What is power, and which discipline sees it best?

Here is the awkward fact economics rarely says out loud: it does not have a word called “power” in its core vocabulary. It used to. The classical political economists—Smith, Ricardo, Marx—wrote about power constantly: bargaining strength, class, the landlord’s grip on rent. The marginalist revolution of the 1870s swept most of that language out and replaced it with marginal products, factor shares, and equilibrium prices. What was left of “power” got re-encoded into four pieces of apparatus, each of which captures something real and none of which uses the word.

Market structure. The first piece is the spectrum from perfect competition through monopoly. A monopolist sets price above marginal cost; the gap is the firm’s pricing power, and economics measures it precisely. The Lerner index reads the markup straight off the price; the Herfindahl-Hirschman index reads concentration off market shares. When an economist says “Amazon has market power,” this is the apparatus doing the talking—and De Loecker, Eeckhout, and Unger’s 2020 finding that aggregate US markups roughly tripled since 1980 is exactly this apparatus reporting that something large has changed.

Asymmetric information. The second piece is what happens when one party knows something the other does not. The insurer who cannot see the driver’s risk, the employer who cannot observe the worker’s effort, the seller of a used car who knows it is a lemon—each holds an informational advantage that translates into the ability to extract surplus. Economics calls this adverse selection, moral hazard, the principal-agent problem. It is power-by-asymmetry, theorized in full, and never once filed under “power.”

Mechanism design. The third piece runs the logic in reverse. If asymmetric information lets one party extract surplus, then designing the rules of the game—the auction, the voting procedure, the contract—is designing who can extract what from whom. The revelation principle tells you when you can build a mechanism that makes people reveal what they know truthfully. Designing an FCC spectrum auction is designing a power architecture; the designer simply has equations for it.

Monopsony. The fourth piece is the one economics rediscovered most recently, and the rediscovery matters. Joan Robinson named monopsony in 1933—a single buyer with power over sellers, the mirror image of monopoly. For decades the competitive-labor-market baseline assumed it away: workers paid their marginal product, full stop. Then Alan Manning’s Monopsony in Motion (2003), and the Azar-Marinescu-Steinbaum concentration work of 2020, put numbers on employer wage-setting power. Suddenly economics had a quantitative grip on something it had told itself for fifty years did not exist.

Two strands of the history-of-economic-thought run straight through this apparatus. The institutional lineage—Veblen’s status competition through Coase, Williamson, North, and Acemoglu-Robinson—is the closest thing economics has to a continuous tradition that kept the word “power” in view; it lives in History of Economic Thought, the Institutionalist tradition. The formal apparatus of asymmetric information and mechanism design that gave economics its rediscovered power vocabulary in mathematical form is the subject of the chapter on Information economics and the game-theory revolution.

See Economics Ch 6 §6.2: Monopoly (the market-power baseline) See Economics Ch 4 §4.6: Information Asymmetry (power by asymmetry) See Economics Ch 12 §12.1: Social Choice and the Revelation Principle (designing the rules) See Economics Ch 21 §21.5: Monopsony (employer wage-setting power)

There is no economics chapter to lean on here, and that absence is the point. Political science treats power as its fundamental unit—the noun the whole discipline is built around—and it has theorized that noun in four successive moves, each developed because the previous one left something unseen.

First face: Dahl. Power as observable, behavioral conflict. A wants outcome X, B resists, A prevails. You can watch it happen and, in principle, measure it. This is the rung economics already stands on: Dahl’s “A gets B to do what B would not otherwise do” is the same shape as bargaining power in a bilateral monopoly. Same precision, same demand that power leave a visible trace in someone’s revealed behavior. A wage negotiation where the employer extracts terms the worker would not otherwise accept is a first-face exercise of power, and an economist and a political scientist would describe it almost identically.

Second face: Bachrach and Baratz. The power to keep things off the table—what they called non-decisions. The city council that never votes on a question because the question never reaches the agenda is exercising power as real as any recorded vote, and it leaves no trace in the voting record. This is where economics starts to strain: mechanism design and the rent-seeking literature reach part of the way, because rigging the rules of the game is a recognizable economic act, but the agenda-setting that happens before any game is specified sits mostly outside what the apparatus tracks.

Third face: Lukes. The power to shape what B wants in the first place. If A can make B desire the outcome that serves A, there is no observable conflict at all. B complies willingly, and the first two faces, both of which require visible resistance, register nothing. Lukes’s example is the worker who never demands workplace-safety regulation because he has internalized individual responsibility as the natural frame. Economics has a translation ready, endogenous preferences, and it is worth seeing exactly why the translation falls short. Endogenous-preference models let preferences move; they do not theorize who shapes which preferences through what mechanism, and that “who-which-how” is the entire content of Lukes’s claim. The economist’s version keeps the math and discards the politics.

Fourth face: Foucault. Here the translation breaks down completely, and it is worth resisting the reflex to make it work. Foucault is not describing A dominating B at all. He is describing power as productive: the discursive field that constitutes A and B as the kinds of subjects who can stand in an A-over-B relation in the first place. Power, on this account, does not merely repress pre-existing desires; it produces the categories, the norms, the very objects we then argue about. When Timothy Mitchell traces how “the economy” became a bounded object that experts could manage, a thing that did not exist as a concept before the mid-twentieth century, he is doing fourth-face analysis: the formation of the field economics takes as given. There is no economics apparatus for this, because economics needs the field already formed in order to start.

The ladder is cumulative, not a sequence of replacements. The first face is still the right tool for an observable bargaining conflict; the fourth does not make it obsolete. What the ladder shows is that “power” is not one phenomenon economics happens to under-measure. It is at least four, and the upper two are not under-measured by economics—they are unaddressed.

Sociology’s distinctive move is to put power in the structure rather than in the person. Three traditions carry the weight, running from the most narrative to the most quantitative.

The power elite (Mills). Power as a function of position within interlocking institutional hierarchies. The chief executive of a major corporation, the four-star general, the cabinet secretary are not powerful because of personal traits; they are powerful because they occupy the commanding nodes of organizations, and those nodes connect. The operationalization is concrete—network analysis of corporate board interlocks, studies of elite circulation, the Domhoff tradition of mapping who sits on which boards. The revolving door between the Treasury, Goldman Sachs, and the Federal Reserve is a structural feature an economist can model as “regulatory capture”—but only after the sociologist has named the pattern that makes the capture legible.

Multi-capital (Bourdieu). Four forms of capital, economic, cultural, social, and symbolic, each convertible into the others through specific, often unnoticed mechanisms. Cultural capital is the load-bearing idea: the accent, the taste, the embodied ease in institutions that the right upbringing installs, and that converts into economic outcomes through hiring, networks, and the soft sorting of who “fits.” This is exactly the place to refuse the economics translation. Cultural capital is not human capital with a fashionable label. Human capital is productivity you carry; cultural capital is a signal of class position that pays off precisely because it is misrecognized as natural merit rather than inherited advantage. When elite-school graduates out-earn equally credentialed non-elite graduates, economics records the gap as an unexplained residual after controlling for skill. Bourdieu’s apparatus names the mechanism the residual is made of. That difference, residual versus mechanism, is the whole stage.

Class as control (Wright). The analytical-Marxism wing, and the one most legible to an economically trained reader. Erik Olin Wright’s Class Counts defines class position as control over productive assets—means of production, organizational assets, skill assets, labor power—and operationalizes it quantitatively, with survey data and a typology that survives statistical scrutiny. Wright’s famous case is the manager: economically a worker, selling labor for a wage, but occupying a contradictory class location because the job confers control over other workers. The category does work that occupation codes and income deciles cannot. Wright is the bridge: a sociologist whose apparatus an economist can engage directly, without first conceding the harder Bourdieusian claims.

Map the three apparatuses against each other and a pattern appears: three places they converge enough to be doing the same work, and three places economics has no analog at all.

Where they converge.

1. Monopsony. The industrial-organization economist using Manning and Azar-Marinescu-Steinbaum and the political scientist using a first-face account of employer dominance arrive at the same phenomenon from opposite directions, and their findings point the same way. Employers have wage-setting power; workers have less exit than the competitive model assumes; the wage sits below marginal product. The convergence is genuine—and it is worth noting it took economics until 2003 or so to walk back to common ground the structural account had occupied for decades.
2. Institutions as rules. The modern institutional-economics line—Coase to Williamson to North to Acemoglu-Robinson—does work that overlaps with both Mills’s positional sociology and the first-and-second-face political apparatus. “Inclusive versus extractive institutions” is a power-distribution claim with the word “power” swapped for “rules of the game.” When AJR say extractive institutions concentrate power and inclusive ones disperse it, a sociologist hears Mills and a political scientist hears the second face.
3. Mechanism design as institution design. Designing an auction, a matching market, or a voting rule under incentive constraints is designing a power architecture. The political theorist who wants a voting rule that aggregates preferences without manipulability is doing mechanism design whether or not they use the word—the Gibbard-Satterthwaite theorem is as much theirs as economics’s. This is the convergence economics is proudest of, and rightly.

Where economics has no analog.

1. Foucauldian productive power. The constitution of “the economy” as a manageable object, the formation of the subjectivities that economic governance then acts on—these are not questions economics can pose, even by stretching. The nearest analogs—Akerlof and Kranton’s identity economics, McCloskey’s rhetoric-of-economics, the social-norms branch of behavioral economics—study how preferences and identities get shaped. None theorizes power-knowledge as constitutive of the discursive field itself, which is the actual Foucauldian claim. This is the clean blind spot.
2. Bourdieu’s convertibility mechanism. Economics measures the residual; sociology names the mechanism that produces it. Closing this gap would require economics to adopt an apparatus for how cultural capital converts into economic outcomes through misrecognition—and that apparatus, built out honestly, would look like sociology, not economics.
3. Third-face preference-shaping at scale. Endogenous-preference economics moves preferences around inside an equilibrium; Lukes’s third face accounts for who shapes which preferences through which mechanism. The gap is methodological, not merely empirical—it is about what counts as a thing worth modeling, and economics’s answer leaves the shaping itself outside the frame.

Notice the asymmetry one more time, because it has been quietly structuring the whole walkthrough. Stages 1 and 4 link out to a history of economic thought; Stages 2 and 3 link almost nowhere, because there is no history-of-economic-thought chapter for the four faces or for Bourdieu. That is not an oversight in the cross-links. It is the finding: the lineages that map power best run outside the discipline whose apparatus this walkthrough started from.

See Economics Ch 18 §18.3: Institutions and Development — the AJR Framework See Economics Ch 21 §21.5: Monopsony (the convergence case) See Economics Ch 12 §12.1: The Revelation Principle (institution design as power design)