Historical Insights into CO₂ Emission Dynamics in Urban Daily Mobility: A Case Study of Lyon’s Agglomeration

Abstract

CO₂ emissions from urban daily mobility play a major role in both environmental degradation, rising economic costs, and sustainability. Reducing these emissions not only advances environmental sustainability but also fosters economic development by enhancing public health, lowering energy consumption, and alleviating the financial strain caused by climate change. Understanding the dynamics of CO₂ emissions from urban daily mobility provides valuable historical insights into environmental impacts and economic costs tied to urban development. This study takes a historical perspective, examining changes in CO₂ emissions associated with daily mobility in the Lyon agglomeration across two decades, drawing on data from the 1995 and 2006 household travel surveys. Our findings reveal that individual factors such as gender, age, employment status, and income significantly influence CO₂ emissions, with males and full-time workers exhibiting higher emissions. Furthermore, household characteristics, including size and vehicle ownership, are critical in shaping emission levels, while urban form variables such as population density and mixed land use demonstrate a negative correlation with emissions, highlighting the importance of urban planning in mitigating CO₂ output. The analysis emphasizes that greater accessibility and proximity to essential services are vital in reducing individual emissions. Based on these insights, we discuss the implications for policy design, suggesting targeted strategies to enhance urban mobility, improve public transport accessibility, and promote sustainable urban development. Finally, we outline directions for future research to further explore the intricate relationship between urban characteristics and emissions, ultimately aiming to contribute to the development of effective climate policies.

1. Introduction

CO₂ emissions from urban daily mobility significantly contribute to environmental degradation and economic costs, challenging sustainability efforts. Reducing these emissions is essential for promoting environmental health and fostering economic development by improving public health, cutting energy consumption, and easing the financial burden of climate change [1,2].

Urban areas, which produce around 80% of global greenhouse gas emissions, are a major part of the problem, with transportation being the fastest-growing source of CO₂ emissions. Addressing energy use in transportation is crucial for meeting the climate change targets set by governments. Investments in cleaner transportation, public transit, and sustainable urban planning will help lower costs, enhance urban health, and support long-term economic resilience.

Reducing CO₂ emissions is a priority globally, and various policies have been established to address it. France, committed to reducing its greenhouse gas emissions, has set ambitious goals, such as a 20% reduction by 2020 and a four-fold decrease by 2050. These commitments are supported by national and local initiatives, focusing on sectors like transportation, which are key contributors to global warming.

The transportation sector is currently the largest emitter of greenhouse gasses and carbon dioxide in France. In 2018, it accounted for 34% of CO₂ emissions, surpassing the manufacturing industry (25%), residential/tertiary sector (22%), and energy sector (17%) [3]. CO₂ emissions make up 70% of the Global Warming Potential among the direct greenhouse gasses considered in the Kyoto Protocol [3]. Additionally, emissions from transportation have risen more rapidly than those in other sectors, increasing by 490% in France between 1960 and 2007 and by 14% from 1990 to 2017, largely due to a 513% rise in road traffic.

To understand these trends, it is crucial to differentiate between the mobility of people and goods, as well as between daily local mobility and long-distance travel. The recent emissions increase is primarily linked to heavy goods vehicles and utility vehicles, while emissions from private cars have slightly declined since 1995 [3]. Therefore, environmental concerns regarding road freight transport need to be paid close attention in both interurban and urban settings [4,5]. Analyzing recent trends in CO₂ emissions related to personal mobility at various scales is essential to understand the factors influencing these dynamics.

The environmental impact of long-distance mobility is substantial. Data from the National Travel Survey of 1994 reveal that 40% of CO₂ emissions from personal transport stem from just 1% of trips exceeding 80 km from home [6]. This distance sees the highest travel increases, especially between urban areas [7,8]. Insights from the National Transportation Survey of 2007–2008 will further illuminate this issue [9,10].

This article focuses on the local scale within a large urban agglomeration, as urban dynamics are crucial to sustainability. Currently, 82% of the French population lives in predominantly urban areas, especially those that are part of major agglomerations [11], which account for most human activities. Most weekday trips for work, education, shopping, or leisure occur within urban settings. While suburbanization persists, recent trends show densification in city centers and nearby areas, along with a decrease in intra-urban car mobility [10]. This suggests a potential reduction in CO₂ emissions.

Instead of quantifying the national impact of changes in daily mobility, this study examines the case of Lyon and its shift in automobile use over the past two decades. By analyzing CO₂ emissions related to daily mobility based on the 1995 and 2006 household travel surveys, we gain insights into the evolution of emission patterns during a period when urban mobility behaviors were different.

Technological advancements and fleet renewal have significantly contributed to reducing gasoline-powered cars and improving environmental standards. However, we also explore the effects of changing mobility behaviors within the context of energy constraints. Our goal is not only to determine whether emissions have risen or fallen over the past 11 years but also to identify differences in mobility behaviors (mode and distance) among various population groups, separating these from socio-demographic changes. This analysis highlights groups with the greatest potential for emissions reduction and informs effective policy measures.

Ultimately, this article discusses the implications of CO₂ emissions from urban daily mobility in Lyon, using historical data to inform urban planning policies. By focusing on individual travel patterns and socio-economic factors, we aim to understand the dynamics of CO₂ emissions and assess the potential impact of social measures aimed at reducing greenhouse gas emissions from transportation, an area that remains underexplored in the existing literature.

2. Theoretical Background

The changing patterns of CO₂ emissions have garnered significant attention from researchers and practitioners. A wealth of literature examines the evolution of transportation-related emissions and strategies to mitigate them (e.g., [12,13,14]). Studies highlight correlations between factors such as population density, land use diversity, and pedestrian-friendly designs with transportation behaviors like vehicle usage and travel distances [15,16,17,18,19]. However, debates continue regarding the specific influences on mobility choices [1,20,21,22,23]. Understanding these factors is crucial for assessing their impact on energy consumption and CO₂ emissions in urban transportation [24,25], highlighting the need for further research to promote sustainable urban development [26].

Research has shown that CO₂ emissions are linked to household appliance ownership, vehicle characteristics (like engine type, fuel, and age), and usage patterns, such as travel distances. Studies within urban areas highlight the role of factors like employment status, income, residential location, and the urban center–periphery dynamic in daily mobility emissions [27]. Additionally, cross-country and city comparisons reveal that economic and social contexts, city size, and urban design significantly affect emission levels [28].

Transportation-related CO₂ emissions are shaped by socio-economic and technical factors, including household makeup, income, education, employment, and urban design. Public transportation policies and urban development strategies can influence private car use and emissions. However, challenges like dispersed living and work locations and complex travel patterns make reducing car usage difficult, despite income-related factors.

2.1. Socio-Economic Factors

The literature continues to debate the importance of socio-economic factors beyond income and household size in influencing CO₂ emissions from daily mobility [29]. Some argue that factors like income and household size are the primary drivers (e.g., [30]), but many studies have found links between emissions and characteristics such as age, employment status, education, household composition, and number of earners, even when controlling for income and household size [31,32]. This suggests socio-economic factors may play a role, but the exact nature of these relationships remains to be fully explored.

Income plays a central role in the debate on CO₂ emissions, with widespread agreement across studies that emissions tend to increase with rising income [29,32,33,34,35]. Research also highlights that CO₂ emissions are more regressive, disproportionately impacting lower-income groups [29,36]. Despite this, studies suggest that when controlling for other factors, the impact of income remains significant but shows less variation in regressivity. This study will examine whether this pattern holds true in the context of CO₂ emissions under investigation.

The link between age and CO₂ emissions is unclear, with studies offering differing conclusions. DEFRA [37] found that emissions rise with age, while Wier et al. [30] observed little impact. Other research, such as Büchs and Schnepf [29], suggests an inverted U-shaped relationship, where emissions increase with age but decline for older individuals who travel less and spend more time at home. Based on this, we hypothesize an inverted U-shaped association between CO₂ emissions and age.

Most studies on CO₂ emissions often omit gender as a variable. However, some research suggests that female-headed households have higher emissions than male-headed ones [37], while others, like Brand and Preston [38], found no significant differences. Since our study uses household and individual data, we cannot distinguish between household heads by gender. Typically, men head households, while women spend more time at home. Thus, we hypothesize that household heads, regardless of gender, may have higher emissions, a theory we plan to investigate.

Earlier studies have highlighted the significance of household composition and size—such as the presence of children or the number of adults—as influential factors in household emissions (e.g., [29,35,37]). However, findings regarding the presence of children have not yielded clear and definitive results to date: while the DEFRA [37] study found a positive association between having children and CO₂ emissions, Baiocchi et al. [32] identified a negative association.

Studies on the link between education and CO₂ emissions show mixed results. Baiocchi et al. [32] and Büchs and Schnepf [29] found a positive correlation but noted that higher education can lower emissions due to increased environmental awareness. However, Brand and Preston [38] reported that those with a university-level education had higher emissions. Lenzen et al. [31] found contrasting results, with higher education being linked to lower emissions in Australia but to higher emissions in Brazil and India, where education is more accessible to the wealthy. We hypothesize a positive correlation between education and CO₂ emissions.

In summary, the literature review reveals that studies on the relationship between CO₂ emissions and socio-economic factors are limited and often yield conflicting results. Recent research by scholars such as Büchs and Schnepf [29] highlights the importance of household characteristics—whether mentioned in this review or not—for understanding CO₂ emissions and informing the equity considerations of mitigation policies.

2.2. Urban Form Factors

The study of how urban form affects energy consumption and CO₂ emissions has gained traction over the past two decades, following the foundational work of Newman and Kenworthy [39]. Researchers are now focusing on the link between urban form and CO₂ emissions from individual or household mobility behaviors. Some studies, such as those by Ewing and Rong [40], suggest that urban form influences greenhouse gas emissions through land use patterns. Factors like density, land use characteristics, population density, urban size and age, housing type, and local climate significantly impact energy demand and emissions. Dispersed residential areas often have larger, detached homes that consume more energy than the smaller, attached units typically found in denser communities.

The layout of urban spaces significantly influences CO₂ emissions. Concentrating populations and economic activities can create economies of scale and reduce energy usage, leading to lower emissions. Studies suggest that increasing urban density can decrease CO₂ emissions by significantly reducing energy consumption [28,39]. For example, Guo et al. [41] found that neighborhoods in Jinan, China, with a high density, mixed land use, and easy access to public transport had lower transport-related CO₂ emissions. Similarly, Qin and Han [13] reported comparable findings in Beijing. However, these studies mainly focus on neighborhood types rather than specific urban form attributes and do not differentiate CO₂ emissions by trip purpose or consider other influencing factors.

To address the research gap, this paper examines how urban form—considering factors like population density, land use, distance from the city center, and residential-workplace separation—affects CO₂ emissions related to individual daily mobility in urban areas of Lyon, France. Our aim is to understand how the different elements of urban form influence CO₂ emissions. We use conventional indicators from the literature, along with additional metrics, to thoroughly analyze this relationship.

We focus first on density, a key topic of debate given that various morphological configurations can yield the same density, yet it remains central to discussions on sprawled versus compact cities [42]. For our study, we define population density as the number of people per total area. Our second metric is concentration, calculated as the ratio of housing units to the total area occupied by buildings, representing the built surface area. Lastly, we consider the urbanization rate, defined as the proportion of built-up areas to the total gross area of the zone.

Accessibility is another crucial metric in this study, reflecting the availability of public transportation. We utilize two variables: bus accessibility, which measures access to bus services, and metro accessibility, indicating access to metro services. We assume that public transport usage increases with density, concentration, and urbanization rate, as these factors enhance public transportation provision and effectiveness.

Functional mixity, another indicator, represents the ratio of jobs to the working population. This metric is often used in planning to align the working population with available job opportunities. Our study examines the correlation between this ratio, employment density, and concentration by considering the working population, number of housing units, and job quantity in each urban zone, identifying areas with job surpluses, such as industrial zones, and those with more workers than jobs.

Lastly, we assess proximity and centrality activities, focusing on the number of associated units. A distinctive aspect of this study is the inclusion of distance from the city center as an urban form indicator. While previous research has primarily analyzed its direct impact, we also investigate how other urban form indicators interact with this variable. Many indicators, including density, urbanization rate, centralization, and proximity, are interconnected with the distance from residential areas to the city center, showing a stronger effect in areas closer to the center.

3. Methods

Estimating CO₂ emissions from transportation is challenging and can be approached in two main ways. The first method uses aggregated data on total energy consumption or vehicle fleet size, along with average kilometers traveled per vehicle. This top-down approach is straightforward and widely used [43,44,45,46]. However, it faces limitations at the urban scale due to data quality issues, such as the lack of reliable information on the vehicle fleet, overall energy use, and average travel distances. Moreover, this method does not directly link mobility behavior and urban form to socio-economic factors, despite evidence that urban characteristics affect travel distances, mode choices, and CO₂ emissions [12,43]. Most existing research analyzes CO₂ emissions at an aggregated level, overlooking variations due to travel purpose, mode, and socio-economic factors.

In contrast, the second approach is a bottom-up method that estimates CO₂ emissions at the individual or household level. This allows for a more detailed examination of how socio-economic characteristics and urban form influence emissions. While this method is common in emissions analyses, studies rarely integrate urban form and socio-economic factors while analyzing individual or household mobility behavior. This gap likely arises from the extensive and detailed data needed on mobility behavior for large populations, which is often not readily available.

Previous research emphasizes the importance of disaggregating CO₂ emissions from household travel surveys, as shown in studies by Orfeuil et al. [7], Gallez et al. [47], Nicolas et al. [48], and Bouzouina and Nicolas [24]. These studies advocate for combining individual characteristics, such as demographic and socio-economic attributes, with travel behavior data, including transportation modes and vehicle types. The aim is to enhance the accuracy of comparisons between the two distinct periods addressed by the household travel surveys.

This paper introduces a new disaggregated approach using multinomial logistic regressions to analyze CO₂ emissions linked to individual daily mobility behaviors for the years 1995 and 2006, tracking changes in emissions over this period. The proposed methodology provides a detailed method for estimating CO₂ emissions from daily mobility, offering insights at a micro-scale level while considering spatial variability.

Our objective is to examine how various demographic and socio-economic factors, urban form elements, and vehicle attributes impact CO₂ emissions related to daily mobility. We utilize detailed travel data from household surveys conducted in 1995 and 2006. Through CO₂ emissions modeling, we aim to analyze the interplay between socio-economic variables, urban form characteristics, vehicle features, and mobility behaviors, highlighting their collective influence on emissions across different travel purposes.

Initially, one might consider using multiple linear regression with the ordinary least squares (OLSs) estimator. However, our model operates at a disaggregated level, focusing on individuals. Since some individuals may not have traveled on the survey day, resulting in zero CO₂ emissions, using linear modeling could lead to dependence between model errors and explanatory variables, violating independence assumptions and skewing regression coefficient estimates. Therefore, it was necessary to categorize the dependent variable (CO₂ emissions) into multiple classes.

We chose a multinomial logit model to identify the factors contributing to the high CO₂ emissions from urban mobility. Given our focus on individual movements, we divided emissions into three categories: low, moderate, and high. This transformation allows us to retain all data in our analysis (avoiding the exclusion of zero values and outliers) and simplifies the handling of household-level cost calculations.

Logistic regression is commonly used to predict or explain a qualitative dependent variable based on various explanatory variables. When the dependent variable has multiple categories (K > 2), the method is referred to as multinomial logistic regression. This technique involves selecting a reference category and expressing the logit of each of the remaining categories relative to it through a linear combination of predictor variables.

4. Individual CO₂ Emissions

CO₂ emissions related to daily mobility are calculated using the methodology from the European MEET program, with specific assumptions detailed by Nicolas et al. [48]. MEET provides emission curves for cars based on parameters like average speed, engine size, vehicle age, and fuel type. Data on vehicle age and fuel type come from household travel surveys, while engine displacement is estimated from fiscal power. The Davisum traffic model is used to recalculate local travel distances and speeds, with emissions adjusted based on whether a vehicle starts from cold. Notably, CO₂ emissions are attributed only to the driver, as household surveys do not capture vehicle occupancy rates.

For public transport, annual travel distances by mode (bus, trolleybus, tramway, metro) were obtained from the Lyon public transport authority (SYTRAL) and used to populate the “220 networks” database. The average bus speed of 17 km/h was applied to calculate emissions using MEET curves, with emissions from electric modes set to zero due to Lyon’s nuclear and hydroelectric energy sources. By combining annual trip data from the Lyon network with average trip distances from surveys, we estimated the average emissions per passenger-kilometer for public transportation, applying this uniformly across all trips. Although this method ensures an accurate overall result, it smooths out variations between individual trips.

Once CO₂ emissions have been calculated, they can be assigned to individuals, households, or residential zones. Our analysis focuses on individuals to understand variations in CO₂ emissions related to daily mobility over the past decade, categorizing them based on key socio-economic and residential attributes. This classification aims to clarify changes in CO₂ emissions by examining average behavioral shifts among different groups and tracking demographic evolution.

5. Research Area

5.1. Case-Study: Lyon Area

Contemporary urban expansion involves the ongoing growth of cities’ spheres of influence and operational areas, with the boundaries of daily activities lacking temporal stability. To analyze the sources of observed disparities accurately, a longitudinal study across different time frames is ideal. However, limitations in data availability from the last two household travel surveys, which may not fully reflect daily activities, support an analysis based on the perimeter defined by the 1995 Household Travel Survey (Figure 1). While this approach provides spatial comparability, it does not account for phenomena extending beyond the original boundary, leaving changes in CO₂ emissions related to peripheral daily mobility unaddressed.

In 2006, the fixed perimeter defined by the 1995 Household Travel Survey included 1.3 million residents, or three-quarters of Lyon’s urban area’s population. Of all trips made by these residents, 97% had both an origin and destination within this perimeter, a figure consistent with 1995 (96%). However, the 3% of trips extending beyond the perimeter accounted for 20% of the CO₂ emissions from residents’ daily mobility, highlighting the environmental challenges created by peripheral travel. Additionally, the drawing sector, which consists of a minimum of 75 households, is the smallest spatial unit that maintains a representative sample at both time points, allowing for the observation of spatial changes between 1995 and 2006 across 87 identical units within the 1995 perimeter.

5.2. Data Sources

This study utilizes household travel surveys conducted under a standardized procedure by CERTU, enabling comparisons across urban areas and tracking changes in mobility behavior over time. The household serves as the statistical observation unit, with individuals aged five and older being interviewed about their travel on the day prior to the survey, which is conducted from Tuesday to Saturday. This approach focuses on daily mobility and its associated emissions for individuals living within the study perimeter, excluding weekend mobility and external traffic.

Our analysis is based on the last two household travel surveys in Lyon, conducted in 1995 and 2006. The first survey included data from 6001 households, comprising 13,997 individuals who made 53,213 trips, representing a broader population of 1,280,000 in 1995. The 2006 survey had a larger sample of 11,229 households and 25,656 individuals who made 96,250 trips, covering a more extensive area that included adjacent urban regions, with a total population of 1,975,260.

Table 1. Socio-demographic and urban structure characteristics of surveyed person.

Variable	Description	Expected Sign
Socio-Demographic
Gender	1 for male; 0 for female	−
Age	Age of surveyed person	−
Driving license	1—if they possess one; 0—otherwise	+
Status	1—full-time active; 2—part-time active; 3—inactive	−
Revenue	1—low; 2—medium; 3—high	+
Education	1—no studies; 2—primary study; 3—secondary study; 4—higher study	+
Position in household	1—reference person; 2—spouse; 3—child; 4—other	+
Household size	Total number of persons in the household	+
Number of active persons	The number of active persons in the household	+
Number of driving licenses	Total number of driving licenses	+
Number of cars	Total number of cars in the household	+
Urban structure
Population density	The ratio between the active population and the gross surface area	−
Urbanization rate	The ratio of built surface area and gross surface area	−
Functional mix	The relationship between employment and the active population	−
Distance to center	The distance to center	+
Concentration	The number of housing relative to the built surface area	−
Employment density	The number of employment relative to built surface area	−
Metro service	Number of metro stations	−
Bus service	Number of bus stations	−
Proximity of activity	Number of establishments providing local activity	−
Centrality of activity	Number of establishments ensuring centrality activity	−
PT gravity accessibility	Gravity accessibility to employment in urban public transport combined with walking	−
Gravity accessibility of a private car	Gravity accessibility to employment in a private car	+

presents the individual, household, and urban structure variables examined as predictors of CO₂ emissions related to daily mobility.

6. Results

6.1. Statistical Test

To ensure the robustness of our models, we conducted several widely recognized stability tests. The Chow Test was employed to assess the stability of regression coefficients across different time periods, allowing us to identify any structural changes in the relationships among variables [49]. Additionally, we utilized the CUSUM test to monitor the stability of the coefficients over time by analyzing the cumulative sum of residuals, a method frequently used in time-series analyses [50]. A sensitivity analysis was also performed to examine how variations in key assumptions impact the model’s results, aligning with best practices in regression analysis [51]. These tests are crucial for validating our findings regarding the factors that affect CO₂ emissions associated with individual daily mobility behaviors. To address heteroskedasticity, we followed the approach outlined by White [52]. Furthermore, we conducted endogeneity tests using the Durbin–Wu–Hausman test, which confirmed that our model does not suffer from endogeneity issues.

The method employed in this study was multinomial logistic regression analysis conducted through correlation coefficient testing. To mitigate multicollinearity, variables that exhibited strong correlations with other variables were excluded (see

Table 2. Descriptive statistics.

	1995				2006
Variable	Mean	Std. Dev.	Min	Max	Mean	Std. Dev.	Min	Max
CO2 per kg	2.13	3.32	0	56.83	1.98	3.22	0	49.46
CO2_per class	1.48	0.75	1	3	1.46	0.75	1	3
Gender	0.52	0.5	0	1	0.53	0.5	0	1
Age	37.65	20.39	5	98	40.61	21.9	5	98
Position in household	1.9	0.89	1	4	1.86	0.88	1	4
Status	2.22	0.95	1	3	2.25	0.93	1	4
Education	1.94	1.15	1	4	2.89	0.84	1	4
Number of active persons	1.2	0.88	0	4	1.18	0.9	0	5
Revenue	2.89	1.41	1	5	2.9	1.43	1	5
Household size	3.28	1.67	1	11	3.08	1.53	1	10
Driving license	0.66	0.47	0	1	0.68	0.47	0	2
Number of driving licenses	1.73	0.88	0	5	1.77	0.84	0	6
Number of cars	1.42	0.89	0	7	1.39	0.86	0	6
Population density	3347.33	4789.93	34.24	33,149.02	4681.99	5799	81.26	64,597.16
Urbanization rate	0.15	0.11	0.01	0.5	0.16	0.11	0.01	0.5
Functional mix	0.49	0.71	0.04	7.4	0.59	0.76	0.04	9.19
Distance to center	7.61	4.96	0	29.98	6.77	4.33	0	22.19
Concentration	2461.79	5063.75	0	38,871.02	11,649.83	11,332.14	635.7	133,424.7
Employment density	8228.06	16,857.14	308.39	143,419.8	13,943.07	48,531.86	740	1,490,756
Metro service	0.25	0.43	0	1	0.87	0.34	0	1
Bus service	0.85	0.36	0	1	0.31	0.65	0	52
Proximity of activity	18.76	12.21	1	74	35.08	28.84	1	174
Centrality of activity	3.31	2.37	0	16	10.4	13.11	0	101
PT gravity accessibility	-	-	-	-	29,034.17	54,231.49	0	605,988
Gravity accessibility of a private car	-	-	-	-	91,347.43	132,178.6	1404	1,422,686

). The number of driving licenses was highly correlated with both the number of active individuals and household size (with a coefficient of 0.75), while the urbanization rate showed strong correlations with population density (coefficient of 0.65) and concentration, which was also highly correlated with the urbanization rate (0.7) and the active population (0.6). Consequently, we excluded the number of active individuals, urbanization rate, concentration, and employment density from the analysis.

To assess multicollinearity, we computed the variance inflation factors (VIFs) for the independent variables. Although age and its squared term are inherently correlated, their significance in the model mitigates concerns about multicollinearity. The VIFs for all other variables remained below 2.35, well within the acceptable limit of 10. Furthermore, we provided the adjusted R-squared to reflect the number of independent variables included in the models.

6.2. Descriptive Statistics

The descriptive statistics for the variables utilized in this study for 1995 and 2006 are presented in Table 2. The table includes CO₂ emissions per kilogram per individual as well as CO₂ emissions per class. We observe that the average CO₂ emissions are 2.13 kg per individual in 1995 with a standard deviation of 3.32 and 1.98 kg per individual in 2006 with a standard deviation of 3.22. Additionally, we note an increase in variables characterizing the urban structure between 1995 and 2006 (for example, population density increases from 3347 in 1995 to 4682 in 2006). Despite improvements in road infrastructure, public transportation networks, and advancements in vehicle technology focused on reducing CO₂ emissions, individual CO₂ emissions remained stable or slightly decreased between 1995 and 2006. This prompts further investigation into the factors influencing individual emissions related to daily mobility, which we will explore in the subsequent analysis.

6.3. The Association Between CO₂ Emissions and Individual Characteristics

The estimated parameters of models 1 and 3 presented in

Table 3. Multinomial logit model of individual daily CO₂ emissions in 1995 and 2006.

	Variable	Estimated Coefficient 1995			Estimated Coefficient 2006
		Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Class 2: medium emissions	Socio-Economics
	Gender	−0.212	-	−0.209	−0.154	-	−0.186
		(−3.40) ***		(−3.34) ***	(−2.37) ***		(−2.56) ***
	Age	0.0506	-	0.0466	0.0538	-	0.0471
		(4.29) ***		(3.91) ***	(4.61) ***		(3.98) ***
	Age2	−0.00070	-	−0.00068	−0.00064	-	−0.00061
		(−5.39) ***		(−5.16) ***	(−5.18) ***		(−4.90) ***
	position	−0.198	-	−0.211	−0.280	-	−0.309
		(−3.76) ***		(−3.96) ***	(−4.86) ***		(−5.30) ***
	status	−0.313	-	−0.320	−0.530	-	−0.528
		(−7.21) ***		(−7.33) ***	(−9.24) ***		(−10.35) ***
	Education	0.0529	-	0.0433	0.0415	-	0.0379
		(1.94) *		(1.57)	(0.75)		(0.84)
	Revenue	0.0818	-	0.097	0.0177	-	0.0099
		(3.71) ***		(4.34) ***	(4.48) ***		(4.30) ***
	Driver license	3.008	-	3.074	3.754	-	3.832
		(18.63) ***		(18.92) ***	(12.80) ***		(12.73) ***
	Size of household	0.016	-	0.00264	0.0871	-	0.0564
		(0.68)		(0.11)	(3.61) ***		(1.99) **
	Number of cars	0.579	-	0.529	0.867	-	0.805
		(14.16) ***		(12.70) ***	(19.18) ***		(16.36) ***
	Number of active persons	−0.188	-	−0.188	0.054	-	0.0055
		(−3.71) ***		(−3.75) ***	(1.97) **		(0.11)
	Urban form
	Population density	-	−0.0202	−0.0176	-	−0.211	−0.254
			(−3.20) ***	(−2.62) ***		(−5.95) ***	(−6.26) ***
	Functional mix	-	−0.0298	−0.0842	-	−0.0945	−0.033
			(−0.75)	(−1.85) *		(−1.73) *	(−0.39)
	Proximity of activity	-	−0.0052	−0.00177	-	−0.0054	−0.0061
			(1.66) *	(−0.52)		(3.82) ***	(−3.85) ***
	Centrality of activity	-	−0.435	−0.0509	-	−0.0150	−0.00224
			(−2.84) ***	(−2.98) ***		(−4.17) *	(−5.63) ***
	Distance to center	-	0.00077	0.00402	-	0.0178	0.0101
			(0.11)	(0.48)		(0.1.46)	(0.73)
	Public transport service	-	−0.0772	−0.0971	-	−0.0718	−0.107
			(1.74) *	(−1.59)		(−1.21)	(−1.65) *
	Accessibility grav. PT	-			-	−0.312	−1.158
						(−1.36)	(−2.59) ***
	Constant	−4.480	−1.490	−3.839	−4.885	−1.392	−4.224
		(−14.33) ***	(−12.54) ***	(−11.27) ***	(−11.69) ***	(−11.01) ***	(−9.53) ***
Class 3: high emissions	Socio-economics
	Gender	−0.549	-	−0.546	−0.460	-	−0.491
		(−8.11) ***		(−7.89) ***	(−6.80) ***		(−2.56) ***
	Age	0.0562	-	0.0417	0.102	-	0.0931
		(3.96) ***		(2.88) ***	(7.42) ***		(6.50) ***
	Age2	−0.00086	-	−0.00075	−0.00122	-	−0.0012
		(−5.42) ***		(−4.67) ***	(−8.13) ***		(−7.62) ***
	Position	−0.356	-	−0.403	−0.365	-	−0.368
		(−6.14) ***		(6.71) ***	(−6.22) ***		(3.22) ***
	Status	−0.359	-	−0.371	−0.456	-	−0.534
		(−7.78) ***		(−7.88) ***	(−10.72) ***		(−12.44) ***
	Education	0.0986	-	0.0682	0.0492	-	0.116
		(3.51) ***		(2.36) ***	(1.06)		(2.46) ***
	Revenue	0.163	-	0.197	0.0666	-	0.0183
		(6.96) ***		(8.18) ***	(2.66) ***		(7.40) ***
	Driver license	6.009	-	6.196	4.476	-	4.866
		(8.43) ***		(8.67) ***	(8.81) ***		(8.32) ***
	Size of household	0.112	-	0.084	0.0162	-	0.0653
		(4.49) ***		(3.28) ***	(6.77) ***		(2.55) ***
	Number of cars	0.725	-	0.580	1.076	-	0.821
		(16.98) ***		(13.13) ***	(22.13) ***		(15.99) ***
	Number of active persons	−0.212	-	−0.227	0.246	-	0.0.0701
		(−4.05) ***		(−4.26) ***	(5.13) **		(1.38)
	Urban form
	Population density	-	−0.541	−0.442	-	−0.271	−0.324
			(−6.55) ***	(−5.16) ***		(−7.73) ***	(−7.66) ***
	Functional mix	-	−0.0096	−0.049	-	−0.116	−0.0147
			(−0.22)	(−0.96)		(−1.98) **	(−0.22)
	Activity of proximity	-	−0.0032	−0.0026	-	−0.0026	−0.0040
			(1.03)	(−0.68)		(−1.77) *	(−2.30) **
	Activity of centrality	-	−0.0274	−0.0375	-	−0.033	−0.0129
		-	(−1.65) *	(−1.98) **	-	(−0.93)	(−3.19) ***
	Distance to center	-	0.0499	0.497	-	0.0516	0.577
			(7.44) ***	(5.93) ***		(4.43) ***	(4.08) ***
	Public transport service	-	−0.189	−0.199	-	−0.0601	−0.121
			(−3.33) ***	(−2.98) ***		(−0.97)	(−1.73) *
	Accessibility grav. PT	-			-	−0.142	−1.175
						(−2.39) ***	(−2.51) ***
	Constant	−7.885	−1.490	−6.993	−7.547	−2.005	−6.978
		(−10.20) ***	(−12.54)	(−11.61) ***	(−12.44) ***	(−15.07)	(−10.11) ***
	R2	0.2509	0.0288	0.2660	0.2694	0.0298	0.2921
	Number of observations	9042	9042	9042	5156	5156	5156
	log likelihood	−8627.89	−11,545.634	−8453.36	−7528.46	−10008.27	−7302.78

Class 1 is the reference class: low emissions. *, **, and *** represent significance at the 10%, 5%, and 1% levels, respectively.

show the effect of individual characteristics on CO₂ emissions. Most of the signs shown were expected. For example, there is a significant negative effect of the gender variable on CO₂ emissions at the 1% level. Since we used men as the reference (see Table 1), this result means that CO₂ emissions increase with male gender rather than female gender. This indicates that male individuals tend to travel further for work purposes, for example, which encourages them to use the car for these journeys and therefore gives them higher emissions. Referring to Table 3, we see that age has a positive and significant effect at the 1% threshold on membership of a medium-emission class as well as a high-emission class (classes 2 or 3). In other words, the probability of individuals’ CO₂ emissions being higher increases with increasing age. This result reflects the tendency of older people to travel shorter distances for different travel purposes, and they are likely to choose transport modes with low CO₂ emissions. This confirms our hypothesis that there is an inverted U-shaped relationship between CO₂ emissions and age.

The parameters of the status variable clearly show the negative and significant effect of the status of the person surveyed on the quantities of CO₂ emissions for 1995 and 2006 at the 1% threshold (see Table 3). Thus, individuals in full-time employment, which is the reference class in this study (see Table 1), are the class of individuals with the highest CO₂ emissions. This means that the probability of belonging to the class with the highest emissions increases when an individual is in full-time employment. Furthermore, personal status is one of the main factors determining and influencing individual CO₂ emissions. This confirms the work of Bouzouina et al. [2] who stated that an active person emits on average more than three times as much as an inactive person. Consequently, full-time working individuals who own a car tend to emit more CO₂ during a work-related journey in a typical working day. This is confirmed by the sign of the individual’s position variable. Generally, the head of the household is a full-time worker. For this reason, the individual’s position variable (head, spouse, child, other) has a negative effect on CO₂ emissions, taking the head as the reference (see Table 1). This result is significant only for the high-emissions class (class 3). In other words, the probability of belonging to the high-emissions class increases if the person surveyed is the head of the household.

Among the individual characteristics used in this study is the possession of a driver’s license. The results of the estimation of models 1 and 3 show that possessing a driver’s license increases the probability of belonging to the class of moderate and high emissions as well, since the parameters of this variable are positive and significant for the moderate and high-emissions classes (see Table 3). This explains that having a driver’s license encourages individuals to use the car for their travels, thereby leading to an increase in individual CO₂ emissions.

Revenue is one of the main factors used in studies of CO₂ emissions at either the aggregated or disaggregated level. The results presented in Table 3 also show the positive and significant effect of income on CO₂ emissions in 1995 and 2006, which confirms previous work. Thus, the probability that individuals’ CO₂ emissions will be higher increases with higher revenue. In addition, people with higher salaries and car owners tend to travel further and use lower-carbon modes of travel for a work-related and non-work-related reason. Overall, young men with higher salaries and car owners tend to emit more CO₂ in these journeys.

This research clearly shows the effect of education on individual CO₂ emissions linked to daily mobility. The results of model 1 and model 3 presented in Table 3 show the positive and significant effect of education, which means that the probability of belonging to the high-emissions class increases with the level of education (see Table 1). This can be explained by the fact that awareness of environmental issues increases with higher education and contributes to lower CO₂ emissions, which confirms the work of Baiocchi et al. [32] and Büchs and Schnepf [29].

In summary, the trends seen in 1995 largely continue into 2006, with gender, age, employment status, possession of a driver’s license, income, and education remaining as significant factors influencing CO₂ emissions. However, there are signs of changing behaviors, particularly among older individuals and the more educated, who may be adopting lower-emission travel modes over time.

6.4. The Association Between CO₂ Emissions and Household Characteristics

Examining household characteristics in relation to CO₂ emissions is crucial for understanding the potential impacts of mitigation policies. This study incorporates the socio-economic factors of households, and the findings, detailed in Table 3, indicate a clear association between these characteristics and CO₂ emissions from daily mobility. Consequently, the expected signs of the variables align with the research hypotheses.

The results of this research show the importance of socio-economic factors in studies of individual CO₂ emissions. In this context, household structure is a determining factor in CO₂ emissions linked to daily mobility. Variables such as the number of people and the number of workers affect CO₂ emissions linked to household mobility.

The estimation results for model 1 and model 3 show the positive and significant effect of household size on CO₂ emissions. The probability of belonging to the high-emissions class increases with increasing household size. In addition, the number of household trips increases with the increase in household size because of the increase in the number of reasons for travel. As a result, the energy requirements of households linked to transport for the purposes of travel also increase with the increase in household size. As a result, household CO₂ emissions increase with household size. This is highly relevant to the design of mitigation policies, particularly if they include per capita allowances.

Generally, the inclusion of the number of employed individuals in CO₂ emissions models allows us to construct our household typology. The number of employed individuals typically increases the number of household trips by expanding the reasons for travel, primarily related to work, accompanying activities, and studies, among others. In this study, we used the number of employed individuals as one of the household characteristics. However, the unexpected and non-significant parameter sign of this variable may be due to collinearity issues that were not addressed by testing correlation coefficients. Thus, the various variables introduced into the models are not perfectly independent: in general, any two variables are not strictly independent. However, this does not necessarily pose problems when estimating coefficients. This scenario rarely occurs and only in cases where there is a logical mechanical dependence between two variables, which is confirmed in our study for these two variables (household size and number of employed individuals).

Staying with household characteristics, this study also introduces the number of household vehicles. According to the results presented in Table 3, it appears that the number of household vehicles has a significant positive effect on CO₂ emissions. Thus, an increase in the number of vehicles tends to increase the probability of belonging to the high-emissions class. What we can infer here is that the increase in the number of household vehicles leads to a higher utilization of vehicles for various travel purposes, consequently increasing CO₂ emissions related to household members’ mobility. This variable is significant in all estimated models, implying its importance in explaining CO₂ emissions; it is one of the fundamental factors contributing to high emissions.

In this study, both household size and the number of vehicles are identified as key factors contributing to elevated CO₂ emissions. This underscores the significant impact of household characteristics on daily mobility emissions. Therefore, understanding household composition is crucial for accurate interpretation. However, mobility behaviors can vary widely between households due to differing locations and travel purposes. Thus, integrating spatial factors into this research is essential for assessing the influence of urban form on CO₂ emissions associated with daily mobility.

The comparison between 1995 and 2006 shows that household characteristics, particularly household size and the number of vehicles, consistently influence CO₂ emissions related to daily mobility. Larger households and more vehicles are associated with higher emissions in both periods, as they lead to increased travel and energy consumption. While the number of employed individuals within a household is expected to contribute to emissions, its effect is not significant, likely due to collinearity with household size. Overall, these findings highlight the importance of household composition and vehicle ownership in determining emissions, suggesting the need for targeted mitigation policies.

6.5. The Association Between CO₂ Emissions and Urban Structure

Much of the literature has examined the relationship between urban form and mobility behavior across various contexts. The debate regarding the effects of urban form on individual behavior remains unresolved. While the question of urban form’s effects on individual CO₂ emissions related to daily mobility is seldom addressed. This study not only focuses on individual and household characteristics to investigate individual CO₂ emissions but also integrates urban form characteristics into the analysis, giving it a unique specificity compared to other studies, particularly at the city level. This constitutes a relevant basis for planning policy design. Through various variables, the estimation results of the different models demonstrate the extremely significant impact of urban form characteristics on individual CO₂ emissions.

The first factor introduced in this analysis is population density, which is one of the fundamental factors of urban form. The estimation results of model 2 and model 3, presented in Table 3, demonstrate the significant negative effect of this variable on individual CO₂ emissions. It is notable that the parameter values of this variable increase with the integration of socio-economic factors (see model 2 and model 3 in Table 3), as well as the significance level becoming 1% in model 3 instead of the 5% level in model 2. This underscores the importance of combining socio-economic factors with urban form in studying CO₂ emissions. Furthermore, population density leads to a reduction in travel distances and frequency, thereby decreasing individual emissions. Consequently, the population density of an area reduces energy consumption and CO₂ emissions, corroborating the works of Newman and Kenworthy [39] on land use and the studies by Dunphy and Fisher [53]. The results of this research confirm that population density is an important factor in predicting travel patterns and relative CO₂ emissions when adding socio-economic and demographic factors.

Mixed land use is a fundamental factor in explaining mobility behavior and mode choice, yet it is seldom utilized in CO₂ emissions studies. In this research, functional mixite, proximity activities, and centrality activities were employed as variables representing mixed land use. The estimation results of the models demonstrate the significant effect of these variables on individual CO₂ emissions. The negative and significant parameter sign of functional mixite indicates that the presence of jobs in individuals’ residential areas reduces CO₂ emissions related to mobility. Functional mixite enables individuals to have jobs in their residential areas, leading to reduced travel distances and energy consumption associated with daily mobility, consequently lowering CO₂ emissions. Moreover, the negative signs of the parameters for proximity activities and centrality activities suggest that the presence of proximity and/or centrality activities decreases the probability of belonging to the high-emissions class. Therefore, these two variables (centrality and proximity) stand out in the models, particularly proximity to public services and educational, health, and social action establishments. This research shows that the presence of services related to education, social action, health, public administration, or associative activities in the household’s residential area reduces individual CO₂ emissions. Indeed, many households have travel motives related to these services, highlighting the importance of having them close to their place of residence. This study demonstrates that the benefits associated with centrality also have positive effects on the environment, particularly on CO₂ emissions. Furthermore, these results provide a relevant basis for planning policy design, and what adds rigor to this research is its focus on a well-defined area (the urban area of Lyon).

Accessibility is a fundamental factor that shapes urban structures. Indeed, accessibility influences individuals’ mobility behavior, determining the number of trips and distances traveled. Some studies suggest that the problem of spatial mismatch primarily stems from issues of accessibility to buses or effective public transportation. The significance of this factor lies in its ability to assess the impact of various indicators on mobility behavior and thus on individual CO₂ emissions. Very few studies have focused on analyzing residents’ mobility and evaluating their accessibility, while those examining CO₂ emissions and accessibility are even rarer. In this research, we utilized distance to the city center, public transportation coverage, and gravity-based accessibility as indicators of accessibility.

The findings of this research indicate that distance to the city center has a significant positive effect on individual CO₂ emissions. Therefore, the likelihood of emitting more CO₂ or belonging to the high-emissions class increases with greater distance from the center. Additionally, distance to the center influences the number of trips and distances traveled, leading to increased energy consumption and consequently higher emissions from individuals. The robustness of this variable compared to other accessibility indicators further confirms that the Lyon urban area is strongly monocentric. Thus, distance to the center effectively explains the number of trips, distances traveled, and consequently, energy consumption and CO₂ emissions.

Public transport service is the second accessibility indicator used in this study, representing the presence or absence of metro and/or bus stops every 5 min (see Table 1). The estimation results show the negative and significant effect of this variable on individual CO₂ emissions. In other words, the probability of emitting CO₂ or of belonging to a high-emissions class decreases with the presence or increase in the number of metro stops and also with the presence of buses every 5 min. Improving the service therefore helps to reduce the CO₂ emissions associated with daily mobility. This result is relevant to the design of planning policies (at the city level), and also to the implementation of environmental and energy policies.

The effect of gravity-based accessibility on car employment is positive and significant for CO₂ emissions from workers in both models (2 and 3). The good accessibility of the road network is positively associated with the number of trips and thus CO₂ emissions. Direct access to major roadways promotes speed and enables reaching distant jobs, but it also encourages car usage for other daily trips. This confirms the findings of Curtis and Headicar [54], who showed that proximity to major roadways facilitates long-distance travel and intensive car usage.

The comparison between 1995 and 2006 reveals that urban structure has a significant impact on CO₂ emissions in both periods. Population density consistently shows a negative effect on emissions, with higher-density areas leading to reduced travel distances and lower emissions. This trend is stronger in 2006, reflecting the growing importance of integrating socio-economic factors with urban form. Mixed land use, such as proximity to jobs and services, also reduces emissions in both years, encouraging shorter trips and less car dependence. Accessibility factors, like proximity to the city center and public transportation, show similar patterns: greater distance from the center leads to higher emissions, while better public transport access reduces emissions. However, the good accessibility of road networks increases car usage and emissions, highlighting the environmental impact of urban sprawl. Overall, the findings across both periods emphasize the critical role of urban planning and accessibility in shaping CO₂ emissions behavior.

7. Discussion

This analysis of the association between CO₂ emissions and individual, household, and urban structural characteristics provides valuable insights into the dynamics of urban mobility, but it is important to recognize that these findings are based on data from two decades ago. The socio-economic factors influencing CO₂ emissions in 1995 and 2006 may not accurately reflect current trends, as attitudes and mobility behaviors have evolved significantly in the intervening years. However, this historical snapshot remains relevant for understanding the mobility patterns in the early 21st century and how they inform contemporary discussions on urban sustainability and transportation policy. This study offers a glimpse into how urban mobility was structured and how certain socio-demographic factors influenced emissions at that time, providing context for current challenges and future improvements.

In terms of individual characteristics, factors such as gender, age, employment status, and possession of a driver’s license were found to significantly affect CO₂ emissions. Men, older individuals, full-time employees, and those with a driver’s license tended to have higher emissions, as they were more likely to use cars for longer, work-related trips. These findings echo earlier research that identified socio-demographic factors as key drivers of emission patterns in urban environments. Given that these data are historical, it is crucial to consider how these trends may have changed over the last two decades. As urban mobility has increasingly shifted toward sustainable modes, such as cycling and public transport, these traditional patterns of car dependency may have altered, particularly among younger generations. Nonetheless, understanding the historical context of these behaviors provides insight into the challenges faced in reducing emissions in specific demographic groups, which can inform policy aimed at fostering more sustainable mobility choices.

Household characteristics also played a significant role in shaping CO₂ emissions related to daily mobility. Larger households and those with more vehicles were found to have higher emissions, as these factors contributed to more trips and a greater energy demand. While this may still be relevant in current contexts, it is essential to acknowledge the shifting dynamics of household structures, particularly with increasing urbanization and changes in family size. Urban planning policies that encourage smaller households, reduce vehicle ownership, and promote shared mobility services could mitigate emissions effectively. However, these strategies must be revisited with contemporary data to ensure their current applicability, as the landscape of household mobility has likely evolved in the face of new mobility options, such as shared car services and improved public transportation systems.

Urban structure—including population density, mixed land use, and accessibility—was found to significantly impact emissions, with higher population density and mixed land use correlating with lower emissions. These findings underscore the critical importance of urban planning in managing CO₂ emissions. Higher-density developments and mixed-use areas tend to reduce the need for long-distance car trips and promote the use of more sustainable transportation options like public transit and walking. However, urban development patterns have evolved considerably since the 1990s, with growing emphasis on densification in city centers and the development of more pedestrian-friendly environments. While these trends may have continued in the present day, it is necessary to reassess the current state of the urban structure to identify whether the historical links between urban design and emissions still hold. More recent urban design strategies, such as transit-oriented development (TOD), should be considered in light of contemporary mobility trends.

The historical data analyzed in this study suggest several policy recommendations that are still relevant today but need to be reframed in light of changing urban dynamics. Urban planning strategies should prioritize higher-density developments and mixed land use to reduce reliance on private vehicles, as this remains a foundational principle for sustainable urban mobility. Policymakers should continue to promote transit-oriented development (TOD), focusing on integrating public transport into residential and commercial areas to reduce the need for car commuting. At the same time, contemporary policies should reflect the ongoing changes in mobility behavior and technology, such as the adoption of electric vehicles and the growth of shared mobility services. Public awareness campaigns should continue to focus on reducing car dependency, particularly targeting high-emission groups such as full-time workers and car-owning households. These groups, identified in the historical data, remain key targets for policy interventions aimed at reducing emissions.

Transportation policies should also continue to increase the coverage and frequency of public transportation services, particularly in areas further from city centers, to offer viable alternatives to car use. However, these policies must be updated to reflect current trends in urban mobility, such as the increased reliance on digital platforms for ride-sharing and on-demand transport services. By taking into account both historical patterns and current trends, these combined approaches can contribute significantly to sustainable urban mobility and the reduction in urban CO₂ emissions.

8. Conclusions

The relationship between individual CO₂ emissions from daily mobility and various factors such as individual, household, and urban characteristics has garnered increasing scientific interest. However, many studies have not fully integrated these key elements, which are crucial for understanding the dynamics of mobility behavior and emissions. This study, focusing on data from two specific time periods—1995 and 2006—provide a historical perspective on how socio-economic factors, land use characteristics, and accessibility influenced CO₂ emissions during that period. The case study of Lyon helps illustrate these dynamics, highlighting how these factors shaped mobility patterns two decades ago. This historical analysis offers valuable insights into the mobility trends of the past, and while the data may not fully reflect current realities, they provide a basis for understanding how past behaviors might inform contemporary discussions about sustainability and emissions reduction.

Using multinomial models, the study examines how personal traits (e.g., gender, age, employment status, possession of a driving license, income, and education) influenced CO₂ emissions from daily mobility. The findings reveal that certain groups—such as men, older individuals, full-time workers, higher-income earners, and those with driving licenses—tended to have higher emissions. Interestingly, the relationship between education levels and emissions was more complex, with both positive and negative associations depending on other factors. Furthermore, larger households and those with more vehicles exhibited higher CO₂ emissions due to increased travel needs, underscoring the importance of considering household dynamics in emission analyses. While the statistical significance of these relationships was evident, the overall explanatory power of the models remained relatively low, suggesting that other unexamined factors may play a role in influencing emissions.

Urban characteristics such as population density, mixed land use, and public transport accessibility were also found to impact emissions. A higher population density and better access to public transportation were correlated with lower individual CO₂ emissions, indicating the importance of urban planning in shaping mobility choices. These historical findings are still relevant today, providing context for current efforts to reduce emissions through sustainable urban design and improved public transport infrastructure. However, it is important to note that the study’s findings reflect the conditions two decades ago, and more recent trends, such as the rise of electric vehicles, shared mobility services, and an increased emphasis on sustainability, would likely alter these relationships.

The study highlights the need for policies that promote alternatives to private car travel, particularly for high-emission groups like men, older adults, and full-time workers. Encouraging the use of public transport, car sharing, cycling, and walking can significantly reduce emissions. Additionally, policies that prioritize higher population density, mixed land use, and better public transport access could help further reduce reliance on private vehicles and mitigate emissions. Creating walkable, bike-friendly communities is another potential strategy to lower emissions and foster sustainable urban mobility. It is crucial that these policies consider equity, ensuring that vulnerable populations are not disproportionately impacted by emission-reduction measures.

While this study offers a valuable historical analysis, future research should incorporate more recent and longitudinal data to better capture evolving trends in mobility and emissions. A longitudinal approach would provide a deeper understanding of how individual and household characteristics impact emissions over time, and how changing urban structures influence mobility patterns. Moreover, incorporating spatial factors related to urban form could offer valuable insights into how different city layouts and infrastructure designs impact emissions, helping to tailor policies to specific urban contexts. Future studies should also evaluate the effectiveness of existing policies aimed at reducing CO₂ emissions, assessing their real-world impact on mobility behaviors and emissions outcomes.

This study underscores the complexity of the relationship between individual characteristics, household dynamics, urban form, and CO₂ emissions from daily mobility. By understanding these factors, policymakers can develop more targeted strategies for mitigating climate change and promoting sustainable urban development. However, due to the historical nature of the data and the limitations inherent in studying trends from two decades ago, it is essential that future research revisits these questions with up-to-date data and considers the impact of emerging trends in mobility, such as the adoption of electric vehicles, shared transportation services, and shifts in urban mobility practices.

The primary limitation of this study is the reliance on data from 1995 and 2006, as no more recent comprehensive data on individual mobility-related CO₂ emissions are available. This limits our ability to fully capture current trends in urban mobility, such as the rise of e-mobility and the increased use of shared transportation services. Additionally, the relatively low explanatory power of the models suggests that other factors, such as individual preferences and attitudes, may influence CO₂ emissions but were not accounted for due to data constraints. Future research should aim to fill these gaps by incorporating more recent data, exploring longitudinal trends, and examining the spatial factors and urban designs that influence emissions. By doing so, future studies can better inform policies and interventions aimed at reducing CO₂ emissions from urban mobility.