Gold Prices, Real Yields and Paradigm Shifts

Just over a year ago, we took a snapshot of the relationship between gold prices and real interest rates as indicated by the yield on inflation-adjusted 10-year constant maturity U.S. treasuries.

That original snapshot was taken on 17 March 2022, one day after the U.S. Federal Reserve started what became series of rate hikes that lifted the effective Federal Funds Rate from 0.08% to 4.83% to combat inflationary forces unleashed by the Biden administration a year earlier.

Because gold is used by investors as a hedge against inflation, rising in value when inflation-adjusted interest rates fall or turn negative, the rising rate environment the Fed has created over the past year should have reduced the price of gold by a substantial amount. When we took our snapshot on 17 March 2022, the gold spot price was $1,944.05 per ounce and the real yield of the 10-year Treasury was -0.72%.

One year later, on 17 March 2023, the spot price of gold had risen to $1,988.11 per ounce, while the inflation-adjusted 10-year yield had swung to +1.29%. That increase occurred despite the Fed’s rate hikes pushed real interest rates to swing from negative to positive.

In between these dates, the price of gold fell for a time, but only while real interest rates were increasing and not by anywhere near as much as the 15-year long relationship between the two would have predicted. But that expected trend stopped after 3 November 2022, when the real 10-year Treasury yield peaked at +1.74% and the price of gold hovered around the $1,629 level, near its lows during the Fed’s rate hike cycle.

After that date, the real yield of the 10-year Treasury fell back while the price of gold escalated. Our new snapshot of the relationship between spot gold prices and inflation-adjusted 10-year Treasuries illustrates these changes.

If the 15-year long relationship from 2 January 2007 through 16 March 2022 between the price of gold and real 10-year Treasury yields still held (shown by the dashed black curve in the chart), the price of gold would have fallen $850 per ounce more than it has. If the price of gold dropped to where it was trending between 17 March 2022 and 3 November 2022, which would be considered recent history, it would be about $350-$400 lower per ounce than it is for a similar real 10-year yield.

But it has risen higher, which raises new questions about what’s driving the price of gold. The staff at Goldmoney, who have a more sophisticated model than we do, think there has been a paradigm shift. In the following excerpt from Part I of their analysis, we’ve added the links to the referenced exhibits, but not the boldface font, which appears in Goldmoney’s original article:

Over the past few months, the gold price has once again detached from the model’s predicted price. And it has done so in a remarkable way. First, the delta between the observed gold price and the model-predicted price has reached an all-time high. Current gold prices are more than $400/ozt over model predicted prices (See Exhibit 2). The previous all-time high was $200/ozt and it only lasted for a short period of time.

Second, this is happening in the most unlikely of all environments. The Fed has been aggressively hiking rates for the past 12 months to fight the highest inflation in over 40 years. The Fed raised the Fed Funds rate from 0% to 4.5% in just 12 months. It is very rare that we see such large rate hikes from cycle bottoms. In fact, this has only happened five times since 1975 that the Fed raised rates more than 4% from the bottom (see Exhibit 3)….

Yet despite all this, gold prices have not just held their ground; they have actually risen! Arguably, it could be that the gold market once again has simply got ahead of itself. Or we really do see a paradigm shift this time.

Before we continue exploring this thought, we must add one caveat here. In our models, we use publicly available data for net central bank sales/purchases. The official data from the IMF is notoriously lagging and incomplete, and we are certain that the reported net purchase numbers are much too low. The World Gold Council (WGC), for example, reports net additions of 1136 tonnes in 2022, more than double the 450 tonnes bought by central banks in 2021. It’s no secret that central banks have been on a buying spree in the second half of last year. But exactly how much gold they added remains a bit of a mystery. That said, even assuming that true central bank gold purchases exceeded the WGC estimates by a massive 50% would bring the model-predicted price only about $70/ozt closer to the observed price. We believe this is partially a shortcoming of our model, as it is based on historical data, and we have not seen a lot of volatility in CB gold purchases in the past. However, we have had years with large central bank purchases before, and we had years with higher overall gold demand from all sectors, and yet this didn’t lead to large distortions in our model. Hence, we don’t think central bank purchases can explain the current huge discrepancy between predicted and observed prices.

Therefore, in our view, the only reason for gold prices to detach from the underlying variables in our model by such a large amount and for such a long time is that the gold market finally starts pricing in that there is a risk central banks, particularly the Fed, are losing control over inflation, which is remarkable given the prevailing narrative that the Fed is willing and able to do whatever it takes to bring inflation under control.

Here’s the bottom line from Part II of their analysis:

We believe that the most likely explanation for the recent rally in gold prices against the underlying drivers of our model is that the market is increasingly pricing in that the Fed, once it is forced to stop hiking, will lose control over inflation. Faced with the choices of years of high unemployment and a crumbling economy or persistent high inflation, the gold market thinks the Fed will opt for the latter. This would mark a true paradigm shift, and from that point on, gold prices may start to price in prolonged high inflation (and our model may not be able to capture this properly).

If true, a lot of models will be as broken as our simple model already is! Then again, in the immortal words of George Box, all models are wrong, some are useful. The trick is to know when they work, because relying on a failed model after its expiration date can lead to catastrophic consequences.

Previously on Political Calculations

• Real Yields and the Price of Gold
• Why Hasn’t the Price of Gold Fallen More with the Fed’s Rate Hikes?

Image credit: Photo by Anne Nygård on Unsplash.