This paper investigates the sources and methods used to construct the aggregate data on consumer spending in the United States, searching especially for imperfections that may have implications for the outcome of empirical work. The paper identifies two such imperfections: sampling error and compositional error. It then presents several examples intended to illustrate that these imperfections may be empirically important and that appropriate remedies for them often can be devised. The paper concludes by suggesting some guidelines for empirical practice.

This paper investigates the sources and methods used to construct the aggregate data on consumer spending in the United States, searching especially for imperfections that may have implications for the outcome of empirical work. The paper identifies two such imperfections: sampling error and compositional error. It then presents several examples intended to illustrate that these imperfections may be empirically important and that appropriate remedies for them often can be devised. The paper concludes by suggesting some guidelines of empirical practice.

This paper analyzes dynamics of production and inventories at the General Motors Corporation during the 1920s and 1930s. The authors begin by examining anecdotal evidence on the nature of the production control system in force during that period. Motivated by that evidence, they then extend the conventional linear-quadratic model of production behavior to take account of annual shutdown. Finally, the authors apply the modified model to newly available data on monthly unit production, sales, and inventories during 1924-40. General Motors appears to have been aiming to maintain a targeted level of inventory relative to expected sales and, secondarily, to smooth production.