Gold Miners - Tayrona Financial 2019-03-12: High Gold Price To Drive Interest Towards Miners


Do Gold Miners Deserve A Relook? High Gold Price To Drive Interest Towards Miners

  • Low real interest rates to keep gold prices elevated. 
  • Gold may appreciate to US$1,420/oz. 
  • Scarcity of world class mines has spurred consolidation. 
  • More interest in junior miners likely. 
  • Favour CNMC for proven operating track record. 

Dovish Fed supports the case for gold

Signalling a pause to rate hikes while waiting for inflation to catch up.

  • Other than speeches by Federal Reserve officials, we referred to the “Semiannual Monetary Policy Report to the Congress” by Jerome Powell dated 26 February 2019. In the report, the Federal Reserve Chair remarked that, “In January, with inflation pressures muted, the FOMC determined that the cumulative effects of these developments, along with ongoing government policy uncertainty, warranted taking a patient approach with regard to future policy changes.”
  • According to the minutes of the 29-30 January 2019 FOMC meetings, such wording was “intended to convey the Committee’s view that a patient and flexible approach was appropriate at this time as a way to manage risks while assessing incoming information bearing on the economic outlook.”

Quantitative tightening to end in 2019.

  • In the same Monetary Policy Report, it was further remarked that “the Committee can now evaluate the appropriate timing and approach for the end of balance sheet runoff.” In fact, the FOMC further released a statement on balance sheet normalisation after the January meeting, asserting that “the Committee is prepared to adjust any of the details for completing balance sheet normalization in light of economic and financial developments.”
  • Some Federal Reserve officials have even remarked that plans to end balance sheet normalisation will be finalized at coming meetings (see Reuter's report). 

Six months before next major rate decision.

  • Since December 2016, the FOMC has raised the target range for the Fed Funds rate approximately every three months. There was a brief pause in the second half of 2017 after the FOMC announced the commencement of balance sheet normalisation (see FOMC's press release). However, the rate hikes resumed in December 2017.
  • On balance, we reckon that the Federal Reserve will likely slow the pace of rate hikes and that the next major rate decision will only be made in the 18-19 June FOMC meeting.

Bottomline… real interest rates to stay flat or drop.

  • Even if the FOMC were to raise the target range for the Federal Funds rate later in 2019, the objective would probably be to match inflation and real interest rates will likely be lower than current levels given the new policy stance of the FOMC.

Negative real rates have been associated with higher gold prices.

  • Figure 3 in attached PDF report shows that periods of negative real rates e.g. 2008 and 2011 have generally been linked to higher gold prices. Conversely, gold prices fell from around US$1,900/oz to US$1,050/oz from 2011 to 2016 when real interest rates reversed from around - 2% to 2% in the same period. Subsequently, gold prices recovered after hitting a low in 2016 when real interest rates fell back towards zero in 2016.
  • Generally, gold prices are inversely related to nominal interest rates and positively related to inflation and crude oil (See Figure 4, 5, 6 and 7 in PDF report attached). The FOMC’s current stance of letting inflation catch up to interest rates is thus constructive for gold prices. In fact, our data suggests two regimes of gold prices as a result of significantly lower interest rates in the last ten years, such that similar levels of inflation are related to different gold prices before and after January 2009.

Gold may appreciate to US$1,420/oz

Quantile regression model.

  • To estimate a theoretical fair value for gold, we referred to a Spring 2013 Journal of Investing article in which Lingjie Ma and George Patterson studied the relationship between monthly gold price and seven factors, e.g. WTI Crude oil price, interest rates, etc, using quantile regression.
  • Unlike ordinary least squares regression, quantile regression does not require certain assumptions such as constant mean and conditional mean zero residuals and thus can be flexibly applied in our study. Quantile regression estimates different coefficients for different gold price levels, which also helps to compensate for the regime shift before and after 2009. Conversely, linear regression may be more appropriate for the study of return data that display mean reverting behaviour.
  • Back then, Ma and Patterson concluded that gold wasn’t overpriced at that time and projected gold prices of US$1,123 or US$1,582/oz under different economic assumptions. It is worth noting that gold prices did indeed drop to around US$1,050 by 2015/2016 as the US economy expanded rapidly and real interest rates rose.

Key changes or variations from reference study.

  • In our study, we replaced some of the variables due to lack of data and to ensure consistency in data frequency. Economic data was also lagged by one month to reflect data availability. The largest change is that we added US M2 money supply to reflect monetary base expansion over time. Whereas Ma and Patterson used data from 1968 to 2008, we only had complete data from July 1983 to December 2018.

Model has acceptable out-of-sample performance.

  • Based on the model coefficients for the 80th percentile of gold prices (tau=0.8), the in-sample predictions of July 1983 to December 2011 have a root mean squared error of US$140.27 while the out-of-sample predictions of January 2012 to December 2018 have a root mean squared error of US$135.33. The in-sample period error would have been lower if we applied a lower tau, to reflect the lower gold price regime.
  • Out of 84 months of out-of-sample data, actual gold prices were within the 90th confidence interval of our prediction for 62 months or 74% of the time. There was a stretch of 15 months when gold prices were consistently higher than our confidence interval. That was over late 2011 to early 2013 when gold prices were elevated. Thereafter, gold prices corrected and stayed within our estimated intervals about 90% of the time or 62 out of 69 months.

Gold may be worth US$1,376/oz – US$1,478/oz.

  • Our model indicates a fair value of US$1,420.13/oz with a 90% confidence interval of US$1,375.96 to US$1,478.17/oz for gold. Our model suggests that gold prices are most sensitive to the level of money supply – a 5% variance changes model gold price by 5.5% – followed by the dollar index and unemployment whereby a 5% variance changes model output by 1.5%. The implication is that moves by the FOMC to maintain the size of its balance sheet will have a positive impact on gold price.
  • Conversely, similar variances in crude oil and the inflation rate, excluding food and energy, affected model output by 0.5% and 0.4% respectively. These percentages are obtained by varying one model input by -5% and +5% while keeping all other variables constant at one time.

Model issues and weaknesses.

  • We noted that the coefficients in our model suggest that gold price is positively correlated with the 3-month US treasury bill yield instead. In fact, the coefficients for CSPX and GB3 may range from negative to positive values based on their 90% confidence intervals.
  • We reckon that this variance could be due to collinearity between some independent variables such as GB3, M2 and DXY. Hence, our model cannot be used to describe the relationship between gold price and individual variables. As we have shown, gold price is negatively related to long term interest rates (Figure 5 in attached PDF report). However, the relationship is less clear when we consider the joint impact of other factors such as the DXY and M2. However, we reiterate that our model appears to be robust based on out-of-sample test results (Figure 9) and that the 90% confidence intervals, which factor in the variance in the coefficients, still indicate 7% to 15% upside from gold price of US$1,286.74/oz as at 4 March 2019.
  • Finally, our model estimates a fair value for gold based on the current values of variables such as UR, ISM, CPIX. Should these values change, the estimated fair value of gold will also change. Without forming predictions about the input variables, our model cannot be said to be predictive of gold price.

Scarcity of future supply has spurred industry consolidation

Recent M&A reflects few new world class discoveries.

  • The need to replace aging mines with high quality discoveries can be seen from recent mergers and acquisitions. In January 2019, Barrick completed the acquisition of Randgold Resources Limited for US$6.5 billion. According to Barrick, the acquisition of Randgold serves to create a portfolio comprising of 5 out of the world’s top ten Tier 1 gold assets and have total reserves of 78m ounces of gold. Barrick defines a Tier 1 asset as a mine with stated mine life of more than 10 years with 2017 production of at least 500,000 ounces of gold and 2017 cash cost per ounce within the bottom half of Wood Mackenzie’s cost curve data (see report). 
  • In turn, Newmont announced in January that it will acquire Goldcorp Inc. for US$10 billion to raise its combined gold reserves to more than 120m ounces. It is worth noting that while Newmont and Goldcorp have a combined production of 7.8m ounces annually, Newmont is targeting sustainable production of only 6 to 7m ounces annually, possibly reducing supply to lengthen mine life (see report).
  • Newmont’s move in turn attracted an US$18 billion bid from Barrick at the end of February to acquire Newmont, before it could complete the acquisition of Goldcorp. According to Barrick, it is not interested in a merger with a combined Newmont/Goldcorp, claiming that Goldcorp only has one Tier 1 gold asset that is also majority owned and operated by Barrick. To date, the Board of Newmont has asserted that the Newmont/Goldcorp combination represents super value creation over a merger with Barrick and both companies have formed a joint venture to combine their operations and assets in Nevada.
  • These transactions imply a scarcity of top tier gold mines and that it is more feasible to acquire rather than to explore for new top tier gold mines. At the same time, miners are relying on synergies from the acquisition of adjacent or closely located mines for growth, rather than starting new exploration projects. As top tier gold mines also become more expensive to acquire, interest may eventually spill towards smaller miners owning Tier 2 or Tier 3 mines.

Top miners account for less of global production.

  • Based on global mine production data from the World Gold Council and reported production data of ten major gold miners, their share of global production has fallen from 31.1% in 2012 to 26.4% in 2018. Production from these ten miners fell by 3% in 2018 while global mine production grew by 0.8%.
  • For 2019, production is expected to decline further as Freeport McMoran’s mine will be transiting from open pit to underground mining which will affect output. Based on the guidance provided, combined production growth is expected to range between -6.49% to 0.13% (or -1.6% to 5.6%, excluding Freeport McMoran). Faster growth at companies such as Newcrest and Polyus will offset slower growth at larger miners i.e. Barrick, Newmont and AngloGold.
  • Newcrest’s production in 2018 was affected by a seismic event that led to a temporary suspension of operations at its Cadia mine in Australia. On the other hand, growth at Polyus will be driven by brownfield projects.

Gold miners to offer higher upside

Gold miners offer higher beta exposure to gold prices.

  • Comparing the daily percentage return of gold, VanEck Vectors Gold Miners ETF (GDX) and VanEck Vectors Junior Gold Miners ETF (GDXJ) from 4 January 2010 to 28 February 2019, we found the two ETFs to have respective beta of 1.77x and 2.16x to percentage changes in gold price. In other words, a 1% increase in gold price generally leads to a 2.16% increase in the unit price of the GDXJ as opposed to a 1.77% increase for the GDX.
  • The share prices of gold miners have retreated by more than 60%, and 80% in the case of junior miners, from their respective highs. Conversely, gold price has retreated by about 30% from its high in 2011. Hence, there is significant potential for gold miners to be rerated higher, especially if gold prices were to break out of their current range and trade above, e.g. US$1,350/oz.
  • Within the SGX, there are 3 listed gold mining companies – CNMC Goldmine Holdings Limited (SGX:5TP), Wilton Resources Corporation Limited (SGX:5F7) and Anchor Resources Limited (SGX:43E). Of these 3 companies, we favour CNMC Goldmine, rating it Overweight with a fair value of S$0.310. See report: CNMC Goldmine Holdings Limited - Tayrona Financial Research 2019-03-12: Getting Back On Growth Track.
  • We also profile Taung Gold International Limited – a Hong Kong Stock Exchange listed company – that was referred to us via word-of-mouth.

Key risks

Change in monetary policy stance.

  • Our thesis on gold is conditioned almost entirely on expectations about FOMC monetary policy. The FOMC is in turn data dependent for its decision making. In the US, strong employment and growing wage growth have been green shoots and may be signs of higher inflation in the future.
  • If inflation accelerates and the FOMC decides to “stay ahead of the curve” and raise rates again, gold price may be adversely affected, especially if interest rates rise faster than inflation.

What if crude oil price drops?

  • The other adverse scenario could be if crude oil prices and inflation drops while interest rates remain unchanged. As gold price is positively related to crude oil price and inflation, an increase in real interest rates will have a negative impact on gold price.
  • Recent soft crude oil prices have dragged on the US CPI.

Liu Jinshu Tayrona Financial Research | 2019-03-12