The dates correspond roughly to the length of the supply glut so far, about a year and a half of data. The scales were adjusted so that they most show 20% increases for each major grid line making the comparisons roughly similar. While the decline in supply is barely noticeable, the drop in WTI price is dramatic and has not let up its pace. These declines have been lead by numerous events like changes in rig utilization and reactions to a possible change in demand. Because of this, the process for determining price should be called into question. Every basic economics course discusses the role of prices in representing accurate supply and demand forces. As price increases, either the object has become more scarce and more desirable, or fewer people may desire it and inventories grow. Typically, the increases and decreases in the equilibrium when looking at the supply and demand model are relatively proportionate which has not been the case in the market for the past year. Hence, enormous amounts of uncertainty introduced a general downtrend in the market as analysts' predictions were always in such a wide range and often capricious. That left investors to do the job of finding the tops and bottoms of oil price when they had really never had that job before. Perhaps no one had realized the importance of OPEC when it came to stable pricing, but now, as the free market mechanism has become erratic and almost nauseating, the contributions from the cartel are almost missed. With power now in their hands, the price has adjusted to include smaller producers from a wide range of countries in the supply channel. My hypothesis is that without the experience and certainty of the investors in deciding how crude oil should be priced, correlations between supply (production) and demand (spot price) will be minuscule. These next hypothesis tests will be conducted in order to test my argument that correlations that should be there due to proportionate changes in supply and demand have either deteriorated or never existed in the first place. The first set of data will analyze weekly WTI spot price and weekly U.S. production over the past year and a half in order to make inferences about domestic connections. All data is from the Energy Information Administration on their official website.
The first thing to notice is the skewed distribution of the spot price data. Points beyond the dotted lines represent data that doesn't reflect the normal distribution. For spot price data, it's very evident that a right tail skew exists with a majority of the data between 60 and 40. This should be expected with the supply glut in effect, but if changes were proportionate like they should be, points should follow the red normal curve. It's important to keep this mind when looking at these tests. Next, we'll look at a scatterplot.
Here we have a comparison between production and spot price with the least squares line showing a negative correlation. Once again, one can observe the concentration of points in the range of $40-$60. Overall a correlation appears evident, but I wish to test the significance of the strength which could be significantly small. The line shows a lot of overpredicted and underpredicted points which represent times where the price should have been higher or lower. Data associated with higher production levels are especially messy and disoriented just like current crude oil trends. This shows the chaotic microcosm of an oversupplied market. Points on the more expensive spectrum, associated with OPEC control, are closer together with relatively smaller residuals. After a hypothesis test, we get a very small p-value that significantly rejects the possibility that there is not a connection. The data generates a linear regression line of:
So we are able to find a significant correlation between the two groups with a moderate strength of R=0.668. The strength is not perfect but relatively reflects what might be going on in the world of supply and demand. Looking at the slope, we observe that every $1.00 price decrease should correlate to a 12,400 barrel increase in production. As highlighted in my earlier article, that proportion seems crazy, but I fear comparisons cannot be made until we experience more from a free oil market. Looking at prices before the supply glut seems illegitimate because of OPEC control. Another argument supporting the untrustworthy nature of this model is that only 44.6% of the variability in production is explained by price. That means the model can only account for half the correlational effects which doesn't seem very strong. Nevertheless, there appears to be a correlation. For a second, though, I'd like to zoom in on the data that defines the smaller price ranges of $40-$60 which were shown to be chaotic. Notice the very weird correlation that emerges.
Wait, I thought we just established a positive correlation between these two variables? When narrowing down on 2015 data, the supply and demand curve inverts. While traditional economic theory can often be incorrect, a total inversion of the supply and demand curve seems irrational. Perhaps a more sensible explanation would be the large amount of bearish events that have caused price declines without consideration of output changes. This just furthers my argument for emotionally charged trading in the oil market.
What have we learned here today? The past year and a half, data I chose to represent the time period where OPEC control had deteriorated, shows a moderate correlation between WTI spot price and U.S. production. Residuals were evident, and data points in the lowest price range $40-$60 were the most chaotic. After zooming into the smaller range which spanned the weeks of 2015, we saw an odd inversion of the supply and demand curve which was clearly an irrational deviation of conventional economic calculation. There seems to be no doubt that the pricing mechanism is not operating like it should. While global production is certainly a confounding variable in these models, the effects are drastic. What does this information mean for the future? Let's extrapolate the $1.00 decrease with every 12,400 barrel a day which should be reversible if the correlation is relevant. As mentioned before, the IEA predicts an 800,000 barrel a day decrease of the next year. If U.S. production was to experience just half of the declines, then we could expect an increase of $32.26 in the price over the same time. With this change, WTI levels could be around $70 when not taking global oil output into account. The model seems fairly reasonable here, but still, the small amount of variability actually explained by the model calls into question its reliability. For final thoughts on predictions and validity, the law of averages may explain a stronger correlation despite higher variability. Because deviations from the average are relatively consistent, the average can still stand with both extremes pulling on it equalling. For predictions, it may not be crazy to forecast larger gains because emotionally charged buyers may irrationally skyrocket the price. The extrapolation mentioned above will never be fulfilled with perfect proportionate increases until the final value is reached; instead, erratic trading may affect exaggerated movements that eventually converge to ideal values. These are the types of outcomes I hope to see in a crude oil market that stabilizes and adjusts to its free-market status.