The Alerian MLP Infrastructure Index (AMZI) is down 27.1% on a total-return basis since peaking in August 2014. Since most MLPs generate fee-based cash flows for handling crude oil and natural gas, the first place to go for an explanation of poor performance is up the energy value chain to their producer customers. The rig count measures the level of drilling activity, so it is considered to be a gauge for the industry’s health (or lack thereof). Baker Hughes publishes this data on a weekly basis.
From 2009 to 2014, the rig count remained relatively stable, and yet the AMZI continued to climb, returning 129.3%. During this time, wellhead spacing got tighter (more wells per acre), laterals got longer (horizontal drilling extended farther) and frac stages got smaller (more fracs per well), resulting in significant increases in production per rig. So while the rig count was fairly constant, production rose significantly during the five-year period.
Given this, it makes sense that over the last 10 years, the AMZI had a lower correlation to the rig count (0.22) and a higher correlation to crude production (0.91).
That high correlation number is quite striking, and I was tempted to label this the “ah-ha” moment and make it the whole point of this piece—that MLPs, traditionally having a price multiplied by volume business model, are heavily dependent on crude production. But when we examine correlations on a rolling 30-month basis, rather than over the entire 10-year period, a different picture emerges.
Over the past 10 years, correlation to the rig count has fluctuated wildly. The correlation approached 1.0 in 2011, only to approach -1.0 a few years later. This volatility makes the rig count an exceptionally poor explainer or predictor of MLP performance.
Likewise, the 0.91 correlation to crude production is actually not indicative of a steady, high correlation, but rather of a generally positive correlation that trended upward through a period of meaningful technological advancement, but then dropped off steadily in the last six months. Crude production could have explained MLP performance before the start of this year, but now, despite a still significant correlation of 0.68, it is not a dependable, single-factor model.
Many investors view midstream MLPs as alternative or defensive investments and appreciate that they are not correlated to any one factor. One element of MLPs has remained stable: distribution growth. In 2009, in the depths and wake of the financial crisis with capital markets closed, midstream MLPs raised their distributions 3%. In 2010, they raised distributions 4.2%; in 2011, 5.6%; 2012, 8.1%; 2013, 7.3%; and in 2014, 6.8%.
For the trailing 12-month period, midstream MLPs raised their distributions 7.5%.
Correlations spike close to 1.0 for almost all asset classes when the markets panic. However, two hallmarks of the energy infrastructure investment thesis—stable baseline cash flows and distribution growth—remain reliable.