MODELING USED CAR PRICES
A recent paper (Stolyarev 2002) will serve to illustrate the puzzlement I feel as an engineer studying the application of mathematics to economics. The paper observes a bimodal distribution of used car sales with peaks at about 3-5 years and at about 10 years. Anyone who has owned a car could probably make some good guesses as to why. Mechanical problems start to surface after 3-5 years prompting many owners to sell. At ten years, most cars are just about finished. Most of us know that. But the paper claims to "explain" these phenomena using a mathematical equilibrium model. Without questioning the soundness of the mathematics employed, one might ask why this phenomenon needs explaining, or why common-sense explanations are inadequate, and most of all, just what has been gained in the end. The work described is in some ways similar to the task of building an engineering model and showing that it successfully explains some data gathered in the laboratory or in the field. One might admire the intelligence and the diligence and the cleverness that went into such a model. One might adopt it as a classroom example or apply it to subjects other than used cars. But the reason engineers undertake such an effort (other than in the classroom) is to improve designs or diagnose failures. No such prospect is offered in the Stolyarov paper. He claims his model "captures the observed resale patterns for autos. " (Stolyarev 2002, 1391) By "capture" he of course means that his model successfully generates the observed bi-modal distribution. No mean feat, to be sure, yet one is left wondering just what has been gained by the capture. He does not claim that his explanation is superior to common-sense explanations nor that his model could be valuable to buyers or sellers of used cars. Perhaps economists who study the paper will develop a better "feel" for used-car markets (or for any market for used goods, as the model is more general than just used cars). I believe any lay person could read the Kirkpatrick crash-analysis paper and without understanding any of the math or engineering, get some idea of the believe any lay person could read the Kirkpatrick understanding any of the math or engineering, get some idea of the practical value of the work. Can the same be said for the Stolyarov used car paper? Could his model provide some benefit to buyers or sellers of used cars, however indirectly?