Job Analysis and Historical Productivities in the American Cotton Textile Industry: A Study in Methodology

Abstract

TN recent years engineering data have been increasingly used in studying the input-output relationship in manufacturing industries. Ideally the fundamental mechanical properties underlying a production process are relied on to answer many questions on input-output relationships. This enables the investigator to exclude such variables as management from a production function, and to isolate the effects of other variables by letting them vary one at a time. In actuality, most of the engineering data are compiled by engineers from their experience with actual practices rather than from carefully conducted experiments. But most of the advantages of the engineering approach remain generally present, and its basic usefulness has been repeatedly asserted.' One of the major road blocks in the engineering approach has been the difficulty of estimating the labor-output relationship. Usually one searches in vain for any labor requirement data in engineering literature; and because of the large variation in the workload among plants, reliance on sample observations is apt to be misleading. More fundamentally, because of the heterogeneity in skill and efficiency the quantity of labor input cannot be measured precisely. It is hardly surprising, therefore, that no serious attention has been given to the laboroutput relationships derived with the engineering approach. In this paper we will attempt to demonstrate some of the merits of a special type of engineering data, based on job analysis, which so far has been almost completely ignored by economists. Compiled mainly by time-and-motion study engineers for such industries as the textile, garment, and machine tool industries, these data give the detailed elements of the jobs in a manufacturing process, and the time required to perform them under specified circumstances. This enables the engineer, as well as the economist, to explain the differences in the performance of individual workers. Ideally the elements of efficiency and skill can be isolated, and it becomes possible for the investigator to study the pure effects of technological change on labor-output relationship, or such problems as substitution between labor and other inputs. The rest of this paper will be divided into five sections. In the first section we will describe a method by which the job analysis data may be used, jointly with other engineering data, in computing historical series of certain input-output relationships. In the next the job analysis data are applied to a problem of substitution between labor and machinery, and the results are shown to have justified, on the whole, the method used in the previous section. Next, labor-output series compiled with two alternative methods are given, to be used in the subsequent section for the demonstration of an important implication in the use of job analysis data. Finally, a brief summary is given.