Saturday, 28 September 2019

New linear regression between the HUI miners index and the Gold Price

Price trends of individual miners are correlated with the gold price. On an aggregated level, a linear regression between the HUI gold miners index and gold may for hold several years. Presently, an emerging linear regression is revealed, fundamentally different from what was observed before.

In order to derive a linear trend between two time dependent variables, it is necessary to eliminate time, retaining only the couples of variables. Here (HUI, Gold) couples are retained, whereby the gold price on the horizontal axis is used to 'explain' the aggregate price level of miners as revealed by the HUI index.

This procedure was first applied to the HUI-Gold price trend prevailing from about mid July 2012 till late summer 2017, when it started degrading. In the below graph these are the blue dots, with the red regression line. The linear regression only became valid after gold failed its last attempt to regain $1800/Oz. Before and after that period of regularity, gold miners were losing ground. From late autumn 2010 till early summer 2012, miners were unable to follow the gold price as it ascended to its all time high. Corresponding points are the orange dots at the upper right end of the graph. After the summer of 2017, miners systematically were dropping below the regression line. The mismatch aggravated with miners moving sideways as gold was progressing reaching the top of its trading range around $1350/Oz. During spring 2018, gold repeatedly topped above $1350/Oz, yet miners ignored gold strength and slid lower. As gold eventually started sliding in May and June 2018, miners initially upheld; yet after the precipitous drop mid August last year, all confidence evaporated.
Corresponding points are the light blue dots in between the two regression lines.

Linear Regression Trends between gold and the HUI miners index
Candidly however, major gold miners were coupling to the gold price again, though at a much less challenging pace. This regression procedure minimizes the sum of squared differences of the dots (observed values) to the regression line. Brown dots and the black regression line show the new linear behaviour relating the HUI to gold:

HUI = 0.2678 . (Gold - $ 680.3)   (correlation 0.95)

Knowing the regression line, a synoptic view can be made using two different Y-axis. The left one is for gold, while the right axis for the HUI is modified to implicitly reflect the regression relationship.
As such the red curve for gold on the left axis coincides with the regression value for the HUI when read on the right axis. The blue curve are the observed HUI values equally read on the right axis.

Synoptic graph of  gold (left hand scale) in red and the HUI (right hand scale) in blue
We start our observation in spring 2017 when previous regression was still valid. During 2017 gold experiences a modest uptrend with higher lows while highs at the top of the trading range are more frequent during winter 2017-18. The HUI graph however moves sideways, challenging and finally invalidating the previous regression trend. This has been reported on repeatedly. See for example: Gold miners persistently lagging the metal (Feb 2018). Both curves in fact first touch late winter and early spring of 2018. However in last effort of 'make believe' the HUI ignores initial gold weakness during late spring. This coincides with the USD strengthening after the hawkish FED intended raising the FED funds rate fourfold in 2018 while drawing the perspective for further hikes in 2019.
Weaker currencies in the producing countries may balance lower gold prices in USD, affecting margins only little. In August, the rapid gold slide toppled that last support mechanism and miners plunged. Amidst all adversity, miners however coupled with gold again with the new regression relationship ultimately dragging the miners from their swamp as gold recovered by end 2018 and started its recent uptrend last summer.

* * *

Are we to be happy about this ? Well, optimism seems slightly inappropriate when we compare the current trend-line to those observed earlier in the 21st century:

Mid March 2002 till Mid March 2008 (the splendid one)
HUI = 0.6183 . (Gold - $ 97.7) with a correlation of 0.946 over the target time interval (6 years)
For a detailed description see: HUI to Gold regression revisited

November 2008 till Nov 2010 (the rapid one)
HUI = 0.5815 . (Gold - $ 377) with a correlation of  0.901 over the target time interval (2 years)
Yet despite the short time stretch, gold nearly doubled over the observation period.
For a detailed description see: HUI to Gold regression: the 2008-10 recovery

Mid July 2012 till August 2017 (the scary one)
HUI = 0.563 . (Gold Price - $ 895) with a correlation of 0.963 over more than 5 years.
This regression relationship guided us through the (scary) gold bear market, the subsequent 2016 boom - bust and much of 2017. For a detailed description see (among others) Linear Regression between the Gold Price and the HUI miners index.

Mid August 2018 till  ???  (the meek one)
HUI = 0.2678 . (Gold - $ 680.3)   (correlation 0.95)

All prior regressions failed in the same way: miners were unable to keep track of a rising gold price. They started lagging gold. The deterioration of the correlation sometimes has been gradual at first but progressive thereby invalidating the regression over time. 

The main difference among the first three regressions was the intercept rising dramatically. This reflects investor's apprehension of what the average level of cash costs is among miners. If the gold price hits the intercept value, the regression line is at zero for the HUI: the gold price durably at or below this level would drive the gold mining industry out of business. Perhaps few mining sites would still be profitable, but they might be dragged along in the default of their mother companies.

The slope of the HUI/Gold regression lines became mildly less steep from 2002 till 2017.
The slope reflects what is the investor apprehension of how much gold a unit of the HUI is capable of producing. The dramatically lower slope now pinpoints that mines are indeed wasting assets. Not only there is a limited quantity of gold reserves, but yearly production also is over the hill. These ideas ultimately seem to be dawning on investors. If few miners produce more, that's due to the mega-mergers they went through. There is little intrinsic production growth of new development projects coming on-line or scaling up. It starts being balanced by reduced production among mining sites nearing exhaustion or facing rapidly rising cash costs turning future production non-profitable.
Any future gold production growth likely will be realized by miners outside the HUI or XAU indices.

No comments:

Post a comment