World Development, Vol. 21. No. 4.. pp. 535-554. 1993. 0305-750x/93 $6.00 + 0.00 Printed in Great Britain. @ 1993 Pergamon Press Ltd Observational Equivalence in the Modeling of African Labor Markets and Urbanization CHARLES M. BECKER zyxwvutsrqponmlkjihgfedcbaZYXWVUT University of Colorado, Boulder and ANDREW R. MORRISON” Tulane University, New Orleans, Louisiana zyxwvutsrqponmlkjihgfedcbaZYX Summary. - This article examines the appropriateness of neoclassical and rent-seeking models of urbanization for the African milieu and demonstrates that the reduced forms of these two models may be quite similar. The models are not observationally equivalent, however, and methods of distinguishing between them are discussed. A demographic cohort shift model of African urbanization also is presented. Its excellent predictive power suggests that migration models that assume migrant homogeneity (i.e., highly aggregate migration models) ignore information that can be useful in predicting trends in migration flows. 1. INTRODUCTION Neoclassical modeling strategies frequently have been used to explain internal migration patterns in less-developed countries (LDCs) at an aggregate level, but alternative models of urbanization are also available - rent-seeking and demographic cohort shift models are two such alternative modeling strategies. This paper presents the assumptions and theoretical under- pinnings of simplified reduced forms of these three models, and provides empirical tests of their applicability to sub-Saharan Africa. In some sense, then, this paper is a test of which of these models best fits African reality. The topic is one of more than academic interest. Africa’s urban population has been growing at an annual rate of 6% or more in recent decades. The problems associated with rapid urbanization are particularly apparent in light of the continent’s economic crisis, and in the associated collapse of governments’ abilities to provide even the most basic public services. Nor is there clear evidence of a deceleration in the urbanization process even given deteriorating public sectors. Given the importance of urbaniza- tion to policy makers, then, it is also critical to understand which model (or models) offer the best explanations for past urbanization experi- ence, and which can be used to best predict the future. There is, however, also a broader purpose which transcends the African continent: to demonstrate that the estimating equations for several classes of urbanization models are quite similar. Thus, results which “confirm” the valid- ity of one model often equally well confirm the validity of a competing model. Stated differently, many “tests” test very little, as the hypotheses they reject or fail to reject are often consistent with a very diverse set of underlying models. This problem is especially severe in macro models of developing countries (and especially for a data- poor region such as Africa), as estimable reduced forms are often but caricatures of complete, structural models. Some authors (alas, including us) have con- ducted tests of one of these urbanization models - typically the neoclassical model - implicitly confident that they are testing only its appropri- ateness. As is demonstrated below, however, “correct” results from a neoclassical formulation are generally consistent with an underlying rent- seeking structure, or an explanation based on shifting demographic patterns. By carefully ex- amining the estimating equations of the various types of models, this paper derives estimating equations that allow the competing paradigms to be distinguished from one another. This is an especially important exercise, since there is little *Final revision accepted: April 20, 1992. 535