For much of the modern history of the oil industry, there has been an implicit hierarchy inside the integrated oil company. Upstream finds the hydrocarbons, owns the reserves and captures the economic rent. Refining, marketing and retail process the product, move it and sell it. The downstream businesses can be valuable and occasionally extremely profitable, but the scarce asset has traditionally been assumed to sit in the ground. Strategic attention therefore gravitated towards exploration, development and resource capture, while the rest of the organisation was frequently described in terms of integration, resilience and support for the upstream engine.
Investors largely adopted the same hierarchy. A major discovery could transform perceptions of an oil company in a way that another refinery or another thousand service stations rarely could. Yet a series of separations undertaken by US oil companies between 2011 and 2013 provides an unusually clean test of what happened once the different parts of that model were allowed to operate independently. Marathon Oil separated Marathon Petroleum in 2011, ConocoPhillips separated Phillips 66 in 2012 and Murphy Oil separated Murphy USA in 2013. Existing shareholders received shares in both businesses.
More than a decade later, the businesses that looked less glamorous at the time have produced some remarkable shareholder outcomes, while the upstream companies have generally been far less impressive. The comparison does not prove that downstream is inherently superior to upstream, and the timing of the transactions introduces a major complication, but it does challenge the assumption that identifying where an oil company generates its operating cash is enough to understand where shareholder wealth is being created.
Marathon is particularly interesting because the reasoning behind the separation was laid out for its shareholders at the time. The company announced in January 2011 that it would split into an independent exploration and production company, Marathon Oil, and an independent refining and marketing company, Marathon Petroleum. Management did not argue that integration was worthless. The combination of upstream and downstream provided diversification, greater scale and some protection against movements in commodity margins. There were financing benefits as well, and separation could reduce some of that balance-sheet capacity. The board nevertheless concluded that the two businesses were sufficiently different that they would function better independently.
Their capital requirements differed, their natural investor bases differed and management believed each would be easier for investors to understand and value on its own. The separation documents also explicitly discussed competition for capital inside the integrated company. Large downstream investments, including the Detroit refinery upgrade, had to compete with opportunities available elsewhere in Marathon, while the upstream business was described as capital intensive throughout the commodity cycle and requiring continuous deployment of significant capital to sustain production and revenue growth. This is the first clue we need to follow.
Fifteen years later, the language reads differently. Marathon was not merely saying that its two divisions deserved different valuation multiples. It was acknowledging that one side of the company faced an essentially continuous call on capital and that projects elsewhere in the group had to compete against it.
The expected outcome in 2011 was not that Marathon Petroleum would become the overwhelming source of wealth for the original shareholder. The logic was much more balanced. Marathon Oil would be free to pursue upstream opportunities without having to accommodate downstream projects, while Marathon Petroleum would be able to invest according to the economics of refining, logistics and marketing rather than competing with exploration and development. Both companies would have dedicated management teams, separate balance sheets, appropriate capital structures and more clearly defined investor propositions.
The market subsequently produced a much more asymmetric result. By 2019, only eight years after becoming independent, Marathon Petroleum said it had returned almost $21 billion to shareholders and generated a total shareholder return of approximately 323%, almost twice the return of the S&P 500 over the same period. Speedway was later sold to 7-Eleven for $21 billion, providing another enormous pool of capital that supported debt reduction and share repurchases.
Marathon Oil followed a very different path. It remained a conventional E&P until ConocoPhillips agreed to acquire it in 2024 for an enterprise value of $22.5 billion, including $5.4 billion of net debt. An investor who owned Marathon immediately before the 2011 separation therefore did not simply own the future MRO. The shareholder owned MRO and MPC, and the downstream security became by far the more important source of capital appreciation. Judging the original decision by looking only at the later performance of Marathon Oil misses most of what actually happened to the shareholder. See also the PetroEquity Signal articles and podcasts on ConocoPhillips, including "ConocoPhillips got bigger, but did it get better?"
Murphy produced a very similar experiment two years later, although the contemporary rationale reveals more explicit concern about valuation. Murphy Oil announced in October 2012 that it would separate its US downstream business, and Murphy USA began independent trading in August 2013. Third Point, led by Dan Loeb, had already been pressing the company to consider structural change. Murphy's own materials argued that the retail business was being followed largely by analysts whose principal expertise was exploration and production, leaving the fuel retail operation underemphasised and poorly understood. The separation was therefore intended to allow investors to value each company against a more appropriate peer group and to give the two management teams independent control over capital allocation.
The initial investment case for Murphy USA was hardly exotic. It sold fuel and merchandise through a large convenience retail network, much of it historically associated with Walmart locations. It did not own a deepwater discovery or a shale position with thousands of drilling locations. Yet independence changed the financial framework around the business. Management could invest in stores, make acquisitions, reduce debt or return excess cash without those decisions being ranked against another offshore development or another exploration programme elsewhere in Murphy Oil.
Murphy USA's subsequent share count tells much of the story. By 2025 the company said it had repurchased more than 29 million shares since separation, equivalent to approximately 63% of the shares outstanding when it became independent, at an average price of around $139 per share. The company did not stop investing. It expanded the store base, improved the network and made acquisitions, but every dollar retained by the business had to compete against the alternative of repurchasing shares or returning cash directly to investors.
Murphy Oil faced a different economic equation. Its producing assets generated substantial cash, but production itself consumed the resource. Maintaining production and reserves required new drilling, development spending, exploration and acquisitions. Murphy has owned good assets and has often been technically capable; the subsequent return divergence is not evidence that its oilfields were poor businesses. It instead exposes a difference between earning high margins on a finite asset and producing high returns per share over a long period. Murphy USA could grow the ownership represented by every remaining share through repurchases while continuing to operate the same underlying network. Murphy Oil could not preserve its productive capacity indefinitely without replacing the barrels it sold. See also the article "Murphy Oil: What Happened to a Great Little Oil Company?"
ConocoPhillips provides another layer because the company effectively argued both sides of the integration debate within a decade. The 2002 merger of Conoco and Phillips Petroleum was partly justified by the benefits of scale, portfolio diversity and integration across the value chain. A larger combined company could balance upstream and downstream exposure, smooth some of the volatility created by commodity cycles and allocate capital across a wider opportunity set. That reasoning was entirely orthodox. Less than ten years later, ConocoPhillips concluded that the advantages of specialisation had become more important and announced the separation of Phillips 66. The 2011 materials again referred to different capital structures, different investor bases and the elimination of competition for capital.
ConocoPhillips would emerge as a pure upstream company positioned for organic growth. Phillips 66 inherited a portfolio that included refining, marketing, midstream and chemicals, and the refining operations were not presented as hidden jewels. Refining and marketing had the highest capital employed and the lowest return on capital among the principal businesses, and management expected to rationalise weaker assets while directing money towards higher-return opportunities elsewhere in the portfolio.
Phillips 66 nevertheless stated from the outset that dividends and share repurchases could form a significant part of shareholder value creation. By 2025 it said it had returned more than $43 billion through dividends and buybacks since becoming independent and had compounded its dividend at roughly 15% annually. The outcome again suggests that the most important change may not have been the discovery of previously invisible operating economics. Independence altered who controlled the cash and the standards against which its use was judged. A refining project no longer had to compete with an upstream development inside ConocoPhillips; it had to compete with every other use of a dollar inside Phillips 66, including simply giving the dollar back to shareholders.
Marathon Petroleum and Murphy USA show similar behaviour. Once these businesses stood alone, management could no longer justify retaining capital because some other division of a large integrated company needed it. Investors could see the operating returns, the reinvestment requirements and the distributions directly. The separation made capital allocation observable.
That helps explain why the phrase "upstream is the cash engine" can be both true and incorrect. Upstream assets can generate exceptional operating cash flow, particularly when commodity prices are high and the underlying resource has low lifting costs. A producing field may be one of the highest-margin industrial assets in the world.
Yet operating cash flow is not the same thing as cash that can be removed permanently from the enterprise without changing its future productive capacity. Every barrel sold reduces remaining reserves. Production from individual fields declines. Leases expire and drilling inventory is consumed. Shale simply makes that observation even more relevant. The upstream company therefore has to decide how much of today's cash should finance tomorrow's replacement barrels. That replacement can take the form of exploration, appraisal, development, acreage acquisition, enhanced recovery or corporate M&A. Each proposal can be perfectly defensible in isolation. Together they create an unusually persistent demand for reinvestment, and they give management a long list of strategic reasons for retaining capital.
A refinery or convenience-store network can also consume substantial capital if allowed, and neither business should be treated as economically effortless. Refineries require maintenance, environmental expenditure and periodic investment, while retailers have to renovate stores, build new locations and defend market positions with lower margins. Phillips 66 itself acknowledged at separation that refining carried heavy capital employed and relatively weak returns. The difference is that the asset does not physically deplete every time it operates. A successful year does not automatically create the requirement to discover another refinery or replace an equivalent number of service stations merely to maintain next year's capacity. That makes the distinction between sustaining capital and discretionary capital clearer. It also means that the cash generated by a downstream asset can become genuinely surplus in a way that is more difficult to establish in an upstream company whose reserve life is continually shortening.
There is, however, a serious problem with drawing a structural conclusion from the three principal separations. Marathon split in 2011, ConocoPhillips in 2012 and Murphy in 2013. Those dates sit almost perfectly at the beginning of an exceptionally difficult period for upstream shareholders. The first phase of the shale revolution generated extraordinary production growth but frequently consumed more cash than it produced. Debt and equity were repeatedly raised to finance drilling programmes, while management teams were rewarded for growth in production and reserves even when per-share returns were weak. Oil then collapsed from more than $100 per barrel in 2014, followed by years of pressure, another sharp deterioration in 2018-20 and the unprecedented disruption of Covid. The sector accumulated impairments and investors eventually forced a wholesale change in behaviour, demanding free cash flow, lower reinvestment and distributions rather than perpetual growth. US refining, logistics and convenience retail experienced their own favourable developments during parts of the same period, and aggressive buybacks magnified the benefits when equity valuations were attractive.
The timing therefore leaves open an important alternative explanation. Perhaps the separations do not reveal a timeless but poorly understood superiority of downstream economics at all. Perhaps they happened immediately before a particularly tough upstream cycle and a much better period for the businesses on the other side of the split?
One useful test would be to move the starting date backwards by roughly a decade and recreate the experiment around 2000, before the commodity supercycle. Oil prices subsequently rose dramatically, upstream earnings surged and large resource owners enjoyed a very different environment. An upstream-heavy portfolio begun at that point might produce an entirely different result. This article does not intend to resolve that question here, but even if the extraordinary post-2011 divergence turns out to be heavily cyclical, the Marathon separation documents identify a characteristic that exists independently of the cycle: upstream requires continuing capital deployment to maintain its productive base. The size and profitability of that requirement vary enormously, but depletion does not disappear when oil prices rise.
So this is where Hess is particularly useful because its history prevents the evidence from collapsing into an argument against upstream reinvestment. Elliott Management (currently sparring with the board of bp) launched a campaign against Hess in 2013 centred on governance, operating performance, portfolio construction and capital allocation. One of its most eye-catching observations was the relationship between Hess's equity value and its annual investment programme. The company had a market capitalisation of roughly $24 billion and was spending around $6 billion a year on capital expenditure. Seen from the shareholder's perspective, management was effectively being entrusted every year with capital equivalent to around a quarter of the value of the company. Hess subsequently transformed itself, disposing of downstream operations and other assets, increasing distributions and becoming a much more focused E&P. Elliott was right to ask whether such enormous reinvestment could consistently earn an adequate return, but the subsequent history also demonstrates why the answer cannot simply be to stop taking geological risk.
Hess already owned 30% of the Stabroek Block in Guyana before the Liza discovery in 2015. What ExxonMobil, Hess and their partners subsequently found became one of the defining upstream discoveries of the era, with enormous resources, highly productive reservoirs, competitive development economics and decades of potential investment.
Guyana ultimately became central to Chevron's acquisition of Hess. Had Hess treated disappointing historical exploration returns as evidence that exploration itself should be eliminated, it could have surrendered the option that eventually transformed the company.
PetroEquity Signal has argued recently that high-impact exploration is more relevant to today's upstream industry than much of the recent discussion around capital discipline and shareholder returns implies. Guyana is an extreme illustration of that argument. The comparison with Marathon, Murphy and Phillips 66 therefore has to distinguish between capital deployed simply because an upstream organisation needs more barrels and capital invested in opportunities capable of materially extending resource duration or creating an entirely new source of value. Both appear in the accounts as upstream capital expenditure, but their economics can be radically different.
Resource duration sits naturally inside that distinction. An upstream company with a large inventory of already discovered, low-cost, long-duration resources has solved a significant portion of its future replacement problem. Capital can be deployed against known barrels with relatively high confidence rather than repeatedly returning to exploration, acreage markets or corporate M&A merely to sustain production. Guyana is an extreme example. A shorter-duration company can report excellent current cash flow while facing a very different terminal-value problem. Its existing assets may generate large amounts of money, but the corporation cannot maintain itself indefinitely without finding somewhere to reinvest. The quality of that reinvestment then becomes at least as important as the cash generated by the legacy assets.
The activist history running through these companies reinforces the role of capital allocation without proving that activists possess superior foresight. Third Point was involved around the Murphy separation. Elliott challenged Hess, later appeared at Marathon Petroleum after MPC had already been liberated from Marathon Oil, and subsequently challenged Phillips 66. The pattern is almost recursive. A large integrated company is separated because the constituent businesses have different capital requirements and investor bases. The resulting company grows, acquires assets and develops its own internal complexity. An activist eventually argues that capital allocation has again become obscured and that another set of assets would be worth more under different ownership or governance. Marathon Petroleum's later sale of Speedway for $21 billion illustrates how the process can continue long after the original upstream/downstream split. Phillips 66 has faced similar debate around its own portfolio.
Those campaigns should be treated as hypotheses rather than verdicts. Separation can destroy genuine synergies, create tax leakage, strand corporate costs and weaken financing capacity. A diversified portfolio can provide resilience when margins collapse in one part of the value chain, while an integrated trading and logistics system can capture commercial value that is difficult to reproduce through contracts between independent companies. An activist can correctly identify weak capital allocation while recommending a transaction that gives away valuable optionality. Hess is a useful warning against treating simplification as inherently virtuous. The stronger conclusion is that management teams deploying large amounts of shareholder capital deserve to be judged on the returns produced by that capital rather than on the strategic language used to defend retaining it.
That question has become unusually current because Elliott is now involved with BP. Its intervention has included pressure for clearer accountability between upstream and downstream, lower spending and stronger free cash flow. BP has since reorganised itself around two principal operating segments, Upstream and Downstream, although this remains an organisational structure rather than a corporate separation. The company continues to emphasise the advantages of integration, particularly the interaction between production, refining, marketing, trading and optimisation. BP may be right. Exxon and Chevron can make similar arguments, and the extraordinary scale of some integrated systems can create resilience and commercial advantages that would be difficult for smaller standalone companies to reproduce. The historical evidence from Marathon, Murphy and ConocoPhillips does not demonstrate that integrated oil companies should be broken apart.
It does impose a harder test on the case for integration. If a corporate structure allows cash produced by one business to be reinvested in another, shareholders need to know whether the recipient business is earning an adequate return on that money. Strategic importance is not enough. Neither is reserve replacement by itself. A dollar used to acquire another barrel has an opportunity cost: it could have reduced debt, repurchased undervalued shares, funded a dividend or simply remained in the hands of the investor. The existence of depletion makes upstream reinvestment necessary if a company intends to preserve itself indefinitely, but it does not make every reinvestment economically attractive. The better the resource base and the longer its duration, the easier that argument becomes. The shorter the duration and the more dependent the company is on continual acquisition or exploration, the greater the burden on management to show that replacing what has been produced actually creates value per share.
The US separations of 2011-13 therefore raise a different question from the one the industry has traditionally asked. Marathon Petroleum, Murphy USA and Phillips 66 did not suddenly discover better geology after becoming independent. They gained their own boards, their own balance sheets, their own investor bases and direct control over the cash their operations produced. In each case management had to decide explicitly whether another dollar belonged in the business or with shareholders. The upstream companies continued to operate in a world where maintaining production itself created a recurring reason to retain capital. Some of that capital undoubtedly created value, and Hess demonstrates how spectacular the payoff can be when it finds a resource such as Guyana. Much of the industry's experience since 2010 suggests that other reinvestment did not.
The cyclical question remains unresolved. Starting the comparison around 2011 may flatter the downstream securities and penalise upstream by capturing an unusually poor period for exploration and production equities. Running the same experiment from an earlier starting point could produce a different hierarchy. That test is worth doing precisely because the initial result is so striking. If upstream dominates during the commodity supercycle but downstream dominates after 2011, the analysis becomes partly about cycle. If the downstream businesses still compound unusually well across different starting points, the structural explanation becomes harder to dismiss.
For decades, describing upstream as the cash engine of an integrated oil company was largely uncontroversial. It may still be the correct description of where the economic rent originates. The experience of Marathon, Murphy and ConocoPhillips suggests that it is not enough to explain where shareholder wealth ultimately accumulates. A business can generate enormous cash and still consume enormous capital. Another can produce less dramatic operating economics while turning a much larger proportion of its cash into dividends, repurchases and per-share growth. The distinction is not really between glamorous upstream assets and mundane downstream ones. It is between cash that arrives and cash that can genuinely leave, between reinvestment that creates new economic value and reinvestment required merely to replace what has disappeared.
The original Marathon documents came surprisingly close to describing the problem before the experiment had even begun. Upstream required continuous capital deployment to maintain production and growth, while downstream investments were competing against those demands inside the same company. The subsequent fifteen years showed what happened once that competition was removed. Murphy and ConocoPhillips produced similar evidence, while Hess demonstrated why the correct response cannot simply be to starve upstream of capital. BP now brings the same questions back into a major integrated company with an activist again asking how capital, accountability and corporate structure interact.
The industry has spent decades asking where the cash engine sits. The historical record suggests a second question deserves at least as much attention: after the cash has been generated, who gets to decide what happens to it, and how good have they actually been at making that decision?