Accurately pinning down the intrinsic price of oil (NYSEARCA:USO)(NYMEX:"CL") is a difficult exercise. Theoretically it can be done -- it has utility, is non-renewable, and has a finite supply, and therefore valuation should be as straightforward as coming up with an intrinsic price range for a stock. But in the crude oil market, it's less cut-and-dried given that most oil traders are speculating on the impact of shorter term events, leading to an idiosyncratic amount of volatility. Econometric models may be more attractive to traders/investors who think on more of a longer-term time scale.
Even though oil price fluctuations of 2%-5% daily are common, the value of crude itself probably isn't changing that significantly, leading to a degree of short-term inefficiency. Accordingly, if the market were to get largely out of whack -- like in 1980, which saw an inflation-adjusted price of ~$110 per barrel, or the massive up- and downswings around the time of the financial crisis -- long-term traders could capitalize on what could be large non-persisting deviations from the commodity's true value.
One way to create such a model is through structural regression analyses. For the purpose of this article, I'll exclusively be using the price of West Texas Intermediate ("WTI"). However, it really doesn't matter which oil price is used as they all largely move in conjunction due to arbitrage.
In terms of the variables that move the market, the most obvious are production and consumption, the elements that make for supply and demand. Beyond that, one can make an argument that mostly all other variables feed into those two to a degree. For example, GDP growth has a clear relationship with oil consumption. Oil is an input into an economy and higher consumption is normally indicative of higher amounts of business activity.
Nonetheless, we can also argue the opposite -- higher oil prices can dampen consumption. Take for example the 2005 to early-2008 period, where economic growth was high, but high oil prices had the effect of reducing its demand and keeping oil consumption low.
For this study, I collected data from 1965 through the current point in 2016 from the following: consumption and OPEC and non-OPEC production (measured in millions of barrels per day), 10-year Treasury yields, US inflation, WTI price per barrel (averaged over the given year), US real GDP growth, OPEC inventory builds measured year-over-year, OPEC and non-OPEC reserves, and certain ratios of those variables. Data was obtained from the EIA website and Bloomberg.
Although I included inflation in the dataset, it really should not be used as an independent variable, as higher oil prices can directly cause inflation rather than the other way around. Oil is a major input in the economy and if input costs rise, then the cost of the end products will as well.
Ten-year Treasury yields could have some level of effect on oil, as many assets are priced off the so-called risk-free rate. This, however, becomes somewhat of a chicken-and-egg problem, as financial market capital inflows and outflows work partially in accordance to how assets are priced in relation to other markets.
Consumption and production, being the fundamental elements of supply and demand, should come in as statistically significant. Being their magnitude has been contingent on population growth over time, introducing a ratio of the two into the regression could also be useful to capture the relative fluctuation in supply and demand on the price of oil.
I tested out a couple dozen different regression models using standard ordinary least squares methodology (also simply known as OLS). Of all models, price was best modeled by an independent variable mix of consumption, production, the ratio of consumption-to-production, year-over-year real GDP growth, and year-over-year OPEC inventory builds. Inventory builds had the highest degree of statistical significance, while the others were significant at a 1%-2% level (see image below). The multiple R-squared and adjusted R-squared came to 0.874 and 0.817, respectively, denoting that the price of oil was anywhere from 81.7%-87.4% explained by the variables thrown into that particular model.
The results can be used to build a linear equation to determine where the price of oil might be relative to current data or any hypothetical scenario. As of Q2 2016, OPEC inventory builds in the US were up approximately 200 million barrels over Q2 2015.
Historically, oil consumption has outpaced production by approximately 1% (source: EIA data). Over the past 2-3 years, supply has surpassed demand, leading to an elongated compression of prices from inflation-adjusted historical norms. Assuming real GDP growth of 1.4%, if OPEC inventory builds remain elevated to the tune of around 200 million barrels over their point last year and consumption and production came in around 96.3 million barrels per day (i.e., supply/demand equilibrium), oil would be priced around $65 per barrel.
The EIA has estimated that global petroleum and other liquid fuels inventory builds will continue to increase into 2017 before inventory draws begin perhaps sometime around the middle of the year. Global inventory builds are projected to average 0.7 million barrels per day (b/d) for 2016 and 0.3 million b/d for 2017. US crude production stood at 9.4 million b/d in 2015, has fallen to 8.7 million b/d in 2016, and is expected fall again slightly to 8.6 million b/d in 2017.
Inventory builds of 0.3 million b/d for 2017 means we could get up to 250 million barrels for the relevant variable used within the regression (OPEC inventory builds), as OPEC-member countries produce about 40% of the world's crude oil. According to the model, at 250 million barrels of year-over-year inventory builds, that would estimate the price of oil at about $57 per barrel, leaving everything else unchanged in the scenario above. Three hundred million barrels of year-over-year inventory build would estimate around $49 per barrel.
This simple model is built based on findings of statistical significance on the variables of consumption, production, a consumption-to-production ratio, year-over-year real GDP growth, and inventory builds. Checking current data against this model suggests that oil is probably pretty fairly valued based on current information.
The model would predict that inventory builds above the current reading in the market, as estimated by the EIA, would place the price per barrel somewhere in the high-$40's to low-$50's depending on the extent of the supply glut. If consumption can keep pace with production (or production brought down to the level of consumption or below) and inventory builds are kept around the level of Q2 2016 or below, oil has the potential to hit a more historically normalized price of $65 or above.
Note that any single model should not be used exclusively to derive oil price forecasts. Any model can fail to include one or more relevant variables and additional variables may eventually work into the picture to cause any given model to fail. A combination of models is best used to estimate prices and cross-check for the sake of greater accuracy.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.