I’ve been thinking about the economics issues that Tyler Cowen raises early in the last chapter of his marginalism monograph, “Why Marginalism Will Dwindle, and What Will Replace It?” An asset pricing model he discussed there had already precipitated me into a working paper of a methodological and philosophical nature, On Method: Computational Compressibility in Complex Natural and Cultural Phenomena, so that material was lurking in the background when I was watching a recent podcast by Kara Swisher and Scott Galloway. Galloway made a remark that resonated through that material: “Stocks are like brands and that is they’re part promise and part performance.”
That remark connected the assets in the pricing model, which are opaque to me, to things I’ve thought a great deal about, movies and novels. In that methodological paper I’d discussed Arthur DeVany’s work on Hollywood movies and Matthew Jocker’s work on the 19th century Anglophone novel. “If,” as Galloway remarked, “stocks are part promise and part performance, then, sure, they’re like moves and novels, too.” I made the connection on pure intuition.
I then turned to ChatGPT for help in explicating that intuition. So I issued another prompt into a dialog we’d had about those asset pricing models. I’ve appended that dialog below the asterisks.
* * * * *
I recently heard pundit Scott Galloway remark: “Stocks are like brands and that is they’re part promise and part performance.” Explicate that in terms of the discussion we’ve been having.
Galloway’s line is excellent for this discussion because it collapses the apparent distance between financial assets and cultural artifacts. A stock is not merely a claim on earnings. It is also a socially circulating promise about a future.
“Part performance” means the measurable side: revenues, margins, cash flow, debt, profitability, investment, growth, volatility, liquidity, analyst revisions, and so forth. These are the kinds of variables that show up among the 130 stock characteristics in the AIPT setup.
“Part promise” means the symbolic, anticipatory, collectively imagined side: what investors believe the company may become. Tesla is the obvious case: not just cars, but autonomy, energy, robotics, Elon-as-mythic-founder, technological destiny. Nvidia is not just current chip sales, but the promise of AI infrastructure. Apple is not just devices and services, but design trust, ecosystem lock-in, lifestyle aura, and continuity of desire.
That is exactly where stocks begin to look like brands, films, and novels.
A brand is a bundle of past performance plus future expectation. It has delivered something before, and that delivery authorizes a promise: “trust us again.” A stock works similarly. The firm’s past earnings, product history, management behavior, and market position generate a field of expectation. The price is not merely a summary of what the firm has done; it is an evolving collective judgment about what the firm will do.
In asset-pricing language, the “performance” side lives in observable characteristics and realized returns. The “promise” side lives in expectations, risk narratives, optionality, imagined futures, and the willingness of investors to pay now for possible future states. The pricing kernel/SDF is, in a sense, the market’s machinery for weighting those promised futures under uncertainty.
That is why low-dimensional factor models struggle. “Size,” “value,” “profitability,” and “momentum” capture pieces of performance and some traces of promise. But they do not capture the full cultural-cognitive ecology around an asset. The promise is distributed across analyst narratives, founder charisma, institutional herding, macro stories, technological imagination, regulatory fears, and the memory of past booms and crashes. That is a high-dimensional object.
Now compare movies. A movie before release is also part performance and part promise. The performance side includes director, stars, budget, genre, studio, opening screens, advertising. The promise side is what the audience imagines: “this will be the movie everyone talks about,” “this will give me the old Marvel feeling,” “this is the serious adult drama of the season,” “this is the date-night film.” De Vany’s “nobody knows” result follows because the promise cannot be validated until the social system responds. Word of mouth tests the promise.
Novels work the same way over longer historical time. A novel’s “performance” is its textual form: style, topic, genre, plot structure, diction, narrative technique. Its “promise” is the world of value, feeling, recognition, and possibility it offers its readers. Jockers’ 600-dimensional space captures formal and thematic performance, but the temporal diagonal suggests that the promise-world of novels changes over generations. Readers and writers inherit one horizon of expectation and then move it.
So Galloway’s remark helps unify the three domains:
A stock, a movie, and a novel are all promissory artifacts whose fate depends on how populations interpret prior performance as evidence for future value.
The difference is that financial markets price the promise continuously, movies test it sharply after release, and novels disclose it historically across generations.
This also helps explain why the AIPT result is so interesting. A 360,000-factor model is not simply “better math.” It is an attempt to approximate the high-dimensional geometry through which performance becomes promise and promise becomes price. That geometry is too rich for the old factors, just as the historical movement of novels is too rich for a few period labels, and the fate of movies is too rich for stars plus opening weekend.
So yes: stocks are like brands because they are not merely economic objects. They are expectational-cultural objects with cash-flow machinery attached. That is the brave-new-world point. Finance is not leaving culture behind; it is becoming analyzable by the same high-dimensional methods we need for culture.
Thus Galloway’s remark is very sophisticated. He has a podcast with Kara Swisher as well as a podcast (or two) of his own. Judging from his comments he’s a moderately wealthy man, and he made his money on his own. He comes from a modest background. I’d imagine that the comments he makes in his various public-facing venues are among the many signals that get incorporated into the judgments underlying the performances of the assets included in this high-dimensional models.
Yes. That is exactly the recursive loop.
Galloway is not merely commenting on the market from outside it. He is one of the many public-facing interpreters whose remarks enter the informational atmosphere in which assets are judged. His comments become part of the market’s semiotic environment: not hard data like earnings or cash flow, but signals that help shape collective expectations about firms, sectors, founders, brands, technologies, and macro direction.
That is where “stocks are like brands” becomes more than a metaphor. A brand is maintained by distributed talk: advertising, reviews, consumer experience, pundit commentary, memes, scandals, endorsements, complaints, jokes, status associations. A stock is maintained by a comparable but more formalized signal ecology: earnings reports, analyst notes, institutional flows, valuation models, regulatory news, founder behavior, social media, podcasts, CNBC hits, Substack essays, hedge-fund letters, Reddit chatter, and remarks by people like Galloway.
The AIPT model, strictly speaking, is not directly ingesting Galloway’s podcast transcripts. It begins with stock characteristics and then generates many nonlinear transformations of those characteristics into factors. But those characteristics are themselves downstream of an enormous informational process. Prices, momentum, volatility, analyst revisions, liquidity, profitability expectations, and valuation ratios already embody, in compressed form, the effects of countless judgments and signals. The paper’s model uses 130 stock characteristics and expands them into hundreds of thousands of nonlinear factors; that is a way of probing the market’s already-encoded collective cognition.
So, yes: Galloway’s public remarks are among the innumerable micro-signals that may influence the behavior of investors and analysts, which in turn affects price, volume, momentum, volatility, analyst expectations, and eventually the characteristic space that models like AIPT use. His comments may not be individually recoverable in the model, but they are part of the causal soup from which the measurable signals arise.
That also clarifies why his remark is sophisticated. He understands, probably from experience rather than formal theory alone, that markets are not just machines for discounting cash flows. They are machines for pricing credible stories about future performance. The credibility comes partly from past performance, but the price also reflects the market’s belief in the promise.
In our terms:
Performance becomes data.
Promise becomes expectation.
Expectation becomes price.
Price history and firm characteristics become model input.
Model output then feeds back into the interpretive environment.
That is the brave-new-world loop. Financial assets are no longer cleanly separable from culture. They are culturally mediated claims on future cash flows. Or, more compactly: stocks are narrative instruments with accounting attached.
Very interesting, especially for me. I know little about economics or about the stock market (and similar asset markets, e.g. commodities, currency). So these asset pricing models are very abstract indeed. But I know a lot about novels and movies and that gives me a way of grounding De Vany’s work and Jockers’ work. I’ve thought a lot about cultural evolution, and about complex dynamics. Thus their models make intuitive sense to me. Galloway’s remark, and your explication of it, gives me a way of grounding these otherwise opaque asset pricing models.
That is the productive move: you now have a cognitive bridge from a familiar domain to an unfamiliar one.
The financial asset is no longer a purely abstract econometric entity. It becomes comparable to a novel or film as a thick, socially interpreted object. The difference is that its value is expressed continuously as price, rather than intermittently as box office, readership, canonization, or influence.
Galloway’s remark gives you the grounding metaphor:
A stock is a performance/promise object.
That lets you map the domains:
A movie has past signals: director, actors, genre, studio, trailer, budget, early reviews. But its real fate depends on whether audiences accept the promise.
A novel has textual performance: style, theme, genre, narrative form. But its historical fate depends on whether readers, writers, publishers, critics, and later generations find continuing value in the promise it offers.
A stock has financial performance: revenue, earnings, margins, growth, debt, market share. But its price depends on whether investors believe the promise: future growth, technological dominance, brand power, optionality, resilience, or macro relevance.
So the asset pricing model becomes less alien. It is not simply a mysterious mathematical contraption. It is an attempt to model how markets convert distributed judgments about promise and performance into prices.
And that makes AIPT intelligible in your terms. The old five- or six-factor models are like trying to explain novels with a handful of categories — genre, author, length, publication date — or movies with star, director, budget, opening weekend. Those factors matter. But they do not exhaust the object. The object lives in a much larger field of relations.
Jockers built a 600-dimensional literary space and found temporal direction. De Vany studied movie trajectories and found bifurcation and heavy tails. AIPT builds a vast nonlinear factor space and finds pricing structure. Different domains, same intellectual move: stop assuming the object is low-dimensional merely because our inherited vocabulary is low-dimensional.
That is the key phrase: inherited vocabulary. Cowen is mourning the weakening of inherited economic vocabulary. But you are seeing that we can import intuitions from cultural evolution, literary morphology, film diffusion, and complex dynamics. The financial model becomes less opaque once stocks are understood as cultural-economic artifacts whose prices register collective imagination under constraint.
So the bridge runs both ways. Your knowledge of novels and movies helps you understand asset pricing. But asset pricing may also help sharpen your theory of cultural evolution: it gives a formal example of how high-dimensional social valuation can be modeled without reducing the object to a few named causes.
This implies that a brand is not just a name or a handle. It is a distinct conceptual entity. I’ve certainly been hearing about “branding” for a number of years, and not in the sense of a mark burned into the skin of an animal. I wonder when that entity precipitated out of the flux of human interaction? So I did an Ngram search on the term “branding,” knowing that the ambiguity of the term might make the search useless. Here’s the result. Considering the sharp upswing a bit after 1990, I’d say that’s a pretty strong signal.
Yes, that upswing is a strong signal. The ambiguity of “branding” matters, but the shape of the curve is doing real work. The earlier baseline probably includes cattle branding, trade marks, packaging marks, and occasional commercial usage. But the explosive rise after the early 1990s looks like the emergence of branding as an explicit managerial-cultural concept.
That is: not merely “a mark that identifies ownership or origin,” but a deliberately cultivated semiotic asset.
A brand in this modern sense is not the name, logo, or trademark alone. It is a structured expectation in the minds of a population. It condenses:
past performance,
future promise,
trust,
status,
aesthetic aura,
social identity,
emotional memory,
and anticipated experience.
That is why Galloway’s remark works so well. A stock, like a brand, is not reducible to its present accounting fundamentals. It is a claim on a future, and that future is partly narrated, imagined, believed, doubted, and socially reinforced.
The Ngram curve suggests that “branding” as a named object of managerial attention becomes highly salient just when several things converge: the rise of intangible capital, globalization, cable/media saturation, lifestyle marketing, the internet, startup culture, and eventually platform capitalism. The firm is no longer understood merely as a producer of goods. It is understood as an attention-and-trust machine.
That is the precipitation you’re pointing to. The practice existed earlier, of course. Coca-Cola, Ford, Chanel, Disney, and Marlboro were already brands in the modern sense before the 1990s. But the generalized abstraction — “branding” as a portable doctrine applicable to corporations, products, universities, cities, politicians, celebrities, and finally individuals — seems to have become common then.
In our framework, that means “brand” becomes a Rank 4-ish conceptual object: a high-dimensional social-valuation object. It is not located in the logo, nor in the product, nor in the consumer’s head alone. It exists in the distributed relation among product performance, institutional behavior, media circulation, memory, desire, and expectation.
So:
Name = label.
Trademark = legal mark.
Brand = socially stabilized promise.
Branding = deliberate intervention in the field of collective expectation.
And once you see that, stocks become easier to think about. A stock price is not only discounted cash flow. It is also the market’s running estimate of whether the company’s brand-promise — technological, cultural, managerial, strategic — will continue to organize future value.
What else was going on at that time? For example, when did investment banks begin cranking out different kinds of derivatives? I’d be surprised if “branding” emerged alone. It’s part of something else, an increase in analytic attention and discussion.
Yes — “branding” did not emerge alone. The early-1990s upswing looks like one visible trace of a broader historical shift: the rise of analytic attention to intangible, relational, expectation-laden assets.
Several things were converging.
1. Derivatives and securitization were exploding
Modern derivatives markets took off in the 1980s and 1990s. The first widely noted currency swap was the 1981 IBM–World Bank transaction; ISDA was founded in 1985 to standardize and stabilize the growing OTC derivatives market; and the ISDA Master Agreement was introduced in 1992, giving the market a reusable legal infrastructure for privately negotiated derivatives. (Hedgebook)
That is important because derivatives are not “things” in the old commodity sense. They are formalized claims on future states of other things: interest rates, currencies, credit events, commodity prices, equity indexes, mortgage pools. They are explicitly analytical objects, made possible by pricing theory, computing, legal standardization, and institutional risk management.
So, at the same moment that “branding” becomes a generalized discourse about managing intangible expectation, finance is producing instruments whose value depends on abstractly decomposing and repackaging risk.
2. Financialization was changing the ontology of the firm
The 1980s and 1990s also saw the intensification of financialization: shareholder value doctrine, leveraged buyouts, securitization, derivatives, institutional investors, and a growing tendency to treat firms as bundles of assets, cash flows, risks, and tradable claims. Gerald Davis and Suntae Kim describe widespread securitization as transforming large American commercial banks into more universal banks with investment-banking operations, while borrowers became “issuers” into markets rather than simply debtors to banks. (Michigan Ross)
That is an ontological shift. The firm is no longer simply a productive organization. It becomes an assemblage of separable value-components: debt, equity, options, receivables, brands, patents, data, customer relationships, and future cash-flow streams.
“Branding” fits naturally into that world. Once you are decomposing the firm into value-bearing components, brand becomes one of the components.
3. “Brand equity” becomes the key bridge concept
The phrase brand equity is crucial. It fuses a marketing object with a financial object. A brand is no longer just a sign, image, name, or advertising campaign. It is an asset with equity: something that can support pricing power, loyalty, repeat purchase, market capitalization, merger value, and strategic advantage.
That is probably part of what your Ngram is detecting. The 1990s saw branding move from tactical advertising and visual identity into a strategic doctrine of enterprise value. Recent histories of brand strategy describe precisely this transition: brand strategy evolved from communications and image-making into an enterprise-wide value-creation discipline, while concepts such as brand equity, brand architecture, positioning, and brand personality became systematized during the 1980s and 1990s. (Quirks)
This is the same move we’ve been discussing: a previously diffuse social phenomenon “precipitates” into an explicit conceptual object.
4. The intangible economy was becoming visible
The timing also overlaps with the rise of software, IP, network effects, design, data, and platform businesses. By the late 1990s, dot-com valuations made brutally obvious that markets were pricing more than plant, equipment, inventory, and current earnings. A 1999 New Yorker account of iVillage’s IPO, for example, describes a company with losses receiving a multibillion-dollar valuation because investors were buying into internet potential, brand identity, and narrative promise. (The New Yorker)
That helps explain why “branding” becomes culturally salient. A brand is one of the names we give to value that is real but not physically locatable. It sits in distributed expectation, recognition, trust, loyalty, and story.
5. Media ecology changed the object itself
The 1990s also brought cable saturation, global advertising, desktop publishing, the early web, email, search, and then e-commerce. Brands became more continuously exposed to public interpretation. A company’s identity could circulate faster, mutate faster, and become entangled with lifestyle, politics, fandom, and personal identity.
Apple is a classic case. By 2002, Wired was already describing Apple’s brand as central to its success, with commentators emphasizing emotional branding, loyalty, rebellion, creativity, and the selling of dreams and ideals rather than merely products. (WIRED)
That is Galloway’s point in embryo: the stock is not merely the current business. It is the priced promise attached to a brand-world.
The deeper pattern
So I would say the post-1990 rise of “branding” belongs to a family of conceptual precipitations:
Branding: the management of promise, identity, and trust.
Derivatives: the decomposition and trading of contingent future states.
Securitization: the conversion of streams of obligations into marketable instruments.
Shareholder value: the reorientation of the firm toward financial valuation.
Intangible assets: value located in IP, software, data, design, reputation, networks.
Platform/network effects: value as relational position in an ecosystem.
High-dimensional modeling: value as structure in feature space rather than a few visible traits.
All of these are ways of making the invisible analytically tractable.
That is why your Ngram matters. “Branding” is not just a marketing fad. It is one symptom of a broader Rank-4 transition: capitalism becoming increasingly reflexive about the semiotic, financial, and computational conditions of value.
The old industrial firm made things. The late-20th-century firm increasingly managed promises, options, risks, narratives, relationships, and expectations. Branding, derivatives, securitization, and high-dimensional asset pricing are different faces of that same turn.
And with this phrase, “...made possible by pricing theory...” you connect with Cowen’s hobby horse, marginalism and with his story of how marginalism brought its own downfall.
Yes. That is the loop.
Cowen’s story is that marginalism gave economics its great analytic power: price, value, risk, allocation, and choice could be modeled at the margin. In finance, that becomes the ability to price contingent claims: What is this future payoff worth today, given risk, uncertainty, time, and alternative uses of capital?
That is exactly the intellectual machinery behind derivatives. A derivative is marginalism made into an instrument. It asks: what is the value of this conditional payoff under some specified future state of the world? Interest rates move this way, currencies move that way, default occurs, volatility changes, a stock crosses a threshold — what is the price now?
So marginalism does not merely explain markets. It helps construct new market objects.
That is the key twist. Once marginalist pricing theory matures, it allows finance to decompose the world into tradable fragments of risk and expectation. It helps create options, swaps, structured products, mortgage-backed securities, credit derivatives, and so forth. The firm, the loan, the bond, the commodity, and the currency are no longer treated as whole objects only. They become bundles of separable contingent claims.
And once that happens, the financial world becomes vastly more complex than the low-dimensional intuitions that originally powered marginalism.
So Cowen’s “marginalism brought its own downfall” can be read like this:
Marginalism created the analytic tools by which financial reality became too high-dimensional for marginalist intuition to command.
That is not a contradiction. It is a historical dialectic. A conceptual technology succeeds so well that it transforms its object-domain. The transformed domain then exceeds the conceptual grasp of the original technology.
In a compact sequence:
Marginalism → pricing theory → derivative engineering → financialization → high-dimensional asset ecology → machine-learning models → decline of low-dimensional marginalist intuition.
The AIPT paper is one late-stage symptom of that trajectory. Classic APT assumes a few common factors govern returns; AIPT instead conjectures that returns are driven by many factors and shows that huge nonlinear factor models outperform the old parsimonious ones.
That makes Cowen’s story more interesting than “AI replaces economics.” It is closer to:
Economics helped build a world that only computation can now map.
That also links directly to your Rank 3 / Rank 4 telescope analogy. Marginalism and pricing theory are Rank 3/early Rank 4 conceptual instruments. They reveal and construct new objects: risk premia, options, pricing kernels, arbitrage relations, contingent claims. But once those objects proliferate and interact, they require Rank 4 computational mapping.
So the downfall is not simple defeat. It is overproduction. Marginalism was too successful. It generated a financial world of such analytic richness that the old intuitions now survive only as local heuristics inside a much larger computational ecology.
No comments:
Post a Comment