Raising the bar (18)

ABSTRACT

This editorial summarizes the papers published in issue 16(4) (2021). The first paper adopts a higher order spatial autoregressive model with endogenous spatial weight matrices. The second paper investigates the existence of the law of one price using regional observations over time. The third paper develops an economic-theoretical model that goes against the common belief that the most productive individuals and firms agglomerate at the core. The fourth paper provides empirical evidence that merger and acquisition deals are more likely to occur between firms in culturally than in geographically contiguous countries. The fifth paper develops a spatial econometric estimator based on the indirect inference principle. The sixth paper examines the investment behaviour of First Nation governments through joint ventures. The seventh paper employs a spatial econometric model with an endogenous spatial weight matrix to construct intraregional input-output models.

KEYWORDS:

• spillovers
• integration
• agglomeration
• productivity
• estimation

JEL:

• C21
• F15
• F23
• J24
• O47
• R15

Spatial Economic Analysis is a pioneering journal dedicated to the development of theory and methods in spatial economics. This issue contains seven papers contributing to the journal’s mission written by authors from all over the world: China, the U.S., Canada, Russia, Sweden, Japan, Italy, Germany, and Poland. It shows the international character of this originally British-Irish journal.

The first paper, by Cheng and Weber (2021, this issue), deals with our own behaviour as researchers. It addresses the extent to which our research productivity is affected by that of colleagues within our own institution and by that of our co-authors. To measure the impact of both colleague and co-author networks jointly, the authors adopt a higher order spatial autoregressive model with two spatial weight matrices. The parameters of this model could be estimated by the spatial two-stage least squares (2SLS) estimator but the spatial weight matrices may be endogenous rather than exogenous. This is because researchers may sort themselves into institutions and into co-author relationships. Although this problem is not new - see previous studies by Kelejian and Piras (2014), Qu and Lee (2015), Bhattacharjee et al. (2016), Cheng and Lee (2017), Delgado et al. (2018), and Corrado et al. (2019) - the authors of this paper develop an innovative estimation strategy. This strategy is based on institutional fixed effects, individual fixed effects, more deeply thought-out instrumental variables (including co-authors’ colleagues from other institutions) and a mobility model measuring whether or not colleagues move to a different institution in a certain period. In short, this paper is a must for any researcher interested in determining causal effects through network effects, social interactions or spatial interaction effects.

The second paper in this issue, by Gluschenko (2021, this issue), investigates the law of one price; that is, whether the price of the same good is the same across all regions or does not differ more than the costs of transporting it from one region to another. This law is investigated for an aggregated good (a basket of 33 basic foods) using monthly data over the period January 2002-December 2019 for 79 (both European and Asian) regions in Russia. Starting from a first-order serial autoregressive model, the author develops several econometric models explaining the price differential between all region pairs in the sample. Based on the results, Gluschenko further distinguishes four forms of market integration between the regions: perfectly integrated, conditionally integrated, tending towards integration, and non-integrated. This paper follows on from two recent studies by Rokicki and Hewings (2019) and Weinand and von Auer (2020), published in Spatial Economic Analysis, which have developed advanced methods to construct consumer price indices at the sub-national level in Poland and Germany respectively. The results of these previous studies can be used to investigate the extent to which the law of one price also holds in these countries. In addition, they offer interesting opportunities to further integrate the methodologies used in these different studies.

This issue continues with a contrarian economic-theoretical model of spatial sorting, developed by Forslid and Okubo (2021, this issue), where the least productive entrepreneurs are drawn to the large (most populated) core region. Although the literature on spatial sorting typically shows that the most productive individuals and firms agglomerate at the core, this paper is inspired by contradictory observations that poor mega-cities are emerging in the developing world and poor individuals concentrate in city centres of the rich world. The basic idea behind this paper goes back to Harris and Todaro (1970). The authors of this classical study show how migration to the city is rational, even though the risk of becoming unemployed is substantial. This behaviour is rational if migration decisions are based on expected earnings and wage levels in the city are higher than in rural areas. The driving force behind the migration decisions of entrepreneurs in this economic-theoretical model is the better consumption opportunities measured by a local price index. In this respect, this paper shows that the determination of local prices - or the evaluation of the law of one price, as investigated in the previous paper of this issue - are relevant contributions to the literature. In short, this paper is a must for researchers interested in agglomeration effects and new economic geography models.

The authors of the fourth paper in this issue (Del Gatto & Mastinu, 2021, this issue) estimate the probability of observing merger and acquisition (M&A) deals between firms belonging to culturally contiguous countries. More specifically, they test whether this probability is positively affected by knowledge-related experience effects fostered by a firm’s investment history in foreign destination countries, as well as in countries that are culturally contiguous to these destination countries. The main idea behind this hypothesis is that, by investing in a given country, the firm increases its capacity to take advantage of further investments in that country or in a country which is culturally similar to that one. According to the authors, a similar experience effect should not be at work in M&A deals directed towards geographically contiguous countries, since these deals are mainly intended to reduce transaction costs. To investigate these hypotheses, the authors build a synthetic measure of cross-country cultural distance, based on linguistic, religious and genetic distance indicators, and distinguish M&A deals with culturally (but not geographically) contiguous countries, and M&A deals with geographically (but not culturally) contiguous countries. The econometric analysis is based on 24,402 M&A deals realized by 17,457 firms located in one of the OECD countries directed to 143 countries worldwide. The empirical evidence strongly points to both the presence of experience effects associated with cultural distance and the absence of such experience effects associated with geographical distance. The policy implication of this finding is that market openness alone cannot counteract protectionism.

Spatial econometric models are generally estimated by maximum likelihood (ML), quasi maximum likelihood (QML), instrumental variables (IV), generalized method-of-moments (GMM), or Bayesian Markov Chain Monte Carlo (MCMC) estimators. Recently, researchers have begun to develop another estimator, based on the indirect inference principle. Initially, Kyriacou et al. (2017) started to consider this estimator for the spatial autoregressive (SAR) model. Then, Kyriacou et al. (2019) and Bao et al. (2020) extended this spatial econometric model with exogenous regressors and heteroskedastic errors. The paper by Bao and Liu (2021, this issue) further extends the latter model by also considering a spatial lag in the error term specification. This model is also known as a SARAR model, a spatial autoregressive (AR) term in both the dependent variable and the error term. Note that the need to consider heteroskedastic disturbances has recently also been advocated by Taşpınar et al. (2019) and LeGallo et al. (2020). This paper is a must for spatial econometricians interested in an estimator that is computationally simpler than the (Q)ML estimator and free of the choice of IV or moment conditions. As an application, the authors explain Airbnb listing prices. By using J-nearest neighbour matrices, they find that the significance level of the coefficient of the spatial lag in the error term in the case of J = 20 is overestimated by the commonly used IV and GMM estimators compared to their proposed indirect inference estimator.

The sixth paper in this issue, by Mirzaei et al. (2021), examines the investment behaviour of First Nation governments (FNGs) in Saskatchewan, Canada. It describes how FNGs often invest through joint ventures with other FNGs. As shareholders in these joint ventures, an individual FNG cannot make investment decisions independently of other FNGs. Spatial econometric methods are used to model this behaviour. Using a Bayesian comparison approach, the authors jointly compare the performance of three potential specifications of the spatial econometric model and ten different specifications of the spatial weight matrix. This approach is one of the latest techniques developed in spatial econometrics to identify which type of spatial lags in combination with which type of spatial weight matrix best fit the data. It is therefore a must for every empirical researcher in spatial econometrics.

The seventh paper in this issue, by Torój (2021, this issue), is a mix between input-output analysis and spatial econometrics. Generally, survey and non-survey methods are used to construct so-called intraregional input-output models from their national counterpart. Spatial Economic Analysis has published several papers dealing with this problem using ever more advanced and sophisticated methods (most recently Temursho et al. (2021), Jahn et al. (2020) and Pereira-López et al. (2020)). As in the previous paper, this model reflects the notion that regions should not be treated as independent entities and that statistical information about nearby regions should be given a higher weight than that of remote regions. The authors show that a spatial Durbin model results in which the spatial weight matrix (W) is endogenous rather than exogenous, as is standard in the spatial econometric literature. This is similar to the first paper in this issue by Cheng and Weber (2021). As already indicated, estimating spatial econometric models with endogenous spatial weight matrices is not new - see previous studies by Kelejian and Piras (2014), Qu and Lee (2015), Bhattacharjee et al. (2016), Cheng and Lee (2017), Delgado et al. (2018), and Corrado et al. (2019). This study proposes to estimate the elements of W by the gamma cumulative distribution function and to jointly maximize the log-likelihood function of the parameters related to the spatial Durbin model and to this distribution function of the elements of W. In this sense, the paper overlaps with earlier work by Gross (2019), who also developed a (different) method to estimate the elements of a spatial weight matrix jointly with the parameters of a spatial econometric model. In short, this paper is a must for all researchers interested in the construction of intraregional input-output models and the estimation of spatial weight matrices.

Finally, the editorial board would like to congratulate Abhimanyu Gupta for the best referee award 2021 for reviews delivered in 2020, and the authors Xun Zhang, Guanghua Wan, Jing Li and Zongyue He of the paper “Global spatial economic interaction: knowledge spillover or technical diffusion?” for the best paper award 2021 published in the 2020 volume (Zhang et al., 2020).