Even Heterodox Economics is Misguided

Earlier this week, Arnold Kling—who taught the first economics class I ever took—wrote a post comparing behaviorism in psychology to Samuelsonianism in economics. In his view, the chief failings of these approaches are an overemphasis on mechanistic models on the one hand, and a “blank slate” view of human nature on the other. For both reasons, he’s rereading Stephen Pinker, enemy of blank slate models and enthusiastic booster of computational, rather than mechanistic, models of cognition.

I argued that computational models are not different in kind from mechanistic ones, they are just more sophisticated. I pointed him towards works in the rhetoric of inquiry tradition, in particular Economics and Hermeneutics, an excellent collection edited by the late Don Lavoie and available for free online.

It seems that he’s now reading that very collection, which is great. However, I remember my own first encounter with this tradition of thought—through Deirdre McCloskey’s The Rhetoric of Economics—and I found it quite baffling at the time.

I think there are a lot of people educated in economics who sense that something is wrong in the house that Samuelson built. However, the tools that most heterodox schools have to offer—and here I include even most versions of the Austrian school—simply won’t help you see some of the fundamental errors of the mainstream view. At best the problems of the mainstream schools are replaced with more nuanced, subtle, and complex models that nevertheless share the same underlying characteristics.

This is the basis of McCloskey’s critique of the neo-institutionalists; they think they’ve moved beyond Samuelsonian limitations, when in fact they’ve simply subsumed the idea of an institution into a Samuelsonian framework. Thus instead of conveying meaning or providing frameworks of interpretation or shaping conjective reality, institutions are treated as structures of reward-and-punishment designed in just such a way to make utility maximizers cooperate with one another.

The writers in the Project On Rhetoric Of Inquiry, as well as hermeneutics and what Lavoie calls “the interpretative turn” in general, can be hard for outsiders of that tradition to get their head around. So this post will be my attempt to introduce one version of that line of thinking. My intended audience are primarily heterodox economists, but the argument will be relevant to economists and social scientists in general, and indeed any scientist or scholar.

It’s Time to Abandon Foundationalism

For a detailed, well researched, fully fleshed out discussion of foundationalism and its alternatives in relationship to economics, you should read the third chapter of Economics and Hermeneutics. It’s an essay by G. B. Madison that goes into the history of thought on the matter and draws heavily on the work of continental philosophers Hans-George Gadamer and Paul Ricoeur. I’m not going to give you anything nearly as thorough, but I hope I can provide something of value by approaching it in a much less technical manner.

Let’s start with Kling:

What I wish to claim is that epistemology in economics is really difficult. It is more difficult than in physics. We have a much harder time testing our theories experimentally. We face insurmountable levels of causal density. We do not have a neat, clean answer to the question “How do you know that?” It appears to me that physicists can answer that question in ways that are much more straightforward and compelling.

There are some things to agree with here, but I want to focus on “How do you know that?” This is a recurring phrase in his post.

As I see it, the instinct of most scholars in an epistemological mood is to reach for some form of foundationalism. At least, scholars that are unfamiliar with the debates on knowledge and interpretation among philosophers in the 20th century.

The elevator pitch for foundationalism is: what if you could prove, once and for all, that you truly and absolutely know something. Imagine there’s a procedure that, if followed, will provide you with fool-proof output. Or better yet, imagine there is a computer that can certify whether something is truth or not. In the seemingly never ending sequence of “we know X because Y, which we know because Z,” foundationalism is the quest to find the bottom, the thing that validates the whole chain.

Problems arise immediately. If we rely on a procedure or a computer to validate our results, what can we rely on to validate the procedure or computer themselves? If the foundational thing, whatever it may be, is what validates everything, then must it validate itself? That seems unacceptably circular. If we don’t care about validating the validator, then the whole thing appears like an appeal to authority like any other—the sort of thing that Kling and others think is wrong with modern economics.

This is, of course, a shamefully abridged version of both the formulation and critique of foundationalism. I strongly recommend Madison’s essay, and its footnotes, if you want to dig deeper. But hopefully this will be good enough for our purposes.

Knowledge From the Inside

In an earlier post on the nature of vast literatures, I wrote:

If a work has been around and widely read for a sufficient length of time, you can safely assume:

• All obvious criticisms have been made.
• All obvious criticisms have been responded to.
• A back-and-forth has occurred and the conversation has moved forward.

What if this dynamic is all that knowledge boils down to?

Here’s what I mean: when we speak of “facts” or “data” as though they are givens, we overlook the role of judgment in creating these things. It is for this reason that R. D. Laing argues, along with Deirdre McCloskey, that we should abandon the word “data” which at root means “something given” and instead refer to “capta,” meaning at root “something seized.”

When the results of a survey come out, we tend to think that interpretation is just involved in how we generalize it, or what we use those results as a proxy for. In fact, interpretation came in at the very beginning, when the survey questions were formulated, and when the respondents were selected. Survey questions are designed with the intention of gaining insight about some larger question the researcher cares about. Judgment calls must be made at several levels:

1. What story are the researchers trying to tell?
2. Are the survey questions relevant to that story?
3. Did the researchers deviate from standard polling practices in any way?
4. Did the researchers do enough to address nonresponse biases and other known problems with standard polling methods?

Question 1 is about theory and, more broadly, conceptual frameworks. By asserting that researchers are trying to tell a story, I do not mean that they are assuming what they’re trying to prove, or anything unsavory like that. Obviously there are plenty of times where we encounter researchers doing just that, but I mean something more respectable. I mean that, after they have gathered all of the answers to their survey, they want to be able to say something like “Republicans are more likely to go to church than Democrats,” or “whether or not someone goes to church has very little to do with their political affiliation.” In pursuing their research, they already have a background understanding of the world, as well as a set of categories, that they need before they can even begin to write any questions. Deirdre McCloskey’s essay, chapter 4 of Economics and Hermeneutics, is a good deep dive into the role of storytelling in economics, for those who want to pursue this aspect further.

Question 2 should seem straightforward to most people. There are, of course, subtleties that people outside of polling might not think of. Simply asking “what party do you consider yourself a part of?” by itself can bias the results in a way that asking whether they affiliate with any part and then asking “if so, which one?” does not. This is something that market researchers and polling companies have talked about in great detail for many years. “All obvious criticisms have been made,” and so forth. There is, in short, a vast literature on this subject, and the best way to intelligently engage with it is to familiarize yourself with that literature.

The result of that literature and the conversation around it are a set of standard practices used by basically everyone. Sometimes the standards are specific things that are done, sometimes they simply place bounds on a healthy amount of approved variation. But the standards are there, and they constitute Chestertonian fences beyond which errors are quite likely.

That said, part of the vast literature on such thing involves the identification of problems which are, to various degrees, ineradicable. For problems such as the possibility of systematic nonresponse, we are usually in a scenario where Thomas Sowell’s motto that “there are no solutions, only trade-offs” very much applies. Researchers have a bag of tricks to attempt to compensate for such known problems, but each trick has risks and problems of its own. The right mix is a matter of judgment, and having it accepted is a matter of persuasion. Each mix is contestable, but so are the standards from the previous step; in fact all four parts are completely contestable.

What does it mean to say that such things are a matter of persuasion, that they are contestable, but that expertise and knowledge still matter?

Consider Kling’s statement about appeals to authority:

There is a slight overlap between Romer’s critique and mine. Romer is saying that economists are choosing models in order to maintain “group cohesion.” I say that they are choosing models based on appeals to authority.

One of the things that I found jarring about The Rhetoric of Economics is that McCloskey argues, among other things, that appeals to authority are natural and necessary. That pointing out that an argument involves an appeal to authority does not invalidate that argument. After the Enlightenment, the idea that authority is involved in knowledge at all is a heresy—is not truth manifest?

But truth is not manifest; we are not capable of seeing a perfect, unblemished Truth. Instead, we must interpret the world and use our judgment, and engage that judgment with the judgments that have been made by others. When you enter any field—economics, sociology, and yes, physics—you are exposed to papers and books by authorities in that field. Moreover, learning in a class means acknowledging that the professor is to some degree an authority on the matter. Learning auto-didactically boils down to deciding which sources are in some sense authoritative.

We would like to think that, if there are authorities, it is because they are resting on firm foundationalist grounds and are capable of perfectly validating their ideas. But of course this is not the case. They are people who have immersed themselves in the ideas, conversations, criticisms, and problems of their field. They have gained expertise in the sense of being able to bring to bear their experience with this conversation, and with the research and scholarly practices, that constitute that field. Their judgment is a far more informed judgment than those who are entering the field for the first time, or outside of the field entirely.

The only “foundation” is the pre-existing literatures and other sources which have (to repeat) made all obvious criticism, and all obvious replies to those criticisms, and moved way further on into far more subtle and thorny terrain. In addition and at least as important are the living practices of the field and the knowledge and ethics embedded in those practices.

If this all sounds terribly like I am claiming that truth does not exist, let me assure you that that is not what I’m saying. What I’m saying is that all human accomplishments occur within the realm of human capabilities. This is as true for knowledge as it is for constructing a skyscraper. There is no external validator, there are just human beings with minds capable of making judgments, informed by conceptual schemes and specific ideas and ongoing conversations in various forms.

The fact that we do manage to build skyscrapers, and airplanes that stay in the air drastically more often than they fail to, and atom bombs, and air conditioning, and computers, is a testament to how good human judgment can be, when properly situated.

But there are no givens. Nothing is handed to us, where knowledge is concerned. All knowledge is political, in Will Wilkinson’s sense:

By political I mean contested, negotiated, and normatively binding. Coercion is a limiting case of rule enforcement and not the essence of the political, in this sense.

So, I made coffee this morning. The exact way this came about reflects a vast set of rules that are political in my broad sense, but also political in a libertarian’s narrower sense. The coordination of the production and distribution of my coffee and my coffee maker depended on a system of property rights that is, in fact, politically defined. And the interface between different national regimes of property rights that got my beans from Guatemala to Chattanooga was probably mediated by politically negotiated trade agreements. And my coffee machine is what it is and cost what it did in part due to politically-determined patent law. And the coffee was processed, and my machine functions, in a manner in part determined by politically-implemented health and safety regulations. And so on and so on. Politics in every drop.

As with coffee, so too with physics, chemistry, and yes, economics. If scholarship and science did not involve normatively binding standards, we would not be able to trust anything as an authority; fraud would be just as likely as honest work. If we did not believe that such standards were contested and negotiated, we would not do things like write books in an attempt to influence our field.

Kling’s argument that economics faces “insurmountable levels of causal density” is a cross-subject comparison problem that Aristotle was aware back way back when the Nicomachean Ethics was written:

Our discussion will be adequate if it has as much clearness as the subject-matter admits of, for precision is not to be sought for alike in all discussions, any more than in all the products of the crafts. Now fine and just actions, which political science investigates, admit of much variety and fluctuation of opinion, so that they may be thought to exist only by convention, and not by nature. And goods also give rise to a similar fluctuation because they bring harm to many people; for before now men have been undone by reason of their wealth, and others by reason of their courage. We must be content, then, in speaking of such subjects and with such premises to indicate the truth roughly and in outline, and in speaking about things which are only for the most part true and with premises of the same kind to reach conclusions that are no better. In the same spirit, therefore, should each type of statement be received; for it is the mark of an educated man to look for precision in each class of things just so far as the nature of the subject admits; it is evidently equally foolish to accept probable reasoning from a mathematician and to demand from a rhetorician scientific proofs.

Bold added by me.

There are different levels of precision we can expect from a given subject matter. Sometimes we make breakthroughs that give us more precision than we thought was possible. But Aristotle and Kling’s point is that some things are irreducibly imprecise beyond a certain level—and I would agree with them.

But that perspective is, of course, contestable.

A Fully Human Economics

As Kling says, the practice of economics is very mechanistic at present. This is reflected in the mechanistic theory of science that they hold onto, as discussed above. But the problems in their theory of science are reflected in their theories, proper.

Consider the heart of this post: persuasion, rhetoric. Economists don’t have much to say about persuasion. Yet by one back of the envelope estimate, persuasion accounts for one quarter of GDP. Is it all waste?

The problem with economics is what Dan Klein has characterized as its flatness. It’s easy to illustrate with game theory. In another post, I attempted to take Jason Briggeman’s game which included the two-sided possibility of coercion and add persuasion into the mix.

It just doesn’t make any sense, in that framework. In these models, you can make it possible for one person to get someone else to take some action. You can add payoffs and costs. But there’s no meaningful way to distinguish between coercion and persuasion. You can do it in an arbitrary way, by having one way with higher costs to the one imposing and lower costs to the one being imposed upon, and call that “persuasion” as opposed to one where the costs-ratio runs the other way. But it adds complexity to the model without really adding any illumination. Adding the possibility of imposing a choice on the other player is interesting, worth fleshing out the way Jason does for its implications. But once you try and contrast it with persuasion, you immediately encounter how limited the world of a cost is a cost is a cost (and a benefit is a benefit is a benefit) is.

Similarly, Klein’s main complaint is that economists reduce all knowledge to mere information. Information in this view is something you either have or you don’t. But in the real world, treating something like information requires judgment. Deciding that something you perceive has implications for the future price of some commodity is an active process, for which you must take responsibility. That is why it’s so socially useful for the price system to put the risk on the back of speculators; the future is not something you have information about or you don’t. It’s good to reward people for being right, and to let them decide for themselves if they trust their judgment enough to risk a loss.

Richard Ebeling’s chapter of Economics and Hermeneutics is quite good on the role of price within an interpretative (rather than mechanistic or computational) framework. But the best, and most accessible chapter is the very last one, Tom Palmer’s “The hermeneutical view of freedom.”

Palmer points out Adam Smith’s tantalizing but fleeting reference to the “faculties of reason and speech” right after the famous passage about our natural tendency to “truck, barter, and exchange.” He quotes the following from one of Smith’s lectures:

If we should enquire into the principle in the human mind on which this disposition of trucking is founded, it is clearly on the natural inclination every one has to persuade. The offering of a schilling, which to us appears to have so plain and simple a meaning, is in reality offering an argument to persuade one to do so and so as it is for his interest. Men always endeavour to persuade others to be of their opinion even when the matter is of no consequence to them.

For Smith, trade was never a mechanistic process. The act of offering payment is itself an act of persuasion.

I think this notion of the centrality of persuasion in human affairs, and in markets in particular, is what economics should strive to rebuild itself around. This does not mean that the insights from economists’ contributions up to this point should be discarded, just that we should seek to find their appropriate context. I’m a big fan of Joseph Heath’s suggestion that agency theory, for instance, is not literally accurate but is a useful tool for identifying “fault lines” where problems are likely to arise. Andy Clark’s work on the institutional contexts in which economists’ assumptions yield accurate predictions of behavior is also extremely valuable.

While economists are increasingly open to insights from psychology, I would urge them to look also into linguistics, and the philosophy of mind and of science. Economics and Hermeneutics is a good guide but McCloskey’s The Rhetoric of Economics is shorter and probably more accessible to the typical economist. Both have footnotes available for raiding, for those seeking to flesh out an interpretative approach to the human sciences.

The greatest mistake of the behaviorists and the Samuelsonians is not that they ignored inborn tendencies or didn’t think in terms of computation. The thesis that computation is equivalent to how humans think is not at all established; see Harry Collins on how even when computers beat us at something like Chess, we’re completely unable to make computers do it the way that a human would. Moreover, inborn tendencies can easily be dealt with from a behaviorist or Samuelsonian perspective—we can just speak of it being easier or harder to train a given individual, or having a higher or lower cost for instilling specific skills.

The real problem with behaviorists is best exemplified in the cheeky comment made by Alfred North Whitehead in his critique of them: “A consistent behaviorist cannot feel it important to refute my statement. He can only behave.”

Behaviorist and Samuelsonian theories cannot adequately deal with the fact that we are a speaking, interpreting, and political animal.

Previous Posts in This Thread:

• Efficient Markets and Communities of Rhetoric
• Is Philosophy Information Efficient?
• Filtering Out the Garbage
• The Arc of the Universe Bends Towards Meh
• The Conversation Behind and Within the Literature