Planning instruments enhance the acceptance of urban densification.

Dense and compact cities yield several benefits for both the population and the environment, including the containment of urban sprawl, reduced carbon emissions, and increased housing supply. Densification of the built environment is thus a key contemporary urban planning paradigm worldwide. However, local residents often oppose urban densification, motivating a need to understand their underlying concerns. In order to do so, we examined different factors driving public acceptance of housing densification projects through a combination of a conjoint survey experiment and different proximity frames among 12,402 participants across Berlin, Chicago, London, Los Angeles, New York, and Paris. Respondents compared housing densification projects with varying attributes, including their geographic proximity, project-related factors, and accompanying planning instruments. The results indicate that the acceptance of such projects decreases with project proximity and that project-related factors, such as the type of investor, usage, and climate goals, impact densification project acceptance. More specifically, we see a negative effect on acceptance levels for projects with for-profit investors and a positive effect when the suggested developments are mixed use or climate neutral. In addition, planning instruments, such as rent control, inclusionary zoning, and participatory planning, appear to positively influence acceptance. Interestingly, a cross-continental comparison shows overall higher acceptance levels of densification by US respondents. These multifaceted results allow us to better understand what drives people's acceptance of housing projects and how projects and planning processes can be designed to increase democratic acceptance of urban densification.

A key challenge for growing cities worldwide is to provide sufficient adequate housing while trying to contain the expansion of their settlement area to prevent urban sprawl. To address this, planning academics and practitioners agree that higher densities in cities are desirable because they can induce positive sustainability effects that protect unbuilt land and biodiversity and help reduce CO2 emissions from mobility and energy through compact development (1–3). Densification projects also increase supply of housing, which can help make housing more affordable (4–7). Additionally, densification may enhance urban diversity, as restricting density through land use regulations contributes to socioeconomically and racially segregated cities (8, 9). Consequently, policymakers around the world have introduced measures to optimize land use by focusing on housing densification projects (10–12).

While there is scant opposition to the abstract planning paradigm of densification, opposition to housing developments on the ground is widespread, especially from residents who are in close proximity to such projects (13–18). This discrepancy between general and local support for densification (13, 16) can be explained by the long-term and large-scale unfolding of positive sustainability effects stemming from densification projects versus the immediate changes that occur at the neighborhood level. Yet, local public acceptance is normatively desirable as well as critical for successfully attaining urban densification and unlocking its sustainable effects, as urban residents are often able to resist, delay, or block densification projects (19–21). Previous studies have thus focused on better understanding how individual-level factors (e.g., age, political ideology, or homeownership) (13–16, 20) and project-related characteristics of the new housing development (e.g., usage types or the type of project investors) (13, 22, 23) can explain public acceptance and opposition to such projects. Moreover, some research investigates how accompanying urban planning instruments—such as providing a fixed amount of affordable housing units for low-income residents—might trigger or mitigate perceived negative effects of housing densification projects (14, 16, 17). However, a systematic analysis of the influence of planning instruments on public acceptance of densification has, to the best of our knowledge, not yet been carried out. In this paper, we thus expand existing research by integrating project-related factors and three common planning instruments (inclusionary zoning, rent control, and participatory planning) in a survey experiment, which examines people’s acceptance of housing densification in their own cities. We also analyze how acceptance of densification varies depending on the respondent’s geographic proximities to a new housing project.

Planning instruments, such as rent control (i.e., regulation of rent increases), inclusionary zoning (i.e., ordinances that require a minimum share of newly constructed affordable housing units), and participatory planning (i.e., involving the public in planning processes), are increasingly implemented in North American and European cities (4, 24). Berlin, for example, introduced a rent cap policy in January 2020 (25), although it was then ruled unlawful by a federal court in 2021 (26). New York City has implemented an inclusionary zoning policy, prescribing developers to include affordable housing units in new and more extensive luxury housing developments (27). In London, the mayor has introduced estate redevelopment ballots in 2018, which aim at insuring that regeneration of London’s housing estates happens only if residents have supported the proposals via ballot (28). In all of these cases, planning instruments are used to 1) mitigate anticipated negative socioeconomic effects of urban densification, 2) find locally anchored solutions, and 3) enhance local public acceptance for housing developments. In these ways, these planning instruments appear to be important measures to tackle the urban housing crises.

To test the effect of different project-related factors and planning instruments on people’s acceptance of housing developments in their cities, we conducted a large-scale survey experiment illustrated in Fig. 1, which combines two experiments: three different proximity frames (randomly presenting three different proximities of the densification project to the respondent’s places of residence) and a conjoint experiment (presenting randomly varying project-related factors and planning instruments). The survey includes responses from 12,402 residents of six cities of global importance: Berlin (n = 1,820), Chicago (n = 2,120), London (n = 2,120), Los Angeles (n = 2,119), New York (n = 2,120), and Paris (n = 2,103). This large-scale survey design allows us to generate a comparative understanding of acceptance and opposition from panel respondents that are representative for the city populations regarding age, gender, and income in six metropolises in Europe and the United States. This advances the urban densification debate, as to date, the academic focus on acceptance of densification has mainly been directed toward the United States.

Fig. 1.

Study design and illustration of experimental survey design with an example for the conjoint tasks.

Theoretical Expectations

Although practitioners, scholars, and the wider public tend to agree about the importance of urban densification—here referring to an increase in housing units through developments in already built-up urban areas, which also results in an increase of population density (29)—concrete projects frequently face opposition by local residents (13, 16). There are different reasons for this local opposition.

Much of the previous research, mostly focusing on US cities and suburbs, has conducted survey experiments to examine individual-level factors that shape attitudes toward housing development projects (13–16, 20). Recent studies have, for example, shown that objectors are more likely to be older, White, wealthier, and homeowners that are politically active and long-time residents (14, 19). Furthermore, opposition tends to be stronger among ideological conservatives than among liberals (14, 15). At the same time, people who live in single-family neighborhoods (13, 14, 30) and people who live in suburbs (8, 14) are more likely to oppose dense developments than people living in more dense and more central locations. The reasons for opposing density are multifaceted. People may dislike denser developments because they are believed to decrease the overall desirability of a neighborhood, increase crime and traffic, lower school quality, and decrease property values (14, 17, 19, 31). Opposition to densification is substantially stronger among homeowners than renters. This difference also persists after controlling for ideology, suggesting that their objection to new housing developments is primarily driven by their fear of property price declines (15, 16). However, some renters also object to densification. Since the majority of cities lack free central building plots, densification mainly occurs through the redevelopment and transformation of older and comparably cheaper housing stock that comes with the danger of displacing lower-income renters (24). Similarly, new housing developments can be interpreted as a sign of gentrification, and renters may be afraid to be priced out (16, 17). In summary, these studies have focused on unpacking individual-level factors driving opposition and reveal that self-interest seems to be an important factor informing individual attitudes toward urban densification. In this paper, we shift the focus more to the project level. We are thus interested in assessing the effect of 1) the projects’ proximity, 2) specific project-related factors, and 3) accompanying planning instruments on densification project acceptance.

Regarding the physical proximity of housing development projects, we expect to find a so-called “not in my backyard” (NIMBY) effect. This concept refers to the phenomenon where people who generally accept new developments reject them when they are built near them (18, 32, 33). Whereas such individuals accept densification in theory, they oppose it at the local level, as they wish to preserve the existing ways of life and the character of their area (31). In line with existing literature, we therefore expect to find that respondents’ acceptance of housing densification decreases with closer proximity to their home (13–17).

Beyond the influence of a housing densification project’s geographic location, scholars point out that acceptance to densification can be driven by what we here subsume as project-related factors (13, 23). To test how specific features of a housing project may influence people’s acceptance of the given project, we have included four project-related factors—also called attributes—in our conjoint experiment. These are the degree of densification, the usage of the building, the type of project investor, and specific climate goals. We consciously decided to forgo specific features of the actual building design (i.e., facade articulation or roofing), as their effect on acceptance of housing projects appears to be rather limited (23).

Regarding the degree of densification, existing literature suggests that increased density generally decreases acceptance based on status quo preferences and perceived negative effects of densification, such as loss of green spaces, changes in traffic volume, gentrification, or changing demographics of the new arrivals (14, 16, 17, 31). Previous studies find that the proposed level of density for housing has an impact on people’s assessment of the given project, with less dense developments favored over more dense developments (13, 17). In line with this thinking, we expect to find decreasing levels of acceptance with increasing density.

As a second project-related factor, we introduce the usage of the new development to our experiment. Planners increasingly advocate that the usage of new housing densification projects should go beyond residential use and move toward more mixed-use buildings and areas (34). Existing research suggests that residents tend to appreciate spaces for shops, businesses, and services and that mixed-use areas may help to preserve the interests of existing communities by diffusing potentially increased competition for access to existing services (13, 17, 35). Building on these findings, we expect mixed-use developments to receive higher levels of acceptance than projects that are built for residential purposes only. However, we also expect that while residents appreciate mixed-use developments, some of the residents might not necessarily want them in their immediate proximity, as new amenities could be a source of potential noise and congestion. Accordingly, we expect to find a general preference for mixed-use that wanes and becomes neutral if the project is located in close proximity.

We included the type of project investor as a third project-related factor. Project investors can be separated based on their for-profit or nonprofit (including governments) orientation. For-profit investors build properties with the intent of selling them or holding them as long-term investments. Previous studies have shown that residents tend to view for-profit housing densification projects more critically than nonprofit projects (22, 36). Accordingly, we expect lower acceptance of for-profit housing projects than projects developed by nonprofit investors.

Under the general promotion of climate-friendly cities, developments in urban areas also increasingly set climate goals (i.e., that the development is climate neutral) (3). To study the impact of this trend, we have thus included climate goals as a fourth project-related factor. We are not aware of another study that has systematically analyzed the effect of climate-friendly developments on people’s acceptance of urban housing projects. However, as urban residents tend to score high on proenvironmental attitudes and show high commitment to environmental issues (37), we expect to find higher acceptance levels for densification projects when they are climate-neutral than if no climate goals are specified.

Our third set of factors includes planning instruments, which refer to measures that are in the hands of local planners or local governments and can serve as ancillary tools that influence how and under what circumstances housing projects are implemented. Such planning tools can be substantive, as in the case of rent control or inclusionary zoning, or procedural, as in the case of participatory planning. In this paper, we included three different instruments: rent control, inclusionary zoning, and participatory planning.

Rent control regulates rent increases by providing rules that specify how much landlords may charge for rent (38). Linked to densification projects, rent control measures could lower risks of displacement of existing residents by serving as a mechanism to ensure that renters living in the surrounding areas would not be priced out by new developments. However, some scholars critical of rent control argue that it may lower investments in housing and fail to deliver quality affordable housing (39). Furthermore, whereas rent control may have benefits for renters in the short run, the loss of rental housing supply due to shifts in ownership can drive up market rents in the long run (40). Despite these potential negative effects, we expect renters to favor rent control as a form of insurance against the risk of rent increase (38) and expect that liberal homeowners and renters alike may also embrace it because it is a tangible regulation of the housing market (16). In our experiment, we therefore expect to find a positive effect on proposal acceptance levels.

Inclusionary zoning requires developers of a new housing project to provide a certain percentage of its housing units at an affordable price for low- and moderate-income households (41). Previous studies on acceptance of densification have examined varying percentages of affordable housing units within the proposed housing projects. Results from these studies indicate differing levels of support depending on the suggested amount of units being reserved for low-income residents (14, 16, 17). Opposition to affordable housing by existing residents may be triggered by the stigma associated with lower-income households moving into their neighborhoods and potentially devaluating the area (42). Taking this into account, we expect to find differing results depending on the proximity frame, with higher levels of general acceptance and lower levels of acceptance if the project is suggested to be in close geographic proximity to the respondent’s places of residence.

As a third planning instrument, we have included participatory planning in our experiment. The involvement of the public in planning processes has been suggested as a means to increase the legitimacy of planning decisions and to find locally anchored solutions (43). Because residents may feel strongly about preserving and protecting their neighborhoods, involving them in the process may give them a greater sense of agency and thus have a positive effect on their assessment of housing developments. Consequently, we expect to find that densification projects with more inclusionary forms of participation will have higher levels of acceptance than housing developments without local resident involvement.

In summary, we expect to find that acceptance for housing densification projects will decrease with greater proximity to the respondents’ places of residence and increase when there are more project-related benefits and ancillary planning instruments in place. We also expect the projects’ proximity to affect the directional effect of the conjoint attributes (Table 1). Consequently, we expect the direction of the effect to be negative (decreasing acceptance levels) when the project is located within the respondents’ own neighborhoods or districts (or borough, side, Bezirk, or arrondissement; hereafter collectively referred to as districts), with higher levels of densification, and with a for-profit project investor. By contrast, we expect to find positive effect directions (increasing acceptance levels) for measures of rent control and public participation as proximity increases. Furthermore, we argue that there is no proximity effect on climate goals, thus showing no effect on acceptance levels depending on project location. Accordingly, we expect climate neutrality to be appreciated, but not a priority, when a project occurs in one’s own neighborhood. Regarding usage, we expect that residents appreciate mixed use, but not all residents want these amenities in their immediate proximity because of potential noise and congestion effects, which can thus fade out a positive or negative effect. Finally, we expect that inclusionary zoning negatively affects acceptance levels when it occurs in close proximity, as people might fear the devaluation of their neighborhood due to an influx of low-income households.

Table 1.

Overview of the conjoint attributes with the expected direction of acceptance on the main and framing effects

	Conjoint attribute	Main effect	Framing effect
Project-related factors	Densification	↓	↓
	Usage (mixed)	↑	=
	Project investor (for-profit)	↓	↓
	Climate goals	↑	=
Planning instruments	Rent control	↑	↑
	Inclusionary zoning	↑	↓
	Participatory planning	↑	↑

Symbols indicate the expected directional effect of the respective attribute on overall acceptance; ↓, less acceptance; ↑, more likely acceptance; =, no effect.

Results

Conjoint Results and Proximity Effects.

In the conjoint experiment, respondents were provided with five sets of two proposals for a housing densification project with randomly alternating attribute values (Table 2). For each set of proposals, respondents were asked to rate each proposal separately. They also had to choose which proposal they preferred in a binary forced-choice task. These results are presented in the .

Table 2.

Conjoint attributes with descriptions and possible values

Attribute	Attribute description	Possible attribute values
Densification	The aim of the development project is to increase the population density...	• By 20% • By 50% • By 100% (double)
Usage	Buildings can be used for multiple purposes. The use of the development project includes…	• Only apartments • Mixed use with apartments and small service businesses • Mixed use with apartments and small craft businesses • Mixed use with apartments and small businesses (entertainment, restaurants, cafes, and bars)
Project investor	The main investor for the development project is...	• Government investor • Nonprofit investor • National for-profit investor • International for-profit investor
Climate goals	The development project contains the following specifications with regard to climate-neutral design of the project:	• No specifications • Climate-neutral project (net zero emissions)
Rent control for the whole city	A rent control is a legal requirement to set maximum possible rents or a prohibition or restriction on rent increases for residential leases. Such a rent control applies regardless of whether a change of tenant occurs or not.	• Rent cap (0% for a certain period of time) • Maximum 5% increase per year • Maximum 10% increase per year • No rent control
Inclusionary zoning	Inclusionary zoning stipulates that a certain share of the project must be affordable for people with low income (poorest quarter of the population).	• No requirements • At least 20% • At least 10%
Public participation	Residents can participate in various ways in the planning of the development project. This can range from information events and participation in the design of the project to a final, binding vote on the development project.	• No involvement of the residents • Public information events on project goals and progress • Involvement of the residents in the planning and implementation of the densification project • Opportunity to have a vote in project decisions

Fig. 2 shows the marginal means of all attribute values of a housing densification project based on the individual acceptance rating task in the conjoint experiment (for tests of statistical differences between the different estimates, see ). Marginal mean for the individual rating can be interpreted as the decimal expression of mean acceptance levels of the respective attribute. For example, the marginal mean for a 100% increase in population density in respondents’ own neighborhoods is 0.514, thus indicating a mean acceptance of 51.4%. This is 9.7 percentage points lower than a 20% increase in population density in other districts (marginal mean of 0.611, 61.1%). The individual rating task allows us to look at potential differences between the three proximity groups. Here, the two housing densification projects have been rated individually and are thus not dependent on one another.

Fig. 2.

(A and B) Estimated marginal means for acceptance rating task (A) and predicted acceptance levels for all possible projects (B) by proximity frame. Error bars show 95% confidence intervals.

Overall, densification project scenarios with only 20% population densification and rent control in the form of a rent cap or of a maximum 5% increase per year received the highest acceptance levels. By contrast, the absence of rent control and an increase of the population density by 100% had a negative effect on respondents’ acceptance of the project. Here, we can also observe an effect of the different proximity frames. We can see that respondents’ levels of opposition toward increased population densities of 100% were stronger if the project was in their own neighborhood or district than when it was located elsewhere in the city (). In addition, we can see a significant negative effect in Fig. 2 on the rating task if the proposed densification project did not include any participatory measures (), if the project was carried out by an international for-profit investor (), and if there were no inclusionary zoning requirements () and no climate goals in place (). Including climate goals had a significant positive effect, while mixed-use tended to be more accepted than apartments alone. Regarding investors, respondents favored nonprofit investors, followed by the government and national for-profit investors. International for-profit investors received the least support.

Regarding the planning instruments, we observed a significant positive effect across all three proximity frames if both rent control and inclusionary zoning were present in the housing densification project proposal. However, the effect of rent control—or rather its absence—tended to be larger than the effect of inclusionary zoning in the respondents’ neighborhoods. Projects that included participatory planning instruments tended to trigger higher levels of acceptance than projects with no resident involvement; however, we did not find significant differences between the different forms of involvement if participatory planning measures were already included ().

The distribution of predicted acceptance ratings for densification projects by attribute and framing characteristics are displayed in Fig. 2. To check whether the proximity frames influenced the average level of acceptance, we calculated a Bonferroni-adjusted two-sided t test for the differences of “other district” and “own district” or “own neighborhood”. We found a significant (P < 0.01) effect for both group comparisons (). The mean acceptance level among the other-district group was 0.582. The own-district group had an average rating of 0.560, and the own-neighborhood group had an average rating of 0.555. The marginal mean of the individual acceptance rating also reflects this finding (Fig. 2). A proposal with the same combination of features faced increasing acceptance the further the proximity from the respondent (for example, a climate-neutral project received a marginal mean of 0.601 when located in another district, a marginal mean of 0.573 in the respondents’ own districts, and a marginal mean of 0.568 in their own neighborhoods).

Overall, these findings support our theoretical expectations of the main effect direction and largely confirm our expected proximity effect directions outlined in Table 1. The largest deviation from our theoretical expectations can be found regarding inclusionary zoning. We did not find any experimental evidence of systematic negative perception of inclusionary zoning across any of the three proximity frames. Our data thus do not show a decrease in acceptance with closer proximities. This contradicts our theoretical expectation. Also, the effects of usage and climate neutrality deviated from our expectations. Regarding usage, residents seemed to generally appreciate mixed-used developments in close proximity. People’s acceptance of housing projects thus increases when introducing craft businesses and entertainment-based mixed use (e.g., restaurants, cafes, and bars) in their neighborhoods or districts compared to when they are located in other districts. Thus, residents appear to prefer amenities nearby. In addition, while climate neutrality generally increases acceptance of densification projects, climate-neutral housing projects have a stronger effect on acceptance when they are built in other districts. This could point to a type of externalization of climate mitigation responsibility of residents in urban planning.

City-Specific Differences.

Comparing the results across the respondents in the six cities revealed interesting differences. Although the general direction of attribute effects did not change much across respondents of the different cities, there were relevant differences regarding effect size. Fig. 3 shows the marginal means of all attribute values based on the individual acceptance rating tasks in the conjoint experiment for each of the cities. We found cross-continental differences, with higher acceptance levels of housing densification developments in the three US cities than in the European cities. Similarly, we found considerable differences in predicted and average acceptance levels across the six cities (Fig. 3), confirming these cross-continental differences. On average, respondents in New York reported the highest levels of acceptance for these types of projects (marginal mean = 0.624), closely followed by respondents in Los Angeles (marginal mean = 0.617), and Chicago (marginal mean = 0.600). Among the European cities, respondents in Paris had the highest average acceptance of the housing densification projects (marginal mean = 0.542), followed by respondents in London (marginal mean = 0.523); respondents in Berlin reported the lowest acceptance levels (marginal mean = 0.472). Due to the nature of the nonrandom sample, it needs to be noted that these results do not allow for drawing inferences on the general population across these cities but only on the respondents from the panel sample, although it is representative in terms of age, gender, and income ().

Fig. 3.

(A and B) Estimated marginal means for the acceptance rating task (A) and predicted acceptance levels for all possible projects (B) by city. Error bars show 95% confidence intervals.

Acknowledging general differences in average acceptance across the six cities, we nevertheless found considerable differences in attribute effect size in the studied metropolises. For example, we found that the degree of densification had the same effect direction in all six cities; yet, there was a higher effect tendency in the European cities and Chicago. This means that the difference between a 20% and a 100% population increase creates a more substantial effect on acceptance levels of respondents from those cities. Similarly, we found differences in the acceptance level of housing densification projects depending on the type of investor. While there were no significant effects on acceptance levels for respondents in the US cities whether the project was run for-profit or nonprofit, there was a significant negative effect on acceptance levels for respondents from London and Berlin when the project was run by for-profit investors, especially international ones. Furthermore, we found different effect sizes for rent control measures across respondents from the six cities. For example, projects with rent caps significantly affected the forced-decision task within European cities, especially in Berlin. This means that here, densification projects that are accompanied by a rent cap are more likely to be favored. Similarly, we found differences in effect sizes linked to inclusionary zoning. From our data, we can derive that incorporating inclusionary zoning measures into densification project proposals results in higher levels of acceptance among respondents in London and Berlin. Meanwhile, the effects of inclusionary zoning were rather limited for respondents in Paris and in the three US cities. These findings point to interesting variations between the different contexts, which might also be informed by the given local legislative and policy contexts.

Due to the inclusion of differing proximity frames in our study, we can also conduct a spatial analysis to examine where housing densification projects tend to get accepted the most in the six cities from the survey respondents. To do so, we only focused on the ratings for the projects located in the respondents’ own districts and neighborhoods. Fig. 4 maps out the predicted average acceptance of these projects and thus displays the geographical distribution of individual acceptance levels for densification in respondents’ own districts across the six cities (). It shows that respondent’s acceptance for densification in their own districts appears to be higher in more central locations than in more peripheral districts. While this analysis reveals interesting spatial patterns, the variance of predicted acceptance, however, remains comparably low within each of the six cities.

Fig. 4.

Predicted average densification acceptance of a project located in respondents’ own district for the six cities studied. , displays specific values for each district.

Discussion

This paper identifies how 1) the projects’ proximity, 2) specific project-related factors, and 3) accompanying planning instruments affect people’s acceptance of housing densification projects in Berlin, Chicago, London, Los Angeles, New York, and Paris. To scrutinize the interaction between these three explanatory factors across cities, we have conducted a large-scale survey containing a combination of proximity frames with a conjoint experiment. Overall, we found that in all six cities, acceptance of housing densification projects decreases with respondents’ proximity to them. We observed a small but significant effect of proximity on acceptance; higher densification nearby (own neighborhood/district) received significantly lower levels of acceptance than projects planned elsewhere in the city. This is in line with previous findings (13–17).

Yet, what this study shows is that opposition to densification is more than simply rejecting higher density within one’s neighborhood and is not solely driven by self-interest. Instead, local acceptance is also shaped by the specific characteristics and unfolding process of the project. We find that project-related factors and planning instruments have a positive effect on acceptance of densification. Urban residents tend to show higher levels of acceptance if a project is mixed use, climate neutral, and developed by nonprofit investors. Moreover, our results show that planning instruments have a positive effect on acceptance across all cities. Of the examined planning instruments, rent control has a slightly larger effect than inclusionary zoning, yet both are significantly positive. Contrary to our expectations, inclusionary zoning has a positive effect across all proximity frames, including when it is proposed in people’s own neighborhoods. This finding is encouraging, as residents seem to not only favor rent control—a policy that tends to be motivated more by self-interest—but also inclusionary zoning, which is especially designed to support low-income households. Therefore, respondents seem to support measures that anchor housing affordability in their cities regardless of whether they would personally benefit from them or not. In addition, there is an apparent positive effect of participatory planning. Our results indicate that people want to be involved in housing projects. Yet, we did not find significant differences across different types of public involvement. In summary, our study holds important findings that show the potential of planning instruments to mitigate public opposition toward housing developments.

Comparing the results of the panel respondents from the six cities, we found intriguing differences, especially across continents. General acceptance of housing densification projects is higher across US respondents and tends to find more opposition from European respondents. Such differences are also present across some specific attributes. For instance, while the overall negative effect of for-profit investors on acceptance levels is in line with existing research (22, 36), we do not see substantial differences between different types of investors from US respondents, but we can detect a strong negative effect of for-profit investors among respondents from London and Berlin. An explanation for this antideveloper sentiment in European cities might be that the growth rationale is less prevalent in the comparably older European cities than in US cities, and developers tend to be more dominant and entangled in North American urban governance regimes than in European ones (44, 45). Regarding planning instruments, our results indicate that measures addressing affordable housing, such as rent control and inclusionary zoning, seem of greater importance for respondents from London and Berlin than in the remaining four cities. There might be several reasons for this. These planning instruments have been implemented or at least discussed in London and Berlin, and the public discussions about their potential to mitigate the housing affordability crises are salient in both cities (46). In the case of Berlin, this might also be due to the comparably higher share of renters in the sample than in the other five city samples (). We would therefore argue that such city-specific differences, inter alia, emerge from path-dependent development trajectories, different governance models and housing systems, and context-specific public housing discourses.

Our findings enrich existing literature in several ways. By systematically analyzing acceptance of housing densification projects in six metropolises, we expanded the scope of previous research to an international perspective, which has mainly focused on the US context. This cross-continental study reveals important differences, which may help expand and guide theoretical discussions on densification. Furthermore, existing research has mostly assessed preferences at the individual level, paying attention to who opposes densification and why. Here, we have shifted the focus to assess acceptance of densification based on specific project characteristics, including a project’s proximity and accompanying planning instruments. Furthermore, our approach presents an elaborated survey experimental design (combination of proximity frames with a conjoint experiment) and thus may advance methodological discussions of how to disentangle and better understand explanatory factors that drive people’s acceptance or opposition of a given topic. Lastly, our results show that the introduced project features and planning instruments all have a positive effect on the acceptance of densification. Only the densification attribute itself has a negative effect on acceptance. The results thus suggest that planners and policy makers have options at hand to implement locally accepted housing densification projects, which can enhance the democratic legitimacy of urban housing development projects. Future research could build on and extend our results in a variety of ways. For example, studies could include different, or more nuanced, project-related factors and other planning instruments, such as land value capture tax, urban development contracts, or community benefit agreements, and assess their effect on people’s acceptance of densification. It would also be interesting to contrast our findings, which are based on rather abstract projects, with studies that work with more tangible development proposals and/or results from local ballots on existing projects. While survey experiments are methodologically rigorous and allow for quite realistic assessment scenarios in situations where multiple attributes are present, we are also aware that our results only yield rough estimates of the extent to which residents are likely to act in real-world scenarios. Additionally, our sample intentionally consists of large cities that are centrally affected by the global housing crisis. Nevertheless, we are aware that their residents tend to share a relatively liberal and cosmopolitan political ideology, and therefore a comparison to smaller or more conservative cities could be interesting. Parallel to expanding the scope and variety of cities, conducting a more nuanced analysis of a single city in order to do justice to these context-sensitive data and to study spatial patterns within cities could also prove worthwhile. Although our respondent sample is representative toward the general population of these six cities regarding age, gender, and income (), to draw inferences on the general population of these cities, random sampling approaches rather than a sample from panel respondents is necessary.

Our findings are nevertheless highly relevant for urban planning research and practice and show which features could be included in housing development projects to increase the chances of being accepted by the local community. We thus put forward specific tools that may contribute to enhancing the democratic legitimacy of urban planning endeavors. Accordingly, we show that planners have options at hand to implement locally accepted housing densification projects and suggest that they are well advised to supplement densification projects with ancillary planning instruments to address housing and land use concerns meaningfully and sustainably in dense and growing cities.

Materials and Methods

To assess people’s acceptance of housing densification, we conducted a survey experiment consisting of two parts, namely, varying proximity frames of the project’s location and a conjoint experiment on the project characteristics. With the proximity frames, we examined the effect of the geographic proximity to a housing densification project on people’s acceptance of this given project. In a conjoint experiment, we then assessed how far project-related factors and accompanying planning instruments drive people’s acceptance of such projects (Fig. 1). Preregistration of the survey, the full questionnaire, information sheet, full replication data, and R code supporting this paper’s findings are available at OSF (Open Science Framework) preregistration https://doi.org/10.17605/OSF.IO/X7GTQ (47).

Proximity Frames.

The proximity frames first present respondents with a hypothetical housing densification project. The project is consciously presented in a rather abstract form, as a more narrowly specified housing proposal, such as an explicit number of housing units, the form of building, or number of floors, could potentially have impeded the comparability across the six cities due to context-specific housing preferences.

Before the conjoint experiment, respondents were randomly assigned to one of three possible treatment groups, each with a different location of the hypothetical project: 1) within their neighborhood, 2) within their district, or 3) within another randomly assigned district of their city. Accordingly, our sample was randomly split into three groups: respondents with projects in their own neighborhoods (n = 4,276, 34.4%), in their own districts (n = 4,054, 32.7%), and in another district (n = 4,072, 32.8%). These proximity frames—which put a bracket around the conjoint experiment explained below—allow us to assess whether the acceptance of housing densification projects differs depending on respondents’ geographic proximities to the project.

Experiment Design.

Independent of respondents’ assignments to one of the framing treatment groups, they received the same instructions for the conjoint experiment. We decided to include a conjoint experiment, as it enables us to measure the influence of these different factors on respondents’ acceptance of the suggested projects (48).

Entering this part of the experiment, respondents were then asked to compare two randomly assigned housing densification projects at a time, which display varying project-related factors and planning instruments. Table 2 provides an overview of the attributes and their possible attribute values.

In total, respondents were provided with five sets of two different proposals for a housing densification project with randomly alternating attribute values. They then answered three questions in the conjoint task, which serve as our primary outcome variables. Specifically, this means that respondents first indicated whether they preferred proposal 1 or proposal 2 (binary forced choice; “Which of the two proposals would you prefer if implemented?” [proposal 1/proposal 2]). After this choice task, respondents were then also asked to rate each of the projects (proposal 1 and proposal 2) separately, indicating whether they would generally accept it or not (“Would you accept proposal ½?” [yes/no]). Here, the two housing densification projects were rated individually as they are not dependent on another. Unlike the forced-choice task, this individual rating task thus allowed us to look at potential differences between the three framing groups because it allowed us to assess the overall acceptance level.

Data Collection.

The survey was conducted in March and April 2021. We obtained the respondents through a nonrandom sample provided by Qualtrics panels. Prior to fielding the survey, we obtained the approval of ETH Zurich’s institutional ethics committee (EK-2021-N-01). Before starting the survey, respondents were asked for their informed consent to participate in the study, including a link to the “information sheet for survey participants”. To ensure that the questions were adapted to the respective context in each of the different cities, we furthermore pretested the survey qualitatively and quantitatively. This included testing the concepts and attributes in four preliminary qualitative expert interviews. We then also tested the survey instrument in advance with a small number of colleagues at ETH Zurich (n = 27), before then having piloted the survey in London with a sample of 209 respondents.

The online sample of respondents was proportional to the population of the six cities (Berlin, Paris, London, New York, Los Angeles, and Chicago). In our sampling, we used hard quotas for gender and age and soft quotas on income deciles based on census data from cities. While these panels are not probability based, they are nonconvenience samples that were actively managed and updated to target respondents who matched census information. In this way, we aimed to ensure that the samples were not extensively skewed toward certain sociodemographic groups. For our survey, Qualtrics built samples that matched the population of voting age in the six cities and preselected respondents from their panels in accordance with the quotas. The sample consisted of 12,402 respondents from six major cities (2,120 in New York, 2,119 in Los Angeles, 2,120 in Chicago, 2,120 in London, 2,103 in Paris, and 1,820 in Berlin). Each survey respondent completed five comparisons of two proposals, yielding a total of 124,020 units of analysis. The data from the pretest in London (n = 209), which was conducted in February 2021 based on the same sampling approach as the main study, were not included in the final sample. For an overview on the sample and sample statistics, see .

Statistical Analysis.

We examined our theoretical expectations by following the analysis of conjoint designs proposed by Hainmueller et al. (48), which recommends the examination of the marginal effect of a given attribute on the choice probability over the joint distribution of all other attributes. In practice, we regressed a dummy variable indicating whether a respondent preferred a particular set of attributes, using cluster-robust Ses to account for within-respondent clustering. We displayed the results in the form of marginal means (49). Given that our analysis combines responses across different contexts, we also ran mixed-effects regressions with the data from all cities independently (). Accounting for conjoint experiment results being sensitive to baseline categories, we furthermore analyzed the individual acceptance rating of each proposal to examine the expectations regarding heterogeneous treatment effects. To check for differences between the marginal means, we also tested for statistically significant differences in effect estimates shown in .