The Marcellus Shale Operator Index, Part 2 - Market Sentiment

Includes: CHK, COG, RRC, SWN, TLM
by: Moshe Ben-Reuven


A Marcellus Index of operators is proposed, involving eight public companies. As of December 2013, these cover 67% of the Marcellus total production and 59% of total wells.

This article complements Part 1, which dealt with production efficiency and pro forma operating margins of the Marcellus Index participants.

An analysis of the financial parameters of the eight Index members is provided, with a view to estimate risk, growth, and P/E.

And now, something completely different. As we know, logic is in eternal conflict with the healthy instinct to gain. We should expect no less from the Marcellus Index, introduced in Part 1, the previous segment of this article.

As of the end of 2013, the selected Index companies have 67% of the total production, and 59% of the operating wells in the PA Marcellus. They are, in no particular order: Range Resources Appalachia, LLC (NYSE:RRC), Cabot Oil and Gas Corp. (NYSE:COG), Chesapeake Appalachia, LLC (NYSE:CHK), Southwestern Energy Production Co. (NYSE:SWN), Talisman Energy USA, Inc. (NYSE:TLM), Anadarko E&P Onshore, LLC (Anadarko Petroleum Corp. (NYSE:APC)), XTO Energy, Inc. (owned by Exxon Mobil (NYSE:XOM)), and Shell Western E&P Inc. (Royal Dutch Shell, Plc RDS-A (NYSE:RDS.A)).

Part 1 has dealt with Logic. Namely, the Index operators' production efficiency, and their respective pro forma operational profit. Please note that Tables 1 - 3, and Figs. 1 - 6, pertain to Part 1 of our article. This was done to avoid confusion, due to frequent references made here to earlier data.

Production efficiencies were expressed as production rate per well, or millions of MBTU/year per well. We have shown a range of efficiencies, from 1.72 million MBTU/y per well , to 0.32 million MBTU/y per well (XTO or XOM). So what? Whereas 1 million MBTU/y production costs COG $3.7 million in CAPEX, the same production should cost XTO $20 million. For RRC, one of the lowest efficiencies, 0.436 million MBTU/y per well, the cost would be $14.7 million.

The pro forma numbers indicate that all of the Index operators have an accumulated deficit, collectively $10 billion, related to their Marcellus operations, for the period 2010 - 2013. The highest pro forma Marcellus-related deficit is $3 billion, held by RRC. The lowest, $0.5 billion, is held by COG.

Comparison is often made with the Barnett, TX shale play, which is in operation since 2013. In Part 1, it was shown that the overall Marcellus production rate per well, exceeds that of the Barnett by a factor of 6. This may be due to progress in horizontal fracking, namely, longer laterals and more fracking stages. As the Barnett is richer per acre of surface than the Marcellus (but smaller overall), the difference may indicate that the Barnett is under-utilized.

With all the foregoing laid aside, the last question in our series is, what is Wall Street's perception of our Index?

1. Market Sentiment

Looking at the stock prices, roughly two types of behavior are observed, considering companies, which have significant income from NG sales. We have compared monthly closing adjusted stock prices with the Henry Hub Spot monthly price moves. These are shown plotted vs. time in Figs. 7 and 8, over the period covering 2010 - 2013.

1.1 Type A Behavior. Figure 7 shows Type A behavior. The stocks of APC (Beta=1.87), COG (Beta=0.71) and RRC (Beta=0.81) with Henry Hub Spot prices over-plotted (10X, to bring to a common price range). The 3 stocks have progressive linear regressions, in opposite trends to the declining HH Spot price trend. Obviously, APC has acquired its high Beta value due to its high amplitude fluctuations about the mean linear, and, more recently, by acquisition of certain toxic Kerr-McGee assets, namely Tronox. COG and RRC are relatively low-volatility in comparison.

Since both RRC and COG appear to have a significant portion of their income from Marcellus NG sales over the same period covered here, as seen in Fig. 3, one wonders about this gravity-defying behavior. For RRC in particular, 2010 - 2013 looks like a long, sunny day. Meanwhile, in its basement, a $3 billion mushroom of long-term debt was growing, as in Fig. 6. For fairness comparison, Per RRC's Dec. 31, 2013 Balance Sheet, the RRC long-term debt was $3.141 billion.

Note that the trend-lines for COG and RRC are virtually parallel; RRC is driven progressively and has the highest gradient in this group, but when converted to growth per share, COG is much higher, (its shares are cheaper) as shown later. Of course, on the shorter timescale the stocks do correlate with the HH price moves. We also note that RRC and COG are two companies with very similar market capitalization, but very different price per share, and, as shown later, very different P/E ratio. APC on the other hand is almost 4X larger in market capitalization; although its Marcellus income is a relatively small percent of its total revenue, see Fig. 3, it has significant overall dependence on NG sales.

1.2 Type B Behavior. In Fig. 8 we see Type B behavior. This is another 3 stock subset, plotting share prices vs. time during the same period, 2010-2013. The stocks of Talisman (TLM, Beta=1.54), Chesapeake (CHK, Beta=1.14) and Southwestern (SWN, Beta=1.42) are shown, with Henry Hub Spot prices over-plotted. This time we have an entrained wave-like behavior, as indicated by the sixth-order polynomial regressions shown. By "entrained," we mean, nearly in-phase, and at the same apparent long timescale frequency of the HH timewise variation.

Also very clear from Fig. 8, that SWN and HH Spot track very closely, with a correlation coefficient of 0.75. Unlike Fig. 7, the behavior of SWN, CHK, and TLM, shown in Fig. 8, is as expected of companies, which depend on NG sales for a significant part of their income. Talisman is the smallest in market capitalization of the entire Index group, at $10.7 billion, and its stock is the cheapest. TLM is also the only operator having a negative P/E. Although TLM's trend-line in Fig. 8 has a negative slope, its prospects do look up in our calculations. The difference between Type A and Type B? Whereas the second must work hard at it, the first is ascending on Wall Street thermals.

2. Index Financial Attributes, or performance parameters, are summarized in Table 4. We have recalculated the values of Beta, defined as:

Beta = COVARIANCE(A, X) / VAR(A) (1)

For the trailing 4-year period, 2010-2013, to be congruent with the dataset we have used herein. The vector "X" is the % Return of the particular stock, and "A" is the % Return vector of the Reference Index over the same period, with exactly the same sample size as X. The % Return is defined as:

X(n) = 1- S(n-1)/S(n) where S(n) denotes the stock price at month "n."

Beta is an indicator of volatility or risk, normally computed using the S&P 500 index as reference, and over the trailing twelve months (TTM) period. For the S&P 500, by definition, Beta=1. Higher volatility is indicated by Beta >1 which is awarded by higher interest rates when risk is factored into capital discount rate. Conversely, for Beta<1, less volatility or risk perceived, thus, lower interest rates. We have found little differences in our computed Beta values relative to the data listed in Yahoo Finance.

Following the concerns raised above as to the Marcellus operations' sustainability of the Index members, all of which, per Fig.6, appear firmly anchored in the Red, the question we ask in the last part of this article concerns growth, or trends going forward. The reported financial data can be used to provide some tentative answers.

Growth can be estimated based on recent-past performance, using the so-called Gordon growth model. This starts with a simple formula for the price per share, P, which has the forward-projected dividend per share, DPS, divided by the difference between the risk-adjusted interest on capital, R, and the forward-projected growth percent, G.

P = DPS/(R - G) (2)

From which, after division of both sides by the earnings per share, EPS, we get the P/E ratio, denoted PE:

PE = [(DPS/EPS)(1 + G)] / ( R - G) (3)

Where the (1 + G) compounding multiplier is used to adjust the payout ratio (DPS/EPS) from the current year value to the forward projected one, for next year.

Obviously, the resulting PE is highly sensitive to the difference in the denominator. A small difference R - G can drive a very large PE. Also, this formulation cannot be used for equities, which do not distribute dividend, like SWN, for obvious reasons. Finally, a clear relationship between P/E and G is evident from Eq. (3): A higher P/E corresponds to a higher growth rate, G, and vice versa-all other things being equal.

Some further definitions for R and G are as follows. The risk-adjusted cost of capital, R:

R = RF *( 1 + Beta*RPF ), (4)

Where RF is the risk-free rate, as measured by, say, 10-Year US Treasury Notes' yield, and Beta is the risk parameter defined above. RPF is the Risk Premium Factor, which is a market-wide parameter defined by the Reference equity Index, in our case the S&P 500, as explained later. The inflation-adjusted growth variable, G:

G= GR + LTI, (5)

Where GR is the "real" growth, namely unadjusted for inflation, and LTI is the Long-Term Inflation, as determined by the current 10-year Treasury Inflation-Protected Securities (or 10-Year TIPS) yield.

The calculation of the RPF, or Risk Premium Factor is easily done with the available S&P 500 fundamentals data, and the foregoing definitions.

For instance, on April 17, 2013: Using the last Q4 GDP rate as real growth, GR=2.60%, the long-term inflation = 10Y TIPS yield, LTI=0.59%, Beta=1.00, Risk-Free rate equal to the 10-Y US Treasury Note yield, RF=2.69%. Now for S&P 500 P/E= 18.59, DPS/EPS=36.23/101.59 = 35.66%. From Eq. (5) above, G=0.0260 + 0.0056 = 0.0316, and R= 0.0269*( 1 + 1*RPF). Substitution into Eq. (3) and solving for the unknown RPF, we find RPF=0.9218. Slightly different numbers appear in Table 5 below, which were used in the calculations.

This completes our set of parameters in the Gordon formula, Eq. (3), and allows calculation of Real Implied Growth Rate, GR for given value of P/E, or vice versa, calculate P/E from an imposed GR value. The parameters are as listed in Table 5.

So, if we have P/E and all other parameters except for real growth, we may use the formula to calculate this growth, GR. Or, conversely, if we impose an assumed growth (e.g., GDP value), and all other specific parameters, we may use the formula to calculate P/E.

In Fig. 9, we have a P/E chart for the Marcellus Index operators, comparing the published values, and calculated ones, assuming the actual growth GR = GDP rate = 2.60% (Q4, 2013).

In Fig. 10, we compare the implied real growth rate for two calculations. One, representing the market sentiment, as reflected in the linear regression gradients for the Marcellus Index operator stocks, see Figs. 7, 8. Two, using the Gordon model formulation to calculate GR, with the parameters in Table 5, and the published values of the stock's P/E, EPS, and DPS.

The overall observation from the foregoing comparisons in Figs. 9 and 10, is that the Gordon model provides relatively more conservative results than Market Sentiment. With the exception of Exxon, XOM and Shell, RDS-A, where the calculated P/E ratios above 20 exceed considerably the published values, see Fig. 9. Indeed, using the published P/E values (with the remaining market data), the calculated growth rates for XOM and RDS-A are considerably below the prevailing Q4/2013 GDP rate of 2.60%. At their respective Beta values of 0.89 and 1.27, higher implied growth rates would be more realistic representations. If one knew nothing else, (and we do not), these calculations would imply that XOM and RDS-A are under-rated equities. Not so Anadarko, APC, which roughly belongs to the same class. APC's P/E is listed as 61, and is calculated as 11.4.

Regarding Fig. 9, RRC, COG and CHK show a similar trend, namely that P/E ratios listed, are considerably higher than the Gordon model values. In particular, RRC's P/E of 126 stands out, again paraphrasing Rick. It is 8 times larger than the Gordon model predicts. Looking at Fig. 10, the Gordon model predicts a GR growth rate of 3.9% annually for RRC, while the stock price moves suggest 11.6% growth. For COG, the difference between P/E listed, and calculated, is somewhat smaller, just a factor of 5. However COG has just a 1/6 of RRC's long-term debt, produces nearly 4 times more NG per well (see Table 2) and has broken even in 2013 in its Marcellus activity.

As to Chesapeake, CHK, the 4-year linear regression of its stock price is slightly negative, -2.5%, possibly reflecting residual animosity toward Mr. Aubrey McClendon. What cannot be denied is CHK's long-term debt of $12.9 billion (2013), relative to a market capitalization of $17 billion, and its accumulated EBIT deficit of $1.5 billion from Marcellus operations alone. Nevertheless, CHK's Marcellus operations seem on the pathway to profitability, if only it stayed the course. That means, adding some 262 wells, and increasing production by 286 million MBTU/y by the end of 2014. The Gordon model predicts a growth rate of 3.5%, very similar to the predictions to its sisters, RRC and COG. Further support for this notion is offered by the relative closeness of the P/E values listed and computed for CHK, 36 and 21.4 respectively.

With regard to Talisman , a negative P/E=-53 was listed, yet the calculations predict a P/E = -7, far closer to zero, indicating a healthier phase ahead. Regarding growth, Fig. 10, market sentiment assigns a negative value of GR= -19% for TLM, while the computed Gordon model value is +5.45%, quite encouraging. Further, TLM's Marcellus pro forma shows a positive EBIT for 2013, one of only 3 in the Index. These lend credibility to the envisioned improvement projected by the company in recent publications. Especially if it focused on improving its poor performance and faltering well expansion in the Marcellus. Carl Icahn may not have been so wrong after all. But again, this Canadian company has $4.48 billion in long-term debt, with about $1.1 billion related to Marcellus operations deficit, see Fig. 6. TLM has many diverse world-wide assets, such as aging north sea crude, which, as added detriment, includes Sinopec (NYSE:SNP) as partner. Thus its actual progress depends critically on new, determined, management, and is difficult to predict.

Southwestern will not be discussed here, and the fortunes of APC, XTO and SWEPI will be only mentioned briefly. These Marcellus assets belong to rather large oil companies, pertaining to a small fraction of the owner's total income. With APC considerably better at productivity, XTO and SWEPI's Marcellus operations seem to be half-hearted efforts, represent 0.1% or lower of total respective income, and firmly holding the lowest production efficiency in the entire Index, see Fig. 4 and Table 2. SWEPI has an accumulated pro forma deficit of $1.46 billion associated with its Marcellus operations, while XTO has about half as much. With certain turbulence faced by Royal Dutch Shell (RDS-A) operations globally, it would not be a surprise if SWEPI was spun off. We stress again that the discussion herein is concerned more with XTO and SWEPI in the Marcellus, rather than the respective parents, XOM and RDS-A, whose stock seems undervalued.

3. Conclusions

The objective of the Marcellus Index was to provide a gauge for its current viability and future. This was done by analysis of published performance, and financial data. Also, to find out the degree of interest, and attitude, of large companies like Exxon Mobil, Royal Dutch Shell, and Anadarko. The following observations are offered.

A pro forma of 4 years operation, since 2010, of horizontal hydro-fracking in the Marcellus, with a total of nearly 5,000 wells, indicates that the sector is firmly in the red. Our eight Index operators, responsible for 67% of total production, and owners of 59% of total wells, have a combined pro forma accumulated deficit of $10 billion by the end of 2013, associated with Marcellus activity exclusively.

The financial review carried out in this Part 2 indicates that whereas some operators are rewarded, other are, to paraphrase Rick, less so. The analyses here should help astute investors make a better decision.

It should be in everybody's interest that the Marcellus is further developed, however, as indicated in our first Marcellus article, a new recovery technology is needed. Whereas such technology typically takes 30 years from the laboratory proof of concept, to the commercial market penetration stage, we do not seem to have such a long time available to cover US current NG needs.

In this respect, the recommendations are (1) that large international oil companies expand, not shrink, their involvement, and provide significant R&D boost, and (2) stop the preparation for LNG exports, and prepare for imports instead.

Disclosure: I am long XOM, RDS.A, APC. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.