No Attention Span Summary: The current spread between the yield of the Alerian MLP Index (AMJ) and the 20 Year U.S. Treasury of 388 basis points (3.88%) compared to a 16 year average of 258 basis points (2.58%) argues that the MLPs in the AMJ index are not overvalued. When/if that spread compresses to less than 100 basis points (1.00%), then we need to worry. The current AMJ yield of 6.29% is towards the lower end of the distribution of AMJ Yields over the same time period, indicating that AMJ Index MLPs are relatively fully valued.
MLP Index Yield versus the Treasury Yield as a Valuation Indicator
The purpose of this AMJ Yield Spread to Treasury ("AMJ Yield STT") analysis is to look at the relative yields of the Alerian MLP index and the 20 Year U.S. Treasury to determine if the spread can indicate when MLPs are relatively overvalued. According to the Alerian website, "The Alerian MLP Index is a composite of the 50 most prominent energy Master Limited Partnerships that provides investors with an unbiased, comprehensive benchmark for this emerging asset class." The top 10 constituents of the index are:
- Enterprise Products Partners, LP (EPD)
- Kinder Morgan Energy Partners, LP (KMP)
- Plains All American Pipeline, LP (PAA)
- Energy Transfer Equity, LP (ETE)
- Magellan Midstream Partners, LP (MMP)
- Energy Transfer Partners, LP (ETP)
- Linn Energy, LLC (LINE)
- Oneok Partners, LP (OKS)
- Kinder Morgan Management, LLC (KMR)
- Enbridge Energy Partners, LP (EEP)
The top 10 currently represent approximately 60% of the total index, so the index is heavily weighted towards the largest MLPs. This analysis uses weekly data starting from December 29, 1995 (the earliest data for the AMJ Index). The 20 Year U.S. Treasury yield is used because it has the highest correlation with the AMJ Yield over the time frame studied. The 20 year U.S. Treasury has a correlation coefficient of 43.9% and a coefficient of determination (R-squared) of 0.192 (many people default to look at the 10 year U.S. Treasury, and that comes in at a slightly lower correlation coefficient of 40.0% and an R-squared of 0.16).
(click images to enlarge)
Yield Spread Analysis
Since it takes two items to create a spread analysis, you should keep in mind that the spread will be affected by changes in both sides of the calculation, and will be more volatile than one single measurement. For example, U.S. treasuries have been on an unbelievable run over the time period analyzed, and eventually, that trend will turn and begin to head in the other direction.
Now, just because they are somewhat correlated does not mean that there is direct causation between their levels: the AMJ Index reflects what all market participants think about the future prospects of its underlying MLPs, and the U.S. bond market reflects what all market participants think about interest rates, inflation, budget deficits, future value of the U.S. dollar, all other macro factors, etc. etc.
So how could the trend of generally increasing interest rates impact the MLP universe? Clearly, since MLPs tend to use a healthy amount of debt financing, rising interest rates will translate into increasing overall borrowing costs, thus creating an increasing drag on distributable cash flow. On the other hand though, as more invested capital comes out of the U.S. Treasury market, where will those investment dollars go? Some may be redeployed into corporate bonds, and some may redeployed into the equity markets with some potentially finding their way into MLP equities.
If investment dollars are leaving the bond market in the aggregate, bond prices will go down and yields will go up, thus reducing the AMJ Yield STT (making bonds look relatively more attractive). If some of those investment dollars find their way into MLPs, then generally speaking, MLP unit prices will increase and the AMJ Yield will go down, thus also reducing the AMJ Yield STT. So typically, as the AMJ Yield STT decreases, one would expect the AMJ Index itself to generally be moving in a positive direction. Here's how those relationships have looked since year end 1995:
The history shown in the chart generally tends to support that thesis -- as the AMJ Yield STT line trends downwards, it appears that the AMJ Index line trends upwards.
The distribution of the AMJ Yield over that time frame based on 869 data points is shown below:
Some observations from the data:
- The current AMJ Yield is 6.29%, which is in the orange colored column.
- The average AMJ Yield is 7.69%, so 50% of the time, the yield is higher and 50% of the time, the yield is lower.
- 94.8% of the time, the AMJ Yield is above 6%, I view 6% as a good reasonable valuation/overvaluation dividing line. The further below 6% the yield is for the AMJ Index, the more I will be concerned that MLPs are generally overvalued as a group.
- The maximum yield was 15.25% and occurred on November 21, 2008 (a really good time to have bought MLPs hand over fist).
- The minimum yield was 5.37%, and occurred on July 13, 2007 (a good time to have sold MLPs hand over fist).
The distribution of the AMJ Yield STT over that time frame based on 869 data points is shown below:
Some observations from the data:
- The current AMJ Yield STT is 388 basis points, which is in the orange colored column.
- The average AMJ Yield STT is 258 basis points, so 50% of the time, the spread is higher, and 50% of the time, the spread is lower.
- 84.2% of the time, the AMJ Yield is 151 basis points or more above the 20 Year U.S. Treasury Yield, which I view as indicating MLPs are reasonably valued, with the comparative valuation becoming more favorable for MLPs as that spread increases.
- 13.6% of the time, the AMJ Yield is 101 to 150 basis points above the 20 Year U.S. Treasury Yield, which is an area where their relative valuation may be viewed as "stretched," but is not necessarily indicative of an impending reduction in the AMJ Index.
- 2.2% of the time, the AMJ Yield is less than 100 basis points above the 20 Year U.S. Treasury Yield, which I view as indicating that MLPs are overvalued, and I would be looking to outright sell down various MLP positions.
- The maximum spread was 1,132 basis points, and occurred on November 7, 2008 (a really good time to have bought MLPs hand over fist).
- The minimum spread was 10 basis points, and occurred on July 6 and July 13, 2007 (a good time to have sold MLPs hand over fist).
Quick Background on the 2007 Peak in MLP Valuations
From 2003 to 2007, the MLP PIPE (Private Investment in Public Equity) market grew like a weed on Miracle Grow and peaked in 2007, with more than $9 billion being raised via the PIPE market that year -- including one deal for Linn Energy in August of 2007 for $1.5 billion. Over the same time frame, the AMJ Index increased about 109%, so the early deals were all very successful. As a result, more and more institutional funds and hedge funds were formed to participate in the MLP market. These deals gradually concentrated ownership in fewer hands while they also helped to drive up valuations, which one could argue made the space relatively overcrowded and overvalued.
You can see the outlier in the second chart above, where the AMJ Yield STT breaks below 100 basis points (1.00%). The Linn Energy deal actually closed after the valuation peak had already occurred. The AMJ Index peaked at 342.5 in July of 2007 (with an all time low yield of 5.37%), and then bottomed at 145.8 in November of 2008, a painful 57% decline. This basically wiping out all of the gains from the beginning of 2003 (but not if you reinvested all of your distributions over that time).
So Where are We Now?
The current AMJ Yield is 6.29% -- only 92 basis points above the all time low -- while the spread to the 20 Year Treasury is 388 basis points -- 378 basis points above the all time low. The AMJ Yield leads me to believe that the MLPs in the AMJ Index are relatively fully valued, but the spread to the 20 Year Treasury makes them appear potentially undervalued. Of course, the spread to Treasury is definitely just one indicator and only a partial influence on MLP valuations. But if MLPs were to move towards the average spread of 258 basis points, then the AMJ Index would move up around 26% to 494. If they were to move to a 100 basis point spread, the AMJ Index would rocket up 85% to above 720 (I would definitely be selling some positions at those levels!).
However, I do not think that is going to happen. It seems more likely to me that there will be a move from the other side of the spread, i.e. rising 20 Year Treasury yields will compress the spread, but that won't necessarily mean that the AMJ Index will fall because the large current spread appears to provide a decent cushion compared to the historical data. Additionally, even as interest rates rise, MLPs have the benefit of growing their distributions, which acts to keep their comparative yield looking more appealing than Treasuries, even taking into account the additional risks. So MLPs as represented by the AMJ Index appear relatively fully valued at the present time but not overvalued, when/if the AMJ Yield Spread to the 20 Year U.S. Treasury compresses to less than 100 basis points, then we need to worry.