Ann Reg Sci (1994) 28:177-196 --The Annals of--- Regi0nal Science © Springer-Verlag 1994 The spatial and temporal dynamics of US regional income inequality, 1950-1989 C. Cindy Fan ~ and Emilio Casetti 2 ~Departmen~ of Geography, University of California-Los Angeles, 405 Hilgard Avenue, Los Angeles, CA 90024, USA, Fax: (310)206-5976 2Department of Geography, Ohio State University, I90 N. Oval Mall, Columbus, OH 43210, USA Received January 1993 / Accepted in revised form March 1994 Abstract. The inverted-U hypothesis which so influenced research on regional in- come inequality is obsolete and does not predict or explain the recent rise in re- gional inequality. We argue that the regional dynamics literature on polarization, polarization reversal and spatial restructuring offers more powerful explanations to changes in regional income inequality. In addition to the conventional ap- proach of measuring systemic inequality, the empirical analysis in this paper em- phasizes inequality variations, which put into focus the interplay between regional dynamics and regional income inequality. The findings highlight the impact of sectoral shifts and global spatial restructuring on the US regional economy, where new cores of growth and renewed growth are emerging. Introduction Kuznets' seminal work in 1955 introduced the inverted-U hypothesis which states that income inequality between individuals tends to first increase and then decline as a nation progresses during the course of economic development. The hypothe- sis has since been extended to explain changes in regional income inequality. Based on international cross-sectional data and national time-series data, Williamson (1965) showed that regional inequality within nations also follows the path of an inverted-U, increasing in early stages of economic development and decreasing in later stages. The initial increase in regional inequality is explained by concentration of income and income-generating factors in selected regions of a nation; and the subsequent decrease in regional inequality is explained by diffu- sion of income and income-generating factors. The increase and decrease in re- gionai inequality are also referred to as regional divergence and regional con- vergence respectively, and their occurrences have been verified by empirical studies of many developed countries (Easterlin 1958; Perin and Semple 1976; Robinson 1976; Smolensky 1961; Williamson 1965). Explanations for the inverted-U in regional income inequality have been strongly influenced by neoclassical logic, which emphasizes the tendency toward178 c.C. Fan and E. Casetti equilibrium as the factor market adjustment mechanism evens out regional dif- ferentials in income (Alonso 1980; Easterlin 1958). Using the US as an illustra- tion, Williamson (1965) argued that the initial directions of labor and capital migration favored the more developed North; later as the nation's interregionat linkages improved, labor and capital migration became less selective and would slow down or reverse to benefit the South. Recent evidence however suggests that changes in income inequality and changes in regional income inequality continue after the last stages of the in- verted-U model. In the US once again personal income inequality has increased (Danziger and Gottschalk 1993). Interestingly, regional income inequality in the US has also increased during the late 1970s and the 1980s (Amos 1988; Braun t991; Coughlin and Mandelbaum 1988; Rowley et al. 1991). The neoclassical logic, on which the inverted-U is theoretically based, does not predict increases in inequality after regional convergence at advanced stages of development. These new trends raise doubts about the predictive and explanatory power of the in- verted-U hypothesis. Until recently, the attention in the literature has been on testing the validity of the inverted-U rather than on formulating a theory of regional income in- equality (Fisch 1984). There have been some attempts of broad theorization (e.g. Lipshitz 1986 and Semple 1977), but they tend to be confined to the issue of equilibrium versus disequilibrium and are not concerned with what may occur following regional convergence in the last stages of the inverted-U. In this sense the highly acclaimed inverted-U hypothesis is more a hindrance than a catalyst to development of theory. While this paper does not attempt to formulate a full- fledged theory of regional income inequality, we wish to elaborate an approach which may be instrumental in future theory development. Rather than focusing on regional divergence and convergence, we choose to look at inequality dynamics and variations, that is, where and when inequality increases and decreases. In- stead of specific positions in the economic development continuum, we highlight the flows of factors of production which underlie changes in inequality. The re- gional dynamics literature provides a ready pool of knowledge and is the most appropriate building block for explaining regional income inequality - the selec- tive growth and decline of regions underlie changes in regional income inequality. We start with a discussion of salient strands in the regional dynamics literature which have important bearing on regional income inequality. Then we explain the importance of focusing on both an overall level of inequality, which we refer to as systemic inequality, as well as the variations of inequality. The purpose of the empirical analysis is to demonstrate our suggested approach, and to illustrate us- ing the US as an example that regional dynamics is a more powerful and mean- ingful predictor of regional income inequality than level of economic develop- ment. Regional dynamics and regional income inequality In this section we attempt to explain regional income inequality in the context of regional growth and decline. The discussion involves three phases - regionalThe spatial and temporal dynamics of US regional income inequality, 1950-1989 179 divergence and regional convergence as predicted by the inverted-U; and the recent increase in regional income inequality. In all three phases the emphasis is on flows of factors of production, especially capital and labor. Polarization as an explanation )'or regional divergence During the course of development in advanced industrial economies, initial ad-