Chapter 1. Why delineate functional areas in all territories?

This chapter provides the rationale for delineating functional areas in all territories, not only in urban areas but also in rural areas. It explains how functional geographies can complement administrative geographies. It discusses how functional areas can enrich the collection and computation of territorial statistics. Finally, it illustrates the potential benefits of the concept of functional areas for regional policy.

    

This chapter outlines the rationale for mapping functional areas in all territories. In OECD countries, the most commonly used type of functional area is based on cities and their economic areas of influence. However, economic linkages that define functional areas exist in all types of territories, urban and rural alike. A comprehensive approach to functional areas, therefore, goes beyond the focus on core-based metropolitan areas, instead aiming to offer the methodology to identify functional areas in all territories and specifically in predominantly rural regions.

To be effective, regional development policy must rely on sound and reliable statistics. A crucial aspect of such statistics is the geographic unit of analysis. In a number of policy domains, administrative boundaries often do not constitute the appropriate geographic scale to fully understand local economies and citizens’ economic reality. Instead, the need for meaningful geographies for analysis and policy requires the creation of several concepts, such as metropolitan areas, labour market areas, daily urban systems or, more generally, functional areas. These concepts have been used extensively by OECD countries with the purpose of complementing administrative areas.

While using functional areas can generate benefits in all territories, their predominant existing definitions tend to focus on cities and their surrounding commuting zones. Despite the work carried out and experience gained in a number of countries, the functional organisation of space in predominantly rural areas is something that still needs to be studied in depth from an international comparative perspective. This chapter explains the complementary nature of functional geographies relative to administrative geographies. It then explains how such geography can also offer an enriching perspective with which to look at regional development. Finally, it discusses how it can thus improve policy design and deliver better outcomes for citizens.

Functional and administrative geographies

Traditionally, administrative geographies have been the foundation of territorial statistics. They provide the framework for the production, analysis and understanding of economic and social geography as well as dynamics across space.

Functional areas cannot replace administrative geographies. In fact, they are generally created by clustering small administrative units; hence, functional areas should be regarded and used as an additional, complementary geography that can provide evidence on socio-economic trends across space and can help inform place-based policy. They can enhance the understanding of key economic trends that unfold on a spatial scale that is not properly captured by small administrative geographies. In fact, administrative boundaries sometimes do not adequately capture or reflect the geographic reality of economic activity (Casado Diaz and Coombes, 2011[1]). Furthermore, they can enrich conventional administrative statistics by offering precise information on policy-relevant areas in a way that facilitates better service provision.

A functional approach can improve the effectiveness of public policies. Economic relations, flows of goods and people, do not stop at the administrative border but inherently connect different areas. These economic interdependencies are particularly relevant to topics such as housing, transport and land use, all of which have external effects on neighbouring territories.

In the case of predominantly urban regions, the examples of Paris and Rome succinctly illustrate the challenges that policy makers face when administrative statistics are used to address such policy areas. In both cases, the administrative boundaries of the municipality do not correspond to the actual extent of the city (Figure 1.1). In fact, the administrative boundaries can often differ drastically from the city’s reality – or urban area. In Paris, the urban area is much larger than the municipality would indicate. In contrast, the boundaries of the municipality of Rome extend significantly beyond the actual urban area of Rome and includes smaller towns and rural areas. What the two capitals have clearly in common is a discrepancy between their economic or functional reality and the respective administrative areas.

Figure 1.1. Mismatch of municipal boundaries – Paris and Rome

Note: Urban areas in the figure above denote areas with a population density of more than 1 500 inhabitants per square kilometre.

Source: (OECD, 2016[2]).

As Figure 1.1 points out, there is often a scale mismatch between the social or economic reality of the geographies and the way they are structured and defined administratively. This mismatch can have significant implications as using the wrong geographic scale to inform policies can yield ineffective policy choices. National and subnational policy makers need an accurate evidence base on the relevant and respective geographic area to address urgent policy challenges and to ensure effective strategic planning. Administrative boundaries are often, however, the result of historical decisions and circumstances rather than a depiction of the current linkages that could define a region, city or labour market area in a functional sense.

The rise of new sources of data and the increasing availability of different subnational data may likely facilitate the delineation and use of functional areas. In many OECD countries, statistical information on key economic, demographic and social factors is becoming increasingly available at a detailed and more granular subnational level. The geographic granularity often goes beyond large regions (TL2), small regions (TL3) or municipalities. In fact, data are often made available at the level of small statistical enumeration areas or even at a regularly gridded scale.

Functional areas can offer information that is more precise for specific issues and thus enrich statistics that are based on administrative areas. For example, since functional areas are defined according to economic and social territorial linkages, they are likely to offer more meaningful perspective on topics such as employment, economic activity (gross domestic product [GDP] per capita) or earnings (see below for a more detailed discussion). As a consequence, they also allow more tailored policy design and evaluation on issues and challenges concerning those topics and can thus contribute to better living conditions for citizens in a particular place (see below). The rationale for functional areas is not limited to large cities or metropolitan areas. It equally applies to all types of territory as dispersed and complex patterns of mobility and commuting yield territorial linkages outside of large cities.

Better data for all territories: A statistical perspective

Functional areas offer a different perspective on statistics that can produce a more accurate picture of actual circumstances than administrative areas. In particular, socioeconomic conditions are better described by functional rather than administrative areas, precisely because functional areas are delineated and based on economic or social linkages across the territory or region.

When collecting data and producing subnational indicators, national statistical offices (NSOs) often face the challenge of choosing the right geographic scale. On the one hand, having the choice between different geographic areas can enhance policy making by looking at the most appropriate level of administrative areas. On the other hand, this choice illustrates the sensitivity of subnational indicators to changes in the boundaries of territorial units being considered. Adjustments to the size of the area analysed can yield significant changes to subnational indicators.

This issue, known as the modifiable areal unit problem (MAUP) and first prominently discussed by Openshaw, raises the question of whether there is an ideal geographic scale for territorial analysis (Openshaw, 1977[3]; Gehlke and Biehl, 1934[4]). Simply put, the MAUP highlights that the “results of any territorial analysis will partly depend on the areas used for that analysis”, meaning that any change to the area considered in the analysis will yield differences in the results (Casado Diaz and Coombes, 2011[1]). The MAUP is particularly pertinent for labour markets and economic indicators. Previous research demonstrates that the choice of subnational areas is relevant for labour market statistics and regional economic indicators such as average GDP per capita, as different scales can lead to significantly different statistics (ESPON, 2007[5]).

Labour market statistics are the primary topic where territorial statistics can benefit from the perspective of functional geographies. Regional economies and therefore regional or local labour markets do not necessarily correspond to administrative units. They are usually significantly smaller than TL2 regions but include various cities and municipalities and can extend beyond as well as across TL3 regions (see Box 1.1 for an explanation of TL2 and TL3 regions). Hence, reporting labour market statistics for established administrative units might yield a better representation of employment, unemployment or labour force participation for a given place than functional geographies would.

Box 1.1. OECD territorial classification

Regions within the 36 OECD member countries are classified by the OECD into 2 territorial levels that reflect the administrative organisation of countries. The 393 OECD large regions (TL2) represent the first administrative tier of subnational government, such as the Ontario Province in Canada. The 2 256 OECD small regions (TL3) correspond to administrative regions, with the exception of Australia, Canada and the United States. These TL3 regions are contained in a TL2 region, with the exception of the United States for which the economic areas cross the states’ borders. For Israel and New Zealand, TL2 and TL3 levels are equivalent. For example, the TL2 region of Aquitaine in France encompasses five TL3 regions: Dordogne, Gironde, Landes, Lot-et-Garonne and Pyrénées-Atlantiques.

Source: OECD (2019[6]), OECD Territorial Grids 2019, OECD, Paris.

In the case of remote or sparsely populated regions, functional areas can constitute a geography that might improve policy makers’ understanding of relevant policy challenges. On the one hand, functional areas in rural regions usually bring together (or cluster) administrative units below territorial level 3 (TL3) that are too small in terms of population to generate reliable and representative statistics.1 On the other hand, in those cases, the higher geographic level of TL3 regions may be too large to represent local labour markets properly. For example, the Canadian version of functional areas, self-contained labour areas, illustrates that significant parts of the country’s rural and sparsely inhabited administrative units jointly form functional areas with a larger population that might yield the critical mass to allow for reliable subnational indicators for that territory (see Chapter 5 for more details). Thus, clustering together small (below TL3) administrative units may generate an adequate compromise for the proper statistical representation of small territorial units.

Functional areas in the form of local labour markets are necessary for collecting and publishing labour market statistics across countries in a coherent and consistent manner, building on comparable territorial units (Casado Diaz and Coombes, 2011[1]). Taking into account economic territorial linkages ensures that labour market statistics are more informative and give a more representative picture of employment in an economically integrated area. For example, considering a large city and its surrounding suburbs and towns connected by non-negligible commuting flows jointly provide a better representation of labour market statistics than considering each of the municipalities encompassed in the local labour market independently. Similarly, TL2 regions appear not ideal for labour market statistics. Many TL2 regions are large and very heterogeneous in terms of labour force participation and employment. For example, US states such as California or German federal states such as North Rhine-Westphalia can display significantly different internal employment patterns. Thus, they will not capture potentially significant and important geographic disparities in employment or labour force participation.

GDP per capita offers the second illustrative example of how functional areas can enrich existing geographies to provide an accurate description of economic realities through territorial indicators. GDP per capita often serves as an estimate of living conditions or income levels. However, such an approximation can be severely skewed if administrative units are used exclusively to generate subnational indicators on GDP per capita. The economic activity in a place is not necessarily reflective of the GDP or income produced by the residents in that place. For example, in Europe, the GDP per capita in capital cities often significantly overstates income levels of their residents, resulting in between 4% and 76% higher estimated income levels (European Commission, 2007[7]).

The question of the suitability of only using administrative areas for assessing subnational indicators on economic issues highlights the need for a more realistic reflection of the territorial dimension of economic activity and economic linkages. The case of Brussels offers a particularly striking example of the bias of subnational indicators induced by using administrative units as areas of analysis. As a study by ESPON (European Observation Network for Territorial Development and Cohesion) shows, the TL2 Brussels-Capital region is among the wealthiest TL2 regions in Europe in terms of GDP per capita, even though the actual disposable income per inhabitant is even lower than in the two other large Belgian regions, Flanders and Wallonia (OECD, 2019[8]; ESPON, 2006[9]). Sixty percent of workers who contribute to Brussel’s GDP do not live in the city but actually commute to work from a different region on a daily basis, which causes income levels in Brussels to appear drastically higher than they are in reality (ESPON, 2006[9]). While income can be generated in a given area, it may still be largely earned and consumed by households in other areas.

Besides greater statistical accuracy for socio-economic indicators, a consistent delineation of functional areas can also help address the challenge of international comparability of subnational statistics. Using the same methodology to identify labour market areas or functional urban areas can yield relatively comparable geographies across different countries. In contrast, administrative boundaries and the size of administrative areas differ vastly across OECD countries. For example, TL2 regions such as federal states in Germany or the US differ substantially in size due to differences in the nature of administrative boundaries. The same is true for TL3 regions across OECD countries.2 Even more granular data based on municipalities might not be comparable internationally as municipalities in certain countries, e.g. Mexico, tend to be considerably larger than in other countries.3

Delineating functional areas also enhances within-country comparability of territorial statistics on socio-economic indicators. For instance, in countries where metropolitan or core-based functional areas are delineated, applying a consistent methodology for functional areas nationally implies that the same approach is used for the whole country and means that the delineation is replicable (Franconi et al., 2017[10]). Ideally, a comprehensive methodology of delineating functional areas should, therefore, generate a partition of the entire country and thus include rural areas or areas remote from major cities.

While functional areas might not be suitable for all statistical purposes, they can complement territorial statistics. No specific unit is ideal for all types of analyses. Instead, the most appropriate geography depends on the purpose of each specific analysis. Functional areas offer an enriching perspective that complements statistics for administrative areas in efforts to accurately capture the spatial dynamics of socio-economic aspects such as employment and GDP per capita.

Better policy design, delivery and evaluation for all territories

In many countries, the focus of functional areas has been placed on urban or (large urban) metropolitan areas. Yet, the concept of functional areas is not exclusive to urban areas but can cover the entire inhabited territory of a country. Economic and social linkages that identify a functionally interlinked area also connect rural areas, villages and towns with each other.

Delineating functional areas for the entire territory is important for fostering social cohesion and economic development across all types of regions and areas by informing the organisation of public service provision and local labour market policy. A set of national functional areas needs to broaden the city-focused perspective to include rural and more remote areas in order to acknowledge their importance for economic growth and development. Understanding the functional connectivity of rural areas enables policy makers to use a targeted and tailored approach to pressing challenges in those areas. For example, problems of public service provision in sparsely populated areas are likely to differ from urban contexts.

Functional areas offer a geography that will likely take into account spatial externalities, as these arise due to economic linkages, which are the defining element of functional areas. Consequently, functional areas can enhance the analysis of policy challenges, the appropriate design of policy action, and the delivery of positive outcomes for residents. Many economic activities create spatial spillovers such as congestion, pollution or effects on housing affordability and availability. These spillovers can cause significant inadvertent negative, but also positive, effects on residents in nearby municipalities, areas or regions. Policies that are based on territorial statistics on administrative units might not appropriately identify these externalities.

The very nature of functional areas implies that they offer a crucial geographic perspective on key subnational policy issues that evolve around territorial linkages. For example, labour market policies can benefit from comprehensive analysis based on information on commuting, which defines people’s access to jobs and economic opportunities.4 Additionally, the economic integration of areas across administrative borders implies that a functional approach to transport planning can help address bottlenecks and identify policy priorities more effectively to alleviate congestion. Furthermore, functional areas reveal information on the geographic patterns of economic opportunities and they can, therefore, be used to examine migration patterns between different labour market areas.

Service provision is another area that can benefit from the information elicited by statistics on functional areas. Commuting flows define a coherent labour market and integrated economic zone, which could also be used as a reference point for assessing access to services. Especially in rural areas that are sparsely populated, providing or maintaining access to public services is increasingly challenging in many OECD countries due to demographic change. To know where and how services in areas with low population density can be best provided, it is essential to understand the functional relationship of neighbouring rural areas because knowing the functional links in non-urban areas yields insights into mobility and transport connectivity in those areas.

References

[1] Casado Diaz, J. and M. Coombes (2011), “The delineation of 21st century local labour market areas: A critical review and a research agenda”, Boletín de la Asociación de Geógrafos españoles 57, pp. 7-32.

[5] ESPON (2007), Preparatory Study on Feasibility of Flows Analysis: Final Report.

[9] ESPON (2006), The Modifiable Areal Unit Problem.

[7] European Commission (2007), Growing Regions, Growing Europe, Fourth report on economic and social cohesion.

[10] Franconi, L. et al. (2017), “Guidelines for labour market area delineation process: From definition to dissemination”, ISTAT.

[4] Gehlke, C. and K. Biehl (1934), “Certain effects of grouping upon the size of the correlation”, Journal of the American Statistical Association, Vol. 29/169.

[8] OECD (2019), OECD Regional Statistics Database, OECD, Paris.

[6] OECD (2019), OECD Territorial Grids 2019, OECD, Paris.

[2] OECD (2016), Comparison of administrative boundaries and the urban extent of cities.

[3] Openshaw, S. (1977), “Optimal zoning systems for spatial interaction models”, Environment and Planning, Vol. A9, pp. 169-184.

Notes

← 1. This is based on the internal analyses of the territorial grid used by the OECD, without prejudice to the national statistical conventions of member countries.

← 2. As pointed out in Box 1.1, there are a few exceptions, i.e. countries where TL3 regions do not correspond to administrative boundaries.

← 3. The issue of contrasts in the way municipal boundaries, and administrative boundaries more generally, are set is also highlighted by Figure 1.1, which compares Paris with Rome.

← 4. Please see Casado Diaz and Coombes (2011[1]) for further examples and explanations of the policy relevance of functional areas.