Very high correlation between gold and US Treasuries

I am a total newbie and only been investing for a few years but I noticed something I have never seen before: there is extremely high correlation between gold and bonds (US Treasuries). I haven't looked at distant history to see how common this is, but it certainly has been rare in the last few years.

What is most strange about this correlation is that both signal different inflationary outcomes. Long-term bonds will not rise if inflation is thought to be a threat. Conversely, many investors buy gold to hedge against inflation (yes, there are some who buy gold for deflation/wars/disease outbreaks/etc but they are in the minority).

Precisely when a lot of brainpower, not to mention ink and bytes, have been spent on the debate over inflation vs deflation, I find it interesting that the market is bidding up both, long-term US government bonds, as well as gold.

Here is a quick look at the relationship between gold, bonds, and stocks. I'm using S&P 500 as a proxy for stocks; the 20-year constant maturity TLT ETF as a proxy for US Treasuries; and the GLD gold ETF as a proxy for gold. Source data is from Yahoo! Finance and S&P 500 dividends may not be accounted for properly (this doesn't impact my conclusion in this scenario).

In statistics, correlation is used to measure the relationship between variables. The R value is supposed to indicate the relationship, either positive or negative, while squaring the R to produce R-squared results in a measure of the influence of one variable over another. Do note that even if correlation is high, it doesn't mean that one variable influences the other; it is possible that a 3rd factor drives those two variables (this is quite common in investing where factors such as profits/economic growth/tax policy/wars/etc drive all assets).

Correlation Analysis - R Value

The R value indicates the direction of the relationship so I always like taking a quick look at that. A negative value means that one variable may negatively influence the other; while a positive R value means both move together in the same direction.

R Value: S&P 500 vs Bonds

The following chart plots the R value between S&P 500 and 20 year US Treasuries:

For some reason, bonds and the S&P 500 moved similarly in 2005 and 2006—do note that this was the period when almost everything went up, which is a hundread-year-type scenario—but they seem to be moving in opposite directions since 2007. Anyone who has been following the markets is probably aware how bonds have remained in a bull market throughout the financial crisis while stocks actually crashed. As the chart confirms, bonds and stocks have largely moved in opposite directions since the stock market crash started in 2007.

From my limited reading, my impression is that stocks and bonds sometimes move together but there is no strong relationship because the bond cycles are much longer, usually lasting 20 or 30 years. For instance, bonds were in a bear market (interest rate rose) in the 50's and 60's, while stocks were in a bull market. In the 70's, both were in severe bear markets. Yet, since the early 1980's, stocks and bonds have been in a major bull markets. The stock bull market ended in 2000, whereas bonds have seen their bull market continue, with bond yields hitting 50+ year lows recently.


It is possible we are witnessing a new long-term trend being established between stocks and bonds, whereby they move in opposite directions (a la 1950's and 1960's). However, I lean towards deflation and don't think bonds have entered a clear bear market yet.


R Value: Gold vs S&P 500

Here is a chart of R values between gold and S&P 500 since 2005:

I have to check but I believe, in the long run, gold and stocks have a low, near-zero or slightly negative, correlation.

Surprisingly, gold and S&P 500 had a positive correlation over the last 5 years. The early part was likely due to "excess liquidity" when almost everything went up. However, the last few years was a surprise to many, when gold collapsed along with stocks, and then rallied hard, during the stock crash that started in 2007. Many passive investors, such as pension funds, were overloaded on gold (as well as commodities) because the correlation was low to negative yet it didn't save them during the recent crash.

The most interesting thing is that the correlation appears to be negative in 2010, so far. If this trend holds, it will move the relationship towards historical norms and it'll be something different from the 2005 to 2009 period. Investors who owned both gold and stocks from 2005 to 2009 didn't pay a price for doing so, although it didn't save them during the crash either. Going forward, though, one of them is going to be a drain on their portfolio.

R Value: Gold vs Bonds

Let's look at the most interesting relationship, gold vs bonds:

This chart only covers a short term but, historically, gold and bonds tend to have a very strong inverse relationship. Gold and bonds are mortal enemies and this should make sense. After all, inflation is lethal to bonds while gold actually prospers; and vice versa.

I don't know why there were strong positive movements in both, gold and bonds, in 2007 and 2010 (so far). I also wonder if there is some comment element driving the change in R value. It should also be noted that the R value is very high (I'll comment on this later).

Charts and price movements don't necessarily repeat but I wonder if 2010 will end up like 2007. In both periods, gold basically was on a large, uninterrupted, bull run. Here is log chart of the gold ETF:

Correlation Analysis - R Squared

If you square the R values, you get R-squared, which is supposed to measure the relationship between variables. If the value is 100%, it means, in theory, that the two variables move together perfectly. However, note that by squaring the R value, you lose the direction of movement. A value of 100% could mean that the variables move perfectly together in the same direction, or perfectly opposite. The directtion is a very important point for investment asset analysis because the direction impacts decisions on the effectiveness of hedging and macro bets. Furthermore, the direction may also signal changes in macro trends, such as bonds moving from a bull market to bear.

Since the economic environment is extremely complex and there are many drivers of asset prices, it is rare to see high R-squared values. I would personally consider anything over 20% to be a very strong correlation.

The following chart plots the R-squared values for the 3 assets discussed above:

The most noticeable trend in the data above is how the correlations became strong across the assets during the stock market crash and economic meltdown that started in 2007. Most of the correlations started approaching 50% and as most investors would be aware, nearly all assets went down together, and then all went up.

The most interesting observation to me is how the correlation between gold and bonds is so high. From an R-squared that use to range around 20%, it has jumped up to roughly 70% in 2010. This is an amazingly tight correlation and I haven't seen anything like this (but do note that I have only been investing for a few years and haven't studied the past that closely).

It is also worth nothing that the bond and gold correlation is really high in 2010 and 2007, which are precisely the years when the relationship was negative (refer to prior chart above).

From a macroeconomic point of view, it is quite bizarre that gold and bonds have rallied so much this year. A bull market in bonds generally implies deflationary or disinflationy tendencies, yet the multi-year peaks being hit by gold has historically implied an inflationary expectation. There is also the possibility that both are being bid up as safe havens but I don't believe that is necessarily the case right now (this was more of the case in 2008). One of these views is going to be wrong big time! I have a feeling this relationship won't hold in 2011 or beyond. We'll see.