Sage Journals: Discover world-class research

Abstract

This study analyses intra-urban consumption flows in Madrid through a spatial analysis using over 689 million anonymized bank card transactions recorded in 2023. By disaggregating data by age and income groups, the research uncovers significant variations in consumer mobility patterns and spatial preferences. Older adults and high-income groups exhibit more localized consumption behaviors, while middle-income groups demonstrate wider mobility across the city. High-income populations concentrate their activity in affluent areas, whereas central districts emerge as common ground with higher levels of social interaction. The study applies network and centrality analysis to identify functional commercial areas and their degree of connectivity, revealing strong links between sociodemographic structure and urban spatial dynamics. Results underscore the relevance of integrating consumption-related mobility into urban planning frameworks to address spatial segregation and promote social mix. The research contributes to a growing body of work leveraging Big Data to understand complex urban behaviors at a granular scale.

Keywords

big data bank card transaction data consumption spending patterns mobility

Introduction

The rise of large retail outlets has threatened small neighborhood shops. Their consolidation on city peripheries has created new urban centralities that challenge the relevance of traditional commercial areas (Guy, 2000). These transformations have reshaped consumption patterns and established a new intra-urban consumption landscape, linked to broader urban restructuring. Factors such as the redefinition of city centers; gentrification (Janoschka et al., 2014); changes in lifestyles, values, household composition, and leisure practices; the spread of new technologies and e-commerce (Jones and Livingstone, 2018); and the growth of activities like the night-time economy, large-scale events, and tourism are among the many transformations that impact the ways consumption takes place in cities (López-González, 2024). In this context, consumer spatial behavior has become increasingly complex. Christaller’s premise of choosing the nearest store is now surpassed by a decision-making process involving numerous factors, such as the consumer, the offering (Guy, 1998), trip utility (Fotheringham, 1988), or the socio-symbolic capital of the spaces (Carpio-Pinedo and Gutiérrez, 2020).

Traditionally, marketing and psychology have been the main disciplines focusing on the study of consumer spatial behavior (Kivell and Shaw, 1980). Research relied on surveys and fieldwork, whose high cost and limited scope confined analysis to specific cases (Schuler, 1979; Vithanage et al., 2023). This situation—exacerbated by the lack of official statistics with an intra-urban and spatial focus on spending—has hindered the development of broader studies on intra-urban mobility and consumption to better explore the relationship between consumption and the city.

The emergence of new data sources associated with Big Data has enabled advances in this research area. These studies often rely on sources that act as proxies for consumption, such as social media data (Carpio-Pinedo and Gutiérrez, 2020; Martí et al., 2019), or more direct sources like credit card transactions. The latter has been used in previous academic research for various purposes (Aparicio et al., 2022; De Las Obras-Loscertales et al., 2025; Hashemian et al., 2017; Lenormand et al., 2015; Sobolevsky, Bojic, et al., 2015; Sobolevsky et al., 2014, 2016; Sobolevsky, Massaro, et al., 2015; Yoshimura et al., 2016, 2018, 2021, 2022); however, only three studies have jointly analyzed mobility and consumption (Lenormand et al., 2015; Yoshimura et al., 2016, 2018), focusing on multi-purpose shopping and differentiating by age, gender, and occupation.

This research makes a fourfold contribution to the international literature using bank card transaction data: (1) it considers differences by both age and income, the latter so far unexplored; (2) it deepens the understanding of urban spatial structure by classifying consumption spaces according to levels of centrality and attractiveness; (3) it identifies consumption functional areas by applying a modularity analysis and characterizes them with indicators such as distance and social mix; and (4) it reflects on the implications for urban planning and social interaction, emphasizing the importance of integrating these dynamics into policy design.

The article is structured as follows: after this brief introduction, Section 2 reviews the related literature, and Section 3 describes the study area. Section 4 presents the data and methodology, Section 5 shows the main results, and Section 6 discusses findings and conclusions.

Literature review: Big Data sources, consumption, and mobility

The emergence of new Big Data sources helps address some of the limitations of traditional consumption studies. Their high spatial and temporal resolution, combined with the large data volumes and sample sizes (Kaisler et al., 2013; Kitchin, 2013), provides a unique opportunity to complement and enhance traditional survey-based approaches (Girardin et al., 2008). In recent years, numerous studies have analyzed intra-urban consumption using social media data from various perspectives. In most cases, these data serve as proxies for commercial activity. For instance, studies based on Google Places data have enabled the identification of the most frequented areas in cities and their temporal variations (Martí et al., 2017, 2019; Santiago-Iglesias et al., 2023). Other widely used proxies include social media platforms such as Foursquare, which have been used to locate popular establishments (Carpio-Pinedo and Gutiérrez, 2020) and analyze visitor activities (Salas-Olmedo et al., 2018). These articles deal with consumption in the city using proxies, but they do not properly study the consumption mobility patterns.

On the other hand, Big Data has been widely applied to urban mobility studies using diverse methods, sources, and purposes (Wang et al., 2020). Several studies have examined spatial behavior across population groups. For instance, Luo et al. (2016) used Twitter data to analyze the radius of gyration by age, gender, and race/ethnicity, finding the largest differences in the latter. Xie et al. (2016) employed GPS trajectories to explore variations by gender, age, and occupation. Mobile phone data have also been used to identify gender-based mobility differences (Gauvin et al., 2020), patterns that have likewise appeared in studies on urban vitality, where greater gender disparities are associated with higher densities of third places (Collins et al., 2024); spaces where differences by age have also been studied (Botta and Gutiérrez-Roig, 2021). However, these studies generally adopt a broad perspective. In fact, mobility studies based on mobile phone data have not been able to differentiate shopping trips from other types of trips (Wang et al., 2020).

To overcome these limitations, the most suitable source for studying urban consumption is bank card transaction records. This data source allows for an accurate measurement of actual spending, the number and location of transactions, and even certain sociodemographic characteristics of consumers. Despite the inherent bias—Yoshimura et al. (2018) point to a possible bias in the representativeness of the economically active population—the increasing use of card payments as a common payment method supports their growing validity. According to the Bank of Spain (Rodríguez Molina and Esandi Erro, 2025), card usage increased from 28% to 32% between 2022 and 2024, with up to 86% of the population now using them, making these records increasingly broad and representative for consumption studies. However, due to the sensitive nature of this information and the difficulty of accessing it, their use remains in its early stages, with limited application to intra-urban research. In fact, most existing studies have employed interurban or regional scales (Hashemian et al., 2017; Sobolevsky et al., 2014, 2015, 2016; Sun et al., 2024).

Using an intra-urban approach, Di Clemente et al. (2018) studied the sequences of credit card purchases to detect patterns of collective behaviour and generate consumer profiles. Yoshimura et al. (2021) examined how retail density and accessibility influence consumption. More recently, Aparicio et al. (2022) examined tourist spending patterns in Madrid, while Carpio-Pinedo et al. (2022) focused on the spatiotemporal patterns of consumption by commercial sector in the same city. Additionally, De Las Obras-Loscertales et al. (2025) explored the spatial distribution of spending in the nighttime economy and its links to urban and sociodemographic variables.

To our knowledge, only a limited number of studies have approached the intersection between mobility, consumption and urban space. Lenormand et al. (2015) examined the influence of age, gender, and occupation on mobility, measured in terms of distance and time between consecutive transactions and the radius of gyration. Meanwhile, Yoshimura et al. (2016, 2018) focused on multi-purpose travel behavior. Yoshimura et al. (2016) analyzed the pre- and post-visit behavior of customers to a retail brand with multiple store locations, aiming to determine whether factors such as store location or seasonal variation influence the number, origin, and type of consumers. Yoshimura et al. (2018) applied urban association rules to analyze relationships between different stores and commercial sectors within shopping trips. However, none of these studies analyze differences across income strata or the implications of consumption mobility patterns for the city’s spatial structure in terms of centrality and functionality.

Area of study

Madrid, with over 3.3 million inhabitants (7.3 million in its metropolitan area), is the largest city in Spain and one of the largest in Europe (INE, 2023). Its economy is highly tertiarized, and commercial deregulation has facilitated the proliferation of diverse retail establishments across the city. Consequently, the historical duality between central traditional commerce and large peripheral malls has evolved into a more heterogeneous landscape with multiple retail formats (Carpio-Pinedo and Gutiérrez, 2020).

The city’s commercial structure shows a sharp contrast between the center—characterized by a high density of neighborhood shops—and the periphery, where density is low except in planned areas such as retail parks and large shopping centers (Figure S1a). On the other hand, Madrid exhibits a pronounced spatial differentiation of the population according to income level (Figure S1(b)). Lower-income groups are concentrated in the southern and southeastern periphery, whereas wealthier populations are found in the center and northwest of the city (Carpio-Pinedo and Gutiérrez, 2020).

Data and methods

Dataset

The dataset used contains anonymized bank card transaction data from one of the main banks in Spain during 2023¹. More than 689.2 million transactions were recorded. The data include card payments at any point of sale (POS) terminal. For each transaction, the amount spent and the location (500 × 500 m grid cells) where the payment occurred are recorded, as well as the postal code of the consumer’s residence and their age group. In this way, for each consumption cell, we have the total expenditure and the number of transactions originating from each of the residential postal codes, disaggregated by age groups.

Since the spatial origin unit in the dataset is postal codes and the destination unit is grid cells, the first step was to transform the origin spatial unit to match the destination unit. Using regular tessellation is more appropriate than administrative units because it avoids potential distortions in distance due to the varying sizes of postal codes and mitigates the problem of MAUP (Modifiable Area Unit Problem (Openshaw, 1979)). To do this, each grid cell was assigned to the postal code covering the largest portion of the cell, and transactions from each postal code were redistributed to the corresponding grid cells, weighted by the estimated population of each age group. Population per grid cell was estimated using census tracts and adjusted based on residential square meters per building from the Cadastre. The selected age groups (≤35, 35–65, and >65) have been defined based on the available data and labor considerations. These ranges capture the differences between young people (≤35), the economically active population with greater family responsibilities (35–65), and retirees (>65), reflecting groups with distinct characteristics and lifestyles.

This information has been enriched with data on average per capita income available at the census tract level (INE, 2022) (Eq 1). Five income strata have been established by quintiles–with Q1 = lowest income and Q5 = highest–ensuring equal representation of spatial units in the Origin-Destination (O-D) analysis, even if population shares differ (Figure S1(b)).

R_{c} = \frac{\sum_{i}^{c} s P_{i}^{c} \times R_{i}^{c}}{\sum_{i}^{S} c P_{i}^{c}}

(1)

where R_c is the average income in cell c,

P_{i}^{c}

is the population of section and in cell c, and

R_{i}^{c}

is the income of population

P_{i}^{c}

in cell c.

Methodology

The transaction numbers and total expenditure data for the city of Madrid have been analyzed to understand the consumption relationships between different areas of the city. Network analysis methods have been applied to identify the spatial structure of the consumption network, calculate centralities, and detect commercial functional areas for the different sociodemographic groups.

Network analysis

Creation of origin-destination matrices

Origin-Destination (OD) matrices have been widely used in mobility studies to identify patterns, estimate trips, or predict them (Mohammed and Oke, 2023). Their cartographic representation serves as a valuable tool for visual analysis, helping to understand complex spatial datasets and uncover patterns and flows (Robinson et al., 2017). In this study, the OD matrices were constructed to represent the spatial relationships between places of residence and transaction locations, disaggregated by population groups, allowing us to analyze the connections between different parts of the city and consumer spatial behavior.

Matrices were generated in ArcGIS Pro using the network dataset to calculate pedestrian distances, providing distance statistics for the entire population and each study group.

Centrality calculation

Centrality measures the relative importance of nodes within a network. The definitions and formulas used to estimate centrality depend on the analysis objective and network properties. Degree centrality, one of the most widely used metrics, quantifies the number of nodes connected to a given node (Freeman, 1979) and has been extended to weighted networks by summing the connection weights (Barrat et al., 2004; Newman, 2004). However, this approach presents limitations, as it considers only the total weights, without accounting for the number of nodes to which a node is connected (Opsahl et al., 2010).

In our directed and weighted network, we aim to measure the input degree centrality, based on the consumption recorded at them—considering that consumption activity is higher the greater the number of transactions recorded at a node—but also accounting for the number of connections established. Therefore, between two nodes receiving the same number of transactions, the one with transactions from a greater number of other nodes will have higher centrality, as this implies greater attraction from different parts of the city. For its calculation, Opsahl et al. (2010) propose the following formula (Eq. 2):

C_{o u t}^{(i)} = k_{i}^{o u t} \times {(\frac{s_{i}^{o u t}}{k s_{i}^{o u t}})}^{α}

(2)

where k is the number of connections received by node i, s is the sum of weights (transactions), and α is a parameter that can vary depending on the purpose of the research. Values of α between 0 and 1 favor nodes with more connections, whereas values above 1 emphasize fewer but stronger links. Given that establishing more relationships is positive in this study, α has been set to 0.5. Calculations were performed using the R package tnet for each population group. To distinguish varying degrees of centrality, the calculated centrality values were classified according to the following criteria:

• Main centralities or outliers, corresponding to the highest outlier values of the distribution, using the interquartile method. Centrality values that are 1.5 times the interquartile range above the third quartile.

• Secondary centralities or Top10, corresponding to values above the 90th percentile of the data.

• Tertiary centralities or Top25, corresponding to values above the 75th percentile of the data.

• Moderate centralities, which correspond to values above the average.

• All other spaces are considered to have low centrality.

Detection of functional areas

Identifying functional areas is a key aspect of urban planning (Zhong et al., 2014). Numerous studies have calculated functional areas from various data sources, methodologies, and purposes, including approaches based on urban morphology, land use, landscape (Xing and Meng, 2018), human activities (Gao et al., 2017), or mobility (Osorio Arjona et al., 2023). In this study, functional areas are defined as a set of nodes that have a stronger link in consumption, measured through the number of transactions recorded. Thus, each generated area is characterized by stronger connections within it than with other areas.

To delineate these areas, the modularity value has been calculated, which is an optimization performed on a network to detect and obtain an adequate number of clusters (Fortunato and Barthélemy, 2007). Modularity was calculated using the algorithm by Blondel et al. (2008) implemented in the Gephi software. This method estimates modularity in large, directed, weighted networks in two phases: first, each node is assigned to a community, and nodes are iteratively moved to the community that maximizes modularity gain; second, a new network is built with communities as nodes and link weights recalculated. In this way, each node shares a modularity value with its functional area (Blondel et al., 2008), without necessarily being spatially contiguous.

This algorithm has limitations, such as fitting issues or the tendency to search for communities of similar size (Peixoto, 2021). Therefore, the detected areas are considered useful for summarizing key patterns for urban planning but not as an inferential method. For functional areas by income strata, due to the observed distribution, matrices representing less than 0.004% of transactions were removed to reduce noise and emphasize the main functional areas.

Results

Our dataset reveals distinct consumption patterns by age and income. Individuals under 35 accounted for 21.5% of transactions, those aged 35–65 for 57.9%, and those over 65 for 20.6%, the latter averaging the highest number of transactions per card (2.7 vs 2.2). Higher-income groups also showed more transactions per card (2.6). Average transaction amounts decreased with both age (€31.1 to €18.1) and income (€27.7 to €23.5), while middle-aged and lower-income individuals exhibited greater mobility, with longer travel distances and more connections (Table S1).

Origin–Destination Matrix

Most interactions took place between nodes located less than 3 km apart (Figure 1(a)), indicating strong spatial relationships between nearby areas. Although these short-distance flows represent only 10.45% of connections, they account for 69.33% of the total transactions. The remaining 30% take place over longer distances, with the number of connections decreasing as distance increases (Figure 1(b)). Notably, some long-distance flows, particularly toward the Centro and Salamanca districts, stand out due to higher transaction volumes and spending, driven by increased average transaction amounts with distance (Figure 1(d)).

Figure 1.

Origin–Destination Matrix. (a) Flows between nodes >3 km apart; (b) Flows between nodes >3 km; (c) spatial distribution of transactions; (d) relationship between average ticket (points) and number of transactions (bars) with distance.

Each age group shows distinct spatial behaviors. Older adults concentrate more activity within 3 km (18% of connections, 84% of transactions), compared to 12% and 69% for aged 35–65, and 11% and 62% for those under 35 (Table S2). Average ticket size remains stable with distance for younger individuals, varies slightly for middle-aged adults, and increases notably for older adults (Figure 2).

Figure 2.

Origin–Destination matrix by groups. (a) Flows by age group; (b) By income.

Income level also reflects spatial patterns. High-income groups (Q5–Q3) connect mainly in the affluent central-northern areas, while lower-income groups (Q2–Q1) connect both within their neighborhoods and in areas like the city center, Salamanca, and peripheral malls. Consequently, higher-income groups make over 13% of connections and 80% of transactions within 3 km, while middle- and lower-income groups travel further (Table S2). Ticket size generally increases with distance, especially among higher-income groups (Figure S2).

Centrality analysis

Centrality reflects the areas that concentrate the highest volumes of transactions and maintain connections with a larger number of other locations. Therefore, highly central areas can be considered consumption centralities, characterized by both high activity levels (more transactions) and a strong ability to attract consumers from across the city.

Using the classification from the Methods section, centrality scores were grouped to define a typology of consumption spaces. Primary centralities are found in the Centro and Salamanca districts, as well as in peripheral shopping centers. These are surrounded by secondary centralities, while tertiary ones are spread across all city districts (Figure 3). This distribution is closely related to the presence of bars, restaurants, and commercial areas (Figure S3).

Figure 3.

Centrality spaces. (a) Centrality values; (b) Classification by centrality types.

Centrality also relates to attractiveness, measured by the distance consumers are willing to travel or the social mixing (Figure 3). Low and moderate-central areas attract more nearby residents (79.1 and 78.1% within 3 km) compared to higher-order centralities (72.1%, 65.6%, and 56.2%, respectively), with shorter average travel distances (3.3 km vs 4.9 km in primary centralities).

Social mixing is highest in primary centralities, especially in terms of socioeconomic diversity (1.29 in primary ones vs 0.44 in low-centralities), revealing a clear center-periphery gradient. In contrast, age diversity is more evenly distributed across (Figure S4).

The spatial distribution of centrality varies by age (Figure 4). For younger individuals, primary centralities are concentrated in the city center (5.7% of cells), the 35–65 age group shows more secondary centralities (2.2%), and those over 65, primary centralities regain importance (4%) with wider city coverage. Centrality values correlate strongly with bars and commercial areas, particularly for young people (Figure S3).

Figure 4.

Centrality by population groups. (a) By age; (b) By income strata.

Income-based patterns also reveal differences (Figure 4). Both the highest- and lowest-income groups concentrate activity in primary centralities (2.8% and 2.7% of cells), while Q4 shows a more balanced distribution. Additionally, the geographic location of centralities differs between groups: high-income groups focus on affluent areas like Salamanca and Chamberí, while lower-income groups are more active in Centro and peripheral zones. The centrality values for the low and lower-middle income strata (Q1 and Q2) are those that show the lowest correlations with the presence of bars, restaurants, and shops, since their highest centrality values are distributed around the periphery of the city (Figure S3).

Detection of functional areas

The functional area analysis identifies 10 communities within Madrid, where internal connections are stronger than external ones. These areas show distinct behaviors based on centrality, social mix, and travel distance (Figure 5).

Figure 5.

Commercial functional areas.

Functional areas with high centrality values like Center-North and Salamanca have a higher proportion of transactions conducted by consumers from other areas (42.96% and 53.51% of the transactions) and have longer average travel distances (∼1 km longer) than areas like East_2 (14.47%) or South (6.81%). They also show higher-income diversity (Shannon index 1.47 and 1.39 vs 0.32 and 0.18), while age diversity varies little across areas (Figure 5).

Age-related differences emerge in the delineation of functional areas, reflecting the distinct spatial behaviors (Figure 6). Particularly noteworthy is the variation in the number of functional areas detected for each age group: 6 functional areas for younger individuals, 8 for those aged 35 to 65, and up to 11 for individuals over 65. These results confirm that older adults have stronger local ties, while younger groups, with weaker local ties and broader connections, form larger and expansive zones.

Figure 6.

Functional areas by population group. (a) By age group; (b) By income strata.

Income-based patterns do not appear uniformly across the city (Figure 6), largely conditioned by the residential distribution of income groups. High-income functional areas are concentrated in the central-northern sectors, low-income groups are mainly in the south while middle-income groups, with greater mobility, form functional areas spanning much of the city.

Discussion and conclusion

Traditionally, the study of consumption-related mobility has faced inherent difficulties due to the data sources used. Mobility surveys do not include a consumption category to classify trips by purpose, so it is necessary to use related categories such as shopping or leisure as proxies, which do not necessarily correspond to consumption. In recent years, mobility studies based on mobile phone data have proliferated, but it is well known that they have not been able to identify consumption-related mobility.

These limitations have been overcome with the use of data from bank card transaction records. This article, using a dataset that combines residential locations and consumption destinations (grid cells), enriched with socioeconomic information, examines mobility patterns associated with consumption across age and income groups. In addition, centrality areas were identified through both the number and strength of connections, and functional areas were delineated for each group.

Findings reveal the intensity of connections across the city, as well as the mobility and consumption radius for each group. Sociodemographic influences on travel distance have been well documented in relation to shopping (Bandara et al., 2022; Theodoridis and Priporas, 2009) and mobility in general (Luo et al., 2016; Xie et al., 2016). Nevertheless, further investigation was required to understand how these factors affect the evolving intra-urban commercial landscape.

Our results align with studies on general mobility using other data sources showing that travel distance decreases with age (Boschmann and Brady, 2013), likely due to fewer work-trips and reduced car usage (Newbold et al., 2005), favoring greater reliance on walking or public transportation. This latter pattern is also observed in Madrid, where older age groups tend to travel more frequently on walking (50% vs 25% among the 30–60 age group) and by public transport (25% vs 20%), indicating shorter travel distances (CRTM, 2019). A key insight is that this reduction in travel distance does not necessarily imply lower consumption activity, but rather a greater sensitivity to distance (Newbold et al., 2005). Indeed, do not usually use bank cards for small expenses, so their average ticket is higher, and have more intense interactions within their immediate surroundings (Figure 2, Table S1).

In contrast, existing findings on mobility differences by income are often contradictory (Xu et al., 2018). Xie et al. (2016) and Moya-Gómez et al. (2021) observed that higher-income groups tend to concentrate their movements in smaller areas, whereas Barbosa et al. (2021) found the opposite in Brazilian cities. Our analysis provides a more nuanced perspective, showing that urban context and spatial structure play a crucial role. Specifically, the lower urban sprawl typical of Mediterranean cities, combined with a dense public transport network, reduces car dependence and the effect of transport costs, unlike other contexts (Yang et al., 2023). This, along with longer work trips and greater distance from main urban centers for lower-to middle-income groups, explains why they travel farther on average than higher-income individuals, who tend to concentrate activities near their homes due to more selective habits (Macedo et al., 2022).

The study deepens understanding of the urban structure and area interconnections. The proposed centrality methodology (Opsahl et al., 2010) offers a broader view of centrality and consumer attraction. The findings align with previous research on Madrid’s urban dynamics (Carpio-Pinedo and Gutiérrez, 2020), and with other analyses of spending in Madrid, such as those by Carpio-Pinedo et al. (2022), De las Obras et al. (2025), or Aparicio et al. (2022), which show a high concentration of spending in the city center of Madrid. Moreover, recalling Christaller’s (1966) notion of spatial reach to distinguish different types of centrality. They also highlight the growing role of shopping malls as new centrality areas, especially for middle-aged groups, while reaffirming the importance of neighborhood-based commerce.

A key contribution is the analysis of centrality across population groups, showing distinct spatial behaviors and perceptions. Among younger groups, the concentration and intensity of central areas suggest a preference for commercially dense spaces with socio-symbolic value (Figure 4), reinforcing the role of space in consumption decision-making (Carpio-Pinedo and Gutiérrez, 2020). By contrast, the dispersed pattern observed among those aged 65 and older reflects convenience, potential mobility constraints, and stronger neighborhood ties.

Income groups analysis also reveals differences. First, there are centralities clearly tied to residential locations. Second, certain districts—such as Chamberí or Salamanca—are primarily frequented by affluent groups, whereas the Centro district functions as a shared space across income levels, underscoring its stronger attraction capacity and role as a hub for social interaction (Figure 4).

This latter aspect is especially relevant for urban planning. Segregation is increasingly measured not only in residential terms, but also through the spaces where daily activities take place (Loo et al., 2024), so promoting interaction in these areas can enhance diversity and social mixing. Understanding mobility patterns and the centrality distribution allows for more efficient transport planning and helps reinforce services in key areas or strengthen less central spaces, contributing to a more balanced urban development.

The study also has limitations, mainly related to the dataset. Our dataset does not cover all payments, as some are made in cash, not by card. This introduces a certain bias, since small expenses are frequently made in cash and may therefore be underrepresented. Credit card data present certain biases—such as an overrepresentation of the economically active population (Yoshimura et al., 2018)—so it is possible that some dynamics may not have been captured. Origin units are estimates based on population weights within each postal code. The lack of a temporal component prevents analysis across seasons, weeks, or days, limiting the study of temporal variability and of social mixing over time, not just space. More precise analysis would require information on trip purpose, its temporal sequence, and transport mode.

Future research could explore drivers of consumption-related mobility behaviors, which is key for policy design. Analyzing the temporal dimension of these patterns and their relation to urban actors like tourism would also be valuable. Given the growing demographic and economic weight of older adults, further disaggregation of this group could provide important insights. Additional promising avenues include deeper study of activity-based segregation and its urban implications. Finally, replicating the study in cities of different sizes and contexts, including the wider metropolitan area, would offer useful comparative evidence, especially regarding socioeconomic differences.

Supplemental material

Supplemental material - Intra-urban consumption flows and sociospatial disparities: An analysis based on bank card transaction data

Supplemental material for Intra-urban consumption flows and sociospatial disparities: An analysis based on bank card transaction data by Julia de las Obras-Loscertales, Juan Carlos García-Palomares, Diego Ramiro-Fariñas, María Soledad Hernández Martín-Caro, Javier Gutiérrez in Environment and Planning B: Urban Analytics and City Science.

Footnotes

Acknowledgments

This article has been developed as part of a research contract between Madrid City Council and the Complutense University of Madrid, entitled “Support service for the statistical exploitation and analysis of spending in the city of Madrid,” with number 511/2023/10901. This article was supported by Grant TourMobility (PID2021-127499OB-I00) and Grant DynCity (PID2023-148339OB-I00) funded by MICIU/AEI/10.13039/501100011033/and by “European Union FEDER and FSE+.” We are also grateful to the FPU Program (University Teacher Training) of the Spanish Ministry of Universities (FPU21/04291).

Julia de las Obras-Loscertales

Juan Carlos García-Palomares

Diego Ramiro-Fariñas

Javier Gutiérrez

Author contribution

Julia de las Obras-Loscertales: Writing—original draft, visualization, software, methodology, formal analysis, and conceptualization. Juan Carlos García-Palomares: Writing—original draft, supervision, methodology, funding acquisition, formal analysis, and conceptualization. Diego Ramiro-Fariñas: Writing—original draft, supervision, methodology, and conceptualization. María Soledad Hernández Martín-Caro: Resources, methodology, and funding acquisition. Javier Gutiérrez: Writing—original draft, supervision, methodology, funding acquisition, formal analysis, and conceptualization.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: as part of a research contract between Madrid City Council and the Complutense University of Madrid, entitled “Support service for the statistical exploitation and analysis of spending in the city of Madrid,” with number 511/2023/10901, and by FPU Program (University Teacher Training) of the Spanish Ministry of Universities (FPU21/04291). Also, was supported by Grant TourMobility (PID2021-127499OB-I00) and Grant DynCity (PID2023-148339OB-I00) funded by MICIU/AEI/10.13039/501100011033/ and by “European Union FEDER and FSE+”

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data used in this study cannot be made openly available because the use of the data is restricted due to privacy concerns.*

Supplemental material

Supplemental material for this article is available online.

Note

Author biographies

Julia de las Obras-Loscertales is a PhD candidate in Geography. She graduated in Geography and Spatial Planning and completed her master's degree in sustainable and smart cities. Her areas of interest are spatial patterns of urban consumption, big data and the city.

Juan Carlos García-Palomares is a professor of Human Geography. His areas of research are urban studies, transport and mobility. To this end, he has relied on the use of GIS, gaining extensive experience in the application of spatial analysis tools, network analysis, spatial statistics and the use of new data sources associated with Big Data. He has been president of the Geographic Information Technology Group of the Association of Spanish Geographers and coordinator of the Doctorate programme in Geography (UCM).

Diego Ramiro-Fariñas is a Professor and Director of the Institute of Economics, Geography and Demography at the Spanish National Research Council (CSIC). He has been President of the Association of Historical Demography (ADEH) and of the European Society of Historical Demography. He is a member of the steering committee of Population Europe, the European network of centres of excellence in demography, and has actively participated in various international panels of the International Union for the Scientific Study of Population (IUSSP). He is also on the editorial board of the journals Demography and Spatial Demography.

María Soledad Hernández-Caro has a degree in mathematics. She is currently head of the Socioeconomic Studies Service, Department of the Government Area for Economy, Innovation, and Financeat Madrid City Council.

Javier Gutiérrez is Professor of Human Geography at the Complutense University of Madrid and director of the tGIS (Transport, Infrastructure and Territory) Research Group. His research focuses on GIS (Geographic Information Systems), transport and urban studies. He has been Director of the Department of Geography, Coordinator of the Master's Degree in Geographic Information Technologies and the Doctoral Programme in Geography at the Complutense University, President of the Geographic Information Technologies Group of the Association of Spanish Geographers and Coordinator of the Diploma in Tourism at CES Felipe II.

References

Aparicio

Hernández Martín-Caro

García-Palomares

, et al. (2022) Exploring the spatial patterns of visitor expenditure in cities using bank card transactions data. Current Issues in Tourism 25(17): 2770–2788.

Bandara

Thalgaspitiya

Hettiarachchi

(2022) Generation gap, gender and shopping enjoyment of shopping mall visitors in Western province Sri Lanka. Asian Journal of Marketing Management 1(02): 01–29.

Barbosa

Hazarie

Dickinson

, et al. (2021) Uncovering the socioeconomic facets of human mobility. Scientific Reports 11(1): 8616.

Barrat

Barthélemy

Pastor-Satorras

, et al. (2004) The architecture of complex weighted networks. Proceedings of the National Academy of Sciences 101(11): 3747–3752.

Blondel

Guillaume

J-L

Lambiotte

, et al. (2008) Fast unfolding of communities in large networks. Journal of Statistical Mechanics: Theory and Experiment 2008(10): P10008.

Boschmann

Brady

(2013) Travel behaviors, sustainable mobility, and transit-oriented developments: a travel counts analysis of older adults in the Denver, Colorado metropolitan area. Journal of Transport Geography 33: 1–11.

Botta

Gutiérrez-Roig

(2021) Modelling urban vibrancy with Mobile phone and OpenStreetMap data. PLOS ONE Mehmood R 16(6): e0252015.

Carpio-Pinedo

Gutiérrez

(2020) Consumption and symbolic capital in the metropolitan space: integrating ‘old’ retail data sources with social big data. Cities 106: 102859.

Carpio-Pinedo

Romanillos

Aparicio

, et al. (2022) Towards a new urban geography of expenditure: using bank card transactions data to analyze multi-sector spatiotemporal distributions. Cities 131: 103894.

10.

Christaller

(1966) Central Places in Southern Germany. Prentice-Hall.

11.

Collins

Di Clemente

Gutiérrez-Roig

, et al. (2024) Spatiotemporal gender differences in urban vibrancy. Environment and Planning B: Urban Analytics and City Science 51(7): 1430–1446.

12.

CRTM (2019) Encuesta De Movilidad De La Comunidad De Madrid 2018. Documento Síntesis. Consorcio Regional de Transportes de Madrid.

13.

De Las Obras-Loscertales

García-Palomares

Ramiro-Fariñas

, et al. (2025) Consumption after dark exploring spatio-temporal patterns of spending in the night-time economy. Cities 159: 105726.

14.

Di Clemente

Luengo-Oroz

Travizano

, et al. (2018) Sequences of purchases in credit card data reveal lifestyles in urban populations. Nature Communications 9(1): 3330.

15.

Fortunato

Barthélemy

(2007) Resolution limit in community detection. Proceedings of the National Academy of Sciences 104(1): 36–41.

16.

Fotheringham

(1988) Note—consumer Store choice and choice set definition. Marketing Science 7(3): 299–310.

17.

Freeman

(1979) Centrality in social networks conceptual clarification. Social Networks 1(3): 215–239.

18.

Gao

Janowicz

Couclelis

(2017) Extracting urban functional regions from points of interest and human activities on location‐based social networks. Transactions in GIS 21(3): 446–467.

19.

Gauvin

Tizzoni

Piaggesi

, et al. (2020) Gender gaps in urban mobility. Humanities and Social Sciences Communications 7(1): 11.

20.

Girardin

Calabrese

Fiore

, et al. (2008) Digital footprinting: uncovering tourists with user-generated content. IEEE Pervasive Computing 7(4): 36–43.

21.

Guy

(1998) Classifications of retail stores and shopping centres: some methodological issues. Geojournal 45(4): 255–264.

22.

Guy

(2000) From crinkly sheds to fashion parks: the role of financial investment in the transformation of retail parks. International Review of Retail Distribution & Consumer Research 10(4): 389–400.

23.

Hashemian

Massaro

Bojic

, et al. (2017) Socioeconomic characterization of regions through the lens of individual financial transactions. PLoS One 12(11): e0187031.

24.

INE (2022) Atlas de distribución de renta de los Hogares. Available at. https://www.ine.es/dyngs/INEbase/operacion.htm?c=Estadistica_C&cid=1254736177088&menu=resultados&idp=1254735976608

25.

INE (2023) Estadística del Padrón continuo. Available at. https://www.ine.es/dyngs/INEbase/operacion.htm?c=Estadistica_C&cid=1254736177012&menu=resultados&idp=1254734710990

26.

Janoschka

Sequera

Salinas

(2014) Gentrification in Spain and Latin America — a critical dialogue. International Journal of Urban and Regional Research 38(4): 1234–1265.

27.

Jones

Livingstone

(2018) The ‘online high street’ or the high street online? The implications for the urban retail hierarchy. International Review of Retail Distribution & Consumer Research 28(1): 47–63.

28.

Kaisler

Armour

Espinosa

, et al. (2013) Big data: issues and challenges moving forward 2013 46th Hawaii International Conference on System Sciences, 995–1004. Wailea, HI, USA, January 2013.

29.

Kitchin

(2013) Big data and human geography: opportunities, challenges and risks. Dialogues in Human Geography.

30.

Kivell

Shaw

(1980) The study of retail location. In: Dwason

(ed) Retail Geogrpahy. John Wiley & Sons., 95–155.

31.

Lenormand

Louail

Cantú-Ros

, et al. (2015) Influence of sociodemographic characteristics on human mobility. Scientific Reports 5(1): 10075.

32.

Loo

BPY

Fan

Moro

(2024) Residential and experienced social segregation: the roles of different transport modes, metro extensions, and longitudinal changes in Hong Kong. Humanities and Social Sciences Communications 11(1): 1439.

33.

López-González

(2024) The right to the city versus the right to tourism in teleological perspective: an ethical conflict between goods. Current Issues in Tourism: 1–13.

34.

Luo

Cao

Mulligan

, et al. (2016) Explore spatiotemporal and demographic characteristics of human mobility via Twitter: a case study of Chicago. Applied Geography 70: 11–25.

35.

Macedo

Lotero

Cardillo

, et al. (2022) Differences in the spatial landscape of urban mobility: gender and socioeconomic perspectives. PLOS ONE Gao S 17(3): e0260874.17

36.

Martí

Serrano-Estada

Nolasco-Cirugeda

(2017) Using locative social media and urban cartographies to identify and locate successful urban plazas. Cities 64: 66–78.

37.

Martí

Serrano-Estrada

Nolasco-Cirugeda

(2019) Social media data: challenges, opportunities and limitations in urban studies. Computers, Environment and Urban Systems 74: 161–174.

38.

Mohammed

Oke

(2023) Origin-destination inference in public transportation systems: a comprehensive review. International Journal of Transportation Science and Technology 12(1): 315–328.

39.

Moya-Gómez

Stępniak

García-Palomares

, et al. (2021) Exploring night and day socio-spatial segregation based on mobile phone data: The case of Medellin (Colombia). Computers, Environment and Urban Systems 89: 101675.

40.

Newbold

Scott

Spinney

JEL

, et al. (2005) Travel behavior within Canada’s older population: a cohort analysis. Journal of Transport Geography 13(4): 340–351.

41.

Newman

MEJ

(2004) Analysis of weighted networks. Physical Review 70(5): 056131.

42.

Openshaw

(1979) Million or so correlation coefficients: three exper-iments on the modifiable areal unit problem. Statistical Applications in the Spatial Sciences: 127–144.

43.

Opsahl

Agneessens

Skvoretz

(2010) Node centrality in weighted networks: generalizing degree and shortest paths. Social Networks 32(3): 245–251.

44.

Osorio Arjona

Ruiz-Santacruz

De Las Obras-Loscertales

(2023) Mapping of functional areas in Spain based on mobile phone data during different phases of the COVID-19 pandemic. Journal of Maps 19(1): 2214804.

45.

Peixoto

(2021) Descriptive vs. inferential community detection in networks: pitfalls, myths, and half-truths. arXiv.

46.

Robinson

Demšar

Moore

, et al. (2017) Geospatial big data and cartography: research challenges and opportunities for making maps that matter. International Journal of Cartography 3(sup1): 32–60.

47.

Rodríguez Molina

Esandi Erro

(2025) Evolución y tendencias en los pagos de los consumidores españoles. Boletín Económico, 2025/T2: 03.

48.

Salas-Olmedo

Moya-Gómez

García-Palomares

, et al. (2018) Tourists’ digital footprint in cities: comparing big data sources. Tourism Management 66: 13–25.

49.

Santiago-Iglesias

Carpio-Pinedo

Sun

, et al. (2023) Frozen city: analysing the disruption and resilience of urban activities during a heavy snowfall event using google popular times. Urban Climate 51: 101644.

50.

Schuler

(1979) A disaggregate store-choice model of spatial decision-making. The Professional Geographer 31(2): 146–156.

51.

Sobolevsky

Sitko

Grauwin

, et al. (2014) Mining urban performance: scale-independent classification of cities based on individual economic transactions. arXiv:1405.4301.

52.

Sobolevsky

Massaro

Bojic

, et al. (2015) Predicting regional economic indices using big data of individual bank card transactions. arXiv 1506: 00036.

53.

Sobolevsky

Bojic

Belyi

, et al. (2015) Scaling of city attractiveness for foreign visitors through big data of human economical and social media activity IEEE International Congress on Big Data, 600–607.

54.

Sobolevsky

Sitko

Combes

, et al. (2016) Cities through the prism of people’s spending behavior. PLoS One 11(2): e0146291.

55.

Sun

Zhong

, et al. (2024) Unraveling changes of spending behavior in pandemic cities: a nationwide study of South Korea. Computers, Environment and Urban Systems 114: 102181.

56.

Theodoridis

Priporas

(2009) Store choice in computer retailing: the case of home users in Greece. EuroMed Journal of Business 4(1): 58–68.

57.

Vithanage

Wattage

Kariyawasam

, et al. (2023) Choice modeling of grocery shopping behavior: a Sri Lankan case study. Cities 143: 104610.

58.

Wang

Zhang

Chan

EHW

, et al. (2020) A review of human mobility research based on big data and its implication for smart city development. ISPRS International Journal of Geo-Information 10(1): 13.

59.

Xie

Xiong

(2016) The correlation between human mobility and socio-demographic in megacity 2016 IEEE International Smart Cities Conference (ISC2), 1–6. Trento, Italy, September 2016.

60.

Xing

Meng

(2018) Integrating landscape metrics and socioeconomic features for urban functional region classification. Computers, Environment and Urban Systems 72: 134–145.

61.

Belyi

Bojic

, et al. (2018) Human mobility and socioeconomic status: analysis of Singapore and Boston. Computers, Environment and Urban Systems 72: 51–67.

62.

Yang

Samaranayake

, et al. (2023) A clustering-based approach to quantifying socio-demographic impacts on urban mobility patterns. Environment and Planning B: Urban Analytics and City Science 50(9): 2452–2469.

63.

Yoshimura

Amini

Sobolevsky

, et al. (2016) Analysis of customers’ spatial distribution through transaction datasets. In: Hameurlain

Küng

Wagner

, et al. (eds) Transactions on Large-Scale Data- and Knowledge-Centered Systems XXVII. Berlin, Heidelberg: Lecture Notes in Computer Science, 177–189.

64.

Yoshimura

Sobolevsky

Bautista Hobin

, et al. (2018) Urban association rules: uncovering linked trips for shopping behavior. Environment and Planning B: Urban Analytics and City Science 45(2): 367–385.

65.

Yoshimura

Santi

Arias

, et al. (2021) Spatial clustering: influence of urban street networks on retail sales volumes. Environment and Planning B: Urban Analytics and City Science 48(7): 1926–1942.

66.

Yoshimura

Kumakoshi

Fan

, et al. (2022) Street pedestrianization in urban districts: economic impacts in Spanish cities. Cities 120: 103468.

67.

Zhong

Arisona

Huang

, et al. (2014) Detecting the dynamics of urban structure through spatial network analysis. International Journal of Geographical Information Science 28(11): 2178–2199.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.30 MB