change?
It seems the most likely outcome will be that ETF holdings will stabilize or continue to fall as the gold price forms a lasting bottom and begins to move higher again. Look for the two to become much less correlated at some point in the weeks ahead and, as shown in prior data, once the correlation falls below 0.8 on a three-month rolling average, it will likely fall dramatically.
Additional disclosure: I also own gold coins