<p> This dissertation is concerned with the problems of constructing testable models of the short-run dynamics of urban labour markets, given the currently available data sources. Bimonthly data for manufacturing production workers in U.S. cities were considered the most complete for the purpose. Twenty-nine cities were chosen (Indiana, 1 city; Michigan, 1 city; New York State, 7 cities; Ohio, 8 cities; Pennsylvania, 12 cities). The period of the study, 1964-73, was chosen to avoid changes in data-base definition and to avoid the impact of the oil embargo of late 1973. </p> <p> Two types of model were estimated for each city, with different specifications for each model. The first model consisted of three linear, simultaneous difference equations determining, for each city, the number of hours worked per period (number of employees x number of hours worked per employee), the supply of hours available per period and the average weekly wage rate. When tested empirically this model was successful in explaining all these variables. The second model consisted of five equations, determining the number of people employed per period, the number of hours worked per employee per period, the size of the labour force, the hourly wage rate and the voluntary quit rate from employment. This model was considered the theoretically superior of the two in that it allowed for employers substituting between the number of their employees and the number of hours worked by each employee per period. This model also proved the more empirically successful of the two models. The models were tested using the Two-Stage Least-Squares estimation technique. It is believed that this is the first time that such models have been tested in order to analyse short-run urban labour market behaviour. </p> <p> It was hypothesised that the level of manufacturing production is a major determinant of labour market activity in each city. Unfortunately no short-run urban manufacturing production data are available either for the U.S.A. or elsewhere. This fundamental deficiency in the data base was overcome by the development of a synthetic urban manufacturing production time-series, using national U.S.A. production time-series weighted by the proportions of each of nine manufacturing categories in each city. The technique cannot be validated directly but the results from the models are consistent with the synthetic series being excellent proxies for the true series. </p> <p> The results indicated that the cities all had labour markets that behaved in remarkably similar ways, despite the fact that the labour forces involved ranged in size from 57,000 to 5,558,400 people. In particular the labour markets all exhibited highly stable dynamic behaviour. This result indicated that the labour markets were unlikely to be the generators of boom or slump in their respective cities. When estimated labour market parameters were mapped there appeared to be only weak spatial groupings of the parameter values. Similarly weak groupings appeared when the parameter values were plotted against labour force size. No firm conclusions could be drawn from the groupings. Originally it was intended to model the inter-urban labour market interactions but this proved impossible. All the results are based, therefore, on the assumption that those inter-urban interactions are weak enough to be ignored in the short-run. </p> / Thesis / Doctor of Philosophy (PhD)