In my last post discussing the Alerian Effect, we talked about what happens to an MLP that is added to the Alerian MLP Index (AMZ) or Alerian MLP Infrastructure Index (AMZI) in a rebalancing. We learned that typically, on the day it is announced that an MLP will be added to an index, it outperforms the index it will be joining. Looking at those same MLPs through the day after the rebalancing takes place yields a bit fuzzier results. On a per MLP basis, the majority of MLPs in both the AMZ and AMZI still outperform their respective indices. Companies added to the AMZI outperform the index during the period by an average of 368 bps. But, those added to the AMZ perplexingly underperform the index by an average of about 100 bps.
In an effort to look at the Alerian Effect from all angles, today we’ll be studying the names that were removed from an index, focusing on the same two windows of time and study period (1Q 2011 through 3Q 2016) as the December 13th post. Unlike the previous post, names removed due to a special rebalancing have been excluded from this analysis, since their removal has been triggered by a corporate action—namely, they have merged with or been acquired by another MLP. Their price movement during the rebalancing procedure will typically be based on the proposed merger exchange ratio.
Since you’ve likely slept many nights since my last post, let me share this calendar again to remind you of our quarterly rebalancing schedule. This is a snapshot of how the quarterly rebalancing events shook out this past September:
Quarterly rebalancings to the Alerian Index Series occur every March, June, September, and December. Results are announced on the second Friday, actual rebalancings take place on the third Friday after market close, and changes are effective at market open the following Monday. Special rebalancings are reviewed on a case-by-case basis; thus, we are unable to offer an easy-to-follow calendar because they are triggered by corporate actions.
On the day of announcement, an MLP that is removed from the AMZ or AMZI usually underperforms its respective index. Constituents removed from the AMZI typically underperform the index by a greater margin (199 bps) than those removed from the AMZ (154 bps). One reason for this, similar to the Alerian “entering constituent” Effect, is that there are significantly more assets tied to the AMZI (about $12.3 billion) than the AMZ (about $3.6 billion), The chart below shows relative constituent performance on the day of announcement versus the indices they were removed from.
Next, let’s take a look at what happens to exiting MLPs through the day after rebalancing. If we simply look at the data in aggregate, MLPs removed from the AMZI continue to underperform the index throughout this time period by an average of 105 bps. Constituents removed from the AMZ outperform the index by an average of 100 bps. This data is a bit misleading, though. If you look at the chart during the March 2016 rebalancing, you’ll see that some names in the AMZ significantly outperformed the index. Teekay Offshore Partners (TOO) and Archrock Partners (APLP) outperformed by a staggering 51.6% and 48.0% respectively. Enable Midstream Partners (ENBL) also outperformed by a significant 24.2%. Our theory as to why this extreme outperformance occurred is that this was during the time we started to see MLPs recover. The AMZ was up 4.6% over this time period and the effect was magnified for the aforementioned names because of their smaller size. However, this explanation is by no means full proof given the notable underperformance of Vanguard Natural Resources (VNR) of -36.1% over the period. If you were to remove these outliers, MLPs exiting the AMZ would underperform through the day after rebalancing by around 162 bps. As a function of the methodology criteria, in most scenarios, the MLPs removed from the AMZ are either smaller and/or less liquid than the MLPs removed from the AMZI. Thus, it may take fund managers a bit longer to get out of certain positions in the AMZ versus AMZI. When the outliers are excluded, the results are a bit more in line with our pre-study guesstimates.
On a per MLP basis, an MLP removed from the indices will underperform on the announcement date and throughout the study period the majority of the time.
The chart below gives us a quick side-by-side comparison of the data for MLPs entering an index versus those exiting.
It’s safe to say that in general, names that are added to the AMZ or AMZI outperform the index and names that are removed underperform the index during the rebalancing time period. However, there is no conclusive explanation for why names, either in the AMZ or AMZI, perform the way they do over the full study period (announcement date to day after rebalancing). On one hand, there are significantly more assets tied to the AMZI (about $12.3 billion) than the AMZ (about $3.6 billion), which could warrant a magnified effect for AMZI names due to the sheer dollar amount of units needed to be purchased or sold. However, on the other hand, if a name is being added or removed from each index, it’s likely to be a name “on the border” from a size or liquidity standpoint. This could warrant a magnified effect for AMZ names due to liquidity to enter or exit a position. The reality is that there could be so many contributing factors that it’s impossible to isolate a week’s long performance down to a single variable—in this case, an Alerian rebalancing.
We will continue to collect and monitor this data. As we amass more data points, it will be interesting to see whether more prominent trends can be identified. Stay tuned for an update around this time next year.