Accessibility, congestion and economic growth
Traffic behaviour, especially information about congestion levels, is considered a key issue in transport management. For many years the scientific literature has analysed congestion from different approaches and considered different indexes, such as level of service, lane occupancy ratesssssss, queues and capacity adequacy, among others (Boarnet et al.
1998; Rao and Rao
2012). Although many studies have addressed different congestion indexes and the desired factors or attributes that should be considered, one of the most recurrent factors is travel time (Levinson and Lomax
1996). Travel time influences users’ decisions relating to route selection, modal choice and practitioners' decisions as it allows for an understanding of road infrastructure improvements. Travel time is one of the main indicators of congestion levels, and therefore is a crucial factor in transport planning.
In simple terms, congestion is when the number of vehicles trying to use a specific road link reaches or exceeds the capacity for which the traffic network was designed, and this situation induces increases in travel time, which result in traffic delays. However, congestion is a very complex phenomenon that generates impacts at different scales: these include first-order or micro-scale impacts to individuals on the vehicles, such as traffic delays, and second-order or macro-scale impacts which affect activities and/or the regional economy (Sweet
2011). Precisely, this two-scale approach leads us to a continuing debate about mobility and accessibility approaches of congestion analysis. Mobility enters accessibility calculations significantly in congestion measures, although it is a ‘mean’ rather than an ‘end’ (Grengs et al.
2010). Congestion is not only a one-off issue associated exclusively with transport systems, but is also a consequence of the distribution and intensity of land use and different social interactions, among others. Therefore, it is necessary to incorporate these components to identify other causes and effects of congestion and determine possible side effects that may call into question some measures adopted to solve congestion problems (Levine and Garb
2002).
Recent studies have addressed the relationship between traffic congestion and the economy, and have produced differing evidence regarding the former's effects on regional economic growth (Marshall and Dumbaugh
2020). On the one hand, several studies have empirically demonstrated that severe traffic congestion decreases employment and income growth (Hymel
2009; Sweet
2014a). Moreover, some of these studies have shown that congestion is detrimental not only to firms, but also to household income (Jin and Rafferty
2017), and that reducing traffic congestion will provide economic benefits in terms of increasing employment and income growth.
On the other hand, some recent papers have stated that traffic congestion certainly has negative impacts, but is also a by-product of economic activity and social interaction (Sweet
2014b). Congestion must not be viewed only as a cost to society because agglomerations of activities frequently give rise to traffic congestion (Mondschein and Taylor
2017). Precisely, agglomeration trends and localisation economy theory are very relevant for understanding congestion and economic growth links. By focusing on firms’ location, certain studies have found that new firms prioritise being located near same-industry firms because the access advantages of these areas of agglomeration outweigh the impedances of traffic congestion (Osman et al.
2019). However, some differences are found depending on the scale of congestion and the sector: as mentioned before, while localised congestion may be a proxy for amenities valued by many firms, regional congestion may be detrimental (Sweet
2014b), especially to office-based firms (Hou
2016). Many of these studies are based on accessibility analyses highlighting the importance of land use and destination attractiveness in transportation planning. Access to employment, for instance, is highly conditioned by travel time delays but also by proximity-based issues (Thomas et al.
2018). Similarly, studies about polycentrism-based policies show that, in general, they may reduce congestion levels, although maybe not accessibility levels. This would depend on the number of new sub-centres, among many other variables (Li et al.
2019).
Besides the differing impacts of congestion on economic growth, most studies agree on the relevance of household income and employment as key variables in congestion analyses (Jin and Rafferty
2017; Mondschein and Taylor
2017; Osman et al.
2019; Thomas et al.
2018). However, to measure and evaluate individuals’ exposure to congestion properly, consideration must be given not only to commuting (employment/jobs access) but also to the whole activity-travel pattern of individuals and households (Kim and Kwan
2019).
Furthermore, it is important to understand the impacts of congestion on economic growth and vice versa. There is a bidirectional causation as economic success can lead to traffic congestion, but when traffic congestion is sufficiently impactful, it has the potential to affect economic activity (Marshall and Dumbaugh
2020). For instance, Jin and Rafferty (
2017) analysed the interrelationship between income, employment and congestion. The first two of these variables are positively associated with the third. Similarly, Mondschein and Taylor (
2017) stated that trip-making spatial patterns are generally associated with income levels, showing that low average trip-making rates are associated more with low-income households and vice versa. Moreover, it is essential to highlight that not only the direct impacts of these socioeconomic variables on congestion but also the different effects of the economic cycles—recessions and recoveries—depending on households’ profile, could help us better understand variations in congestion levels. Depending on the income level and the kind of employment sectors, individuals and households are not equally affected by economic crises: for instance, low-skilled workers with lower salaries and temporary contracts were precisely the profile that suffered the most among employment losses during the 2008 financial crisis (Lallement
2011). On the other hand, highly educated middle-aged individuals are more resilient to economic cycles (Doran and Fingleton
2016).
Apart from these economic-related variables, congestion levels are both temporally and spatially influenced. Many studies on traffic congestion have focused on peak-hour periods during working days as the most relevant scenarios to be analysed. However, long-term spatial and temporal analysis is needed to fully understand congestion patterns for both commuting and non-commuting trips (Zhao and Hu
2019). As Weber and Kwan (
2002, p 226) stated: ‘
the temporal dimension is very important to accurately assessing individual accessibility’. Precisely, the next section addresses how new data sources are offering wider opportunities for travel time and congestion analysis from the spatio-temporal perspective.
New data sources for measuring traffic travel times and congestion
Nowadays, the extensive use of different devices and the resulting data revolution have led to a new generation of interdisciplinary accessibility models. Measuring travel time and congestion now benefits from advances in geospatial technology and the availability of massive geo-located data, which are characterised by their high temporal and spatial resolution. As Geurs and Osth (
2016, p 295) stated: ‘
It seems with advances in geospatial technology, internet technology, and growing abundance of detailed spatial data and real-time transport datasets, the field of accessibility modelling is thriving.’ All this information offers many possibilities, especially for continuous accessibility analyses, which are based on examining temporal variations in accessibility (Chang and Cheon
2019; Zhao and Hu
2019), and are using real-time driving information, open-source mapping, and public transit supply data (Geurs and Östh
2016; Järv et al.
2018).
Concerning traffic, new applications using Floating Car Data (FCD), such as Google Maps Traffic Overlay, AutoNavi, Waze, or TomTom Live Traffic, show real-time traffic information for users and allow for the collection of information such as traffic volume, average traffic speed and actual journey times (Bartosiewicz and Wiśniewski
2015). Google Maps is the most extended application that can calculate optimal routes for different transport modes. Private vehicle travel times are based on tracks’ historical data combined with real-time traffic patterns from mobile phone records. Using Google API services, researchers can apply this data, computing OD travel time matrices for different times and days of the week (Dumbliauskas et al.
2017), which allows us to analyse the impacts of congestion in different temporal scenarios (García-Albertos et al.
2019). Also, thanks to GTFS files, API services allow interested parties to analyse the level of coverage of public transport networks, average speeds, and line overlaps (Hadas
2013), as well as to compute travel time matrices according to time slots, which can be used in dynamic accessibility studies (Boisjoly and El-Geneidy
2016; Fransen et al.
2015; Pritchard et al.
2019; Stępniak et al.
2019). AutoNavi is the largest Chinese web mapping, navigation, and location-based services provider that offers digital maps and real-time traffic information. Based on accumulated massive traffic and travel data of millions of AutoNavi map users, certain studies have identified it as a useful data source for measuring traffic congestion (Li et al.
2019). On the other hand, Waze is a mobility-oriented social network that allows users to obtain real-time traffic information, such as optimal routes, traffic speed, travel times, low-speed points, among others. As a contribution to this social network, users can also provide certain information related to traffic jams, car accidents, and road works, for example. All this information can be downloaded through Waze’s API and applied to specific urban studies from crowdsourced data related to traffic accidents (Angeles Perez et al.
2018; Santos et al.
2017).
TomTom provides detailed information about road networks and traffic and offers different products. One of them is the TomTom Multinet product, which is implemented in some studies both at the European(Ibáñez and Rotoli
2017), national (Moya-Gómez and Geurs
2018) or city levels (Schio et al.
2019). This database provides a homogenised network base for accessibility analyses. Then there is the useful TomTom Speed Profiles product, a digital network for private transport, which includes the average speeds of vehicles for each road link every five minutes. This historical data, obtained by different devices, including browsers and mobile phone GPS, enable dynamic accessibility analysis that considers the effects of congestion. In the literature, certain studies are now considering TomTom’s historical information (links’ speed profiles) in their quest to measure travel times at different times of the day or even to analyse the effects of link failure in transportation networks (Cui and Levinson
2017). Condeço-Melhorado et al. (
2016) use the TomTom database to calculate internal travel distances for European NUTS-3 regions. Other studies are more oriented toward analysing daily variations in speed profiles for automobiles, which allows for an assessment of congestion impacts on accessibility (Moya-Gómez and García-Palomares
2015,
2017; Moyano et al.
2018) or analysing changes in spatial–temporal job accessibility during different periods (Moya-Gómez and Geurs
2018; Pritchard et al.
2019). Also, TomTom is now offering a new web service for developers, Traffic Stats,
2 which collects FCD from TomTom navigation devices (in-dash systems and apps). These devices send anonymised data to TomTom servers in real-time and allow subscribers to derive historical traffic information.
In addition to network performance, the analysis of daily accessibility should also incorporate the effects of city dynamics (García-Palomares et al.
2018), a reason for congestion. Social networks such as Twitter or Flickr provide massive geo-located information about their users at different times of the day. The literature has used this information to understand and map specific mobility patterns (Luo et al.
2016; Salas-Olmedo and Rojas Quezada
2017). Moreover, this information reflects the temporal location of the main activity areas, in which there is a concentration of workers, tourists and/or residents, which can serve as a dynamic measure for a city’s activity hotspots. Accordingly, such dynamic data sources combined with other transport-oriented spatiotemporal information have become invaluable for their contribution to accessibility studies (Moya-Gómez et al.
2018; Moyano et al.
2018). However, the use of this spatial data in accessibility analysis is still in its early stages (Condeço-Melhorado et al.
2018), although its potential for transportation studies is considerable.