Models, truth, and analytic inference in economics

A popular view of models among economists and philosophers alike is that all models are false, but some are useful. Models are frequently treated as convenient fictions, idealizations, stories about credible worlds, or "near enough" to the truth. But such a understandings pose serious questions, among them: if models are false, how is it that they are so useful? how can they have any bearing on what is actually the case in the world? how can we evaluate them empirically? How can we develop them for greater precision? for understanding how models related to the world, how they can successfully support scientific investigation? The paper argues that these and related questions reflect a fundamental confusion: models are, in fact, useful only to the degree that they are instruments for stating truth. The confusion arises from a failure to understand how models relate to the world analogically. Analogies are fundamentally incomplete and perspectival, so that the truths that state are necessarily piecemeal But models may nonetheless be apt. A critical distinction is drawn between accuracy and precision in modeling. Modeling is related to Charles Peirce's analytical inference. And the application of analytical inference in economics is illustrated with a historical case-study of Lawrence Klein's early econometric models of the U.S. economy.