Raising the bar (22)

ABSTRACT

This editorial summarizes the papers in issue 17(4) (2022). The first paper combines input–output modelling with priority weighting to analyse supply-chain impacts of disasters. The second paper examines skill-based functional specialization of value chains in Brazil using interregional and international value-added measures. The third paper questions the common belief that agglomeration economies are the driving force behind cluster formation using an agent-based model. The fourth paper applies modern instrumental variables techniques to measure the impact of forced migration flows from Venezuela to Colombia on house prices. The fifth paper explores the impact of an ageing population on per capita labour income, consumption and wealth at the regional level using a multivariate spatial econometric model. The sixth paper examines the impact of neighbouring countries on migrants’ aggregate decisions to remit based on an advanced spatial econometric origin–destination model.

KEYWORDS:

• Disasters
• input–output analysis
• agglomeration
• migration
• ageing
• flow data

JEL:

• D57
• F6
• F22
• F24
• R23
• R3

Spatial Economic Analysis is a pioneering journal dedicated to the development of theory and methods in spatial economics. This issue contains six papers contributing to the journal’s mission. The first two deal with input–output (IO) analysis; the third with agent-based modelling (ABM); while the last three estimate econometric models, focusing respectively on instrumental variables (IV), spatial vector autoregressive (VAR) modelling and spatial autoregressive origin–destination (OD) modelling.

This issue opens with Manfred Lenzen’s and Mengyu Li’s (Li et al., 2022, in this issue) Spatial Economic Analysis Annual Lecture given at the Regional Studies Association 2021 Winter Conference. This is normally held in London but was delivered online in 2021 due to the Covid-19 pandemic. The paper combines disaster analysis based on IO modelling with priority weighting to analyse supply-chain impacts of disasters. In their literature review, the authors briefly summarize more than 30 articles previously published on this topic, including Bouwmeester and Oosterhaven (2017). It is worth reminding readers of this article in particular because it analysed the economic impacts of natural gas flow disruptions between Russia and the European Union (EU) – a current topic of interest. Following Bouwmeester and Oosterhaven, the authors combine a multiregional IO and a constrained non-linear programming (MRIO-NLP) model to evaluate the impacts of the mid-2014 drought and drop in oil prices, and the subsequent drop in, respectively, crude oil and electricity production in Venezuela in the late 2010s. The key innovation is the use of decision weights, which give priority to specific gains and losses. One can use equity weights or distributional weights taken from healthcare planning and policy (Round & Paulden, 2018) based on the age and health condition of people (Gu et al., 2015), or try to extract them from other sources. Since the choice of weights is crucial for the outcomes, the authors pay attention to the distinction between gains and losses in the objective function, the magnitude of the decision weights and the interpretation of these weights. Since the number of disasters due to climate change is expected to increase, the methodology set out in this paper provides valuable ideas for future research.

The second paper, by Sanguinet et al. (2022, in this issue), analyses skill-based functional specialization of value chains in Brazil, considering both interregional and international value-added measures. Its contribution is to explore regional patterns of specialization using a novel approach to measure the skill intensity of value-added contributions to regional and global value chains based on occupational data and building upon previous work by Mudambi et al. (2018). The authors define the regional dimension by the 2015 interregional input–output (IRIO) model developed by the Regional and Urban Economics Lab at the University of São Paulo (NEREUS-USP) (see their online appendix with supplemental material), which covers 27 regions (26 states, plus the Federal District of Brasilia) and 67 industries. The results reveal a central role for the São Paulo region and specialization in highly sophisticated functions in the Southeast, while the rest of the country lags behind.

The third paper, by Van Roekel and Smit (2022, in this issue), criticizes the common belief of many researchers and policymakers that agglomeration economies are the driving force behind cluster formation. Using an ABM, the authors show that herding behaviour can create the same agglomeration patterns as agglomeration economies. Following Baddeley (2010, p. 261), herding behaviour is defined as ‘the phenomenon of individuals deciding to follow others and imitating group behaviours rather than deciding independently and atomistically on the basis of their own, private information’. Since this paper challenges the work of several previous studies, many researchers and policymakers, such as one of the reviewers of this paper, are likely to question the definitions, scenarios and ABM considered in this paper. However, the authors are among the first to admit that any model is an oversimplification of reality and that their results indicate what is possible rather than what is true in the real world. This paper warns the reader that cluster formation may have more and/or different causes than just agglomeration economies. In this way, herding behaviour could be thought of as one of many positive externalities from spatial proximity, like other positive externalities such as knowledge spillovers contribute to clustering behaviour and agglomeration economies (Bond-Smith et al., 2018; Bond-Smith & McCann, 2020). It is therefore a must-read for every spatial economist active in this field.

The fourth paper, by Forero-Vargas and Iturra (2022, in this issue), deals with a socially relevant problem. Just like the aforementioned paper by Li et al. (2022, in this issue), it deals with the economic and social crisis faced by Venezuela. Since 2013 this crisis has led to massive population migration to neighbouring countries, with Colombia being the most affected. During this time more than 3 million people have left Venezuela, 1 million of whom settled in Colombia. In 2022, similar kinds of forced migration flows are taking place from Ukraine, especially to Eastern European countries due to the Russian invasion in February 2022. This topic is also closely related to that of Conigliani et al. (2021) in the previous issue,¹ who investigated the impact of natural disasters caused by climate change on forced outmigration flows in 14 South and Southeast Asian countries over the period 1990–2016. Forero-Vargas and Iturra investigate the impact on house prices; they find that a population increase in Colombian cities measured by the ratio of immigrants to residents causes a 1.25% increase in house prices. Importantly, they note that the ratio of immigrants to residents might be endogenous due to omitted variables, that is, immigrants can choose destinations with better amenities, which in turn drive up prices, or due to reverse causality, that is, immigrants may choose destinations with lower housing costs. When the authors do not account for omitted variables and reverse causality, house price increases are in the range of 1.64–1.96%. To account for endogeneity, the authors first introduce two IVs, which allow them to test whether their instruments are both strong and exogenous. One of the instruments is the current level of immigrants predicted by using the distance between spatial units of host and origin countries as well as the proportion of native population living in the foreign country in the past, as suggested by Caruso et al. (2019). In addition, the authors demonstrate that these IVs should do more than just explain house prices, based on the so-called exclusion restriction, since this restriction does not appear to be fully satisfied. To this end, they continue to apply the latest IV techniques developed by Conley et al. (2012) to provide a confidence interval for the IV estimates under a certain degree of violation of the exclusion restriction, as well as Lewbel (2012) who proposes the use of residuals from the first-stage regression as internal instruments to avoid using any external information. The referees of this paper praised the use of these modern IV techniques and indicated that this advanced approach also serves as a model for future studies.

The fifth paper in this issue, by Giannini et al. (2022, in this issue), sets out a three-equation spatial econometric model explaining per capita labour income, consumption and wealth in the working population. The dependent variable observed in one unit in each equation can be affected by two factors. The first relates to the dependent variable itself observed in other units at the same moment in time, though not by one of the two other dependent variables. The second factor relates to the time and space–time lags of all the dependent variables observed in their own and neighbouring units. The model also includes control variables as well as a spatial lag and a unit-specific random effect in the error term. It analyses the macroeconomic dynamics of an increase in the average working age in Italy. For this purpose, the authors use impulse response functions. The model is estimated based on biannual data of 102 Italian regions over the period 1993–2016 (T = 12). It is important to point out that the use of multivariate spatial econometric models, including models that allow the dependent variable in one equation to affect the dependent variable in another equation at the same moment in time, is increasing (see Yang & Lee, 2019; and Elhorst & Emili, 2022, for recent contributions). This paper fits within this trend.

The sixth paper, by Laurent et al. (2022, in this issue), examines the impact of neighbouring countries on migrants’ aggregate decisions to remit. The analysis explicitly considers origin and destination spatial effects, also known as a second-order spatial autoregressive model, considered for the first time by Brandsma and Ketellapper (1979) and further developed by LeSage and Pace (2008) in the context of OD models. Unlike Nowotny and Pennerstorfer (2019), the aspatial macroeconomic literature studying the impact of networks on migrants’ location decisions has not accounted for the role of neighbouring countries or regions. The paper is relevant for spatial economic researchers for the following reasons. First, it aims to assess the extent to which the inclusion of ‘neighbouring countries’ affects remittance estimates from both an origin and a destination perspective. Second, it estimates the relative importance of alternative explanations determining remittance streams: the ‘altruism motive’, the ‘investment motive’ and ‘financial frictions or transfer costs’. Migrants are altruistic if they derive utility from their family’s utility, which may depend on the family’s level of consumption. Migrants invest if they send money back home to purchase assets and ensure their maintenance, or they may remit money in preparation for their future return to their country of birth. Remittances may also decrease due to financial frictions or transfer costs. Third, they estimate the prediction performance of the model in both the absence and presence of spatial dependence and find that the mean absolute prediction error reduces by 31–44% when accounting for spatial dependence. Fourth, and this is a remarkably strong contribution, they provide graphs of the local origin, destination and network effects that result when a shock occurs in one of the explanatory variables of remittances – in this case, electricity use per capita and population. They find that a shock to an origin characteristic has a larger impact on the remittance flows relative to an equivalent change in destination characteristics and that there is considerable heterogeneity across countries in their local network effects, especially within advanced economies. This heterogeneity depends on the degree of a country’s centrality, identified by eigenvector centrality, meaning the influence a country has on the given neighbourhood structure (following Bonacich, 2007). Finally, it attempts to estimate the expected impact of the ongoing Covid-19 pandemic on remittance streams. Overall, this paper serves as a model for future studies based on flow data.

With this issue, we close volume 17 and the calendar year 2022. This year we published 26 papers, of which six were open access and four were editorials. We also announced a new section, ‘Replication Studies’, devoted to short papers that replicate or extend published empirical results and discuss their sensitivity to relevant changes in the model, estimation method and/or interpretation (Ditzen & Elhorst, 2022). The journal’s impact factor further increased from 2.159 in 2021 to 2.317 in 2022.

Notes

1 See https://www.tandfonline.com/toc/rsea20/17/3?nav=tocList/.