Involvement in exploring a substantial discovery and bringing it too development is an exciting opportunity. There is nothing more valuable to a company than taking an unknown prospect, acquiring the land for low cost, and turning it into a cash generating machine. Finding such a play is the motivation for many teams. It can be the pinnacle of success and a career. Furthermore, there is nothing more fun than winning by creating value.
However, far too often big shale discoveries have failed to deliver the growth and returns that seemed eminent based on the large and low risk hydrocarbon accumulations. Plays proved far more complex than initially perceived, and a consistent path to development and returns was often derailed in delineation.
After discovery comes heightened expectations, spending announcements, and a period of high grading and data collection. Characterized as delineation, this stage helps to quantify the uncertainty regarding performance, in-place volumes, yields, capital and other inputs in the economic model. A re-calibration of the P50 of inputs can trigger a realization that the play may under perform original projections. If the re-calibrated economic model fails to justify further spending, an impairment is likely.
A good example of this cycle may be Apache’s discovery of Alpine High in 2016. Alpine High was a substantial discovery, that was well documented in publications such as Forbes magazine. The find was material, to say the least, with 75 TCF and 3 billion barrels of oil. However, in 2020 it was announced that Apache would cease drilling in the play as Bloomberg reported on APA’s decision regarding a $3 billion write down.
Impairments are not uncommon. All companies and plays may be susceptible, especially in volatile commodity markets. Given the huge economic stakes, it is a worthwhile to examine some aspects of how impairments occur. Particularly for large shale discoveries where there is little geologic risk of the substantial resource in place.
In the following article, we look to explore some pivotal questions regarding large shale plays and the progression through the delineation stage with a focus on the following:
- How are big shale discoveries different than big shale developments?
- How does uncertainty in inputs lead to continued investment in big discoveries?
- What is the role and cost of data in determining the uncertainty?
- What role does loss aversion and early expectations play in decisions?
- Is there a different approach that can reduce the substantial uncertainty of data sparse plays, and quantify the probability of the potential before a lot of capital is spent?
Table of contents
Finding Big Tuna vs Getting it In the Boat
Finding resource plays must be separated from developing them and bringing them to market with competitive full cycle returns. I remember a great metaphor by a well-known shale pioneer, “We caught the big tuna, now we have to figure out how to get it in the boat”.
As we have seen in the shale oil and gas revolution of the last decade, discoveries have been plentiful but full cycle returns have been elusive. There have been considerable shale plays that never lived up to the resource potential and ultimately fell short of development. Perhaps the Monterey Shale in California best exemplifies this as the EIA estimated 23.9 Billion-barrels of oil-in-place. I have worked some large accumulation frontier plays as well (Canol Shale, Muskwa Oil Shale, 2WSP Shale). All contained substantial hydrocarbons, were explored with a Hz technology, and eventually ceased spending during delineation. Even many successful shale plays that have good single well returns, market access, and friendly local jurisdictions have failed to produce consistent results and sustained free cash flow.
This pattern has been repeated so often that investors grew tired of the inability to produce sustainable free cash flow from large shale discoveries. They are no longer going to reward finding a big tuna or even getting in on the line. The market wants it in the boat and sold at a high price with receipts. As we now know more clearly than ever, finding a large accumulation of hydrocarbons does not guarantee good returns, even in well developed shale resources that have no conventional geologic risk. In fact, far from it.
The Danger of Big Discoveries
The early stage pursuit of organic growth in shale is particularly perilous. Finding a resource play and acquiring land at low cost usually only occurs when there is limited data and sparse results. How could you acquire cheap land if everyone knows the value? You can’t. These data poor plays offer substantial upside and possible returns, but they also come with more risk. Similar to the market, higher returns are obtainable with higher risk.
The risk is not associated with the accumulation or quantity of hydrocarbons such as a conventional play. Assessing the geological risk in conventional terms (trap, source) produces a high probability of success. The risk is essentially the uncertainty in being able to produce the volumes economically at full cycle.
After substantial resources in tight rock are found to exist, the delineation period begins. This period is characterized by acquiring more data and verifying inputs. At any stage there is churning of economic models which require several prognosticated inputs (type curve, forward price estimates, capex, opex, yields) in order to demonstrate the play viability and upside of further investment.
These inputs have considerable uncertainty, which is usually quantified with data. However, due to the lack of data during the delineation stage the uncertainty may not be known, or it may be very high.
The danger comes when production driving variables in the economic models are estimated and presented deterministically, and growth possibilities are explored. As documented in a previous post, bias exists in forecasting. Especially when teams have economic incentives to present possibilities that demonstrate upside and growth. In many cases upside is given an over-weighting vs the uncertainty of the inputs or the risks (which may or may not be known).
Large accumulations provide a road to large growth, and substantial upside. In many ways, the upside is captured on paper and presented confidently. I think Burton Malkiel (author of classic investing text “A Random Walk Down Wall Street“) would characterize this potential growth as building “castles in the sky”. Early stages of resource plays provide the possibility of upside, a transformational opportunity, and growth. But are the uncertainty and risks equally explored?
the Broken Feedback Loop
Big discoveries in oil and gas exploration are often worked in silo’d teams. This is done to protect confidentiality in the land acquisition phase. Due to this confidentiality and related isolation, the exploration team often has minimal technical input, peer review, or external feedback. This is problematic as this is the stage with the greatest unknowns and largest uncertainty. It is exactly at this stage that peer review would be of the most value but is often not an option.
It is also problematic if the team is convinced, since they have been designated to work such a valuable project, that they have little to gain from other (external or internal) experts. Constrained and unwanted feedback is a double-edged sword.
In either case, the result could be that economic models, and the inputs, are not thoroughly vetted. The uncertainties and risks may not be thoroughly challenged. This can lead to proceeding down a road that has a low probability of success that will likely become clear when more data is available.
It has been proven that groups are more adept at predicting uncertainties of outcomes than individuals. Ask a group of experts to guess a value of a random question, such as “What is the average tonnage per well pumped in the Permian?”. The average of the answers provided by the group will likely be closer to the correct value. The single answers provided by individuals will vary wildly. Collectively groups are often right at approximating the mean, but the group’s individuals are often more wrong.
High Cost of Data: The Powder Keg of (potential) Impairment
As mentioned the early stages are a dangerous time in the pursuit of a large resource play. Why? Because the lack of data and uncertainty around key inputs in the cashflow model. The type curve, price, capex, opex, yeilds are all considerable inputs in the model. Due to the lack of data, they may be presented deterministically. Inputs in any model have uncertainty. However, for plays in delineation that lack data, the uncertainty is not known.
Consider the economic model as a product of these inputs. Does the model (which is the product) represent a high or low probability scenario? If the uncertainty of the inputs is unknown, the product of said inputs is definitely unknown. Not knowing the probability of the inputs is precisely why this is so dangerous. Capital must be spent to acquire the data to produce a distribution, and the resultant probability.
For example, in order to know that a type curve is a P50, a distribution of valid data points is needed. In order to know the P50 lifetime yield of a gas-condensate well, you need a distribution of wells with considerable production. These examples could continue for porosity, pressure, saturation, pay, IP, maturity, completion capex, drill capex, or any production driver the team is quantifying. Distributions need data. Shales are complex plays with many unknowns, which means they need large amounts of data.
However, the capital being spent to get the data, and to validate the assumptions, is absolutely known and very significant. At this stage, large capital expenditures begin. 100’s of millions are deployed to lease land and acquire data, test pad concepts, build roads, build pipelines and facilities, drill wells, test production, etc. Full cycle returns are poor in the short term, but they are expected and justified by the projected long term growth and future free cash flow. Payout is often modeled to be obtainable within a few years, but at what probability? Not understanding the economic model input probabilities, and the resultant product in the full-field development model is extremely dangerous when spending significant dollars. If the inputs are off, the cashflow projection will be substantially off.
The main crux is that dollars are spent to get data to validate the assumptions in the economic model. However, the economic model is being used to justify the spending to collect the data. This is a precarious position.
This creates growth in book value, which is certain. The resource quantity may be certain. However, the free cash flow growth and value of the reserves are much less certain. This creates a powder keg for a potential future impairment.
It is after the dollars are spent that the inputs are more known as data has been collected. P50’s, for all inputs, are re-calibrated. It is now time to see if the play ranks.
- Were the assumptions made prior to delineation valid?
- Are the undeveloped reserves reasonably certain and do they offer a good return?
- Is the company committed to development?
These questions have to be answered. If the re-calibrated model results are poor and development capital is pulled, an impairment is inevitable.
Big resources lead to big expectations, big development plans, big growth profiles, and an intensive capital spending profile. What begins slowly with maps quickly escalates to spending significant dollars. The danger comes from being heavily indebted into a play before you truly know the performance, risks, and uncertainty of the development.
Loss aversion is a defined concept in behavioral psychology that I believe may be a powerful force when oil & gas companies are heavily invested in a specific play. Loss aversion affects investors that seek to avoid a loss, particularly if there is possibility that the loss may be regained. This occurs in the market all the time as investors hold a ‘loser’ too long. It occurs in oil & gas plays as well. A company may continue to invest in a play if a scenario exists that shows that the play will return acceptable metrics. The scenario may have a low probability that may or may not be known. However, the scenario represents the possibility that the loss can be avoided.
Loss aversion can have a drastic effect by pushing forward alternate scenarios with different capex assumptions, type curves, yields, opex, etc. These often deviate from the original economics promised, but still provide a way forward to profitability, and meeting company economic hurdles. If these scenarios have a low probability and under-deliver, you have now added to the potential loss as capital spending was continued based on the revised scenario. A hard look at the probability of new (or past) assumptions should be considered, before proceeding (with caution) with additional economic runs.
A Different Approach?
At this point it must be asked what is a better approach? How can we understand and verify the plan inputs, the uncertainty of those inputs while constraining capital spending? Can the delineation phase capital be minimized? Essentially, how can we get to commercial development or play exit sooner and cheaper?
Can we use data driven models from plays in development to predict the outcomes from similar plays that are data sparse and at the early stages of delineation or appraisal? Can learnings from plays that have scaled to full field development be transferred to plays in appraisal? This topic will be explored in future posts.
- At the height of the shale revolution, the announcement of large resources were common. The in-place estimates were reasonably certain. However, economic returns and successful development were much less certain. Big resources only guarantees the possibility for big growth. There is no guarantee that the big tuna can be reeled in.
- Big shale discoveries can lead to substantial expenditures before key inputs in the economic model have quantifiable risk. Data is required for delineation and to quantify the uncertainty of the economic inputs. Large capital spending in this phase can be hazardous. The accumulation of substantial book value before the economic viability of play is certain to create the possibility of a hefty impairment.
- The silo’d nature of teams can be detrimental. The opportunity for feedback can assist in determining the range of outcomes and quantifying the uncertainty earlier in the delineation process. This could have a dramatic impact of future capital expenditures.
- Loss aversion behavior can have a hugely detrimental impact in that it can drive continued capital spending in a play with low probability of success.
- The industry has never been better positioned to explore new shale plays with ability to combine the vast amounts of data and geologic characterization, across all plays, to predict uncertainty and outcomes.