Measurement Costs and the Market for Variable Quality Commodities

Yoram Barzel (University of Washington)
Aurora Stephany (University of Washington)

Abstract: Observing the characteristics of a good is at the root of any transaction. As pointed out by Barzel (1982) in his paper about measurement costs, the problem of measuring the attributes of a commodity is much more central to the economic problem than the widespread assumption of zero transaction costs would lead to believe. Measurement is always costly, and some resources will be dissipated, and possibly, mutually beneficial trade will be forgone in the absence of measurement information. In this paper, we develop a model in which a seller has a batch of goods of varying quality. Pricing each good individually is prohibitively expensive, so they are all sold at the same price. In the absence of trust between buyer and seller, the former doesn’t trust the latter with the selection of units, because the seller would gain by giving him the worst units in order to improve the remaining distribution. So the buyer will spend some resources “picking and choosing”, that is, inspecting items until finding an acceptable one. This inspection is costly and will result in trade below the optimal levels. We prove several results: 1., in such circumstances the good will be necessarily sold at a price above the average value, with buyers still participating in the market if they can inspect until they find items from the top of the distribution. 2. the distribution will decay over time, and the seller will be forced to lower the price. 3. The greater the dispersion in the quality of the goods, the larger the difference between average value and initial price, and the larger the dissipation associated with the lack of trust. 4., when there are different types of buyers, the ones with lower cost of inspecting (and with lower willingness to pay) will drive out the ones with higher inspection cost and higher ability to pay. When this problem is serious enough, the market for the commodity may not exist. Empirical tests for the implications of this model are still under development.