What is it about gold prices? Many people seem to believe they are impossible to predict, or even understand. At her Senate confirmation hearing in November, Janet Yellen said, “I don’t think anybody has a very good model of what makes gold prices go up or down.” Ben Bernanke also said last year that “nobody really understands gold prices, and I don’t pretend to understand them either.” While many factors influence the price of gold, PIMCO believes there is one that can explain the majority of changes in gold prices over the past several years: changes in real yields.
To understand how, it helps to start with a simple example. Pretend there was an asset that had no risk of default and a real – that is, inflation-adjusted – value that varied over time but did so around some constant level. In other words, this asset has no credit risk and in the long run maintains its purchasing power. How much would investors pay for it? Whatever the amount is, it would likely vary over time with the level of real yields available in very high quality, nearly “default-free” assets (such as U.S. Treasuries). That is, when real yields on other such assets are high, investors would likely want a bigger discount to the long-run estimated real value of the hypothetical asset. Conversely, when real yields are low, the opportunity cost of owning the asset drops and investors would likely be willing to pay a higher price relative to the asset’s long-run estimated real value.
In essence, this guides how PIMCO thinks of gold. And the market seems to view gold this way as well; over the past several years, gold prices have been heavily influenced by the level of 10-year U.S. real yields (see Figure 1).
Recent history has changed the behavior of gold prices
With the launch of gold ETFs in 2004, gold has increasingly become a liquid financial asset. But over much of history, the price of gold was either fixed or gold was a relatively illiquid physical asset held by a small minority of investors. Today the marginal price of gold is largely set by financial demand, as over $70 billion of gold is held by ETFs, and investors choose to buy or sell gold ETFs by comparing the expected real return on gold to that of other liquid financial assets. This means that the future behavior of gold is more likely to resemble the past several years rather than the 1970s or some other period.
To quantify the relationship between real gold prices and real yields, we can regress the price of gold from 2006 to 2013 (we used the logarithm of the real price of gold in our model) against the 10-year real yield from the Treasury Inflation-Protected Securities (TIPS) market. (In our view, this regression is appropriate since gold and real yields are co-integrated and there is an economic rationale for believing they should be.) Based on our study, the regression shows that, all else equal, a 100-basis-point (bp) increase in 10-year real yields has historically led to a decline of 26.8% in the inflation-adjusted price of gold. In other words, over the past seven years gold has had a real duration of 26.8 years. (Note that this is solely an empirical duration that describes the way that gold has traded. Since gold has no cash flows, its duration does not need to be constant, and there is nothing magic about the 26.8 number. Just as the correlation between stocks and bonds varies over time depending on changes in macroeconomic variables and investor risk appetite, the real duration of gold may also change in the future.)
Using this framework, consider the 15% price drop in gold in mid-April following talk of Fed tapering. This move predated the sharp move higher in yields in the fixed income market by two weeks. Over the month of May, 10-year real yields rose 57 bps. Even though the markets moved at different times, the size of their moves over this period was remarkably consistent with the historically observed 27-year real duration. In hindsight, we believe the move in gold gave an excellent early warning of both the direction and magnitude of the move in rates.
Another potential use for the empirical duration of gold is to see how gold prices have changed over time after controlling for moves in the level of real yields. By computing a real-yield-adjusted gold price we can look at a gold price that adjusts for the fact that the opportunity cost of owning gold varies over time. The real-yield-adjusted gold price is calculated by adjusting the gold price by a discount factor based on a 26.8 gold duration and the level of real yields (see Figure 2). If real yields explained all the moves in gold prices, we would expect this real-yield-adjusted gold price to be completely static and never move. This would mean that all moves in the inflation adjusted price of gold were fully explained by a change in the discount factor that links today’s gold price with the real-yield-adjusted gold price. While the real-yield-adjusted gold price does move around in Figure 2, it does so over a smaller range than the inflation-adjusted price of gold does. This means that although real yields don’t explain all the moves in the gold price, they do seem to explain a significant portion of them. In particular, since 2006 the real-yield-adjusted price of gold has fluctuated within a much smaller range than does the price of gold. In other words, most of the changes in gold prices can be explained by viewing gold as a real asset with 27 years of real duration.
Notice the large run-up in gold prices in 2005. The gold ETF “GLD” was launched at the end of 2004, so in 2005, gold had a new source of very large investor demand, and this created a structural break in the real-yield-adjusted gold price. While real yields explain much of the movements in gold prices, large structural changes in the market can have large impacts on the valuation.
The perception of gold as a “safe haven” asset also has some influence on gold prices. During the credit crisis and the bankruptcy of Lehman Brothers, many market participants expected gold to do very well. Yet gold prices actually declined during the second half of 2008 as the credit crisis intensified. Why? During the credit crisis we saw a spike in the level of real yields, which puts downward pressure on gold prices. But the real-yield adjusted gold price actually rose sharply following the Lehman Bankruptcy. This shows us that while there was a flight-to-quality bid that increased the real-yield-adjusted gold price, the impact of higher real yields was larger.
Looking at the value of gold today, we see that, adjusted for real yields, the price of gold is very similar to the pre-crisis value of gold – despite the fact that nominal gold prices have risen over 50%. This suggests that any premium in the gold price following the credit crisis relating to investor risk aversion has now been removed, which seems sensible given the recovery in the equity markets. This real yield-adjusted-gold price, which adjusts for real yields and an investor’s opportunity cost of holding gold, is a useful barometer for assessing the valuation of gold across different regimes and relative to other assets.
Yes, other factors can and do affect the valuations of gold, such as the launch of new products to access gold (such as gold ETFs), prior gold returns, investor risk appetite, central bank purchases and government policies such as India’s efforts to limit gold imports. Global real yields and the value of the U.S. dollar versus other global currencies are others. However, one could argue that real yield differentials between countries also influence relative currency levels, so there could be an offsetting effect between these variables.
Nevertheless, we believe that real yields are the single most important factor. As gold increasingly becomes a financial asset, when real yields rise, gold prices should fall if they are to maintain a given level of financial demand relative to investors’ other opportunities. Similarly, when real yields fall, we expect the price of gold to rise. Investors should be aware of the relationship between gold and real yields because it has important implications for how they think about the role of gold in their portfolio in an asset-allocation and risk-factor framework. Additionally, controlling for the level of real yields allows for a purer picture of what we believe the underlying value of gold is, and it can help investors better determine their allocations to gold within their portfolios.
To be sure, it is challenging to predict the future path of real yields. Looking ahead, we expect the Federal Reserve to move very gradually in reducing accommodative policy and for 10-year U.S. real yields over the next several months to be relatively steady around current levels, which would be neutral for nominal gold prices.