I thought this week I might step outside my usual area of interest and take up positive, predictive economics, and the normative influences on that, philosophy of science, that sort. There are a few moderately interesting issues there, as well.
In an academic economic context, if one expresses any interest in the normative aspect of economic theory, one will quite likely quickly be shuttled to discussions in philosophy of science, under the conceit economics is a science and that’s where such issues are discussed. Let’s just play along today. As I’m normally concerned with normative neoclassical “welfare” economics and related bad economics in the conservative style, let’s first take up the issue of whether neoclassical economic theory, used for empirical prediction, is a proper “science.” I find that question, about neoclassical economic theory specifically, rather easier to discuss than whether “economics” is a science because “economics" is a pretty general category, and some parts of it may be rather more scientific than others.
So let’s consider what to make of neoclassical economic theory ostensibly used for the positive purpose of predicting empirical economic phenomenon. I typically characterize it as an “engineering model,” but others have it a “science.” Let’s briefly discuss. I suppose a proper scientific theory should reflect or incorporate what we know about the world, which is what leads to its accretive quality, its ability to grow piecemeal as research takes up this or that part of the puzzle and then synthesizes those data and results. Attempting to account for any awkward divergence of reality from what particular scientific theories might present, express, suggest, predict is what fuels much research in any real science. But the final test of a scientific theory must involve successful empirical prediction.
An “engineering model” used for predictive purposes seems to me to stand in a different relationship to reality than a scientific theory. For one thing, it makes no presumption that the “simplifying assumptions” on which it is based reflect reality. They’re arbitrary, made up. For another thing, incorrect empirical prediction is not necessarily a problem because one is making an ostensibly self-conscious decision involving simplicity, tractability, and predictive ability. Predictive failure does’t necessary imply any presumed need to revise a model. When it comes to engineering models, the evaluation is a softer matter of advantages relative to other engineering models for that purpose, and unless one is presenting a better, sufficiently simple, tractable theory, there may be no perceived problem with the existing model. I would suggest those are the primary reasons neoclassical economic theory, price theory, doesn't seem to exhibit the piecemeal, accretive quality of a real science. Indeed, quite the opposite. It has existed in the same basic or general form for literally decades.
Importantly for my concerns and interests, engineering models give researchers relatively more leeway than proper scientific theories to construct models they find useful as they subjectively weigh criteria such as descriptive realism, simplicity, and predictive results. This greater leeway of researches using arbitrary engineering models to pick and choose models and elements of models they find useful in the moment clearly allows much more scope for normative interests and biases to be reflected in such models than in proper scientific theory. I noted before the basic form of neoclassical economic theory has remained the same for decades. It raises the question, is it because of its combination of predictive ability and simplicity, success relative to competing models for the very practical purpose of prediction? Or may there just be something about the model itself some researchers find appealing? Its elegance, simplicity, beauty, intuitive appeal? Or indeed its use in normative or evaluative economics, in applied neoclassical “welfare” economics, often lumped with the positive form? Or, going yet a step further, might it involve its rhetorical usefulness for promoting anti-democracy bad economics in the conservative style, the dodgy normative theory based on flawed interpretations of normative neoclassical welfare economics? Within economics, there does seem a curious fascination with, or exaltation of, one particular, arbitrary, engineering model, a curious lack of interest in presenting engineering models with entirely different false but simplifying assumptions to be used for predictive purposes. And certainly it seems unusual, notable the same neoclassical economic theory can be used for normative, evaluative purposes in the guise of neoclassical “welfare” economics as well as positive, predictive purposes. It’s not dispositive, but suggestive, surely.
This has led some or let’s just say many to suppose it folly to try to draw any distinction between positive and normative economics in the context of neoclassical economic theory. And, I see the issues, the difficulties, that generate such sentiments. But I must disagree. As I’ve argued many times, I feel a useful distinction can be drawn between positive economics, at least ostensibly about empirical prediction, which may have normative influences, and normative economics, about evaluation, goals, values, ethics, appealing storytelling. The big controversies about economics are about normative economics, ethics, values. Although economists may also be bad at empirical prediction, may slavishly adhere to models inadequate to the task for some old reason, may pretend to be scientists, that’s really a side issue.
