The Rhetoric of Economics

McCloskey’s trilogy on the rhetoric of economics — “If You’re So Smart,” “The Rhetoric of Economics,” and “The Vices of Economists” — treats economics as a subject for literary criticism throughout. I really would not recommend the first volume. Its fragmentary, notebook-like criticism quickly becomes tiresome, and the ornate language cannot hide how weak the overall structure is. The second book is much better. This post is a summary of that second volume.

Rhetoric, put simply, is persuasion with a purpose. In ancient Greece, rhetoric was framed around argument, structure, and style. In contemporary economics, that repertoire becomes richer: facts, stories, logic, and metaphor. Unlike abstract natural language built only to describe, a social science essay is really “writing with intent,” and that gives economics a distinctly human, rhetorical character.

In this post, “the author” always refers to the author of The Rhetoric of Economics, Deirdre N. McCloskey.

A table for the structure of rhetorical analysis:

Four rhetorical angles: fact, story, logic, metaphor.

What is rhetoric, and how do you do it?

The anonymous voice of authority.

Economics is not the voice of God. It is people using language to persuade other people to believe a certain claim. Yet although it is human language, it still has to pretend that it speaks for the laws of history, anonymously casting itself as science itself.

No need to applaud me, audience. Today, history is the one speaking.

That is why economics so often describes factory production and profit formulas in an eternal present tense, even though these forms of organization really ought to be analyzed within a historical setting.

Techniques of creation.

First, the poet and metaphor.

Extend game theory to nations, and the market becomes an arena of competition. Economists then address their own country as if saying, “We are going to lose the game because we have problem X.” Metaphor makes the model feel more necessary, and more persuasive. Networks become webs, families become small firms, children become durable goods…

Second, the novelist.

Economics needs to learn how to build stories.

X was poor at first, but then embraced capitalism, and so X became rich. Building a story means arranging related events through descriptions of states, actions, and internal connections: causation, temporal sequence, and both positive and negative turns.

Each ending of a story marks a new state. In economics, a perfect or imperfect ending becomes equilibrium or disequilibrium. A good story needs a curtain call, not just a summary of the past, but a forecast of the ending state. The line can always be extended further.

A text without action or state is not a story.

Economists place so much emphasis on beginning with exogenous variables because, within a story system, that works as a beginning. If you start with endogenous variables, the story feels as if it begins halfway through.

That also reflects a more speculative concern. If you pin all of history’s bad debts on the most obvious character, the ending is not very illuminating.

Performance.

A novel does not imitate reality. A novel creates reality. The economics papers that leave the deepest mark are often the ones written in difficult, knotted sentences. Keynes’s The General Theory of Employment, Interest and Money is an example. When writing a paper, try to create both the “authorial audience” (which knows the story is fictional) and the “narrative audience” (which does not know it is fictional)¹.

Economists themselves need to realize that they are not always stating facts. Sometimes they are creating an audience.

Every article contains an implied author, an implied reader, a piece of history, and a form.

The literary side of economics: science, literature, art

The author’s goal in writing this book is to probe the boundaries between economics, science, and text. In her view, science is only one way of speaking, not a separate domain of truth.

• Science needs text in order to be expressed.
• Scientific resources for experimentation are limited, so people often judge the value of testing a theory by the beauty of the theory itself.
• Text persuades in ways that are not so different from literature and public speaking.

So science, literature, and art overlap rather than standing in simple opposition, as they often do in more traditional ways of talking.

I still have reservations about that myself.

Economics relies on the following methods of argument:

The first three come from British scientific argument. From a literary angle, the author expands that list much further.

Statistical tests, analysis of specific markets, experiments on theory, introspection (if X happened, what would I do?), thought experiments (based on experience, what would someone else do in situation X?), case studies, folk wisdom (merchants), the traditional knowledge of the academy, the symmetry of laws (for example, defining both a glorious age and a garbage age), pure definition, analogy…

For example, an economics paper cannot simply describe the motion of objects. It has to portray human behavior, and that has something in common with literature. Even when the author is portraying a type of person rather than a particular person, what the reader imagines while reading is still a concrete human being, not an abstract movement of objects. That kind of characterization overlaps with literature.

Naming is part of rhetoric, and it is also part of shaping an image. The word “preference,” for example, carries a strong literary sense of desire. Most economic terms have a similar shade to them, Hayek’s endogenous order, Adam Smith’s animal spirits…

Rhetorical devices in economics

Take Samuelson’s use of mathematics as an example. When he uses statistics and mathematical models, he says we; when he discusses economic propositions, he says I.

Because Samuelson says in the preface that “economic propositions are open to discussion and open to refutation,” “I” better shows a posture of discussion, while “we” conveys authority (as if what is being said here is already consensus).

If we actually ran an experiment, we would find that most people do not care about mathematical proofs or statistical detail, but the mere presence of mathematics strengthens persuasion.

John Stuart Mill’s paper on expectations stresses the position of the economics professor: criticizing people’s sluggish response to market expectations from the standpoint of an economic researcher.

The author’s criticism of economics as prediction is interesting. Economic predictions are always conditional predictions, and conditional predictions are always consistent with experience in what they mean. So how could they ever produce something genuinely beyond expectation?

I’ve heard similar views from others too. Econometrics identifies causality through exogeneity, so how could we expect econometrics to predict exogenous events?

Fogel uses a strategy of sacrificing the small to attack the large, because he is rebutting the mainstream conclusion that “railroad construction promoted American economic growth.” For x, y, and the coefficient, maybe y is small; maybe x is large; maybe we overestimated the coefficient, and that is what produced the conclusion we have today. His paper uses those hints of uncertainty to cast doubt on the theory.

Fogel

Fogel was the first to compare rail transport with river transport when estimating the railroad’s contribution. He used quantitative historical calculation and counterfactual reasoning, which later helped inspire counterfactual causal inference. His point was that earlier views had overstated the railroad’s contribution.

After his report, economics also began using quantitative history to reshape the paradigm of economic history.

Dave Donaldson was inspired by this work and measured market access through railway expansion. He then built a general equilibrium framework around railways and market accessibility, making comparative advantage empirically measurable, and won the 2017 Clark Medal.

Coase²’s paper speaks in the voice of a lawyer, laying out theoretical possibilities and then rebutting them one by one. It reads like the structure of ancient Greek oratory: introduction, narration (facts), classified exposition (disputed claims and undisputed claims), proof, refutation, conclusion. That is why the classic paper “The Nature of the Firm” is full of adversarial phrasing.

Other rhetorical techniques include appeals to authority, verbal suggestion (using language to imply a competitive relationship), metaphor³ (zero-sum games, children as durable goods), and parable (metaphor + story)…

Literary interpretation can really stretch very far…

Does quantification need rhetoric?

Two questions: how big is big? And so what?

On “how big is big”

Some people believe that once economics turns a problem into mathematics, purely mathematical thinking can separate it from the range of ordinary human judgment. But is that really true?

How big is big? How big counts as big, and how small counts as small, is still a process of trying to persuade other people. You still need people to compare and judge before reaching a conclusion. So the process still contains rhetoric.

Of course, if you have studied advanced econometrics⁴, you realize that the logic of hypothesis testing is to construct low-probability events, “Type I error” and “Type II error,” in order to test significance. That approach is widely treated as basic practice, but it is still only a relative standard.

I agree with Professor Zhentao Shi’s view of econometrics teaching: “How large is large enough to be credible” is not only a statistical question, but also a philosophical one. Advanced econometrics mainly forces people to start thinking about this.

For example, Bayesians and frequentists have argued over this for years, and Bayesians themselves split into subjective and objective camps. It all looks like mathematical deduction, but the construction of the basic unit of credibility never escapes a worldview. The old saying that everything can turn into philosophy really is true.

For example, the classic point is that statistical significance and economic significance should be distinguished.

On “so what?”

An economics paper has to speak to a particular audience, and the society it describes is itself a fictional construction. Different groups see the world differently. If you want their recognition, you have to respond to the part they already treat as “common sense.” Those are the topics more likely to draw attention.

What econometric work should watch out for

Distinguish statistical significance from economic significance.

Put simply, a paper’s argument cannot rely on statistical significance alone.

You need to spell out assumptions about the sample distribution, the real-world meaning and functional form of coefficients (for example, whether they are elasticities), analysis grounded in shared real-world understanding, a strict definition of significance, and a sufficiently natural transition from statistical significance to real-world impact. Statistical meaning, policy meaning, scientific meaning, and economic meaning are different things. (This part comes from records in the AER from the 1980s on mistaken uses of statistical significance.)

The author’s later attacks on the modernism represented by mathematical statistics go on and on, but in the end the principle is simple:

Many people treat falsification as science, but falsification also involves arbitrariness. Even a great deal of evidence cannot guarantee that everything has been covered. And when a theory is tested in economics, what that usually means is simply that many people are interested in it. So in the end, the author argues that any rule-bound methodology may be wrong⁵. We should therefore turn against anti-rhetoric and against modernism.

Could rhetoric become a discipline of the future?

Two trends in economics are obvious right now.

1. Basic mathematical statistics is increasingly prized. In my advanced econometrics course, I have seen students from the history department taking it as an elective, while older generations of economists keep moving further toward mathematics courses. The deeper crossover between math and economics is becoming normal. More scholars of mathematical economics keep entering the field, and the bar keeps rising.

2. On the other hand, once mathematical statistics is used widely enough, its legitimacy starts being questioned. The tool-like role of mathematics really does seem to be giving some ground back to rhetoric. A paper’s central idea, its story, and how illuminating it is are becoming the things people care more about.

This trade-off is as hard to choose between as the difference between “arriving at a conclusion correctly” and “arriving at the correct conclusion.” I genuinely do not know which direction will become the mainstream in the end.

If we examine science from the single standpoint of “persuasion,” then every methodological rule stops mattering. Under the author’s extremely loose definition, worldview may turn into a kind of function-oriented utilitarian thinking.

But rhetoric stands on the opposite side of modernism, and it unquestionably offers a fresh way to examine science. Innovation needs room for wild thinking, and a lack of restraint can provide exactly that springboard. That is why I think The Rhetoric of Economics is absolutely worth reading.