Spatial macroeconomics

ABSTRACT

This special issue on spatial macroeconomics aims to bridge the divide between spatial and macroeconomics. Defined in the introduction, spatial macroeconomics explores the interactions between economic activity and geographical space. The issue comprises eleven papers authored by a total of 32 researchers. These papers were selected through a combination of solicited submissions and an open call for contributions. Four papers within this special issue delve into spatial macroeconomic theory. They cover topics such as agglomeration economies for innovation, a neoclassical spatial general equilibrium growth model, the spatial sorting of heterogeneous workers and the impact of national industrial policies in strategic industries on trade. Additionally, seven papers offer empirical studies that encompass a wide range of methodologies. These include general equilibrium models, input-output-based analyses and econometric models. The empirical research addresses various topics, such as the impact of trade on productivity, the trade-off between efficiency and equity, fiscal assistance, local and nationwide fiscal multipliers, forced human displacement during wars and the spatial diffusion effects of renewable energy resource deployment.

KEYWORDS:

• General equilibrium
• models
• regional impacts
• policy
• economic development

JEL:

• C
• D58
• E6
• F1
• O1
• Q4
• R

1. INTRODUCTION

Macroeconomics is generally considered a-spatial. As it deals with the behaviour of aggregate measures of economic activity, concern over where this activity takes place is overlooked. However, the magnitude of measures such as gross domestic product (GDP), aggregate supply or total factor productivity, is inherently tied to where they take place. In this way, developments in the spatial economy impact the macro economy and concomitantly, developments in the macro economy influence the spatial economy. This interdependence is the hallmark of what we term ‘spatial macroeconomics’.

A central tenet of macroeconomics is that aggregate national activity is independent of the micro units of which it is comprised. Just as the individual firm in a production network or the individual share in an investment portfolio cannot affect aggregate performance, so by analogy, individual spatial units cannot influence the macro economy. The central limit theorem ensures that these granular units remain in the tails of the distribution. This granularity condition underlies not only macroeconomics but also asymptotic theory in statistics. If this condition did not apply, statistical distributions would not be asymptotically normal and would instead tend to be fat-tailed. In a fat-tailed distribution, individual units can be conceived as affecting a whole system. If the macro economy is considered a ‘system of systems’ (Donaghy, 2021), these effects are amplified.

The subprime mortgage crisis of 2007–08 that hit the US caused reverberations across the global economy. It also generated a fallout in macroeconomics (Beenstock, 2022a). Macro-economic policy failed to predict the crisis and responded with a suite of measures such as zero interest policy and quantitative easing that only exacerbated matters. This policy response in turn reflected a crisis in both macroeconomic theory and macroeconometrics. The former paid scant notice to the relationship between inflation rates and interest rates. The latter relied on the workhorse dynamic stochastic general equilibrium (DSGE) model that ignored spatial general equilibrium and assumed that the marginal products of capital and labour are equated in equilibrium.

While the subprime meltdown caused a reconsideration of macroeconomics, it had less noticeable effects on spatial economics. For example, the advent of the crisis did not evoke a reassessment of the way interest rates, returns to capital and national savings affected the distribution of foreclosures. Similarly, the role of macroeconomic policies in exacerbating regional disparities was not given due consideration. As Beenstock (2022b) has noted, macro indicators such as national interest rates can intensify regional gaps according to the level of capital intensity in a region. National exchange rates can impact regions according to the latter’s dependence on the production of traded goods. National price levels can affect regions differentially according to regional variations in the demand for money. Macroeconomic policy can therefore be as pertinent as spatial economic policy in determining the extent of regional gaps.

Against this background, the current special issue on spatial macroeconomics is an attempt to bridge the divide between spatial and macroeconomics. While a full-fledged synthesis of the two fields is beyond the remit of this thematic collection, a reconsideration of common foci and overlaps in theory, methods and empirics is certainly in place. In contrast to many special issues, the theme of ‘spatial macroeconomics’ is not motivated by a conference, commemorative event or research project. Rather it is driven by the desire to fill a knowledge gap in an area previously unaddressed. To that end, we have both solicited papers from both targeted authors and via a public call. The response has been encouraging indicating a real desire to explore hitherto uncharted territory.

1.1. Spatial macroeconomic theory

Spatial economic theory inherently deals with spatial transaction costs, spatial externalities and spatial sorting, all typically at a microeconomic scale. It is from this spatial micro-foundation that spatial economists and economic geographers extend the study of economic behaviour to spatial decisions. Urban and regional economic theories aggregate this spatial micro-foundation to the city or regional level, often by making parsimonious spatial assumptions about the regional nature of economic activity, rather than explicitly modelling the underlying spatial microeconomic mechanisms. In much the same way, macroeconomics has traditionally made overarching assumptions about the aggregate nature of economic activity rather than explicitly modelling the underlying a-spatial microeconomic mechanisms. The similarities between spatial economics and macroeconomics make their slow fusion somewhat surprising, but also indicate the significant potential for new insights.

This common modelling approach has led both fields to make significant advances by explicitly modelling these micro-foundations, such as the optimising behaviour of representative agents. Macroeconomic theories of growth recognised that economies of scale for innovative firms were required to endogenise growth (Romer, 1990). Similarly, economies of scale combined with transport costs endogenise uneven spatial development (Krugman, 1991), while local characteristics and geography also incentivise location choices (Roback, 1982). This approach of advancing an explicit micro-foundation has continued in both fields in parallel. But such developments are not always in isolation of one another. Regional economics has frequently borrowed approaches from macroeconomic models of growth. For example, Rappaport (2004) shows how shocks in a neoclassical growth model influence migration flows from a small to a large region. Macroeconomics has also taken spatial perspectives from the field of international trade. Models of growth and trade have long acknowledged the effects of international openness on aggregate economic performance (Edwards, 1993). And trade models can explain differences in inflation rates between developed and developing countries (Ghironi & Melitz, 2005). Yet such findings from international trade do not always acknowledge the spatial mechanisms that lead to such results or extended their results to regional models within countries.

Spatial transaction costs and spatial externalities generate external increasing and decreasing returns to scale that imply concentration and dispersion forces that influence spatial sorting and interdependent location choices (Proost & Thisse, 2019). Increasing returns to scale and monopolistic competition also provide a foundation for studying the aggregate economy (Matsuyama, 1995). The interesting challenge for spatial macroeconomic theory is to fuse parsimonious assumptions about increasing and decreasing returns to scale from an a-spatial context in ways that reflect reality in the spatial economy when such assumptions imply both intended and unintended spatial forces (Bond-Smith, 2021).

The theoretical papers in this special issue examine macroeconomic phenomena while explicitly recognising and emphasising the role of spatial economic mechanisms: a micro-foundation for agglomeration economies for innovation in a model of endogenous growth without scale effects; the relationships between regional output, aggregate output and the interest rate by modelling neoclassical growth in spatial general equilibrium; the implications of housing constraints for productivity and inequality through the sorting of heterogeneous skilled workers; and operationalising the notion of strategic industries in a trade model where one location has a strategic advantage.

1.2 Empirical spatial macroeconomics

Empirical research in spatial macroeconomics has traditionally studied the distribution of economic activities such as aggregate production, consumption, investment and trade across geographical areas, and the effect of this distribution on overall macroeconomic performance. A central theme in this research is the study of regional economic disparities and their implications for global economic growth and regional convergence (Baumol, 1986; Corrado et al., 2005; Mankiw et al., 1992; Quah, 1997; Sala-i-Martin, 1996; among others). Researchers have investigated why certain regions experience higher levels of economic activity and growth, while others lag behind examining factors such as natural resources, human capital, infrastructure and institutional quality that contribute to regional disparities (Islam, 1995; Jones, 2016). When defining specific drivers of local growth and convergence, researchers have estimated models in which spatial factors are captured by agglomeration economies where firms and individuals benefit from spatial proximity (Ottaviano & Puga, 1998; Ottaviano & Thisse, 2004; Redding, 2010). Abstracting from the specific economic forces that characterise the spatial estimation of structural models, a more agnostic approach is to use reduced form setups such as global vector auto-regressive (GVAR) models to quantify the international macroeconomic linkages among countries/regions and provide insights into the transmission mechanisms of shocks (Dees et al., 2007).

In recent years Spatial Economic Analysis has published articles that use spatial econometric techniques to measure macroeconomic outcomes such as business cycle synchronisation and spatial asymmetries in monetary policy effectiveness and macroeconomic volatility among regions/states (see, among others, Cainelli et al., 2021; Capasso et al., 2021; Fiaschi et al., 2017). Overall, the empirical papers in this special issue further contribute to the empirical field of spatial macroeconomics by addressing various challenges and using innovative methods to understand the complex interactions between economic forces, policy interventions and spatial dynamics also in light of the challenges posed by major crisis events such as the recent pandemic. They highlight the need for integrated approaches that consider both macroeconomic aggregates and localised impacts in order to inform effective policymaking.

The empirical papers in this special issue employ various methodologies to investigate these intertwined effects: the fusion of dynamic microsimulation and spatial regional macroeconometric models to analyse economic policy impacts in the UK; interregional general equilibrium models to examine the effects of fiscal policy through fiscal multipliers at both local and national levels; high-resolution satellite data to explore the economic impact on the host regions of human displacement during wars; dynamic spatial general equilibrium models to evaluate the macroeconomic impact of the EU cohesion policy; multiregional input-output models to investigate the spatial effects of energy policies; finally, two research studies in this special issue use spatial instrumental-variables models and directed graph theory to examine the macroeconomic spillovers of federal aid in the US and of local productivity shocks in the EU.

2. THEORETICAL STUDIES

Bond-Smith (2024, in this issue) builds a quantitative spatial model (Redding & Rossi-Hansberg, 2017) to incorporate agglomeration economies for innovation into a spatial model of endogenous growth without scale effects. In this model, the productivity of innovation effort in each firm depends upon its proximity to the innovation efforts of other firms. Bond-Smith pays particular attention to scale effects, because increasing returns are required to endogenise growth but also imply concentration forces. By using a Schumpeterian model of endogenous growth without scale effects (Peretto, 1998) increasing returns to scale in the ideas production function do not create any unintended spatial forces (Bond-Smith, 2021). This allows the specific assumptions about agglomeration economies and land rents to drive the spatial equilibrium. The model generates a geographic concentration of innovation in large cities (Audretsch & Feldman, 1996), although it does not guarantee it, meaning that models of this nature could be calibrated to accommodate differing rates of innovation in different cities. This simple spatial micro-foundation offers the potential for a rich and detailed economic geography that simply specifies city locations and initial productivity levels in otherwise continuous space. Other spatial foundations could easily extend the model to understand the macroeconomic implications and trade-offs that result from various other spatial mechanisms.

The paper then explores the macroeconomic implications of including a spatial agglomeration mechanism for innovation effort, but the approach could also be used for any other spatial mechanism. In this model, macroeconomic outcomes are related to the spatial distribution of economic activity but not its scale. Agglomeration economies for innovation enhance the return on investment in innovation efforts. This means that greater concentration of economic activity increases growth rates since innovation efforts generate greater improvements. Spatial concentration also enhances productivity levels by freeing up labour from innovation and shifting it to production. However, greater returns to innovation effort consequently increase the required rate of return, deterring entry but increasing innovation efforts per firm. This means that both aggregate productivity levels and growth rates are functions of the spatial distribution of economic activity. The model demonstrates how spatial economic policy affects macroeconomic outcomes. Urban planning and spatial migration patterns have inherent macroeconomic implications for both transitional and long-run macroeconomic conditions. The model is an interesting departure towards examining economic models in continuous space from existing two region models of endogenous growth without scale effects (Bond-Smith et al., 2018; Davis & Hashimoto, 2015a, 2015b; Minniti & Parello, 2011).

Beenstock (2023, in this issue) takes a more traditional approach to macroeconomics with a neoclassical growth model, extended to spatial general equilibrium using Roback (1982). Neoclassical growth is naturally paired with Roback due to a common foundation based on perfect competition, enabling a tractable theory that elegantly explains the relationships between regional economic performance, aggregate performance and the interest rate. Crucially, the neoclassical growth model explicitly includes capital which is often assumed away for parsimony in new economic geography (NEG) and endogenous growth models. Whereas many models in spatial economics focus specifically on regions and places, Beenstock’s model is distinctly macro, fully embracing the theme for the special issue.

In the initial specifications, where savings rates are the same in both regions and capital is perfectly mobile, the spatial equilibrium affects GDP but has no effect on other macroeconomic variables. This changes as spatial heterogeneity and frictions are added to the model. When savings rates vary between locations the rate of interest also depends upon the spatial equilibrium. Importantly, frictions in capital mobility lead to feedback from the macro economy on the spatial equilibrium, meaning the spatial and macro economy are interdependent. The model demonstrates a powerful approach for integrating spatial- and macro-economic theory. Beenstock suggests a number of opportunities for future extensions in this spatial framework. These opportunities for spatial macroeconomics are notable since a-spatial macroeconomics has already made a number of such advances with Heterogeneous Agent New Keynesian models (Kaplan et al., 2018).

Truffa and Montecinos (2023, in this issue) examine how spatial sorting of heterogeneous workers in general equilibrium affects productivity and inequality. The extent of spatial sorting is limited by bidding wars for a constrained housing supply. This means that cities with high amenities attract higher skill workers along the lines of Richard Florida’s (2002) creative class, but in order to access those amenities workers try to out-bid each other for the limited availability of housing as in the Superstar Cities approach of Gyourko et al. (2013). The article’s focus on the contribution of spatial sorting to macroeconomic outcomes is the reverse of Hsieh and Moretti’s (2019) study that focuses on the spatial misallocation caused by housing constraints. Truffa and Montecinos’ model offers greater richness with heterogeneous skills and productivity between workers, interacting spatially due to agglomeration externalities, meaning that productivity differences emerge endogenously through spatial sorting and the attractiveness of amenities.

They calibrate the model to US data in order to estimate how much spatial sorting contributes to aggregate outcomes. The sorting of high-skill workers into big cities combined with agglomeration externalities increases aggregate productivity by 1.9%, and by 30% in those superstar high-amenity cities. Yet constraints on housing supply mean that spatial sorting increases housing costs in those cities by 20–40%, while smaller cities experience housing prices that are 30% lower due to the loss of high-skill workers. Spatial sorting also exacerbates wage inequality by 7.5% as higher-income workers out-bid low-skill workers in housing constrained cities and the remaining low-skill workers are compensated due to their diminishing local supply. Interestingly, place-based policies to increase housing supply by only 1% in constrained cities increase productivity by 0.2 to 0.4% but further exacerbate wage inequality.

The fourth theoretical paper by Colantone et al. (2023, in this issue) considers how macroeconomists can use general equilibrium trade models to examine the recent trend towards national industrial policies in strategic industries. General equilibrium trade models provide an international spatial perspective on aggregate economies but have so far not considered the role of strategic industries. The authors propose to define a strategic industry as one that generates positive nation-wide externalities and justifies an industrial policy to maximise national welfare. They also introduce a locational advantage to one country in the strategic industry in the form of comparative advantage and market access.

As trade gets freer, countries with a locational advantage achieve greater gains from trade as strategic industries relocate to the more advantageous location. Conversely, countries with a locational disadvantage achieve smaller gains, although the effects differ depending upon the form of competition. The gains from trade for the country with a locational advantage increase with monopolistic competition, while inverse effects occur for the country with a locational disadvantage. This highlights the importance of the form of competition in trade models, which has previously been less understood.

3. EMPIRICAL STUDIES

The paper by Dai (2024, in this issue) underlines the role of regions in shaping aggregate output in macroeconomics. Looking at inter-regional trade flows and knowledge spillovers, it studies the empirical pervasiveness of local productivity shocks at the regional level on productivity at the macroeconomic level. The paper provides an empirical procedure to identify the pervasive regions in inter-regional research and development (R&D) spillovers and trade and test their influence on macroeconomics as common factors. Spatial panel data for 272 EU NUTS 2 regions (1991–2019) is used to estimate the extent of this effect and to tease out the relative contributions of economic and geographic centrality. Regional pervasiveness is identified on the basis of the ranking of interaction matrix outdegrees from trade flow matrices, as frequently applied in network analysis.

In terms of this special issue, the paper combines economic geography, macroeconomics and spatial panel data econometrics and makes a series of contributions to spatial macroeconomics. First, it introduces the macroeconomic notion of granularity into economic geography. It tests the granularity hypothesis that posits that macroeconomic activity is unaffected by individual units in a system. As is well known, the celebrated Lindeberg–Levy (CLT) condition empirically breaks down in various contexts such as city size distributions, firm size distributions and production networks where individual units such as firms and sectors can influence entire production networks (Acemoglu et al., 2012). The paper by Dai extends this work from the domain of firms and sectors to that of regions. It uses spatial panel data econometrics to investigate the notion of granularity at different spatial scales, both theoretically and empirically.

The paper also contributes to the study of granularity. It introduces the use of a non-granular, endogenous spatial weight matrix in spatial panel data econometrics focusing on empirical research of productivity spillovers across EU regions. This extends work on endogenous weight matrices published in Spatial Economic Analysis (Delgado et al., 2018; Kubara & Kopczewska, 2024) highlighting the importance of economic and not just geographic distance. From a policy perspective, the paper identifies key productivity generating regions in the EU and their highly concentrated production networks. Methodologically, a framework is presented that unifies both the notions of spatial dependence and common factors in spatial economic analysis. This is done in three steps. First, the extent of a regional productivity spillover is measured by estimating inter-regional trade flows through gravity equations. Then, the most pervasive regions in terms of technology spillover are identified based on total factor productivity (TFP) regressions. Finally, the aggregate effects of the pervasive regions on their national macro economies over time are considered common factors in the regional production function.

The results show a highly centralised dependence of productivity on domestic and imported R&D capital stock over time. For some years, this dependence is reflected in a power law coefficient of around 2, which supports the granularity hypothesis of aggregate regional productivity. A few technological leader regions account for this dominance. For example, the total technological multiplier of the leading region (Ile de France) is 2630 times larger than that of the most technologically laggard region (Aaland, Finland). The top ten most pervasive regions in the EU explain about 12% of TFP growth and play a significant role as common factors in the production function of almost every other region. Thus, a 1% increase in R&D stock in Ile de France will generate average TFP growth of 0.01% in the other NUTS2 regions. The lesson emerging from this paper is that macroeconomics cannot afford to overlook the role regions play in the national economy.

The next paper by Barbero et al. (2024, in this issue) conveys a message similar in spirit. This paper couches the popular efficiency-equity trade-off in terms of the relationship between macroeconomic and spatial economic outcomes. For example, assistance to poorer regions (equity outcomes) can come at the expense of higher national efficiency in terms of GDP or returns to investment. The existence of such a trade-off challenges alternative pathways to economic growth such as the trickle-down effect (Aghion & Bolton, 1997). Barbero et al. simulate shocks to EU cohesion policy in different regions and observe their effects with respect to public expenditure and investment, R&D, transportation infrastructure and labour productivity. They utilise the European Commissions’ well-known spatial CGE model (RHOMOLO) developed by the JRC-Seville team (Brandsma et al., 2015) and reported in an earlier paper in Spatial Economic Analysis (Di Pietro et al., 2021). This is a dynamic, recursive general equilibrium framework for analysing macro-wide impacts of regional shocks. As such, this paper is very much a spatially-inspired contribution to the theme of spatial macroeconomics. It illustrates the way spatial economists believe that the spatial economy matters for aggregate outcomes in both short-term demand, such as public expenditure, and longer-term supply, such as transportation infrastructure or labour productivity.

The RHOMOLO model is a scenario and policy-testing tool. It generates place, time and sector specific outputs for 10 NACE 2 classifications (sectors). Barbero et al. use this tool to simulate policy shocks relating to 105 NUTS2 regions in cohesion fund-recipient countries for a 20-year period. To estimate the equity efficiency trade-off, a calibrated steady state is perturbed with respect to region-specific policy shocks in one of the above policy areas. While these shocks are all of the same monetary magnitude, the different interregional trade flow structures, wage differentials, migration costs and so on, generate dynamic endogenous aggregate effects in variables such as GDP, prices consumption, imports and exports over time, regions and sectors. The magnitude of region-specific shocks is expressed in terms of regional multipliers, national multipliers and their associated spillovers. The latter form the link between the spatial effects and the aggregate macroeconomic outcomes. For example, an investment shock may generate a large national multiplier without exacerbating regional imbalances. This rising-tide-lifts-all-boats outcome can occur through the spillover effects of investment in a (rich) target region not being locally contained. In this way, spatial versus macroeconomic outcomes are also mapped onto the equity-efficiency trade-off. The simulations in the paper do not support the trickle-down thesis of economic development. Aggregate returns to investment in rich target areas are in fact locally contained and this agglomerative outcome tends to intensity with increasing elasticity of substitution.

The paper notes that identifying the conditions necessary for the equity-efficiency trade-off to emerge is a nuanced issue. The equity-efficiency dichotomy is often reconciled in real-world policy settings which in practice can generate both equity and efficiency outcomes. Their relative shares depend on the intensity of initial conditions, spillovers and inter-regional trade patterns. The paper underscores an inescapable fact for macroeconomics: policy shocks occur in real places and in the advent of a shock, labour and capital will move from low to high TFP places. Conversely, spatial economics cannot overlook the reality that the movement of labour and capital depends on national interest, savings and investment rates.

The paper by Clemens et al. (2023, in this issue) looks at the surprisingly limited propagation effects of large-scale fiscal assistance to states and local governments in the wake of COVID-19. Given the experience of the federal relief packages that operated in response to the global recession of 2008 and that generated cost-effective aggregate economic activity, it might have been expected that similar positive outcomes would result from the large-scale stimuli operated in response to the global pandemic. However, contrary to expectations, the paper finds a constrained response for both direct and indirect (spillover) impacts on neighbouring jurisdictions and for aggregate economy-wide impacts represented by macroeconomic variables such as per capita employment, wages, GDP and personal income. To bypass the endogeneity issues inherent in estimating the relationship between assistance and economic outcome variables such as employment, and to address the spillover effects of a state’s economy being influenced by that of its neighbours, the paper conveniently exploits the fact that many small states are over-represented in Congress. The authors instrument for the aid that a state’s neighbours might receive based on this overrepresentation.

From the perspective of spatial macroeconomics, this paper seemingly offers a further study of how macro-economic outcomes depend on initial spatial (equilibrium) conditions, the extent to which the productivity of capital is state-specific and imperfectly mobile and the limited impact of spillovers. However, the COVID-19 context offers a further insight into the relationship between the spatial and the macro economy. The pandemic and its associated constraints on mobility and interaction would seem to have truncated some of the linkage mechanisms between the spatial and the macro economies. This could explain the restricted growth-generating impacts of the COVID-19 aid package despite the fact that it was three times the size of economic stimuli disbursed after the global recession over a decade earlier.

The global economy, including the UK, was severely disrupted by the COVID-19 pandemic, which affected supply chains, labour mobility and production dynamics. Brexit further exacerbated these effects by reducing global trade and possibly changing regional ties, particularly between Northern Ireland and the Republic of Ireland. These events highlight the need for novel modelling techniques to understand macroeconomic dynamics at the regional level and guide policy decisions.

Bhattacharjee et al. (2024, in this issue) explore these complex and interconnected effects at the macroeconomic and regional levels in the UK in the aftermath of major recent crises. Although previous efforts have focused on improving macroeconomic models after the Great Recession, they often overlook the regional or spatial dimensions. The authors bridge this gap by integrating spatial heterogeneity and spillovers into macroeconomic models while considering the distributional impacts at the household level. The need for this integration is emphasised by the heterogeneous spatial impact of recent events and the cost-of-living crisis.

To this end, the study presents NiReMS, a model developed by the National Institute of Economic and Social Research (NIESR), which accounts for agent heterogeneity at the regional level in the UK by combining dynamic microsimulation with a regional spatial macroeconometric model. NiReMS, unlike conventional approaches, uses aggregate national input-output tables and estimates latent spatial weights to capture interdependencies between regions. By utilising dynamic microsimulation, it also integrates heterogeneity at the household-level, allowing a more refined examination of individual and household behaviour. The paper proposes a hierarchical approach to modelling spatial interdependence, considering both core-periphery relations and local deviations within the network. By tracking individuals in time and by considering both regional and national economic dynamics, NiReMS can provide short-run projections of the impact of various shocks and government policies on regional economies and household welfare. The integrated micro-macro model facilitates the analysis of consumption patterns across different groups and allows counterfactual policy experiments, such as quantifying the consumption losses due to the recent crises and the consumption gains of more recent policy interventions – energy cost caps and benefit payments.

In the context of global crises, including the recent COVID-19 pandemic, governments have implemented fiscal expansion measures to stimulate economies and mitigate job losses. However, quantifying the effectiveness of these fiscal packages throughout the business cycle and the magnitude of the fiscal multiplier is not clear as it varies because of alternative theoretical approaches, calculation methods and data sources. For example, the analytical and functional differences between Neoclassical and New Keynesian models (Gali, 2015) contribute to varying multiplier estimates. Empirical approaches such as vector auto-regressions (VAR) and dynamic stochastic general equilibrium models (DSGE) further complicate estimation because of different functional structures (Auerbach & Gorodnichenko, 2012; Blanchard & Perotti, 2002; Smets & Wouters, 2007; among others). In addition, methodological choices and the numerical structure of empirical studies influence multiplier outcomes. The choice of data can also lead to divergent results.

The study by Haddad et al. (2023, in this issue) uses a NUTS-2 interregional general equilibrium model calibrated with data of the Greek economy during the 2010–2013 crisis to quantify local and nationwide fiscal multipliers. Understanding the localised impact of fiscal policy is crucial for policymakers, as it provides insights into the effectiveness of their actions within a spatially integrated system.

The Greek case represents an interesting example because of its heterogeneous economy and different regional impacts. To this end, the study examines local fiscal multipliers linking regional output changes to regional government spending. The authors employ a static multiregional and multi-sectoral computable general equilibrium (CGE) model to analyse the short-run impacts of increases in government spending. By calibrating the model with different datasets representing distinct stages of the 2010–2013 crisis in the Greek economy, the study aims to understand how the different data at hand influence multiplier estimates. The selected years, 2010 and 2013, capture different phases of the crisis, allowing a comprehensive analysis of its impact. Ultimately, the study seeks to provide policymakers with valuable insights into the effectiveness of fiscal policy interventions in a spatially integrated system, drawing on the unique case of Greece.

Nsababera et al. (2023, in this issue) analyse the macroeconomic impact of forced human displacement during wars. While existing literature has explored the short-term effects of refugee camps on labour markets, consumer markets and household consumption, few studies have examined their long-term effects on urbanisation and local development due to data limitations. This study addresses this gap by examining the long-term urbanising effect of refugee camps in Tanzania from 1985 to 2015, using high-resolution satellite data combined with spatial economic data. By implementing a spatial difference-in-differences strategy, it investigates the persistent urbanising effect of refugee camps on the surrounding areas. The analysis reveals a modest but significant effect of refugee camps on the built-up area up to 100 km outside the camp boundaries, with the impact diminishing with distance. In particular, the built-up area within 100 km of a refugee camp increased by about 6% during its lifetime, and this increase largely persisted after the closure of the camp, especially in rural localities.

Matching camp locations with regional GDP data during the period of camp operation shows a faster growth in local economic activity in the regions that host camps than elsewhere. However, the duration of the pre-camp period limits the identification of trend growth rates. Furthermore, matching the data with household consumption and employment data during camp closures suggests a shift in employment away from non-agricultural activities post-closure. These results tentatively suggest an ‘urbanisation without growth’ phenomenon associated with refugee camps, which is in contrast with the traditional urbanisation processes linked to industrialisation and mining activity.

Finally, López-Bernabé et al. (2024, in this issue) apply a multiregional input-output model to study the spatial diffusion effects of renewable energy resources deployment. The macro economy considered in their analysis is the world, with micro units consisting of seven regions: Spain, the European Union (excluding Spain), China, the rest of East Asia, North American Free Trade Agreement (NAFTA), Brazil, Russia, India, Indonesia, Australia and Turkey (BRIIAT) and the rest of the world. Within each region, three main sectors are identified: manufacturing, construction and services, further subdivided into a total of 35 sectors.

The model is used to measure the effect of policy changes in one sector on another within the same or another region with respect to employment and gross-value-added (GVA). Specifically, the research investigates the domestic and cross-border effects of implementing various energy technologies in Spain. These technologies include conventional energy sources such as nuclear, coal-fired, gas-fired and diesel-oil-fired power plants, as well as renewable energy sources like biomass and waste-fired, wind, solar photovoltaic, solar thermal and hydropower plants. The motivation behind this analysis is to address climate change challenges and promote sustainable energy development.

Based on this model and four scenarios, the study derives a series of policy recommendations tailored to Spain’s context. The findings suggest that conventional energy technologies tend to generate slightly greater increases in employment and GVA compared to renewable energy alternatives. However, the authors propose that the benefits of renewable energy could be enhanced if Spain reduces its technological dependence on other regions through more effective industrial policies. Moreover, among the renewable energy options considered, hydropower plants emerge as the most promising in terms of their potential impact.

4. RETROSPECT AND PROSPECT

As noted in the introduction, while full integration between macroeconomics and spatial economics is beyond the scope of this special issue, the papers collectively identify potential pathways for further interdisciplinary collaboration. Inevitably, this collection has attracted contributions from spatial and international economists rather than from mainstream macroeconomists. This asymmetry is evident in the empirical studies presented, which predominantly examine the impacts of spatial economic factors on macroeconomic indicators such as aggregate supply, productivity and GDP. Conversely, less emphasis has been placed on exploring how macroeconomic conditions, including interest rates, exchange rates and general price levels, influence the spatial economy. Nevertheless, this area of research appears in the literature, albeit not macroeconomic, where the effects of macroeconomic variables such as exchange rates and monetary policy on regional variables such as house prices have attracted scholarly interest (Kim & Wang, 2024; Lin & Robberts, 2024).

Given our belief that spatial macroeconomics is characterised by a reciprocal relationship between spatial economics and macroeconomic issues, the question arises as to how such a synthesis can be formulated in retrospect. At the outset, the micro-foundations for any benchmark macroeconomic model would need to be based on aggregate behaviour of prototype ‘representative agents’. The contribution by Bond-Smith (2024, in this issue) outlines the micro-foundations for agglomeration economies in innovation and goes some way towards this goal. The next stage would be to describe the accounting mechanics of the aggregate flow of funds in the economy and the decision-making behaviour related to the operation of national monetary and fiscal systems. These are invariably missing in the canonical DSGE models of macroeconomics. Finally, the aggregate behaviour underlying investment, trade, government spending, consumption and labour supply and demand would need to be specified. In this special issue, Beenstock (2023) has shown how GDP and interest rates depend on the spatial general equilibrium and how the spatial distribution of labour and capital depends on the short-term financial frictions related to interest rates.

Another issue missing from the current contributions relates to integrating spatial economics with the two main macroeconomic approaches to economic growth: secular macro (long-term) growth on the one hand and cyclical (short-term) growth on the other. Both these approaches tend to compartmentalise macroeconomics and spatial economics ignoring any common ground between them. The DSGE models of long-term growth and real business cycle models of short-term growth do not assume any spatial differentiation in the economy. While the papers in this issue have addressed integrating spatial economics with long-term aggregate economic growth, the spatial aspects of business cycles have not been addressed.

Moving forward, we identify three challenges for integrating spatial economics with macroeconomics. The first challenge concerns the limitations of the workhorse DSGE models used in macroeconomics. These models traditionally assume perfect mobility of labour, capital and savings rates. However, reconciling this assumption with the realities of spatial economics – where labour mobility is limited by differences in quality of life, the productivity of capital varies by location and household savings are often invested in geographically distant markets – is complex. Although DSGE models assume that the marginal products of labour and capital are equal to their respective prices at equilibrium, this macroeconomic equilibrium is often decoupled from the spatial general equilibrium, reflecting a significant theoretical decoupling by design.

A second challenge is to rejuvenate the momentum originating from the new international trade theory towards the new economic geography (NEG). As early as 1979, Krugman noted that patterns of migration could be subjected to the same analytic models as international trade, thereby highlighting the interconnectivity between global and national macroeconomic dynamics and the significance of spatial factors (Krugman, 1979). In the same vein, the current challenge for national macroeconomics is to recognise the critical role of regions that comprise the national economy, which means that Krugman’s conceptual journey is not yet over.

The third challenge is purely speculative at this stage. While cryptocurrencies and fintech are generating a sea-change in national macroeconomics, their impact on the spatial economy has yet to be explored. As households increasingly invest in Bitcoins, Ethereum and the like, variation in their wealth – particularly in how these changes influence consumption patterns, equity investments and housing markets – could exhibit significant regional disparities. Therefore, integrating spatial economics with macroeconomics is essential to fully comprehend the potential consequences of these emerging trends.