From the start of 2014 until now the market has seen a strong US dollar bolstered by a continually recovering US economy and downward pressure on oil prices as the result of a supply glut. The general rule of thumb has always been that currencies have an inverse relationship with commodities that are priced in that same currency. This is because strength in the currency means that commodities priced in that currency become more expensive, acting as a headwind for demand in those goods. To the naked eye, it appears that the inverse relationship has held from the start of 2014 until now, but in order to drill down and quantify this relationship I wanted to calculate that correlation over various frequencies to see the strength of the relationship between the US dollar and WTI Oil, and between the US dollar and a basket of commodities.
The dollar (USD) has strengthened significantly since January 2014 on the back of a more stable US economy and the Federal Reserve attempting to return to more normal monetary policy. U-3 unemployment held above 6.5% for the first few months of 2014 before gradually grinding lower to a current rate of 4.5% (Figure 1). Core CPE, the Fed’s preferred inflation metric, rose in the middle of 2014 before hitting a low for the period of just over 1.3% in July 2015. The measure then began to make upward progress, returning closer to the Fed’s 2% target in recent months (Figure 2). With the Fed’s incredibly accommodative policy in place in the wake of the recession, the economy was able to return to stable levels. With the dual mandate under control, the Fed was able to give the US economy a vote of confidence by gradually raising the Federal Funds rate three times in December 2015, December 2016, and March 2017. As expected, the USD increased over this period as investors gained confidence in the US recovery and took advantage of higher rates of return in the world’s largest economy. The U.S. Dollar Index (DXY), which tracks the performance of the USD relative to movements against the currencies of its trading partners, shows the strength of the dollar over this period. In particular, the index saw a 25% gain from mid-2014 to early 2015 as the Fed made its intentions clear of returning rates to a historically normal level (Figure 3). Additionally, the index saw a rise from $95 to over $100 following the Trump election in November, as the market expects that Trump’s pro-business policies and attempts to repatriate corporate incomes into the US will help to boost inflation and keep unemployment near its natural rate, causing the Fed to normalize at a faster rate than the market was previously anticipating.
This same period saw the crash of crude oil and the inability of WTI prices to rebound to their former highs (Figure 4). The story of the oil market seems to be more supply and demand driven than many other asset classes. Demand for oil has remained fairly constant over this period, growing at a mild rate, while innovations in drilling technology, particularly fracking in US shale, has helped to flood the market with a glut of oil. Initially, non-US oil producers ramped up supply over this period in a price war with smaller US shale producers in the hope that they could survive a lower-for-longer oil market. Non-US producers believed their lower cost of production would allow them to take market share from smaller producers. This oversupply has forced some producers out of the market, pushing prices up from their low in early 2016, but producers have been able to slash costs to operate at a lower rate, and bring rigs back online as oil hovers around its breakeven price, keeping a ceiling on current prices for the time being.
I used the correlation function in Excel for weekly and monthly data to quantify the relationship between the USD and WTI Oil, and the USD and the US Commodities Index (USCI). I originally intended to include daily data as well, but the assets traded on different days so it was difficult to match up changes across different schedules. Initially I expected a strong negative correlation of around -0.7, but I was impressed by the actual relationship being much stronger. As shown below, the weekly and monthly correlations for the USD and WTI were -0.9148 and -0.9227, respectively. The weekly and monthly correlations for the USD and USCI were -0.9528 and -0.9585, respectively.
I expected that the change in frequency to a broader time period, from weekly to monthly, would yield a higher correlation because it would help to eliminate some of the short term “noise” that these investments would face. Additionally, it’s not surprising that the correlations are higher for the USCI as the composite index helps to eliminate the noise associated with daily trading of oil, and gives a better picture of the effect of the USD on all commodities priced in the dollar. Since the R2 in a bivariate regression is simply the square of the correlation between the two variables, we also have simple econometric models from these correlations. The R2 values are shown below.
These high R2 values indicate that a significant portion of the variations in both WTI oil and commodities priced in the dollar are explained by the variation in the dollar, which isn’t surprising given the already-known rule of thumb concerning their relationship. There’s certainly omitted variable bias in this simple bivariate regression, particularly announcements regarding current supply levels, attempts at cutting prices, and changes in the rig count. It’s still a supply and demand equation, and these R2 values are likely picking up other variables that account for the changes in price. But the high correlation still tells us that this model has a pulse, and that there is certainly a relationship between these assets.
These correlations are essentially a confirmation of something that we already knew: demand for a good drops when that good’s currency strengthens because the good becomes relatively more expensive. That drop in demand leads to a drop in price to offset the slowing quantity demanded. What this research did tell us is that this relationship is much stronger than most would probably think, with correlations above -0.91 for both WTI oil and a basket of commodities priced in the USD. This relationship in particular holds over longer periods of time, indicating that temporary mispricing by the market could be acted on as a sort of arbitrage opportunity. Given the current abundance of automated algorithmic trading which is created to take advantage of high correlation relationships such as this, this is probably not something the casual investor would be able to profit off of. But it does help to explain longer term trends and can provide support for arguments regarding the future movements of these asset classes.
Looking forward, the market is still anticipating an accelerated pace of rate hikes compared to what it was expecting prior to the election. Some of that optimism has worn off as the market isn’t as hopeful as it was in November that President Trump will be able to pass pro-business legislation in a timely manner to stimulate inflation. There still seems to be more downside risk to the USD as the market is expecting some sort of accommodative fiscal policy, and a failure to deliver by the administration could see the USD dropping and provide a tailwind to commodities. On the other side, the market seems to be fairly optimistic that OPEC will be successful in its compliance and will see its members cutting production across the board, reducing the current oversupply and providing a floor for prices moving forward. Again, the market appears to pricing in some optimism and there appears to be more downside risk than upside to oil prices. Given what we know from the correlations above, something has to give for the longer term price movements between the DXY and commodities, and I personally believe that the current expected pace of rate hikes is too optimistic given the Fed’s tendency to lean dovish in the face of any negative data concerning the US economy. Since there’s more downside risk than upside risk to the dollar, I believe we’ll see a tapered pace of rate hikes compared to the current expectations which should provide a cushion to move commodity prices moderately higher through the end of 2017 and into mid-2018, holding external shocks to these individual commodities constant.