Journal of Regional Science

ABSTRACT Standard spatial autoregressive models rely on spatial weight structures constructed to model dependence among n regions. Ways of parsimoniously modeling the connectivity among the sample of N=n² origin‐destination (OD) pairs that arise in a closed system of interregional flows has remained a stumbling block. We overcome this problem by proposing spatial weight structures that model dependence among the N OD pairs in a fashion consistent with standard spatial autoregressive models. This results in a family of spatial OD models introduced here that represent an extension of the spatial regression models described in Anselin (1988).

We provide a comprehensive overview of the strengths and weaknesses of different spatial econometric model specifications in terms of spillover effects. Based on this overview, we advocate taking the SLX model as point of departure in case a well‐founded theory indicating which model is most appropriate is lacking. In contrast to other spatial econometric models, the SLX model also allows for the spatial weights matrix W to be parameterized and the application of standard econometric techniques to test for endogenous explanatory variables. This starkly contrasts commonly used spatial econometric specification strategies and is a complement to the critique of spatial econometrics raised in a special theme issue of the Journal of Regional Science (Volume 52, Issue 2). To illustrate the pitfalls of the standard spatial econometrics approach and the benefits of our proposed alternative approach in an empirical setting, the Baltagi and Li (2004) cigarette demand model is estimated.

A Malmquist‐Luenberger productivity index is employed to account for both marketed output and the output of pollution abatement activities of U.S. state manufacturing sectors for –1986. The index allows us to decompose the change in productivity into measures of change in efficiency and technical change. By accounting for the change in emissions, average annual productivity growth is 3.6 percent, whereas it is 1.7 percent when emissions are ignored. We also find adjusted productivity growth improved after 1977, and “Frost Belt” states with rapidly growing manufacturing sectors have significantly higher rates of productivity growth than “Sun Belt” states with slow growing manufacturing sectors.

Transport infrastructure investment is a cornerstone of growth‐promoting strategies. However, the link between infrastructure investment and economic performance remains unclear. This may be a consequence of overlooking the role of government institutions. This paper assesses the connection between regional quality of government and the returns of different types of road infrastructure in the regions of the European Union. The results unveil the influence of regional quality of government on the economic returns of transport infrastructure. In weak institutional contexts, investment in motorways—the preferred option by governments—yields significantly lower returns than the more humble secondary road. Government institutions also affect the returns of transport maintenance investment.

Abstract Many theoretical models analyze the effects of decentralized environmental policymaking. The predictions range from a race to the top, a race to the bottom, or no effect. However, little empirical evidence exists to resolve this ambiguity. This paper fills the void by examining the impact of decentralized environmental policymaking in the U.S. under Presidents Ronald Reagan and George H.W. Bush. For abatement expenditures, Reagan's decentralization had no discernible impact before the mid‐1980s, but by the mid‐1980s the data are consistent with decentralization leading to a race to the top. No statistically significant effect is found on nitrogen oxide or sulfur dioxide emissions.

An explanation for the rank‐size distribution for human settlements based on simple stochastic models of settlement formation and growth is presented. Not only does the analysis of the model explain the rank‐size phenomenon in the upper tail, it also predicts a reverse rank‐size phenomenon in the lower tail. Furthermore it yields a parametric form (the double Pareto‐lognormal distribution) for the complete distribution of settlement sizes. Settlement‐size data for four regions (two in Spain and two in the U.S.) are used as examples. For these regions the lower tail rank‐size property is seen to hold and the double Pareto‐lognormal distribution shown to provide an excellent fit, lending support to the model and to the explanation for the rank‐size law.