The Impact of the URBANA Initiative: Socio-Economic Conditions, Self-Perceived Health and Quality of the Built Environment—A Time-Trend Analysis of Repeated Cross-Sectional Data

Abstract

This chapter analyses the potential impacts of the URBANA programme between the periods before and after its implementation. The impact was studied using a quasi-experimental design based on secondary data from the complete series of the Spanish Living Conditions Survey. The impact of the programme on household socio-economic conditions, self-perceived health and the quality of the built environment was studied. In addition, the degree of vulnerability of the neighbourhoods in which the programmes were implemented was taken into account, as the impacts could be conditioned by this contextual dimension. The results suggest some improvements in urban context and health, specifically among those neighbourhoods with higher levels of vulnerability before the intervention process. Therefore, the chapter shows ‘neighbourhood effects’ due to the differences between the selected neighbourhoods and the importance of the selection process in policy design.

Introduction

In recent decades, integrated urban regeneration initiatives have emerged as one of the main strategies to improve the economic, physical, social and environmental conditions of the city, and as an effective way to address socio-economic imbalances between urban areas (Roberts, 2000). One of the main objectives is to bring the living conditions of certain neighbourhoods closer to those of the whole city, thereby reducing the accumulation of disadvantages affecting different dimensions of quality of life (Van Gent et al., 2009). The central objective is to change the living conditions of the residents and to improve the structure of opportunities that the context of the neighbourhood can offer to inhabitants. Furthermore, integrated area-based policies can be seen as interventions that promote ‘structurally transformative agency’. These programmes typically not only deploy interventions aimed at changing the structural elements of the most vulnerable areas but also often involve interventions aimed at mobilising the agency capacity of residents that contribute to structural modifications and social changes in urban places. For example, this is the case when integrated area-based policies are designed to activate mechanisms that seek to foster community participation for vulnerable populations to gain control over key aspects of their quality of life, such as actions developed with the aim of building neighbourhood community capacities as a way to improve safety, more equitable access to healthy leisure spaces, foster mutual support networks and increase a sense of identity with the neighbourhood. In short, not only by redistributing the provision of goods and services between areas of the city but also by acting on what people can effectively do and be in the nearby context in which they live, in other words, by acting also on their individual and community capabilities (Abel & Frohlich, 2012). At least theoretically, these would be the inspiring principles of the integrated interventions developed in the URBANA programme that has been implemented in provincial capitals and municipalities throughout Spain with more than 50,000 inhabitants during the 2007–2013 period (De Gregorio Hurtado, 2017; Ministerio de Economía y Hacienda, 2007).

The previous chapters have dealt with the evaluation of the impacts of urban regeneration policies using aggregated information at the level of Homogeneous Urban Areas. In this chapter we intend to present another evaluative approach using household and individual data from cross-sectional surveys, more specifically, through the information provided by the Living Conditions Survey (ECV), the primary source for the European Union Statistics on income and living conditions (EU-SILC) in Spain. The ECV is a statistical operation that has been carried out annually in Spain since 2004, following harmonised criteria for all countries in the European Union. It provides information on the income, quality of life and social exclusion of some 180,000 households and more than 400,000 people between 2004–2017 (Table 9.1). After asking Spain’s National Institute of Statistics (INE) for the census section code of the households and individuals interviewed, a quasi-experimental study was designed to explore trends followed in different socio-economic indicators by households and individuals residing in urban areas targeted by the URBANA programme and in a sample of paired urban areas acting as controls.¹ In order to ensure sufficient sample sizes, the data were pooled biannually to conduct an initial evaluation of trends in various socio-economic indicators before, during and after the intervention periods (Fig. 9.1).

Table 9.1 Sample sizes of the living conditions survey waves

Fig. 9.1

The logic of the quasi-experimental design for trend assessment using cross-sectional data

Measuring Programme Objectives: Household Socio-Economic Situation, Personal Health and House-Environmental Problems

Bearing in mind the major areas of objectives pursued by the URBANA initiative (improving the physical, social, economic, environmental and governance aspects of the areas involved), three indicators were calculated to assess their potential impact based on the information contained in household and personal questionnaires from the various waves of the ECV survey. Firstly, an indicator related to the household socio-economic situation was derived based on the information provided by six items indicating whether the household: could afford to pay for a holiday away from home, at least one week per year; could afford meat, chicken or fish meal (or equivalent for vegetarians) at least every two days; could cope with unforeseen expenses; whether it could make ends meet right to the end of the month; whether total household expenditures were a heavy burden on the household; and, finally, whether the household was at risk of poverty. Negative responses to socio-economic status were coded with a value of −1 and positive responses with a value of 1. Finally, the indicator was computed as the sum of all these items and takes values ranging from −6 to 6, with higher values indicating a better household socio-economic situation.

Secondly, another indicator of perception of the environment and housing was calculated based on five items, namely: whether the home suffered from noise problems caused by neighbours or from outside (traffic, factory business); pollution, dirt, or other environmental problems caused by industry or traffic; crime, violence, or vandalism problems in the area; whether the house had problems in terms of leaks, damp on walls, floors, ceilings or foundations, or rot in the floors, window frames or doors, and whether the temperature of the house was adequate during the winter. Similarly, negative responses were coded with a value of −1 and positive responses with a value of 1, and therefore, the overall indicator takes values ranging from −5 to 5, with positive values indicating better perceptions of the environment and housing situation.

Thirdly, the self-perceived health was selected as an indicator of the quality of life. Respondents were asked about their overall state of health, choosing an answer among five options: very good (1), good, fair, bad or very bad (5). The coding was reversed, and then higher values indicate a better perception of the informant’s overall state of health.²

Multiple linear regression models were used to explore change trends in the three selected indicators, considering as independent variables the ‘intervention’, which identifies cases situated in an experimental or control area; the variable ‘period’, which identifies the biannual period in which the survey was conducted, as well as a series of control variables for each indicator detailed in the results section. Then, an interaction effect between the intervention and the period in which the survey took place was introduced in regression models.³ This interaction will indicate if there are significant differences in the evolution of indicators between the control and intervention areas. The models were calculated using the Generalised Linear Model procedure. For this analysis, the experimental group encompassed 23 urban areas (N_households = 2531 and N_individuals = 5558) and the control group 105 matched urban areas (N_households = 9645 and N_individuals = 21,334).

Is There an ‘Average’ Impact of the URBANA Projects?

Regarding changes in the socio-economic situation of households, the results show a slightly negative trend in the intervention neighbourhoods (Table 9.2: β = −0.097, p-value ≤ 0.05). However, the socio-economic situation of households appears to remain constant throughout the period considered in the control areas, as shown in Fig. 9.2. The first graph shows that during the period before the implementation of the URBANA programme, there appeared to be no significant differences in the socio-economic conditions of households between the intervention and control areas, but the most negative evolution observed in the intervention areas from 2008 onwards shows a degree of amplification of these differences. Therefore, these results do not support the socio-economic situation of all households living in URBANA areas improved after the interventions. Regarding the perceptions about the quality of the urban environment and housing, a positive period effect is observed; in other words, this indicator seems to have improved in both types of areas over the period considered (β = 0.140; p-value ≤ 0.001, result not shown in tables). However, this positive evolution appears to be less intense in intervention areas than in control ones (Table 9.2: β = −0.073; p-value ≤ 0.01).

Table 9.2 Trends in indicators for intervention and control areas

Fig. 9.2

Predicted trends in intervened and control areas

As shown in Fig. 9.2, households’ perceptions have evolved positively in both types of areas. However, the improvement trend is slightly higher in the control areas than in the intervention ones. It is, therefore, also not possible to conclude that the URBANA programme has positively impacted the perceived quality of the urban environment and housing. Finally, the results for self-perceived health also show a positive period effect, perceived general state of health appears to have improved for the whole sample in the study period considered. However, there are no significant differences between intervention and control areas (Table 9.2: β = 0.004; p-value > 0.05). The findings of our analyses  do not support an impact on the self-perceived health of the whole population living in the intervention areas (Fig. 9.2).

Do Impacts Depend on the Properties of the Context in Which the Intervention Takes Place?

The question posed in the previous section assumed that all initiatives developed within the framework of URBANA projects should have positive effects on social, economic and environmental aspects for the overall population residing in the areas concerned. However, the lack of evidence regarding ‘average’ impacts does not imply that the programme might not have generated differential impacts for population sub-groups. For example, impacts could be conditioned by the characteristics of the urban context in which the intervention takes place, or they may be restricted to specific groups of residents whose characteristics might make them more prone to the improvements the interventions seek to promote (see Chapter 6). To illustrate this argument, we examine whether the programme’s effects are conditioned by the degree of vulnerability of urban areas. To this end, the same model was applied to the sample of households and persons residing in the most vulnerable areas, defined by their unemployment rate.⁴

Firstly, the results show that, even though the analysis is limited to the most vulnerable areas, there are no differences between the evolution of the socio-economic conditions of households in the intervention and control areas (Table 9.3: β = −0.091; p-value > 0.05). Figure 9.3 shows that the estimated trend in control urban areas remains constant, and although a slight negative trend appears to be seen in the experimental areas, these are not statistically significant. Secondly, the analyses show that the perceived quality of the urban environment and housing has improved more intensely among households in the intervention areas compared with control areas (Table 9.3: β = 0.124; p-value ≤ 0.05). The second graph of Fig. 9.3 shows that this perception is higher among households in the control areas, but the trend shows that this perception remains constant in the control areas, whereas it improves moderately in the case of households in the intervention areas. Something similar occurs regarding self-perceived health. Although the residents of the control areas have a better self-perception of their health, which even seems to improve slightly, the trend of improvement of this indicator among the residents of the vulnerable intervention areas seems to be stronger (β = 0.058, p-value ≤ 0.001). Therefore, these ‘explorations’ regarding the relative improvement of perceived environmental quality and state of health in the intervention areas suggest that the programme potentially has a positive impact in the most vulnerable contexts.

Table 9.3 Trends in indicators for intervened and control areas with a high level of socio-spatial vulnerability

Fig. 9.3

Predicted trends in intervened and control areas with a high level of socio-spatial vulnerability

Some Brief Reflections About the ‘Average’ and ‘Heterogeneous’ Policy Impacts of EU Urban Integrated Strategies

The main objective of this study was to explore the impact of the URBANA programme by applying a research design based on controlled trends analysis based on cross-sectional data for households and individuals. Although these initial results do not seem to provide conclusive evidence regarding the impacts of the programme, they help illustrate some of the possibilities offered by the use of cross-sectional survey series in evaluating such policies, which was one of the main objectives of this chapter.

From a methodological point of view, when compared with aggregate data sources, such as population censuses, cross-sectional data series have the advantage of providing a broader range of measurements on processes and phenomena related to programme objectives. In addition, if government agencies give the researchers georeferenced information, as is the case in this research, this type of data facilitates the assessment of trends using multiple measurements from the relevant indicators. On the other hand, by applying a quasi-experimental design logic, the georeferenced cross-sectional data allow these trends to be studied in a controlled way, thus improving causal inferences about the possible impacts of interventions. These georeferenced data also facilitate the integration of information between various sources at different levels, for example, regarding the characteristics of households, individuals and urban areas. This integration makes it possible to explore the direct and indirect impacts of interventions, in other words, whether they produce significant changes in the living conditions of the entire population residing in the intervention areas and/or whether they have the ability to interact with other individual or contextual properties, generating moderating or mediating effects (Navarro, 2016). In short, the integration of contextual information allows the evaluation process to consider the multilevel nature of the mechanisms whereby impacts of this type of policy might occur, placing the phenomena on their corresponding scale and exploring how they interact with each other.

Obviously, the use of cross-sectional survey series is not without problems. Among the most relevant limitations, it should be noted that this type of information raises difficulties in considering residential changes that have occurred in the intervention areas within the analysis, which would have consequences in terms of attributing the results to the programme evaluated (as shown in the previous chapter).

From an evaluative point of view, the results presented indicate no evidence of an ‘average’ impact on the programme with regard to the socio-economic conditions of households, perceived quality of the housing and neighbourhood environment and their health. Even the households’ socio-economic situation shows intervention areas have declined since 2008 despite the intervention process. In this regard, it cannot be ignored that certain external factors, such as the economic crisis experienced in Spain, present a great challenge for evaluating the socio-economic impacts of this type of programme. Along these lines, studies suggest that the recent economic crisis is an important disruption factor in assessing the impact of urban regeneration programmes implemented during the crisis period and also that the Great Recession did not affect all territories in the same way. However, this circumstance also offers a unique opportunity to test whether urban initiatives have the capacity to moderate the negative effects of the crisis, as other studies suggest (Mehdipanah et al., 2014; Zapata-Moya & Navarro Yáñez, 2021).

Assuming that the effects of the economic crisis have been more intense in the most vulnerable areas, leading to a relative loss of resources for residents of these areas, the decision was made to conduct the evaluation based only on information from the households and persons surveyed from census sections with high unemployment rates. Results show interventions appear to produce a relative improvement in the quality of the urban environment and in health among households and people living in these areas, which might be interpreted in line with the hypothesis that such interventions could moderate the consequences of the crisis.

Finally, our findings that programme impacts may be conditioned by the degree of vulnerability of the context also seems consistent with the resource substitution hypothesis, which suggests that when multiple resources are available, quality of life depends to a lesser extent on the presence or absence of a specific resource since the resources are interchangeable with each other (Ross & Mirowsky, 2006). Consequently, the impact of interventions could be greater for those with fewer alternative resources before the intervention, such as households and individuals residing in the most vulnerable intervention areas.

However, it was not the objective of this analysis to obtain a conclusive answer to the questions asked. The idea was to illustrate the possibilities that arise from using transversal information in evaluating such policies and provide some policy evidence in the case of the URBANA Initiative in Spain. This exercise provides previously non-existing policy evidence, and potentialities and limitations of the research design using existing data have been indicated. Both issues show evaluative exercises about urban initiatives promoted by the EU should consider the importance of certain context properties or individual characteristics in greater detail. This issue raises the challenge of multilevel controlled evaluation if we aim to improve causal inference and unpick the mechanisms that produce (or do not) the desired impacts of this type of urban policy.