The current stretch for yield among investors has allowed MLPs to issue shares at a low cost of capital to fund new investments. This has allowed them to raise their dividends at above average rates. This, in turn, attracts more investors to issue more shares to, and finance more investments. A virtuous circle. However, this cycle will eventually create a bubble if left unchecked.
My definition of a bubble, as opposed to mere overvaluation, is when just a reversion to the mean could result in a permanent loss of capital. In this instance, this could be caused by a tightening of the credit markets, sufficient rise in rates, or lack of new investors. In order to achieve an actual bubble, there must be what Charlie Munger would call a lollapalooza effect of multiple biases acting in the same direction. Only one of which needs to be interrupted for the lollapalooza effect to come to an end. The current investing environment supports MLP outperformance due to the confluence in ultra-low interest rates, loose credit markets, investors hungry for income products, growth in US energy infrastructure, and recent above average returns of MLP products attracting new investors.
While there are many types of MLP investment vehicles, such as dry bulk shipping and coal, one of the most common uses of the MLP structure is pipeline operators, and since I happen to own shares in Enterprise Products Partners (EPD), this article will focus on pipelines, because it is the business I am familiar with.
Pipelines are very reliable businesses. Often described as a toll road operating business, they rent the use of their pipelines to oil and gas companies that need to move their product to market, and get paid by volume shipped. There's more to it than that, but that sums it up pretty nicely. Furthermore, business fundamentals are quite good, as the US has seen a large build out in energy infrastructure and has room to grow.
MLPs are required by law to pay most of their cash flow in the form of dividends, so they can't really finance growth with internal cash flow. Instead, they must issue shares. Companies that can issue the most shares at the lowest cost of capital have a meaningful cost advantage over competitors when investing in new projects. That cost advantage quickly begets an advantage of scale. The companies that issue the most shares at the highest prices become the largest operators, and thus have access to the lowest cost of capital (issuing more shares). A virtuous circle.
The table below lists the year end share count of some of the larger MLPs. If one used a larger sample size, you would see some variance, but the trend is pretty simple. Share counts go up.
The willingness of the market to purchase these additional shares is what ultimately allows these companies to grow their dividends, make acquisitions, grow their earnings, their debt, and keep the ball rolling. When companies can issue more shares at a cost of capital that is lower than the rate of reinvestment, they make more money, and everybody wins.
It has obviously been working. So why can't it work forever?
At some point the law of large numbers must come into effect. The nature of the MLP structure is that no one can afford to stand still. Currently there are two large acquisitions in the MLP world. Kinder Morgan (KMI) is acquiring Copano (CPNO) for $5 Billion, which represents about 12.5% of KMI's current market cap. The much smaller LinnCo (LNCO) is acquiring Berry Petroleum (BRY) for around $4.3 Billion, which is more than 280% of LinnCo's current market cap. MLPs need to grow assets in order to grow dividends, and when new projects aren't enough to move the needle, they resort to acquiring each other.
Both of these acquisitions are being funded by issuing shares. As long as these companies can raise capital at federally subsidized rates, thanks to artificially low interest rates, they will rightly continue to do so. But at some point, no acquisition will be large enough to maintain the current growth rate, and then distribution rates (that means dividends) will also have to slow down.
The law of large numbers is one problem, but the more obvious problem is cyclicality in the credit markets. The simple need to continually issue more shares exposes MLPs to undue credit risk. To understand how MLPs are dependent on share issuance, let's take a quick look at Enterprise's financial statement.
Enterprise Products Partners
year ending 2012
Less Distribution to Partners
Retained Cash Flow
In 2012 Enterprise's cash flow was $3.5 Billion (net income plus non-cash depreciation) but it spent $3.6 Billion on capex. So where is the money for dividend payments coming from? Issuing new shares.
And that's how you start a bubble.
Investors over-reaching for yield prompt MLPs to over-reach for assets. All of which is funded by far too low interest rates and far too easy access to credit. If credit markets tighten, or if interest rates revert to the mean and MLPs can no longer raise capital at a cost below the rate of reinvestment, or if investors lose their appetite for further share dilutions, then distribution growth will stop, and the party comes to an end.
Apologists will point out that the majority of this capex is really for funding growth and not for maintenance, so really the dividend is coming from income. This is sophistry pure and simple. Whether you are paying your dividends and borrowing to pay capex, or paying capex and borrowing to fund your dividends, either way you are reliant on access to additional debt to fund your ongoing business concerns. But no one has ever actually tested this theory. Even so, if asset growth stops, what happens to the dividend and share prices? These shares were purchased by income investors with expectations.
As alluded to earlier, as part of the rules defining an MLP, these companies are required to pay out all distributable cash flow not needed to for current operations and maintenance of assets. For example, Enterprise assumes an operating lifespan of 25 years for equipment, so each year those investments are depreciated by 1/25th. However, they apply that rate to net PP&E but gross PP&E is about 25% higher. Thus possibly understating true costs, if future depreciation exceeds the historical rate.
What happens if a pipe's useful life is shorter than 25 years? One research economist suggests that shale gas plays in the Eagle Ford have an expected lifespan of only 16 years (other projections range from 20 to 64 years), which is less than the depreciation schedule of the pipeline being used to transport that gas. So potentially, some of these pipes will sit unused before their productive lives comes to an end.
Unconventional gas wells have much steeper decline rates than conventional wells, with some wells only lasting a few years at best. Pipelines are built to accommodate operations at full capacity, but those rates are only sustainable for a brief time. That means pipeline volumes will run considerably below capacity for the vast majority of that asset's expected lifespan, and that revenues will drop off sharply as production rates decline after the first few years. How are these assets going to be accounted? On 25 year depreciation schedules?
As a final caveat, I don't think there is currently a bubble, but I see a clear path towards one, and a sell strategy is an important part of disciplined investing. We are talking about uncertain future events still years in the making, but there will come a day when MLPs will need to issue shares just to pay for capex on existing assets. When the market realizes this, share prices will drop and funding will dry up.
As a parting thought, current dividend yields in the MLP space aren't attractive enough to justify the potential credit risk. While MLPs are supposedly desirable for their high dividend yields, the three largest pipeline operators by market cap, Enterprise, Kinder Morgan and Enbridge (ENB) all have dividend yields lower than large integrated oils such as ConocoPhillips (COP), Royal Dutch Shell (RDS.A), and BP (BP). Now of course you will argue that the pipeline companies are superior to the integrated energy names because they will grow their distributions at higher rates over time. But BP and Conoco don't need to access credit markets annually to pay dividends; they can fund that with internal cash flows and still make investments or buy back shares. As a matter of fact, Conoco spun off its midstream assets to cash in on the current market sentiment.