Saturday 29 February 2020

Dynamic vs. static analysis, and what the US shale and technology sectors have in common

Over the past decade, few industries have incinerated as much shareholder capital as the US shale oil and gas (O&G) sector. Attracted by the substantial technology-driven structural growth opportunity the sector promised, and its putatively low costs and attractive well-level IRRs much-touted by management, investors initially flocked to the sector, but have since been badly burned, as losses have mounted; debt levels skyrocketed; and share prices plummeted.

Investors were right about one thing though - US shale has been a remarkable secular growth story. New technology has unlocked vast new reservoirs of oil and gas, and propelled the US from being a significant net importer of O&G to now being a growing net exporter (the US is still a net importer of oil, but is now a net exporter of gas, primarily via newly-commissioned LNG projects), while also restoring the US to the station of the world's largest oil producer. However, while investors were right about the growth potential, they have been wrong about the profitability outlook, and as a result, have realised considerable losses, this tremendous structural growth story notwithstanding.

So what has happened exactly? Given that well-costs are low and IRRs putatively so high (often claimed to be >50%), why hasn't the sector delivered better outcomes for investors? I hope to explain why in this post, and in the process also explain (1) the perils of static vs. dynamic analysis; and (2) why the experience of the shale sector actually has a lot of relevance to the much-hyped, loss-making tech sector - sectors most believe have nothing whatsoever in common. I believe the latter could well be heading for a significant fallout on par with what has/is now happening in shale.

The fundamental problem with respect to the shale companies has been this: while it is true that the shale companies have attractive well-level unit economics (i.e. if you compare the marginal costs of well completion and production vs. the marginal cash flow such wells generate from product sales, the IRR on that marginal capital spending is indeed attractive), there are also very considerable support costs that need to be set off against that 'contribution margin'.

You need to buy or lease land; drill out that land and prove up reserves; build a bunch of processing and logistics infrastructure (or sign contracts with third parties to provide such infrastructure); and you need to finance all of that expenditure upfront, often through the use of a significant amount of interest-bearing debt. In addition, you need to absorb all the various corporate central head office costs associated with running a large organisation and publicly listed company - a C-suite doesn't come cheap. All of this costs money. Consequently, while it is true that the well-level IRRs touted by management have been high, the all-in returns on capital after all of these support costs (which are relatively fixed) are included have been very poor, and often below the cost of debt. This is why the well-level IRRs management have touted to investors for so long have borne little to no relation to their reported GAAP earnings/ROICs.

So what do you do if you have attractive marginal unit economics with high incremental IRRs, but large fixed support costs that are overwhelming those unit-level returns? You increase volume. This ought to work right? After all, the marginal unit economics are demonstrably attractive, whereas the support costs are relatively fixed (or at least scale less rapidly than volume). As you ramp up your drilling & completion and increase production, the amount of contribution margin ought to rise, and economies of scale and group profitability will kick in, right? As volumes rise, group ROIC should trend closer to well-level IRRs, as fixed costs shrink as a proportion to total revenue, right?

Wrong. The problem with this perspective is that it relies on what I will call 'static analysis' rather than 'dynamic analysis'. Static analysis holds everything else in the world constant, and assumes nothing changes in the future or is any way influenced by the behaviour of either yourself or any of your current or future competitors. If you use static analysis, you will conclude - seemingly reasonably - that if you drill out more wells and produce more oil and gas at high incremental IRRs, while support costs remain relatively fixed, you will inevitably become very profitable.

The problem is that in the real world, the world is dynamic not static. All else is not held constant, and the actions and reaction-functions of economic actors focusing on the micro-level factors in their own businesses can change emergent macro factors in unanticipated ways. In the case of the shale sector, what actually happened dynamically is that as individual shale O&G producers pushed to ramp up volumes in order to enhance fixed-cost absorption, the supply of oil and gas rapidly increased, which resulted in a significant decline in O&G prices. Every single producer held every other producers' behaviour constant, and assumed only their own would change, and what was rational at the level of the individual turned out to be irrational at the level of the collective.

Producers were initially thinking, we are producing from 50 wells, and if we get that up to 100 wells, based on the great unit-level well economics we are currently seeing, and our current support cost structure, we will be able to make decent returns, so lets push production up to 100 wells. However, by the time production hit 100 wells, commodity prices had fallen, such that contribution margins per well had fallen to offset the volume gain, with well-level IRRs falling from (perhaps) 60% to 30%. So what do they now do? They say, well gee it's unfortunate - we've worked really hard and made great operational progress, but the 'external environment' has proven less favourable than expected. Never mind - if we are able to pump 200 wells instead of 100, we will be able to make up for the decline in well-level IRRs, which after all are still quite attractive. We just need to grow a bit more than we thought to 'make it up on volume'. So they ramp production up further. The problem? This further exacerbates the level of oversupply, and pushes commodity prices even lower. This is how the entire shale industry has found itself - to borrow Jim Chanos' wonderful phrase - on a 'treadmill to hell'.

This perfectly explains why US shale O&G producers have remained so hell-bent on increasing production volumes over the past decade, despite falling commodity prices (it is to a large extent because of falling commodity prices). A lay observer might be inclined to ask, why on earth are shale producers rushing to produce as much of their non-renewable reserves as fast as possible in such a low pricing environment? Why not just sit on the valuable resource and wait for a better pricing environment? The answer is that they have very considerable fixed costs (including debt servicing), and they will therefore bleed to death if they wait. They need whatever contribution margin they can get to try to defray these fixed costs. Contrary to popular opinion, it is therefore not purely a function of the companies being run by a bunch of 'oil men' who just want to 'drill baby drill' because they like drilling. It's due to economic necessity, because the alternative is bankruptcy.

Now what I find most interesting about the above is that there is another sector at present suffering from exactly the same economic forces, and the fallacies associated with static vs. dynamic analysis: the loss-making technology/software sector - particularly Unicorns funded by VC (note, this does not include technology/software companies with business models that are proven to be profitable, although such companies can still be harmed by the actions of other industry actors, in much the same way profitable O&G companies have been harmed by the shale sector). There are many listed examples as well, however, such as companies like Wayfair, Uber, and Carvana. (Indeed, my recent post on Afterpay uses a dynamic analysis to counter the prevailing bullish narrative that is based primarily on a static analysis; most of the considerable criticism I received for the piece from APT bulls on Twitter also used arguments grounded in static analysis).

While on the surface, fast growing but loss-making tech companies might seem to have little in common with US shale producers, the economic parallels are actually quite striking. The only difference is that the market is currently red hot on fast-growing tech companies, regardless of the current state of the P&L, because investors are still using static rather than dynamic analysis to assess their prospects, much as they were many years ago to assess the prospects of shale producers, prior to the sector's recent implosion.

The similarities are as follows: Investors - at the direction of company managements - are focused almost entirely on revenue/volume growth and the unit economics, or contribution margin associated with additional volumes. The level of support costs, including corporate overhead, R&D, and marketing etc, are generally ignored, as is the aggregate level of group losses such support costs are occasioning. This is how companies like Wayfair have been able to multi-bag despite reporting larger and larger losses as they have grown. The support costs are seen by investors as being largely irrelevant, as it is believed they will shrink in relative size over time as revenue growth and contribution margin most assuredly continues to rise. Indeed, it is often argued that you should value companies like Wayfair on a multiple of gross profit (even Bill Miller has made this argument).

Uber, for instance, advertises extensively in its presentation materials the great 'contribution margin' it makes on its ride business. However, notwithstanding the putatively fabulous unit economics the company enjoys, the company still loses about US$1bn a quarter a decade or so after its founding. It turns out R&D, marketing, customer service, and the various other support costs necessary to sustain operations are quite expensive (incidentally, there are almost no companies in the world that don't have robustly positive gross margins and trade at low multiple of gross profit; investors outside of tech do not look at gross profit multiples because they are aware that without the associated SG&A costs, those gross profits and contribution margin could not be delivered).

Like with shale, if one uses static analysis, you can create a bull case for these stocks despite the current state of their P&Ls, because you can argue that as these companies scale volumes and contribution margin off a support cost base that will grow less rapidly than revenue, significant operating leverage will kick in down the track. If you hold everything else constant, including in particular the actions not only of current but also future competitors, then the bull case is plausible.

The problem though is that - as was the case with shale - the world is dynamic not static, and you cannot hold everything else constant. It is as naive for these companies and their investors to believe that the rest of the world - including the many well-resourced incumbents many of these companies seek to disrupt - will sit idly back and allow the said company to capture a large TAM ('total addressable market') uncontested, as it was for shale producers to believe that their own significant ramping up of volumes would not impact end commodity prices.

In reality, as more tech companies with great 'unit economics' but large overall losses attempt to drive revenue/volume growth to increase 'contribution margin', recoup fixed costs, and hence pursue a 'path to profitability', what is likely to happen (and indeed already is) is that competitive intensity will significantly increase and drive a decline in contribution margins that offset that volume growth. This can manifest not just in the form of outright price discounting, but also escalating customer acquisition costs, and the increasing prevalence of various promotional offerings (including 'free delivery' and the like), which are sometimes included below the gross profit line (e.g. Uber's 'driver incentives'). While the companies no doubt couch such offers in terms of 'enhanced customer value' and 'driving engagement', in reality it is simply a form of discounting; the phrase 'buying market share' has long existed in business, and this iteration is no different.

There is no surer sign this is happening that when the companies continue to push out their estimate for when they will reach break even/profitability, or see profitability/margins decline even though volumes are rising, as has recently occurred with Wayfair. And even hitherto profitable companies can be adversely affected. Take Latin American e-commerce company MercadoLibra, for instance. A few years back, the company was robustly profitable and making operating margins of some 20%. Today it is more like -10%. Why? Because there are several very aggressive competing e-commerce companies attempting to ramp volumes to recoup fixed costs, and just like in any industry, it is hard to be smarter than your dumbest competitor. It is no different from profitable incumbent oil companies suffering declining profitability because a bunch of shale companies are desperately trying to ramp up volumes and 'contribution margin' to cover their fixed costs.

The only difference is that instead of investors seeing what is happening for what it is - a price war that is a symptom of excess supply/competition and ugly industry economics akin to the airline industry in eras past - they are instead content to believe the fiction that this is primarily an 'investment in future growth' and being 'long term greedy'. In reality, these companies are losing money for the same reason shale producers are - there is too much supply because too many companies and too much capital jumped into the sector to capitalise on 'secular growth'.

After the 2000-03 bust, people widely questioned how it was that so many smart investors managed to convince themselves that the normal laws of economics did not apply to internet companies. Well, something very similar is happening today. Tech investors are content to pay outrageously high prices for technology companies because they believe they are immune from the normal prosaic forces of supply and demand, and that both competition (both current and future) and a rapid influx of capital into the industry will not drive down returns in the way it has in the past in every industry and in every instance it has ever occurred, just as the laws of economics would predict.

LTV/CAC metrics provide deceptive comfort to companies/investors about the wisdom of incurring hundreds of millions of dollars of losses in order to grow, because they assume that the LTV is locked in rather than merely a hypothetical based on a static analysis of how things look today, rather than a dynamic analysis of how LTV could look in the future, and evolve in an environment of increasingly saturated markets and more and more competitors desperately trying to ramp up volumes to recoup fixed costs. It is exactly analogous to shale O&G companies assuming that their efforts at rapid volume growth would have no impact on the end commodity price. After all, the LTV of their wells were very high relative to their WAC (well acquisition costs). So what could go wrong? Why shouldn't investors value them based on LTV/WAC metrics?

As I have argued for some time now, I think a fairly epic bust in the technology sector is coming, that could well rival the scale of the bust, although with the worst effects being seen in the VC/private space rather than the listed space (although the latter will also likely be badly affected). In many cases, valuations have reached utterly absurd levels for loss making companies with unproven business models, which rely on static analyses that are highly questionable in a dynamic world. As I have argued in the past, consensus expectations for these companies are that they continue to enjoy rapid revenue growth and rising profitability, whereas I think in reality what we will see is decelerating top line growth and falling profitability/widening losses. For those that are paying attention, it is apparent that this is already starting to happen.

Furthermore, there are signs that an incipient reversal in the liquidity flywheel* that has propelled the growth of the VC 'asset class' in recent years is already underway, as Softbank has failed to raise external capital for its Vision Fund 2. If the liquidity flywheel decisively turns, a fairly catastrophic bust could happen in the VC space, and it remains to be seen if it will be of sufficient size to tip the US (and perhaps select other Western countries as well) into recession. The magnitude of the percentage losses realised by tech investors will likely be as large as those realised by shale O&G investors over the past cycle, in my view, if not larger.


*A 'liquidity flywheel' is the name I give to the dynamic where fund inflows result in capital being deployed by those funds in a manner that contributes to the price of the assets they already own being pushed up, thereby contributing to strong apparent returns. The strong reported returns then act to drive additional inflows in a self-reinforcing fashion. This is how you end up with worthless, loss-making companies like WeWork attaining a valuation as high as US$47bn. 

Liquidity flywheels create an 'unstable equilibrium' that can go on for a long time, but eventually turn, and when they do, a discontinuity event happens and the whole movie is then played out in reverse (and usually on fast-forward). Liquidity flywheels are a fundamental contributor - perhaps the most important contributor - to boom-bust asset price cycles in markets.