A little more than four years ago, theprice of gold bullion traded at an all-time high price of more than $1,900 per troy ounce. With America’s credit rating having recently been downgraded, the
U.S Federal Reserve promising to keep interest rates at zero for nearly two years and Europe apparently on the verge of fracturing, it felt like gold’s inexorable rise would continue forever. Since then, however, the shine has come off of this most precious of metals, leaving some investors to wonder when it will regain its glitter.
It sometimes seems as if there are as many views on gold as there are stars in the sky. Depending on whom you speak with, you can hear gold described as merely another commodity to be dug out of the ground, or revered with an almost religious fervor as the cornerstone of any prudent portfolio, and every shade in between. Indeed, in discussions with investors over the years, we have heard all of the following used as justifications for holding gold:
- An inflation hedge
- A deflation hedge
- A U.S. dollar hedge
- A hedge against weakness in other fiat currencies
- A stable asset in times of financial market volatility
- A safe harbor against fragilities in the financial system
- Something to buy when central banks print money willy-nilly
Before discussing the outlook for gold, it is important to define the rules of engagement by identifying those factors that seem to have some instructive power in explaining the evolution of the gold price. In doing so, it is important to check one’s biases at the door; just because one does not believe in the hyperinflation thesis, for example, does not mean others do not. And if their behavior has a quantifiable impact upon the gold price, it needs to be considered.
We identified a list of seven factors that are at least occasionally credited with driving the gold price and defined them via market prices or statistics where we have time series dating back to 1990 (see “Driving gold,” below).
We acknowledge at the outset that this list is far from exhaustive, and misses important factors like Indian jewelry demand, retail Chinese buying and actual gold mined. However, the historical data required to analyze these factors to a statistically significant degree is simply not available, and we can keep these drivers in mind when it comes time to consider the future of gold.
One of the oldest and most important maxims of statistics is that “correlation does not imply causality.” In other words, just because there appears to be a statistical relationship between two variables does not mean that one influences the other. To mitigate this, we ran two separate regression studies: One defining the level of gold in terms of the level of the factors listed, and another explaining changes in the level of gold in terms of changes in the independent variables. We can be reasonably confident that any variable exhibiting a strong relationship to gold in both of these studies really is an important driver.
The results of the first regression, on levels, are set out in “Regressing gold fundamentals” (below).
On the face of it, this looks highly encouraging. The r-squared is very high, suggesting that these variables explain 95% of the level of the price of gold. The p-value column, which represents the probability that a statistical relationship can be ascribed to chance, only gives two factors: the VIX and ISM Manufacturing, which are non-statistically significant relationships.
However, when we actually plot the output of the model versus the gold price, something odd happens (see chart in “Regressing gold fundamentals”).
As you can see, the model is considerably more volatile than the gold price, and on two occasions nearly pegs the price of bullion at zero. It seems hard to believe, therefore, that these factors really do explain 95% of the price of gold.
When we regress the changes in the gold price with changes in the variables, we get what looks to be a more realistic output. In this study, our model only explains 23% of the change in the price of gold, which seems more believable. Tellingly, only three of our factors: The VIX, ISM Manufacturing and Central Bank Gold Buyers, now show up as statistically significant (see “A better regression,” below).
Although we clearly have not found the holy grail of gold price determination, we can nevertheless distill a number of direct and inferred insights from our regression exercise:
- Gold is clearly sensitive to real bond yields, with a coefficient in the expected direction. The higher that real bond yields go, the lower the gold price should go and vice versa. As an aside, this relationship looks to be particularly strong in the post-crisis, quantitative easing era.
- To some extent, gold is just a commodity like any other. That is the message of the relationship with the S&P/GSCI Industrial Metals Index, which exhibits a generally strong relationship during long periods.
- While there is a strong statistical relationship between gold and CPI, during the time period covered by our study it appears that gold is more sensitive to deflation than inflation,
- as the relationship bears a negative coefficient. Part of this, of course, might well be mapping the central bank reaction function to deflation/disinflation scares, which is also caught via the real yield factor.
- This is in direct contrast to the experience of the 1970s, when gold rose sharply in tandem with realized (and expected) inflation. This in turn suggests that the drivers of gold are not stable through time, but depend on a prevailing narrative.
- Somewhat contrary to expectation, there is little discernible relationship between the gold price and financial market volatility.
- While there may be a tenuous relationship with the U.S. dollar, it appears that this relationship is captured more effectively via other industrial metals.
- Although central bank gold purchases probably impact the price of gold, particularly on a short-term basis, the relationship does not appear to be sufficiently strong to be a central plank in the gold investment thesis.
Armed with this knowledge, what can we now say about the price of gold moving forward?
Our model suggests that we need to address the outlook for commodity prices, inflation and real interest rates. Although commodities have fallen precipitously and reached oversold levels on a technical basis, at this juncture it is difficult to see a sustained bounce back to levels of two or more years ago any time soon. Trend growth in China continues to fall and the economy is retooling in favor of a consumption rather than investment-led growth model.
The upshot is that Chinese demand for industrial commodities is likely to remain subdued, and so is the price.
To be sure, given the monetary aspect of gold, one can argue that bullion should trade at a relative premium to base commodities. We would agree with that, but also observe that that monetary premium already appears to be in the price. Indeed, if we plot the relative change in gold and base metals since the end of 2001 (the date of China’s accession to the WTO) we find that industrial metals commanded a premium during China’s most rapid investment phase. Since the financial crisis, only gold has traded at a strong premium.
A look at the gold/silver ratio, which is trading near 20-year highs, makes it difficult to escape the same conclusion. As such, the monetary allure of gold is already in the price and that marginal changes in the price are more likely to be driven by the commodity outlook than this premium. On this basis, we find that the outlook for gold is relatively poor (see “Gold bugged out,” below).
On the inflation front, the outlook may be a little better, at least according to our regressions. With the world confronting an excess of supply and a deficit of demand in not only commodities, but also manufactured goods, it is difficult to see a secular rise in global inflation in the medium term. To be sure, certain emerging market countries facing currency crises, such as Brazil, may experience high levels of inflation, but on a longer-term basis, the inflation outlook remains dim. As a result, we can probably expect to see another year or two of quantitative easing from the European Central Bank and Bank of Japan, and probably more easing measures from China. The prospect for money-printing is an argument in favor of somewhat higher gold.
However, the flip side of low inflation is that real interest rates tend to be higher, which is a negative for gold. This is particularly the case in the United States, where not only is the Fed no longer buying Treasuries, but other central banks are actively selling as their foreign exchange reserve holdings go down. All else being equal, this should lead to higher real rates—a negative for gold. Incidentally, a reduction in forex reserve holdings from the world’s emerging market central banks also would suggest less demand for gold from that cohort. Although they seem unlikely to sell their gold holdings, it also looks unrealistic for them to add to any substantial degree. As such, central bank demand is unlikely to act as support for gold during the next couple of years.
What about factors not captured in our model, such as retail demand from India and China? Indian demand is certainly a constant in the gold market, but it’s just that, a constant, steady source of demand with little prospect for explosive long-term growth. Additionally, the various measures being proposed by the government—a new lending program in which gold is pledged in exchange for Indian currency—is a negative for the price should it ever materialize.
Chinese demand for gold does appear more volatile. However, it seems reasonable to expect that that the evolution of China’s economy to a more demand-based model could see more buying in the future—particularly if expectations emerge for a sharp devaluation of the Chinese currency. Collectively, this current environment offsets each other and the future of gold in each country is grey, not shiny.
Ultimately, however, the price of gold is cyclical, just like that of commodities, currencies, inflation, etc. After the break-up of the Bretton Woods system in the 1970s, gold embarked on an explosive bull market that lasted until the Fed got serious about extinguishing inflation. Thereafter, gold drifted sideways-to-lower for more than two decades before globalization, the emergence of the so-called BRICs and a commodity cycle took hold just after the turn of the millennium.
That commodity cycle appears to have peaked a year or two ago, which may have dampened the prospects for explosive upside in gold. Indeed, it’s worth noting that in real, inflation-adjusted terms, gold peaked at roughly the same price as the 1970s bull market (see “Gold apples to apples,” below). Its subsequent descent has been nearly as swift as it was then, and that’s with the benefit of easy global monetary policies and emerging market demand.
However, gold is more currency than commodity and it was not the commodity bull market that pushed gold above $1,900 in 2011; it was the belief that easy monetary policy and quantitative easing—what many analysts, not just gold bugs, saw as de facto printing of money—would lead to a huge spike in inflation, perhaps even hyperinflation. That inflation never materialized. While some see the huge Fed balance sheet as a sort of inflation sword of Damocles hanging over the market, investors will want to see actual proof of inflation before chasing gold back to those levels.
That being noted, we find it difficult to escape the conclusion, therefore, that for this cycle at least, gold’s time to shine has passed. That’s not to say that there isn’t a place for gold as a small core holding in any investment portfolio. But the expectation of the type of explosive gains seen from 2002-2011 may have to wait for the next commodity super cycle, which is likely many years away.