Main

Cities are increasingly recognized as pivotal actors in sustainable development (SD). With a striking 57% of the global population living in urban areas1, their impact is undeniable. Despite occupying only about 3% of the planet’s surface, cities account for more than 80% of global economic activity2 and they are responsible for three-quarters of worldwide resource consumption and greenhouse gas emissions3. Such figures have prompted international entities to acknowledge the crucial role that cities play in global SD policy, such as the United Nations sustainable development goals (SDGs), the New Urban Agenda of Habitat III and the Paris Agreement. As such, the international discourse has shifted from a perspective that frames cities as primary contributors to environmental destruction and social inequality to an approach that values the potential of cities in creating a more sustainable future4,5,6. This is especially visible in climate change policy as cities have been formulating locally impactful policies but also making a global difference in times of stalled international negotiations7. Indeed, cities are now considered essential in realizing not only the urban-centric SDG 11 but the entire array of SDGs.

Cities across the globe have been developing innovative and progressive urban sustainable development (USD) policy plans. For example, with its so-called ‘city doughnut’ plan, Amsterdam aspires towards a city that thrives socially and ecologically while acknowledging its global responsibility to coexist with the rest of the planet and its inhabitants8. Similarly, Helsinki is striving to achieve carbon neutrality by 2030, has provided extensive public data on its SDG progress and is leading as the first European city to implement a voluntary local SD review9. These ambitious USD policy plans not only address local SD aspects but also shape national and international SD agendas, thereby endorsing a uniquely urban perspective on SD6.

Nonetheless, when assessed with respect to their potential to contribute to a sustainable future, the policies that cities enact may often be found to be deeply ambivalent10. Vulnerable residents may suffer as a result of urban ecological transformations through processes such as ‘climate gentrification’11,12, a narrow focus on ecological sustainability can marginalize local equity concerns13, and mainstream economic growth policies may be disguised as SD-driven programmes14, a process that can be called SDG-washing.

The scope and substance of USD policy plans varies, ranging from the protection of biodiversity and the promotion of circular economies to reducing poverty15,16. This complexity is compounded by the reality that USD should not be implemented in a technocratic manner if it is to be socially accepted and effectively adopted but rather should be coproduced with urban residents17. Social acceptance and democratic legitimacy are particularly critical, given the urgency of sustainability transformations; simultaneously, SD is often criticized as technocratic and depoliticized6,18,19. As such, our aim in this study is to examine residents’ preferences with respect to USD policy issues and to compare these preferences with existing USD policy plans.

We conducted a preregistered, original survey with a total of 5,800 respondents from eight systematically selected European cities: Antwerp, Frankfurt, Helsinki, Lisbon, Manchester, Marseille, Milan and Valencia. The survey comprised a combination of framing and several choice experiments. At the beginning of the experimental component, survey participants were randomly framed with one of two different SD conceptualizations—either the triangle or the nested SD concept (see Fig. 1 for the two SD frames and for the 17 USD policy issues). This framing experiment placed a bracket around the subsequent choice experiment and resembles a randomized control trial. The survey participants were then required to rank 17 USD policy issues that we identified in the literature as topics currently discussed under the label of USD (Supplementary Table 1). In the choice experiment that followed, the survey participants were asked to assess two USD policy plans that consisted of six randomly selected USD policy issues out of 17. In the next analytical step, we compared the residents’ USD preferences with real, existing USD policy plans. We collected and thematically coded 166 USD policy plans from the eight cities mentioned above to assess how each city prioritizes different USD policy issues, based on the frequency with which these issues are mentioned in the plans (Methods).

Fig. 1: Overview of the three-part survey experiment.
figure 1

Visual representation of the three main components of the survey experiment. Framing experiment (left): here, two frames, represented as ‘triangles’ and ‘nested’, were randomly presented to the respondents. Each frame was received by half of the sample. Ranking exercise (middle): this part of the figure shows the 17 USD issues that respondents were asked to rank. The issues are grouped into three SD domains: environment, society and economy. These groupings are shown here for clarity. Respondents ranked the issues independently and these three SD domains were not presented during the actual experiment. Choice experiment (right): this section provides an example of the choice experiment task. Respondents were presented with two proposals, each containing six USD policy issues randomly selected from the 17 issues. The issues shown in this example are for illustration only and the actual proposals varied because of the random selection of issues.

Full size image

We selected eight cities from the sample of all 86 cities in Western and Northern Europe with more than 350,000 inhabitants (Methods). We decided to select cities that represent four types of diverse European contexts in terms of ecological vulnerability and financial capacity (Fig. 2). For our study, ecological vulnerability refers to the potential adverse impacts of climate change on cities, particularly heatwaves, droughts and flooding20. Financial capacity, meanwhile, denotes the ability of a city to finance and implement USD policies21,22. We selected two cities with a similar ecological and economic context to see if potential patterns evolving from this context are stable and we ensured that we not select cities from the same country.

Fig. 2: Systematic case selection.
figure 2

Depiction of the 86 European cities based on GDP per capita (y axis) and vulnerability score by ref. 20 (x axis). The eight cities highlighted in green are our selected cases. The horizontal dashed line depicts the mean GDP per capita of the 86 cities. The vertical dashed line depicts the mean vulnerability index score of the 86 cities.

Full size image

USD policy issues and SD concepts

Cities subsume various policy issues under the umbrella term of SD15,16,21. For example, cities combine several policy issues in planning for compact urban settlements23,24 or they may coordinate health and environmental concerns with housing provision25. Cities also gather existing yet dispersed and ad hoc initiatives under the umbrella term of USD.

Given the broad range of policy issues encompassed by USD, we aimed to identify a comprehensive yet manageable list of specific USD policy issues to examine in this study. To this end, we combined a review of academic literature on urban sustainability with an analysis of various recognized USD frameworks and indicators. The result is a list of 17 distinct USD policy issues, which we categorize into three broad SD domains: environment, society and economy (Supplementary Table 1 and discussion in Supplementary Information).

Our study also recognizes and examines the impact of different conceptualizations of SD and USD approaches because they frame what lies within the scope of SD and form the basis for measurements and indicator frameworks26,27. We therefore apply a rough distinction between two distinctive conceptualizations of SD: the triangle and the nested approach (for example, ref. 26; Fig. 1 and discussion in Supplementary Information).

The triangle conceptualization approaches SD through ecological, social and economic domains28,29, also referred to as the ‘triple bottom line’ approach of sustainability30. There is no explicit prioritization or hierarchy of USD domains. The quest is to find synergies and/or a balance between the three USD domains. This conceptual SD understanding is also reflected in the UN preamble to its 2030 agenda, which frames the SDGs as ‘a plan of action for people, planet and prosperity’31. Given the criticisms of a dominance of economic goals through market-led implementation of SD and the danger of diluting sustainability measures through consensus building32, scholars further developed the triangle framework towards identifying trade-offs between the three potentially conflict domains because sustainability measures cannot maximize all three goals of SD33.

By building on these criticisms, the nested conceptualization of SD rejects the assumption that the three domains are equal and advocates for a more ecological-centric perspective, in which human and economic activities are nested within the biophysical environment26. One application of this SD understanding is the so-called ‘doughnut model’34, which integrates planetary ecological boundaries and the social foundations of SD. The model operationalizes the environmental limits with the planetary boundaries at the planetary scale35 and the social foundations with the 11 social priorities stemming from the Rio+20 Earth Summit34,36. The planetary boundaries should not be exceeded and the social foundations should not be undershot. Thus, SD can occur only within both boundaries.

Residential preferences and potential discrepancies

Based on these more conceptual considerations, we aimed to examine city residents’ USD preferences. The literature of sustainability studies recognizes the importance of democratic acceptance and the potential divergence between residents’ USD preferences and enacted USD as they work with the concept of negotiated understandings of sustainability that may be in contrast to predetermined definitions of SD28,37. The political science literature distinguishes between input (for example, participation in policy-making), throughput (for example, transparency of the process) and output legitimacy (for example, effectiveness of policies)38,39. These three types of legitimacies are connected. It is expected that cities engaging more participatively in their USD policy formulation process are likely to enact more extensive USD policies40. Although we do not go into the details of these types of democratic legitimacies and we do not examine the effect of participation on USD policy preferences, we will examine whether USD policy aligns with public preferences.

To explain potential democratic discrepancies, the urban planning and urban governance literature suggest that different rationalities are at work between planners and residents. We would expect disparities between the more technocratic expert rationalities and residents’ everyday rationalities. Policy-makers, planners and academics all too often base their interventions or recommendations ‘on values, beliefs or rationalities of those for (or with) whom they plan, which frequently do not hold’41. The technocratic character of policy-making processes and limited public participation can lead to divergences between governmental policies and residents’ democratic demands40,42. This discrepancy may also be caused by complex urban governance arrangements in the negotiation and implementation phase of USD policies, which entails interaction between various involved actors, each with different perspectives and interests10,43. It might also result from powerful interests influencing policies towards their own (economic) interests, overshadowing social or ecological aspects44,45. Furthermore, USD policy adoption and implementation may not keep pace with residents’ evolving priorities, thus creating time-lag discrepancies. As such, we do not anticipate that residents’ USD priorities will be directly and linearly implemented in USD policy plans. However, a clear understanding of this potential mismatch between resident preferences and USD policies—the extent to which USD policy aligns with public preferences (output legitimacy)—will probably yield valuable insights for enhancing public support for sustainability measures and fostering democratic urban policy-making.

Therefore, our analysis is guided by two expectations. Given the distinct prioritization of environmental USD policy issues in the nested SD concept and the absence of such hierarchical distinctions in the triangle SD conceptualization, we expect that residents who are presented with the nested SD concept are more likely to give preference to environmental USD policy issues than residents who are presented with the triangle SD concept. Second, given that policy-makers and residents tend to operate according to different rationalities41, we expect differences between residents’ USD policy issue preferences and the priorities enshrined in actual existing USD plans.

Results

We will first present the findings pertaining to residents’ USD policy preferences based on the ranking exercise and then of the choice experiment. Next, we will present the results of our analysis of 166 existing USD plans from the eight cities and compare them with the survey findings.

Ranking exercise

We asked the survey participants to rank the 17 USD policies in terms of their importance to them, ranking the most important issue at the top and the least important at the bottom. The results indicate that USD policy issues that may be categorized as aiming to secure basic human needs are deemed most important. The highest ranked USD policy issues are (both frames combined): (1) cost of living (6.18), (2) public health (6.63), (3) education (6.96), (4) poverty (7.48), (5) unemployment (7.94), (6) water and air quality (8.25) and (7) wealth and income equality (8.69) (Fig. 3). These top seven USD policy issues are separated by a steady interval, while the overall largest gap (0.663) in the top of the ranking is discernible between the issues ranked seventh (wealth and income equality) and eighth (climate change mitigation). Against our theoretical expectations, the SD conceptual framing does not appear to matter much to residents, except for two USD policy issues—biodiversity and water and air quality—in which we can observe a significant difference in the expected effect (see also two-sample t-test results in the Supplementary Information and Supplementary Table 3).

Fig. 3: Ranking results.
figure 3

The average rankings of the 17 USD policy issues as mean values ± s.e. as evaluated by the survey respondents (n = 5,800 total and n = 2,900 for each of the two frames). The USD policy issues are differentiated according to the USD domains, represented by distinct symbols and by the survey frame, denoted by varying colours. See also two-sample t-test results in the Supplementary Information and Supplementary Table 3.

Full size image

Choice experiment results

The choice experiment yields similar results to the ranking exercise. When the choice experiment included USD policy issues, such as the cost of living, education and public health, they consistently achieved higher acceptance levels (Fig. 4). Conversely, USD policy issues, such as biodiversity, circular economy and certain sociopolitical issues (for example, integration of minorities, discrimination and democratic participation), tend to reduce the likelihood that residents will accept the randomized USD policy plans.

Fig. 4: Choice experiment results.
figure 4

Illustration of the AMEs from the rating task in the choice experiments. These effects are categorized according to USD domain, depicted with unique symbols and by the survey frame, signified by different colours. Error bars represent standard errors (s.e.) of the AMEs. The data are presented as AME values ± s.e. The analysis is based on a total sample size (n) of 5,800 respondents, with 2,900 respondents for each of the two frames. Each respondent rated three pairs of proposals with six randomly presented attributes, resulting in a total of 34,800 observations. The study includes two framing groups, focusing on comparing different proposals within these frames.

Full size image

These preference patterns remained stable, irrespective of the conceptual SD framing. They also remained stable in the results of the choice experiment with forced choice answers (Supplementary Fig. 2) as well as among the eight cities (Supplementary Fig. 3). This stability of USD policy issue preferences across different framings, different methods and analyses (ranking and choice experiments both with and without forced choice) and different cities contributes substantially to our understanding of public perspectives on USD and underscores the presence of USD policy preferences among urban populations. The limited and inconsistent impact of SD framing on the experimental survey findings also suggests that framing strategies may not be as effective in shifting public opinion on USD policies as was previously believed.

City differences

The results also show that the average acceptance of USD policy agendas is relatively high, at 72.61%. Results across the respondents from the eight cities are generally stable but they reveal some interesting differences, albeit not in the pattern that we would have expected. The two cities with the significantly highest acceptance levels of USD are Lisbon and Valencia (Fig. 5). Both cities exhibit a comparably low gross domestic product (GDP). Lisbon’s ecological vulnerability is relatively high20, while Valencia’s is relatively low. The comparably rich and ecologically vulnerable cities, Milan and Antwerp, exhibit average USD acceptance levels. These patterns are also supported by the experimental survey results pertaining to the 17 USD policy issues (Supplementary Fig. 3). Additionally, it is evident that the conceptual SD framing does not significantly influence public USD acceptance except for small differences in Frankfurt and Antwerp.

Fig. 5: Acceptance of USD policy agendas in eight cities.
figure 5

The overall average acceptance of the randomly composed USD policy agendas in the choice experiment across the eight selected cities is shown. For each city, the figure shows the average derived from the two distinct survey frames, allowing for a direct comparison of strategy acceptance within each city’s context. Error bars represent s.e. of the AMEs. The data are presented as mean values ± s.e. The analysis is based on a total sample size (n) of 5,800 respondents (900 in Manchester and 700 in each of the other seven cities). Each respondent rated three pairs of proposals with six randomly presented attributes, resulting in a total of 34,800 observations.

Full size image

Overall, the results of the survey analysis indicate that residents prioritize USD policy issues that may be described as aimed at securing basic human needs. Meanwhile, environmental USD policy issues that are typically prioritized in literature on sustainability were associated with lower preference ratings. Hence, one key take-away is that basic human needs must be secured to ensure a more substantial and environmental-oriented USD agenda. We will now compare these residents’ USD preferences with analysis of the existing USD policy plans to determine whether the residents’ preferences are aligned with USD policy plans.

Analysis of existing USD policy plans

Our analysis of the 166 USD policy plans revealed that the most highly prioritized USD policy issues are (1) education and (2) biodiversity. Separated by a considerable gap, these are followed by (3) public transportation and (4) urban green spaces and, following another considerable gap, (5) poverty and (6) water and air quality (Fig. 6). Education and biodiversity are the most frequently mentioned USD policy issues in each city, whereas wealth and income inequality, cost of living and integration of minorities had a frequency count of zero in several cities (Supplementary Fig. 7). A democratic discrepancy emerges with respect to residents’ USD preferences—in particular, a considerable overconsideration of biodiversity, public transport and urban green spaces and a remarkable underconsideration of cost of living, wealth and income equality, unemployment and public health. Residents’ USD preferences and actual existing USD policy plans, however, are exceptionally well aligned with respect to education, poverty and water and air quality. Robustness tests that control for when the USD plans have been introduced and were active show that the results are stable regarding USD policy issue prioritization patterns over time (Supplementary Fig. 5 and Supplementary Table 5).

Fig. 6: Frequency of mentioned USD policy issues in USD policy plans.
figure 6

Illustration of the frequency with which each of the 17 USD policy issues is mentioned in the 166 USD policy plans.

Full size image

Comparisons of the different USD policy plans across the eight cities reveals that the USD plans for Lisbon mention by far the most USD policy issues in terms of absolute numbers, ahead of Valencia and Helsinki (Supplementary Fig. 6). This suggests that neither ecologically vulnerable nor comparably rich cities engage in more active USD policy-making, at least when it is measured on the basis of issue frequency in USD policy plans. We can also observe that Lisbon, Valencia and Helsinki mention more diverse USD policies than the other cities. Manchester, Marseille, Frankfurt and Antwerp do not appear to address USD holistically but focus on specific USD policy issues (Supplementary Fig. 7). Overall, these findings corroborate studies from US cities showing that USD policy plans do not promote sustainability in a profound and balanced way15,16. Yet, particularly Lisbon is an interesting exception, given the high USD acceptance by its residents combined with the high USD policy activity in plans, suggesting that Lisbon’s USD activities are backed-up by its residents. Lisbon’s ambitious USD agenda thereby aims to unite its post-2008 financial crisis economic recovery with a robust ecological transformation, aiming to achieve carbon neutrality by 205046.

Discussion

We analysed residents’ USD policy preferences and compared them with existing USD policy plans in eight systematically selected European cities. On average, we observed a relatively high acceptance rate of randomized USD policy agendas (72.61%). Thus, residents appear to support urban endeavours in pursuit of USD policy-making and generally support diverse USD policy issues. However, more sobering results emerged with respect to the preferred content of USD policy agendas: the two more environmental USD policy issues—biodiversity and circular economy—and the three sociopolitical USD issues—integration of minorities, discrimination and democratic participation—showed comparably low acceptance levels. Furthermore, we found no evidence to suggest that different conceptual framings of USD influence residents’ preference formation.

Regarding potential democratic discrepancies between residents’ USD policy preferences and the issues that are prominent in USD policy plans, our analysis revealed a match for education throughout the eight cities, which is a particularly notable finding, given that education is an SDG in itself as well as a facilitator for SDGs in general (in the sense of education for SD47). Moreover, we observed a match in the USD policy issues poverty as well as air and water quality, albeit to a slightly lesser degree.

However, our analysis points to widespread democratic discrepancies, primarily in relation to those aimed at securing basic human needs. A remarkable relative underconsideration of cost of living, wealth and income equality, unemployment and public health was evident in the existing USD policy plans compared to residents’ USD policy preferences. By contrast, a relative overconsideration of biodiversity, public transport and urban green spaces. Although cities mostly pursue strategies relating to long-term USD policy issues (such as education, biodiversity, public transport and urban green spaces), residents prioritize issues aimed at securing their essential and everyday needs, such as cost of living, public health, poverty and unemployment. These findings are largely stable across the eight diverse European cities.

This key finding highlights a discrepancy between residents’ more everyday socio-economic concerns and preferences and the more long-term and ecologically focused USD policy plans. The planning literature suggest that this discrepancy may be explained by the competing rationalities of a technomanagerial and market-oriented system of planning, often in alliance with other powerful actors, such as profit-driven developers, versus the rationalities associated with coping with everyday life on the part of the residents41,45. This key finding also resonates with the sustainability literature which mobilizes concepts of environmental justice, social sustainability or socio-ecological transformation12,13,17,19. They stress the importance of just sustainability transformations, a process that aims to pursue ambitious SD interventions while alleviating or at least not widening socio-economic inequalities. The crucial implication of this study seems to be the importance of securing basic human needs if cities want to pursue (more) profound USD policy plans focused on key, long-term environmental SD issues, such as climate change, biodiversity or renewable energy.

In spite of these insights, our study has several limitations. First, our analysis of USD plans is based on publicly available plans and therefore does not reflect all ongoing urban sustainability initiatives, particularly those not documented in the plans or those emerging from grassroots movements and community-led projects. Furthermore, differences in data availability and transparency across the eight cities may have influenced our ability to capture the full scope of USD policy plans. Second, our frequency analysis can showcase the coverage of USD policy issues in plans but tells us little about the implementation and financing of these plans. There might be a substantial gap between the plans and what is actually implemented. Similarly, we looked at a particular measure of USD output legitimacy. Yet, the concept of democratic legitimacy is broader and also incorporates input and throughput legitimacy38,39. Third, our study relies on survey data to infer the residents’ preferences. Although this method has several strengths, it is inevitably associated with potential biases; for example, respondents may interpret questions in different ways. Additionally, owing to sampling limitations, the respondents of our survey may not fully represent each city’s population. To partially account for this issue, we ensured that the sample of respondents was proportional to the populations of the eight cities in terms of age and gender and we implemented a soft quota on income deciles (Methods). However, despite the soft quota on income, our sample is slightly skewed towards higher income deciles. Finally, there might be a time lag between the date of the USD policy plans and our survey study, although the results are largely robust across plans and cities (Supplementary Figs. 5 and 8 and Supplementary Table 5). In addition, the survey experiment may have been influenced by current events or media coverage at the time of the survey.

On the basis of these analyses, we suggest that urban policy-makers and planners must take residents’ priorities seriously and incorporate essential, everyday human needs as a central aspect of USD, particularly if they wish to pursue a profound and more long-term-oriented USD policy agenda. To enhance the legitimacy of USD policy-making, residents’ everyday realities must be studied and addressed, particularly those relating to people’s economic security (for example, cost of living, unemployment and public health), as these tend to fall short in existing USD plans. Along these lines, it seems beneficial for planners and policy-makers to have more direct encounters with diverse everyday realities, which should promote more sincere participation and collaborative urban planning41. We would therefore expect that USD policies codesigned with residents through profound participation to have a better alignment with public priorities and thus potentially ease these democratic discrepancies. Yet, this assumption has to be tested.

Our findings suggest that urban planners and urban policy-makers are well advised to prioritize social issues alongside environmental issues in USD policy-making, pointing to the importance of socio-ecological sustainability transformations. It seems to be important to residents that local equity concerns are not marginalized by the primary focus on ecological sustainability. Although we observed only limited influence of SD concepts on residential preferences, it might nonetheless be beneficial for urban planners and policy-makers to work with newer SD concepts that bridge socio-environmental considerations and emphasize the need to secure basic human needs. This SD conceptualization is prominently advocated in the doughnut model34,36, in the framing of sustainability in the Anthropocene48, and in safe and just conceptualizations of planetary boundaries35. While these conceptualizations are able to capture the importance of essential and everyday needs, they seem to not (directly) impact USD policy preferences of residents.

In general, the article suggests that SD policy-making must be democratic and avoid amplifying socio-economic inequalities. This means that securing basic human needs is not simply of secondary priority or morally desirable but is in fact democratically demanded and the basis for pursuing more comprehensive and profound USD policy-making.

Methods

Our research aims to dissect the complex interplay between residents’ preferences and actual (USD) policies in different European cities. This requires a multimethod and a comparative research approach. To achieve this, we collected comprehensive data from eight systematically selected European cities. We designed an original survey to assess residents’ preferences for different facets of USD and coupled this with a collection and analysis of all publicly available USD policy plans for each city. The robustness of our method is strengthened by the choice experimental design of the survey, ensuring that the derived results hold under various framing conditions, thus reducing the impact of potential bias.

In the sections that follow, we elaborate on our case study selection, data collection process, survey design and analysis and, finally, the coding of USD policy plans.

Case selection

The study relies on analyses in eight European cities: Frankfurt, Helsinki, Antwerp, Milan, Manchester, Valencia, Marseille and Lisbon. These cities were systematically selected on the basis of the sample of all 86 cities in Western and Northern Europe with populations of more than 350,000 inhabitants. Only incorporating cities in Western and Northern Europe excludes cities in the following European countries: Albania, Belarus, Bosnia-Herzegovina, Bulgaria, Croatia, Czech Republic, Greece, Hungary, Kosovo, Moldova, Montenegro, North Macedonia, Poland, Romania, Serbia, Slovakia, Slovenia and Ukraine. We also did not include Turkish and Russian cities as only small parts of both countries geographically belong to Europe. Our case study sampling approach is motivated to ensure comparability between cities in terms of similar types of democratic local political systems (Western and Northern Europe) and in terms of size and capacity of their administration (populations >350,000 inhabitants).

The case selection was structured along the two dimensions of ecological vulnerability and financial capacity. To measure financial capacity, we included the GDP per capita of the metropolitan region49,50 as these data were available for all the 86 cases. For ecological vulnerability, we used the vulnerability index of ref. 20 based on the three environmental risks—heatwaves, droughts and flooding. We aimed to select according to capacity to act (GDP as a proxy) and urgency to act (ecological vulnerability as a proxy). We also ensured that we did not select two cities from the same country to capture a variety of national SD policy agendas, local autonomy and national intergovernmental frameworks. Moreover, we did not select the largest European cities (>2 million inhabitants)—namely, Berlin, Madrid, London, Paris and Rome. Thus, the selection criteria were comparable European cities (for population size) with varied ecological and economic contexts. We also want to select four city pairs with similar ecological and economic contexts to see if potential patterns evolving from this context are stable. We pursued this highly systematic case selection procedure to enhance the generalizability of our findings and to generate insights from diverse European cities beyond the very large cities or the European SD champions, such as Amsterdam, Barcelona, Paris or Stockholm.

For the actual selection of cases, we discussed different possible combinations considering our multifaceted criteria outlined above. We especially discussed potential pairs of cities with a close proximity in each of the four quadrants (Fig. 2), meaning that they have similar economic power and ecological vulnerability levels. Thereby we also analysed the descriptive statistics of our selected cities and against the overall dataset. For instance, the mean vulnerability index of these cities (4.31) slightly exceeds the mean (4.12) of the overall dataset, indicating a nuanced higher risk profile. Similarly, their mean GDP per capita (38,928) is marginally lower than the dataset average (40,281), presenting a mix of moderately affluent cities. This also means that we did not opt for statistical outliers but we aimed to capture a comprehensive snapshot of the European urban environment.

By applying these case selection criteria, we selected two cities representing each quadrant. Frankfurt and Helsinki with high financial capacity but relatively low ecological vulnerability; Milan and Antwerp with high financial capacity yet high ecological vulnerability; Manchester and Valencia with low financial capacity and low ecological vulnerability; and finally, Marseille and Lisbon with low financial capacity and high ecological vulnerability (see the relative close pairwise distances in Fig. 2).

Survey research

The survey relied on a sample selected from among the 18+-year-old population of the eight cities. The sample of respondents is proportional to the populations of the eight cities in terms of age and gender and we implemented a soft quota on income deciles (meaning that each income decile had to be filled by 5–15% of respondents). The sample is thus representative for age and gender but slightly skewed towards higher income deciles (average income decile of 6.4). The study sample comprised 5,800 participants (900 in Manchester and 700 in each of the other seven cities). Participants were invited through online panels coordinated by IPSOS. The survey was conducted from September to November 2022. The study protocol received ethics approval from the ETH Ethics Board (approval no. EK-2022-N-152). The study was preregistered at https://doi.org/10.17605/OSF.IO/CZ3AK before data collection to ensure transparency and accountability.

To provide a detailed context of the sociodemographic and political orientation of the respondents, we have included a comprehensive overview table (Extended Data Table 1). This table summarizes key metrics such as age, gender distribution, employment status, political ideology, city connection and preferences about various dimensions of sustainability. The overview data indicate a balanced gender representation and an age distribution that is relatively consistent with the urban population demographics. A total of 70% of the survey respondents are employed. The sample is also balanced with regard to political ideology and political interest, whereas there is a tendency towards prioritizing environmental protection and social justice over economic wealth. The latter is also in line with our findings from the experimental part of the survey. Survey participants were asked to complete the online survey and they used a median response time of 19.9 minutes. The questionnaire included questions on the participants’ sociodemographics and general political attitudes. The second part of the survey study comprised a combination of several framing and choice experiments. Participants randomly received a different SD frame (see the two different conceptualizations of SD in Fig. 1), placing a bracket around the subsequent choice experiment. This resembles a randomized control trial. The two SD frames were chosen on the basis of existing literature. As emphasis framing tends to have limited effects in survey experiments50, we ensured to select two very distinctive frames.

In the experimental part of the survey, participants were first asked to rank all 17 USD policy issues identified in the literature. After the ranking, the participants entered the choice experiment, in which they were asked to assess two USD policy agendas that each consisted of six USD policy issues randomly selected from the 17 USD policy issues (Extended Data Fig. 1). Respondents were required to perform six iterations of comparisons, yielding a total sample of 69,600 observations (2 × 6 × 5,800). Brief two- to three-sentence descriptions of the USD policy issues were provided before the experiment (Supplementary Table 2) and we created an information button so that the survey participants could remind themselves of the meaning of each USD policy issue. The survey questionnaire is available in the OSF repository at https://doi.org/10.17605/OSF.IO/Y9HMP (ref. 51).

Before fielding the survey, we conducted several pretest interviews with students, during which they went through the survey while we noted down any questions or elements that they identified as confusing. Following this, we conducted further quantitative pretests with 42 students and researchers in our institute before conducting a soft launch of the survey with 200 residents in Manchester. This comprehensive pretesting process ensured the clarity and effectiveness of the survey.

In our analysis of the choice experiment data, we estimated the average marginal effects (AMEs) using a two-sided test to evaluate the impact of each USD policy issue on residents’ preferences52. The AMEs allowed us to discern which USD policy issues were of greater relative importance to residents. This statistical method calculates the average change in the probability of the outcome variable for each unit change in a predictor variable, while holding all other variables constant. Our approach using the AMEs focused on the overall average effect across all attribute levels and subgroups. This methodological choice was guided by our research aim of understanding the general preference trends among residents towards various USD policy issues. A comprehensive codebook, including full descriptions of the statistical parameters of each covariate used in our analysis, has been uploaded with the replication data and can be found at https://doi.org/10.17605/OSF.IO/Y9HMP (ref. 51). This provides an additional resource that explains our methodology and facilitates replication and further analyses.

Thematic coding

We qualitatively coded 166 USD policy plans in the eight cities (see Supplementary Table 4 for a list of all the 166 USD policy plans). Initially, we collected and reviewed 219 plans and ultimately included and coded 166 plans, having excluded plans without explicit reference to USD, SD or sustainability; plans not authored by the city government (for example, by regional or metropolitan authorities); or reports that were rather evaluative or diagnostic. The goal of the analysis was to determine the extent to which cities are prioritizing different USD policy issues by detecting how many times the particular USD policy issues are mentioned in the plans (that is, frequencies of USD policy issues).

We searched the eight city and departmental websites along with basic text searches on Google. Keywords for collecting the USD policy plans were the USD policy issue names, the associated search terms (Extended Data Table 2) and the city names. This search took place as a general search through Google but also then on the website of the city. Upon finding a document, it would be downloaded and coded if it contained any USD policy issue under study. We searched both in English and in the official languages of the cities. We did not limit our search to the main USD plan of the city (that which the city put forward as their main or comprehensive USD plan) but searched for any strategy that involved any of the 17 USD policy issues of interest. Although the cities might not always have identified a main or comprehensive plan, this often became clear upon reading the plan and based on its location on the website and how the city presented it. If the plan was available both in English and the official language of the city, we used the English strategy to facilitate the coding process. This was only the case for Helsinki.

We thematically coded the 166 USD policy plans53,54 to identify USD policy issues. We developed a baseline glossary of words associated with each USD policy issue in our research team (Extended Data Table 2). We developed this glossary for each city during the translation process and with an iterative coding approach. During the translation process, we translated each term into the official language of the city to ensure that we developed a glossary that captured the issue in each language. During the coding process, we iteratively further developed the glossary for each city by reading the core USD document and developing a corpus that suited the unique approach of each city to USD policy and planning. For example, Lisbon used the term ‘citizen councils’ in reference to democratic participation while Marseille used the term ‘citizen assemblies’.

We used MAXQDA to conduct the thematic coding and we measured the absolute and relative frequencies of USD policy issues in the USD plans. We uploaded all USD policy plans to MAXQDA and used the ‘text search & autocode’ function’ provided by MAXQDA. No AI tools were involved in the coding process. Upon searching, we would read each instance of the identified USD policy issue and its surrounding context to ensure that it was addressing the issue. If not, it was not coded as an instance of the USD policy issue in the plan. One researcher of the team was mainly responsible for the coding but we jointly discussed the first coding iteration and any potential ambiguous coding decision in the team. We generated absolute and relative frequencies of the USD policy issues. The relative frequencies are relative to the total page count for each city.

Extended Data Table 3 provides an example of coded issues for each USD policy issue from USD policy plans of Manchester and Helsinki. The table provides examples only from these two cities because their USD policy plans are in English. All the coding data and the supporting R code is available in our replication files51.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.