This project consists of two related empirical studies of the behavior of business inventories and their relation to economic fluctuations. The first deals with the role of work-in-progress inventories in smoothing demand fluctuations. The production smoothing properties of inventories in production-to-order industries are tested using U.S. industry data and a linear quadratic model. The project provides estimates of both inventory carrying costs, a final good and total production costs and demand function parameters for durable good manufacturing. The second part investigates differences in inventory management across firms and industries, particularly testing for the presence of (S,s) inventory policies in certain industries, such as retail trade, as opposed to buffer stock models. The study uses a relatively untapped data source: the quarterly small business survey from the National Federation of Independent Business. The NFID data allows the project to test several simple inventory models in time series and cross section and provide information on the aggregate results used in previous work. This project is important because nearly all empirical tests of models of inventories have focuses on inventories of finished goods at manufacturing firms which produce goods "to stock". Production-to-stock industries are those whose products are not individuals, i.e., they can be produced before orders are received and then sold out of inventory. These industries are the least interesting from a business cycle standpoint since their demand and employment fluctuations are small. This project studies inventories and production of firms which produce "to order". These firms generally produce big ticket durable goods, such as machinery, metals, automobiles and other transportation equipment, and they experience large movements in demand and employment over the business cycle. The results of the research will identify the key elements in allowing firms to adjust to demand disturbances or business cycles.