Raising the bar (7)

ABSTRACT

This editorial summarises the papers published in issue 13.1 so as to raise the bar in applied spatial economic research and highlight new trends. The first paper adopts a scale neutral approach to investigate the spatial mechanisms that cause regional innovation and growth. The second paper claims that population-weighting when calculating indices of regional inequality might lead to inconsistent outcomes. The third paper estimates the effect of distance between family residence and higher education institution on a student's academic performance, thereby accounting for endogenous regressors. The fourth paper shows an inverted U-shaped relationship between economic development at region of origin and the propensity to migrate using a multilevel approach. The fifth paper provides spatial econometric evidence of price competition between sellers of used books on Amazon.com. The last paper estimates a hedonic housing price equation and parameterizes the spatial weight matrix to determine how far back in time buyers, sellers and realtors are looking at the housing market.

KEYWORDS:

• scale effects
• inequality
• education
• migration
• pricing

JEL:

• C23
• D63
• I2
• R12
• R23

Spatial Economic Analysis is a pioneering journal dedicated to the development of theory and methods in spatial economic analysis. This issue contains six papers contributing to these theoretical and empirical developments.

The first paper in this issue by Bond-Smith, McCann, and Oxley (2017) offers a well-rounded theoretical contribution on a topical issue in regional growth theory. It adopts a scale neutral approach to investigate the spatial mechanisms that cause regional innovation and growth, thereby relying on counterfactual scale effects. This is achieved through a careful design of spatial spillovers in the wake of state-of-the-art growth theory. In contrast to previous works, the ensuing policy analysis concludes that the spatial concentration of economic activities can be growth enhancing even in the absence of scale effects. According to the reviewers, the analysis in this paper is carefully executed with detailed policy implications for topics such as R&D subsidies, peripheral innovation subsidies, policies to retain industries in peripheral regions and policies encouraging interregional knowledge spillovers. This paper fits within the tradition of new economic geography, about which Spatial Economic Analysis has published more papers, including a recent contribution by Noblet and Belgodere (2016), who provide new insights into how regions and countries specialize in certain production activities and their relation to the size of these activities.

Whereas the first paper in this issue is theoretically based, the second is a methodological contribution. This paper, written by Gluschenko (2017), challenges the work of many previous studies, too many to mention here or even in the paper itself. The author claims that the common practice of population-weighting when calculating indices of regional inequality, such as the Theil, Gini and the Williamson coefficient, might lead to inconsistent outcomes and at least requires rethinking. To illustrate this, he provides several simple numerical examples, as well as mathematical formulae. In addition, he shows that population-weighted indices violate the anonymity principle, the principle of transfers and do not have unambiguous maxima. According to one of the reviewers of this paper, the question of whether or not to weight is more an issue of whether regions are considered ‘individuals’ or ‘groups’. The reviewers agreed that this is an interesting piece of work that might change ways of thinking about the determination of regional inequality measures in the future.

The third paper in this issue is by Vieira, Vieira, and Raposo (2017), and contributes to the literature on the determinants of academic performance and the role of space by estimating the effect of distance between family residence and higher education institution. The empirical analysis is based on a selected panel of individual students who graduated from the University of Évora, Portugal, over 2000–11. Distance is expected to have negative effects, partly because of homesickness, although the ultimate sign is uncertain since its impact might also be non-linear. The time needed to complete a course of study, which is also expected to have a negative effect, is treated as a potential endogenous explanatory variable that should be instrumented. Interestingly, this negative effect is similar to the impact of time on the market and a house's selling price (Dubé & Legros, 2016), and this setup is similar to a series of studies that have appeared in Spatial Economic Analysis, showing that not all explanatory variables can be treated as exogenous and that the estimation procedure should be adjusted for this. One recent example is Haapanen and Böckerman (2016), who determine the causal effect of education level on migration within Finland's 19 NUTS-3-regions.

In Vieira et al. in this issue, the authors find evidence of a negative and significant distance decay effect, which is supported by several robustness checks. Only when adding the square of the distance variable to the model does this outcome seem to change. After a certain point, the negative effect seems to vanish, but to learn more about this, the reader is referred to Table 9A in the online appendix.

Migration keeps inspiring economic researchers, as recently shown by Spatial Economic Analysis's virtual special issue on migration (Jordan & Elhorst, 2016). The fourth contribution to the current issue by Cazzuffi and Modrego (2017) continues this trend by examining internal migration decisions in Mexico. The sample consists of 6,316 working-age individuals taken from 135 of Mexico's 2,456 municipalities over 2002–05. The authors investigate the behaviour of so-called stayers and first-time migrants, of whom there are 421 in the sample. The theoretical model is standard: an individual migrates if the expected welfare at the destination exceeds its counterpart at home and the individual is able to cover the migration costs. This leads to a logit specification where the decision to migrate depends on individual characteristics, as well as characteristics of the places of origin and destination. A multilevel model is adopted to account for the fact that these characteristics are measured at the municipality level rather than the individual level. Interaction effects between individual and municipality characteristics are also included and the multilevel model is found to outperform simpler models. The authors find empirical evidence in favour of an inverted U-shaped relationship between economic development at origin and the propensity to migrate. Readers interested in this study should also consult the recent work of Clemente, Larramona, and Olmos (2016), who extend the standard gravity model with an unknown threshold parameter, which needs to be estimated and above which the response parameters of the explanatory variables are different than below it. This setup is used to show that a potential migrant is only willing to take the decision to migrate to another region if the benefits outweigh the costs to a sufficient degree.

The next contribution to this issue by Wang (2017) is quite an unusual paper for two reasons. First, it focuses on a topic not usually associated with spatial econometric techniques. The author tries to identify factors that affect the listing price of used books sold on Amazon.com. While the condition of the book is one such factor, the author also investigates whether an individual book seller competes with other sellers who supply the same book of the same quality at the same moment in time on the internet. To investigate this, he estimates a spatial autoregressive model and tests whether the spatial autoregressive coefficient is negative and significant. This fits within a small though slowly growing literature that has also tried to find evidence of competition effects, which started with Pinkse, Slade, and Brett (2002). For example, Griffith and Arbia (2010) offer three examples of negatively spatially autocorrelated phenomena, all based on the notion of competitive locational processes. They conclude that if the manifestation of a certain phenomenon in one area occurs at the expense of its neighbouring areas, then negative spatial autocorrelation is likely. Elsewhere, Elhorst and Zigová (2014) test for and find empirical evidence in favour of competition effects among economic departments in Germany: a research unit's productivity negatively depends on that of neighbouring research units weighted by inverse distances. Finally, LeSage, Vance, and Chih (2017) apply a heterogeneous spatial autoregressive panel data model to explore pricing competition, next to cooperation, by duopoly pairs of German fuelling stations. This evidence of negative spatial dependence in a series of empirical studies makes clear that economic-theoretical work on spatial econometric estimation should not ignore negative values of the spatial dependence parameter when investigating the small sample properties of their estimators using Monte Carlo simulation experiments.

The second reason why this study by Wang (2017) is unusual is because it applies a semiparametric adaptive estimation method developed by Robinson (2010) based on the observation that the distribution function of the disturbances is not normally distributed. This is an interesting alternative to the maximum likelihood, instrumental variables, generalized method of moments, and Bayesian estimators that are used commonly within this literature.

The last contribution to this issue is by three experienced researchers in real estate economics: Dubé, Legros, and Thanos (2017). The authors propose a spatio-temporal autoregressive (STAR) model in which housing prices at a particular point in time are explained by comparable sales of other houses within a certain spatial and temporal distance, contemporaneous sales of other houses within a certain spatial distance (at the same moment in time), and a set of control variables describing housing and neighbourhood characteristics. Interestingly, this model encompasses many models of empirical interest that have been applied in previous studies. The first variable in the STAR model is treated as being exogenous, since the corresponding spatial weight matrix is lower triangular and thus unidirectional. The second term is treated as being endogenous, since the corresponding spatial weight matrix is multidirectional as in a regular spatial autoregressive model. Since the first term is exogenous, the authors take the opportunity to parameterize this spatial weight matrix by one single parameter. By doing so, they are able to determine empirically how far buyers, sellers and realtors are looking back in time, which they find to be eight months. According to one of the reviewers, the incorporation and parameterization of the temporal dimension in spatial econometric models is of crucial importance and deserves much more attention in future research.