More Correlations: Oil and the Dollar (and Gold)

Yesterday I posted an article on the short-term statistical correlation between Oil and US Equities. In that article I referenced the connection between Oil and the US Dollar, and one reader asked that I post the rolling correlations for those assets as well. Here they are, with Gold thrown in for good measure.

Also, as a clarification, I did not mean to imply that I think input prices don't impact our economy over time; merely that this is difficult to detect using one basic statistical measure on a day-to-day basis.

ETF proxies include the United States Oil Fund (USO), the PowerShares DB US Dollar Index (UUP), and the SPDR Gold Shares (GLD), as follows starting with USO vs. UUP (click charts to enlarge):


Similar to US Equities, while the daily correlation between Oil and the Dollar has been relatively weak, the Ratio of Inverse Closing Changes has certainly been on the rise. Below is GLD vs. UUP:


As shown, the inverse correlation between Gold and the Dollar has been significantly stronger and more persistent as compared to Oil. While the rotational commodities crush looks more overdone by the day, gold bugs who believe the Dollar will continue to gain strength might bear this in mind.

 

Comments

[User 244491]

What you write is true in normal market conditions,not when Dow Jones is heading for 4000-5000 next year.
There are many heroes writing on SA,better show us your monthly statement so we can judge if you put your money where your mouth is or you are just another piker taken out on a margin call when all commodities crashed.Show the statement hero.

[Whidbey]

The correlations are somewhat better of lagged and fitted. It takes a while for traders and retail buyers to believe a move and then react.

[Respirate]

Thanks again, Jeff.

Your work seems to show that while Oil vs. USD are more strongly correlated than Oil vs. Equities, the relationship is relatively weak over time. This clearly refutes the news headline writers and business show talking heads who blindly claim a strong correlation, or even a direct causal relationship.

Headlines commonly implying that a rising dollar causes a corresponding drop in the price of crude are convenient and certainly sound authoritative, but are misleading. As you've shown, the market relationships aren't that simple, especially over time.

[Pretzel Logic]

Thanks for the excellent charts. I just had to chuckle about this sentence, though: "gold bugs who believe the Dollar will continue to gain strength might bear this in mind."

That's almost an oxymoron. Most of the goldbugs I know think the dollar is suitable only for toilet paper, and will NEVER get their heads around a fiat currency -- even in the face of overwhelming technical evidence that the trend is reversing, at least for the short/intermediate term. :)

[CLH]

Where is the correlation?---The The stock market goes up first and then the dollar. When the dollar goes up the commodities (including oil) tank. Everyone knows when the dollar rises gold heads for the dumpster.

This seem simple to me and has been going on since the begining of capitlism.

By the way Im a gold bug sometimes ---I entered gold 2 weeks after gold started its bull in 2000. I sold gold in March of 2008. Many of you will not live to see another gold bull. The last gold bear lasted 20 years. Many followed it down from 800 to 250. Will you?

[JBP]

CLH,

I bought tech in the mid 90's and sold in 99, then bought a bunch of condos in Miami and sold them all in 2004, then I bought gold and sold at $1020 in March just like you. I then bought DZZ (the stock you mention just about every time you post) and sold Friday and made a killing. Sounds almost as ridiculous as some of your statements on this board....

[hrant]

inflation is going up ,not just here but in europe and China. I do not think it is just about oil . question : what are u guys doing to protect yourself ? Are these TIPS for real ?

[User 53641]

I don't know the exact correlation procedure you used, but if you correlated the actual prices, your results are probably spurious, that is, they don't mean much. This is a common error in such an analysis. I apologize for using a statistical term but this can happen if the two time series are "non-stationary&q... that is, they do not have a statistical mean as would be the case, for example, of the diameter of ball bearings being produced by a machine. Virtually all pricing time series( stock prices, interest rates, oil, wheat, etc.) are non-stationary so such a correlation analysis must be based on the changes or, statistically speaking, the first differences.The proper procedure would be to cross correlate these changes over time. Futhermore, using this procedure, you might answer an interesting question--is there a lead/lag relationship between changes in the value of the dollar and changes in the price of oil. It is widely thought that a weak dollar leads to higher oil prices and watching the market, one would think that this is the case. This makes mathematical sense because oil is quoted in dollars. However, one might also make the case that the cost of all the foreign oil we buy (over $600Billion/day at current prices) weakens the dollar in the long run by flooding the global market with more and more dollars. So, the question is--which one is the dog and which one is the tail?

Finally, when performing such an analysis, the common perception is that more data is better. Statistically, this is true but when dealing with economic time series, the more distant data such as back to the early 1990s may reflect a completely different underlying era and thus the statistical relationships may be distorted. Fewer observations using more recent data may be more relevant and meaningful than a large number of observations going back in history.

Appreciate your effort!