“Nobody really understands gold prices,” said Ben Bernanke when asked in July 2013 about that spring’s 25% plunge in bullion prices. “I do not pretend to really understand them either.”
Bernanke’s slur against the analyst’s art could have been tempered, perhaps, if the then-chairman of the Federal Reserve had pointed to real interest rates.
The yield on 10-year inflation-protected US Treasury bonds, which had been showing an increasingly strong negative relationship with gold prices, had just leapt at a record pace in the second quarter of 2013. This response was based on expectations that the Fed itself was about to end quantitative easing as the global financial crisis receded and start raising short-term interest rates. Non-yielding gold could not compete.
Yet despite gold’s underlying and intuitive relationship with real rates then growing stronger still in the decade since, the precious metal failed to sink when 10-year Treasury Inflation-Protected Securities (TIPS) yields jumped by the same record percentage in spring 2022. Instead, bullion prices fell barely 6% to confound what has become a mainstay of analyst models, and gold in non-dollar terms actually rose. So maybe the former Fed chair was on to something. Maybe no one knows anything.
Alternatively, Bernanke could have pointed nine years ago to speculative and investor flows into gold, or rather out of gold at that time. Added to the net change in bullish contracts held by Managed Money traders in Comex gold derivatives, spring 2013’s flight from gold-backed exchange-traded funds (ETFs) totalled the equivalent of 440 tonnes, a then-record quarterly outflow which shrank that combined net long position by more than 16% between the April and June.
Correlation isn’t causation of course, and here again counter-examples have piled up to defy any analysts or traders looking to this factor as a one-shot solution. April to June 2018, for instance, saw the combined net speculative length of Comex and ETF gold shrink by both a greater weight and a greater proportion than it had in Q2 2013, yet the dollar gold price slid by less than 6%. Gold prices then rose by more than that percentage in the first quarter of 2020, yet Comex and ETF length combined was left unchanged overall as the COVID-19 crisis struck.
April to June 2018 saw the combined net speculative length of Comex and ETF gold shrink by both a greater weight and a greater proportion than it had in Q2 2013, yet the dollar gold price slid by less than 6%.
Tail Wags Dog?
None of this is to say that gold prices won’t return to showing a strongly negative relationship with TIPS yields, nor that investors and speculators won’t cut their positions if real rates continue to rise. Notable highs and lows in that combined net long position have tended to coincide with significant turning points in the gold price (see Chart 1) and gold’s 6% price drop this spring coincided with heavy outflows from Comex and ETF positions. Moreover, the precious metal has fallen further against all major currencies so far in Q3 as the US Fed (now under Jerome Powell) has pushed ahead with steep hikes to try curbing inflation from four-decade highs, finally now followed by the Bank of England, ECB and many other rich nations as well as the vast majority of emerging economy monetary authorities.
In September 2022, and with the price of gold now falling as real rates rise and the dollar soars to two-decade highs against the rest of the world’s currencies on its trade-weighted index, gold shows its strongest inverse correlation with US TIPS yields since summer 2020 on a rolling 20-day basis.
Reading minus 0.92 on the r-coefficient, it would read minus 1.00 if gold and TIPS yields were moving exactly opposite each other. That makes it currently stronger than gold’s average positive correlation with the silver price of the last five years.
ETF holdings in the meantime continued to shrink over the summer, and Managed Money traders in Comex futures and options briefly turned net bearish as a group in July, the first time since spring 2019.
To better judge the importance of these investor and speculative flows to gold, it may be useful to identify those segments of demand which show a negative connection with prices. While inflation leads the red herrings among external factors (as James Steel of HSBC shows in his article ‘Gold, Food and Fuel’ in this issue of the Alchemist), the size of gold jewellery demand shows an inverse relationship with the gold market’s direction, as does retail investment demand for small bars and coin as well as central bank gold buying.
To better judge the importance of these investor and speculative flows to gold, it may be useful to identify those segments of demand which show a negative connection with prices
Global Gold Jewellery Demand
Over the last 10 years, and based on the invaluable quarterly data produced by Metals Focus for the World Gold Council’s Gold Demand Trends, global gold jewellery demand by weight (and gross of household selling) has moved in the same direction as dollar gold prices just 35% of the time. Seasonality around Chinese New Year and Diwali in India may confuse that picture, yet global small bar and coin demand (net of selling) shows the exact same lack of co-movement quarter to quarter. So too does central bank gold demand, suggesting that official sector demand, like retail investment and jewellery, responds to price instead of driving it.
This throws out something of a paradox for gold, because the investment flows which so clearly map if not drive gold prices higher or lower account for only a fraction of final demand. Compared to total ETF inflows over the last 10 years, in fact, jewellery and technology purchases, net of recycling flows back to the market, have been 10 times as large; that ratio grows to 11 if we add the change in Managed Money positioning to those ETF flows, because Comex speculators now hold a slightly smaller net long position than they did in mid-2012. Small bar and coin purchases have also been 10 times the size of the last decade’s inflows to the net long position, and central bank buying has outweighed it more than four times over.
A solution to this paradox can be found in tracking the size of Comex plus ETF flows as a proportion of the total net physical demand across time (see Chart 2). On a quarterly basis, the combined net long is very much more volatile than any other element of visible demand, swinging as wide as nearly 100% (Q1 2016) to more than 100% negative (Q3 that same year). What stands out more clearly still, however, is how the size of the combined net long’s quarterly change repeatedly moves in the same direction as bullion prices.
Tail wagging dog? Perhaps. But while solving the mystery of gold pricing may continue to defy a quick sound bite analysis, and while the size of investment plus speculative flows doesn’t show any kind of consistent relationship to the size of price swings, it’s plain that the behaviour of gold ETF investors and Comex speculators, although marginal to physical demand across longer time frames, tends to map if not drive the market’s direction.
Global gold jewellery demand by weight (and gross of household selling) has moved in the same direction as dollar gold prices just 35% of the time.