The Network of Global Corporate Control

Imagine opening the hood on the world economy and finding not a jumble of parts, but a machine with a shape. Not random, not purely national, but something global, organized, and—crucially—concentrated. That’s what Vitali, Glattfelder, and Battiston set out to see back in 2011. They weren’t satisfied with country-by-country snapshots of who owns whom. They wanted to map how corporate control actually flows worldwide because competition policy and financial stability don’t stop at borders. If there’s a hidden architecture, they argued, that’s where systemic risk and the real power to steer markets might concentrate. They went in with three possibilities on the table. Maybe firms are basically isolated, and control doesn’t travel far. Maybe they clump into separate blocs, a few regional clusters here and there. Or maybe there’s one giant connected component with a tight core and a sprawling periphery. To test that, they pulled from the Orbis two thousand seven database and identified forty-three thousand sixty transnational corporations using the Organisation for Economic Co-operation and Development definition. Then they followed the money. A recursive search traced all the ownership pathways that emanate from, and point back to, those firms. Out of that came a global ownership network with six hundred thousand five hundred eight nodes and one million six thousand nine hundred eighty-seven ownership ties. Not just the big listed companies. The web. Now, ownership is not the same as control. You can own a slice and still be along for the ride. So the authors formalized both ideas. They encode ownership in a matrix—call it W—where each entry says how big a stake owner i holds in firm j. That’s the direct piece. Indirect ownership rides along chains: if owner i owns part of firm j, and firm j owns part of firm l, then owner i has some effective stake in firm l. You can think of the simple, two-step case as the product of owner i’s stake in firm j and firm j’s stake in firm l, applied to firm l’s value. Then comes control. They link ownership to control with a familiar rule of thumb: if you clear a majority threshold in a firm, you effectively control it. That gives a direct control measure C i j. The big move is to let control propagate through the network. They define network control for each actor as the value you control directly—your stakes in other firms, multiplied by those firms’ economic value—plus the value you control indirectly via the network control of the firms you already influence. In plain language: it’s the sum of your immediate sway and the sway you gain because the companies you influence also influence others. They anchor all of this to operating revenue as the value proxy, so "how much you control" is expressed in terms of concrete economic activity. There’s a catch. Real corporate networks have cycles and cross-shareholdings—firm A owns part of firm B, firm B owns part of firm C, and firm C loops back into firm A. If you just keep summing influence around those loops, you can wildly inflate control. Vitali and colleagues acknowledge this and build on their earlier work to damp down those feedback effects in large, loopy graphs. They also stress-check their conclusions with three different models of how ownership translates to control, referred to as L M, T M, and R M. The headline patterns held across all three. So what does the global map look like? It’s a bow-tie. Not metaphorically—structurally. There’s a largest connected component that hoovers up most of the economic action, and within it, a small, dense core sitting at the knot of the bow. Around that are "IN" feeders streaming into the core, an "OUT" section that the core reaches out to, and a set of tubes and tendrils tying the pieces together. The largest connected component alone accounts for about ninety-four point two percent of total operating revenue among these transnational corporations. That’s the bulk of the global corporate economy, tied together. If you peer into the bow-tie, the core is startling. The strongly connected component—the bit where every firm can reach every other along some ownership path—has one thousand three hundred eighteen nodes. That includes two hundred ninety-five transnational corporations and one thousand twenty-three other participated companies, bound together by more than twelve thousand links. It’s not just connected; it’s thick. A typical core member is linked to about twenty other core members. And here’s the kicker: roughly three-quarters of the ownership of firms in the core is held by other firms in the core. That makes it a self-reinforcing cluster where influence circulates internally and compounds. All of that structure matters because of what flows through it: control. When the authors computed network control—this direct-plus-indirect sway over operating revenue—concentration jumped off the page. Only seven hundred thirty-seven top holders accumulate eighty percent of the total control. That works out to roughly six-tenths of one percent of all holders capturing four-fifths of the influence. Compare that with operating revenue itself, where you need about four point thirty-five percent of top revenue generators to hit the same eighty percent threshold. Control is more concentrated than revenue. Much more. And the core isn’t just dense; it’s dominant. Vitali and colleagues identify a group of one hundred forty-seven transnational corporations within that core that collectively hold nearly forty percent of total network control over global transnational corporation value. Nearly four out of every ten dollars of influence sit in a tiny, tightly connected cluster. Inside that cluster, internal control is almost complete—core firms largely own each other, reinforcing the group’s cohesion. That’s why the authors, a bit provocatively, describe it as a "super-entity." Who are these core players? In functional terms, they’re largely finance. Roughly three-quarters of the core are financial intermediaries—banks, asset managers, insurance firms. That composition helps explain the dense cross-ownership and the rapid propagation of influence inside the knot. Location within the bow-tie also shapes your odds of sitting among the top controllers. Pick a transnational corporation from the core at random, and there’s about a fifty-fifty chance it’s a top holder of control. Sample one from the "IN" feeder section and the chance drops to around six percent. Position matters. Let’s pause for what that means. The map isn’t just a curiosity; it’s a funnel. The "IN" feeds the core, the core radiates to the "OUT," and the tubes and tendrils stitch peripheral firms into the fabric. Because control compounds along direct and indirect paths, a densely connected core can punch far above its weight in sheer numbers of firms. And because so much of the core’s ownership is held internally, shocks or strategic moves can reverberate through that group quickly. The authors are cautious about the pitfalls. They show how naive control calculations can overestimate influence in the presence of cycles and cross-holdings, and they offer a methodology to bound that effect in large networks. They replicate the main findings across three ways of mapping ownership to control. They also discuss data limitations: this is Orbis two thousand seven, which means a particular snapshot in time, and it spans jurisdictions with different corporate governance and reporting regimes. Even so, they argue the key ownership relations are governed by investor protection levels in ways that justify cross-country comparisons, and the qualitative patterns—bow-tie topology, a dense core, heavy concentration—don’t hinge on a single modeling choice. If you’re wondering how unusual that concentration is, the authors put it bluntly: the distribution of control in this network is more unequal than typical wealth distributions in national economies. That’s not the same as saying these firms sit in a smoke-filled room and dictate outcomes. Control, as defined here, is a formal measure of how much operating revenue an actor could influence through ownership chains. But when only a microscopic fraction of holders command eighty percent of that influence, you’ve got a structure that’s primed for coordinated effects and sensitive to stress. This is where the implications come in. A small, tightly interconnected core of largely financial institutions isn’t just relevant for antitrust debates; it’s central to systemic risk. In a bow-tie, the knot ties everything together. That enables coordination and possibly efficiency, but it also creates channels for contagion. If a shock hits one core player, densely woven cross-holdings and near-complete internal control give it many avenues to travel. On the other side, a policy nudge or a strategic shift by a core actor can cascade outward because the "OUT" section—the periphery largely controlled by the core—covers the majority of operating revenue in the largest component. So what should we take away from the map itself, leaving policy to the policymakers? First, the global corporate network is not a loose patchwork. It’s a single, giant component with a clear bow-tie form and a very tight knot. Second, ownership is the substrate, but control is the currency. When you account for how control compounds along chains—measured against a real quantity like operating revenue—you see concentration that the raw ownership counts can hide. Third, where you sit in the structure matters. The core is a privileged neighborhood. There’s a temptation to leap straight to prescriptions. The authors mostly resist that and stick to the evidence. Still, the natural next step is surveillance of structure. If the core is the machine’s knot, then monitoring who sits in it, how intertwined they are, and how that composition shifts over time is essential. Watching concentration—like the tiny set of seven hundred thirty-seven holders with four-fifths of control—and tracking the core’s ties into the "OUT" can help anticipate where shocks might travel fastest. And because roughly three-quarters of that core are financial intermediaries, stress in finance plausibly reverberates faster and farther than stress elsewhere in the network. The broader point is simple. When we talk about corporate power, we often talk about size. Big firms. Big revenues. Vitali, Glattfelder, and Battiston remind us to talk about position. Where you are in the web—how many paths lead to you, and how many extend out from you—can matter more than how large you look in isolation. In a world shaped like a bow-tie, the knot commands the flow.