We are discussing MLP structures and why IDR payments
make LP units in MLPs a horrible investment for investors. In this Op-Ed post, we will walk you through
IDR payments, Distributable Cash Flow and the intricate ways in which GP’s will
take advantage of LPs. We then opine on various end-game scenarios for the
average MLP. We will conclude with an
empirical exercise that proves our position is correct beyond the shadow of a
doubt.
Review of Incentive Distribution Rights
Previously,
we have discussed the structure of Incentive Distribution Payments or
IDRs. In review, IDRs are payment
structures in Master Limited Partnerships (MLPs) that create an incentive for
the General Partner (GP) to increase distribution payments to the Limited
Partnership Units. The structure of IDRs
is actually fairly simple. Take the
example of Kinder Morgan (KMI) – the first and largest MLP, which employed the
GP/LP structure until just recently.
Typically in an MLP the GP owns a 2% stake in the LP
and receives compensation for managing the operating partnership’s assets via
the IDR. As can be seen in KMI’s IDR
table, at first there is an equitable split of distribution proceeds – 2% to
the GP, 98% to the LP. That is, until
the per unit distribution exceeds $0.33063, at which point the GP receives 15%
of each marginal distribution dollar.
This continues until the distribution exceeds $0.43125 per unit, at
which point each marginal distribution dollar is split 50/50%.
In
theory, IDRs were a means to align incentives between GPs and LPs and to create
an underlying incentive to grow the assets – and therefore the distributions –
of the partnership. With regard to the
IDR structure itself, one would think that incentives are properly
aligned. As long as the partnership is
not structured in such a way that achieving $0.43125 per unit in distributions
is exceedingly easy, it seems fair to compensate the GP for properly and
prudently growing the partnership’s assets.
However, IDR payments alone do not tell the whole story.
Review of Distributable Cash Flow
One would expect there to be rigid accounting in LP
structures that convey value based on distribution or dividend payments. For instance, in the MLP structure, LP
investors would want to be very certain that the proper amount of money is
distributed to the unit holders or else the partnership may find itself out of
resources. Since, there is a very clear
incentive for the GP to make large distributions, the LP will want to account
for what it considers “distributable” over a certain period of time.
KMI as the first MLP came up with a novel metric that
would dictate the level of distribution in a period – Distributable Cash Flow
(DCF). KMI defines DCF this way:
“Kinder Morgan’s Distributable Cash Flow (DCF)
has historically been an accurate proxy for sustainable cash flow. The main difference between net income and
DCF is net income is reduced by GAAP depreciation and DCF is reduced by
sustaining capital.”
After reading KMI’s definition of DCF, you might be
thinking “What is sustaining capital?” and “How is sustaining capital any
different than depreciation?”
First,
sustaining capital varies greatly by the specific asset characteristics (age of
pipe, pipeline diameter, amount of compression, size of compressor units,
utilization of the system, number of high-consequence areas on the system,
geography, historical maintenance of the pipe, and others). Second, technological advancements have
significantly improved and are expected to continue to lower maintenance
costs. Third, GAAP depreciation is based
on the book value of assets. There can
be a meaningful disconnect between book value and actual value of an asset
depending on several factors including if an asset has been owned for a long
period of time or if it was recently acquired.
So at its core, DCF is meant to mimic cash from
operations without some of the semantics regarding book value and useful life
that exist in GAAP accounting. We get
that, we’re cool with that. You take
income, add back D&A and then subtract sustaining capital. This sustaining capital figure is supposed to
recognize the amount of money that was capitalized per GAAP accounting but
really should have just been expensed.
Let’s say there was a section of pipeline that was old
and rusty and needed to be replaced.
Under GAAP, fixing that piece of the pipeline would be a capital project
that would be capitalized and depreciated over time. Under the sustaining capital regime, this
project would be expensed because it simply “sustains” the fixed assets of the
partnership. Now let’s say there is a
brand new section of pipeline that is built.
Under GAAP that new section of pipeline would be capitalized and
depreciation. Under the sustaining
capital regime, a new pipeline would be seen as a “growth” project and would
not be considered sustaining capital.
OK, now that we have sustaining capital straight, you
might be thinking, “So what?” To this
point, DCF sounds like a very reasonable metric off of which to base distributions. Alas, it is not. By using sustainable capital rather than
D&A to determine the distributions, the LP is effectively giving away
D&A dollars (that is, money that was previously spent 98% by the LPs) to
the GP on capital that he never contributed.
Under the sustaining capital regime, does the GP need
to generate profits to get paid? No. The
GP simply needs to generate DCF
You might be thinking, “Who gets to determine what is
sustaining capex and what is not?” The GP, of course. The GP can determine which capital projects
it would like to consider sustaining (I’ll be calling this maintenance capital
expenditure from here on out) and what is not (I’ll be calling this growth
capital expenditure from here on out).
Therefore, the GP has almost complete control over what DCF will be in a
given period of time.
We hope we are all seeing the many issues inherent in
Distributable Cash Flow.
Distributable Cash Flow + Incentive
Distribution Payments
So far we have established how IDRs work and their
purported role to align incentives between GP and LP. We have also examined how Distributable Cash
Flow is supposed to a “better” metric for the amount of cash that can be
distributed in a given period of time.
Mixing IDRs and DCF is a magical combination for GPs
to extract fund from LPs. Let’s say we
have a growing MLP, one that has not yet maxed out its IDR payments. Let’s also say that the GP has found an
investment project that will generate a positive IRR over its lifetime. LP
units are issued to fund 98% of the project and 2% of the project costs are
fronted by the GP. Since the pipeline is
new and shiny, it requires very little maintenance capex, which boosts every
unit holder’s DCF payment.
Rinse and repeat this process a couple of times.
Voila! The GP has maxed out its IDR
distribution and is now earning 50% of every marginal dollar of DCF… Lest we remind
you that the GP only put up 2% of the funds to fund the project, but we
digress.
We hope we all can see what the incentives of the GP
are given the IDR and what capacity he has for manipulating his share of the
distributions via the DCF. Moreover, the
GP is assuming far less risk in this structure than is widely
acknowledged. The GP is fronting 2% of
the project expenditure, but reaping in many cases 50% of the proceeds that do
not include the cost of the project that was undertaken! Let me say that again. The GP gets back cash flows that do not
account for the cost of a project. Said
another way, the half of the 98% fronted by the LP (49%) might as well be
directly transferred to the GP!!!
Example MLP – Sustaining Capital Regime
vs. GAAP Accounting Regime
In normal GAAP accounting, 100% of depreciation would
be recognized as an expense before any profit and/or distributable figures
would be calculated. In our example,
depreciation is $10, which represents the $100 cost of the project and its 10
assumed useful life. In reality (and
especially in an industry like this one), depreciation is a real cost. At some point this pipeline will need to be
replaced as it will literally rust away. We find the “proper” GAAP DCF amount
to be $10 per annum. Split 50/50, both
the LP and the GP receive $5 in distribution.
In the sustaining capital example, we find a different
means of accounting for DCF. Since
sustaining capital largely ignores growth capex, we will only deduct maintenance
capex to determine DCF. Naturally over
time the pipeline will age and will require a greater degree of maintenance vs.
growth capex. And so, we adjust the
percentage of maintenance higher over time relative to growth capex. What we see are far larger distributions to
both the GP and LP. In the GAAP
accounting example, distributions totaled $10.
In our first year under the sustaining capital regime, the total
distribution is $17.50, with $8.75 for each of the GP and LP.
Please note two things here: First, as the maintenance
capex increases on the pipeline over time, the distributions are reduced
relative to the early years. Also note
that the distribution in this example exceeds the operating income in our
example. Lest you doubt this example, we
would like to add that this is not uncommon in real-life MLPs. It is surprisingly common to see MLPs run
even GAAP losses while maintaining their distributions.
But we would like you to think very hard about what
these two things we called out mean for the structure of MLPs. If the MLP is distributing more than it
generates in cash and DCF will slowly decline over time, what incentive is
created for the GP? The incentive
created is to look for more projects that will have higher DCF in the early
years and boost the distribution to both GP and LP. The other incentive created for the GP is to
classify almost all capital expenditure as growth capex.
Over the 10 year life of the pipeline in our example,
distributions under the sustaining capital regime exceeds GAAP DCF by $52. How is this funded (certainly it’s not funded
organically)? The extra amount
distributed represents what would have been an expense to recognize the wear
and tear on the pipe over the course of the year. Who paid for the pipe in the first place and
should receive the benefits of that cash with first preference? Well, the GP
paid for 98% of the pipeline, so deserves 98% of that capital back. Alas $26 of the LP’s money ended up being
distributed to the GP.
We hope it is abundantly clear now the problems that
exist with the sustaining capital regime vs a far more logical (and respectful
to the LP’s capital) GAAP accounting regime.
The Economics of the MLP for Both GP’s
and LP’s
It should be clear by now that the GP receives
superior economics in the MLP structure, which in and of itself is not an
indictment of the business structure.
Many partnerships exhibit similar structures. For instance, hedge funds and real estate
investment trusts have GP’s that act as fiduciary for investment money and are
compensated through incentive structures.
However, the degree to which GP’s in hedge funds and REITs are often
capped or constrained by provisions like hurdle rates, high water marks and the
like. The degree to which the GP
benefits at the expense in the MLP is what is shocking… and there are no
provisions like high water marks to protect LP investors.
See the graphs above, at just a 9% project IRR with a
50%/50% split, the GP stands to achieve a 225% ROE. Even at a 85%/15% split, the GP still
achieves a 68% ROE vs 8% for the LP.
This egregious split creates what we believe are fatal
issues for the partnership. First, there
is an incentive for the GP to undertake low IRR projects that the LP’s never
otherwise would fund. Supposed the LP’s
wished to only invest in projects with 15% expected IRR. Unfortunately for them, there is no oversight
of the GP by LP board members or something similar. And so, the GP is free to undertake even
negative IRR projects. He even has an
incentive to do so as long as there is marginal cash flow.
Are the GP’s doing this, undertaking low and negative
IRR projects just to boost DCF? Sadly, we think so. The below table shows that IDR fees for some
of the major MLP’s routinely surpassed 30-50% of net income
(remember, this is a fee being paid to a 2% owner). In some cases the IDR fee surpasses 100% of net income. Now we have proof of projects that are financed by LPs, yet are money losers and still lead to GP profit.
(remember, this is a fee being paid to a 2% owner). In some cases the IDR fee surpasses 100% of net income. Now we have proof of projects that are financed by LPs, yet are money losers and still lead to GP profit.
This scheme can continue so long as LP unit holders don’t
realize the impact of the DCF split.
This scheme can continue as long as LP unit holders remain content, are
willing to purchase new units and there are more projects that are NPV
positive. Once one of these conditions
is no longer met, the MLP reaches a steady-state that is not as steady as you
might think. In the steady state, the cash flows will slowly decrease and the
MLP would work its way back down the IDR distribution waterfall; that is, in a
perfect world.
What’s the End Game?
First Potential End Game – LP shares
Tank:
We don’t want to throw around the P-word, but it kind
of looks like that. This whole scheme is
predicated upon more and more funding, debt or equity, especially when low and
negative IRR projects are undertaken. As
long as the LP units retain the same price or continue to go up in value, it is
beneficial for the GP, who will use the “overvalued” units to fund new projects
and acquisitions. This is beneficial
because overvalued units lead to less dilution, which will boost the DCF/unit
metric, which in turn ensures that the near-50%/50% distribution is
maintained.
Since many of the MLP’s distribute more than they make
via net income, eventually investors will catch on. The result will be either a decrease in the
price of the LP unit or a gradual decrease in the size of distributions,
perhaps both. This will likely not happen
in an orderly fashion in the markets.
Second Potential End Game – The
Courtroom:
The part of this arrangement that is most damning is
that the GP gets paid via the DCF D&A dollars that he didn’t commit. We spent a great deal of time talking about
“sustainable” cash flow vs. GAAP operating income. The GP benefits a great deal from this in the
early years of projects when there really isn’t that much capex work put
towards sustaining infrastructure. As
the infrastructure ages, distributable cash flow should fall because more money
will be put toward sustaining vs. growing.
At this point, the damage will already have been done, the distributions
will fall for both GP and LP and the GP will have already reaped massive
benefits. The LP investors will achieve
the equitable split of 98%/2% just in time for cash flows and DCF to dwindle
away. We envision an extremely litigious
end to most of the MLP structures.
Third Potential End Game – GP Takeunder:
The LP is used in these schemes much like a piggy bank
– simply to fund projects. LP’s
surprisingly are more than happy to participate despite their weak economic
participation in the partnership.
Eventually when the partnership distributions begin to decline due to
greater sustaining capex needs or a dearth of positive NPV projects, the LP and
GP units will decline in price. It is at
this point one would think that distributions would fall back to a more
reasonable IDR split for the long run.
Sadly, the GPs remain “the smartest guys in the room” (hat tip to Rich
Kinder’s past at Enron). We have seen
many GP’s simply buy out their LP partners when the economics of the
partnership begin to wane – lower oil prices, fewer available acquisitions, far
worse investment projects. For instance,
Kinder Morgan’s GP (KMI) famously bought out its LP partners (KMR, KMP, EPB)
when oil prices declined in 2015.
What was the outcome of this? Well, the former KMI GP now controls 14% of
the combined entity vs 2% under the MLP regime.
Via the MLP structure, KMI was able to siphon off billions of dollars
from its LP investors via IDR and DCF and then eventually used that capital to
acquire its LP investors’ units, thus bumping its ownership percentage in the
combined entity by 700%.
Quick
Aside: Think of the return differences
between the GP and the LP. The GP
returned 700% in a period in which the LP units declined 50+% from their peak.
What’s the end game?
Well it appears all three of the outlined potential outcomes are
suboptimal for the LP investors. There
is no outcome we envision – barring a larger MLP (i.e. a greater fool)
acquiring the LP units for a premium – that is advantageous for the LP.
The Counter Argument
The argument has been made in class many a time that
GP and LP represent simply separate and distinct claims on various cash flows
and therefore the market is able to risk-adjust properly. We went ahead and tested this claim with
market data.
We identified all the currently trading GP/LP MLP pairings
in the public markets. We then pulled all
of the trading history going back 10 years and calculated rolling 6-month
returns. That is to say, as long as both
the GP and LP in a pairing were public at the same time, we calculated 6-month
return for every single day up to the most recent 6-month period.
The argument made in class would suggest return parity. That is to say, the market should correct for
any differences in how cash flows are divided.
Well, the results are in. The GP is – as an economist
would say – strictly Pareto efficient.
The GP returns outperforms its LP pair 56% to 44% in simple performance;
The GP averages 7.1% 6-month returns vs. 5.9% for LPs; The GP median return of
9.1% exceeds the median LP return of 4.6%.
These are staggering differences in favor of the GP, which indicate that
public market investors do not properly recognize the disadvantage of the LP
units.
We will concede that the GP has a greater distribution
of returns – its standard deviation of returns is 32.7% vs. 24.6% for LPs. However – and this is the real dagger in the
GP/LP argument – the GP has a far higher Sharpe Ratio vs the LP (0.28 vs.
0.20).
We can safely say given the data that the GP strictly
dominates the LP in an MLP structure and that the market does not correctly
risk-adjust the economically disadvantaged LP units. It is only a matter of time before Mr. Market
realizes that the LP is always an inferior investment vehicle versus the GP and
ceases to fund LP unit sales. Therefore,
in the long run we see no sustainable situation in which a GP unit and an LP
unit can trade freely on equal footing in the same market.
