Do jobs follow people or people follow jobs? A meta-analysis of Carlino-Mills studies

The issue whether ‘jobs follow people’ and/or ‘people follow jobs’ has recently emerged as one of the leading themes in regional and urban science. Much of the interest herein stems from alleged inconsistencies in the empirical evidence, which naturally raises questions as for the reasons why. Arguably, the nature of causality differs across space as well as time, while speculations have been rife about a number of methodological issues that may play a crucial role in shaping the research outcomes. In this paper a preliminary attempt is described to clarify these matters, by focusing on an articulate literature of 37 so-called ‘Carlino-Mills studies’. Specifically, a statistically supported literature review, referred to as ‘meta-analysis’, is presented in which the study results are evaluated and systematically related to a variety of study characteristics that underlie these results. By listing 308 study results reported in this literature, it is revealed that the empirical evidence is conform popular belief highly inconclusive, albeit that most of the results point towards ‘jobs follow people’. The findings of the meta-regression analyses indicate that the spatial setting of the study, the adopted model specification, and variables measurement in particular affect the research outcomes that indicate the jobs-people direction of causality. No evidence is found that the examination of data referring to a particular time period, population and/ or employment group make much of a difference.