Scoio-spatial segregation and economic activity:

Overview

Motivation
Concepts
Analysis
Expected results
Take-away message

Inequality of What? between Whom?

Research on inequality should seek to discover the structures that prevent people from leading the kind of life they ‘have reason to value’, and that the problem with inequality is that many people do not have this option (Sen 2006:35).

Key to understand not only where inequality is located but how it is spatially distributed.
Income inequality \(\not=\) income segregation. Hence, has significant implications for social policy.

Spatial segregation exacerbates inequalities of opportunity, well-being and outcomes.

Income segregation can:

limit residential choice.
constrain opportunities to access relevant goods and services.
serve to concentrate poverty in disadvantaged neighborhoods.
reduces contact between groups.

Motivation:

Critique of the study of segregation in academia and policy realm.

Literature ignores that segregation creates areas of transition between relatively homogeneous zones, this is, social boundaries (Kramer, 2017; Legewie, 2018).

Limitations in conceptualization and logic of literature:

\[ Inequality \longmapsto Socio-spatial\: procesess \longmapsto Segregation \longmapsto Social\: boundaries \]

  + Following Jacquez et al. (2000) and Jacquez (2010) boundaries: the locations where variables change rapidly or represent edges of homogeneous areas.
  + Tilly (2004) “any contiguous zone of contrasting density, rapid transition, or separation between internally connected clusters of population and/or activity”. 
Limitations of the current conceptualization and measurement of segregation.
Limitations of binary logic that dominates current models.

Evidence of social boundaries

Dean et al. (2019) and Smith et al. (2022) suggest that there is an elevated risk of crimes in areas joined by social boundaries [UK: migrants & ethinicity].
Legewie & Schaeffer (2016) and Legewie (2018) find that neighbourhood conflict is more likely to occur in areas co-located between two homogeneous communities. [USA: ethnicity]
Křížková et al. (2021) and Klinger et al. (2017) suggest that more social mix and less segregation, in general, are related to fewer neighbourhood conflicts. [Czechia: migrants]
Olner et al. (2023), find that household moving decisions vary according to the proximity to social boundaries (the first study to include asymmetric effects, to our best knowledge).[The Netherlands: ethnicity]]

Research objectives

Interrogating the relevance of areas of transition between relatively socioeconomic homogeneous zones.
Social boundaries as composite forms and effects of segregation as they (de)link spaces and regulate the movement of goods, people, and ideas across these boundaries.

Main research questions:

Do social boundaries exist in the cities of interest?
Do social boundaries mediate the presence, type and intensity of economic activity and co-location of business?

Potential contributions

Conceptualize and interrogate the relationship between social boundaries and economic activity.

Enhance methodological framework to detect and analyse social boundaries.

Move beyond average effect of segregation, by analyzing whether the interaction between social class and neighborhood structure produces different effects on economic activity.

Estimation strategy I:

Identify and map boundaries.
Estimate the effect of boundaries on the location of economic activity.

Bayesian worldview

Observed quantities
- data
- covariates
Unobserved quantities
- parameters
- latent states
- missing data

\[ \underbrace{\underbrace{Pr}_{\text{Probability of}} (\overbrace{\theta}^{\text{Unobserved qtys}}| \underbrace{y}_{\text{Observed qtys}})}_{Posterior} \; \propto \; \underbrace{P(B|A)}_{\text{Likelihood}} \; \cdot \; \underbrace{ P(A)}_{\text{Prior}} \]

Bayesian Hierachical CAR framework

\[ Y_k \sim \text{Binomial}(N_k, p_k); k = 1,...,n \]

\[ \ln\left(\frac{p_k}{1 - p_k}\right) = \beta_0 + u_k \]

\[ u_k|u_{-k}, W, \lambda, \tau^2 \sim N\left(\Sigma_{k \neq l} \frac{u_l}{1 - \lambda + \lambda w_{kk+}}, \frac{1}{\tau^2(1 - \lambda + \lambda w_{kk+})}\right) \] \[ \beta_0 \sim N(0,b) \]

\[ \tau^2 \sim \text{gamma}(e,f) \]

\[ \text{logit} \sim N(0,100) \]

Identify and map boundaries, how Bayes CAR works

Source: Legewie. 2018.

Estimate the effect of boundaries: Geographically weighted matrix estimator

\[ \beta_j(u_i, v_i) = \left[ X^T W(u_i, v_i)X \right]^{-1} X^T W(u_i, v_i)y \]

Estimate the effect of boundaries: Considering atributes relatedness

similarity weighted matrix:

\[ d_k(i,j) = |x_k(i) - x_k(j)| \]

\[ d(i,j) = \frac{1}{m} \sum_{k=1}^m d_k(i,j) \]

\[ W_s(u_i, v_j) = \exp(-d(i,j)^2) \]

Similarity and geographically weighted matrix

\[ \gamma = 1 - \alpha, \quad \alpha \in (0,1) \]

\[ W_{GS}(u_i, v_i) = \alpha \cdot W_G(u_i, v_i) + \gamma \cdot W_S(u_i, v_i) \]

Data and variables

Variable used to determine boundaries	Dependent Variable	Main independent variable	Control Variables
Proportion of SEC	Number/type of businesses by census track	Measure of social boundaries (edge intensity)	Demographic variables
			Social structure of neighborhood.
			Measure concentrated disadvantage (factor).
			Build environment and land use
			Controlling for the effects of structural characteristic and the conventional physical boundary measures of highways, parks, and rivers.
			Spatial lags for ethnic categories and diversity index.

Challenges and limitations

There are significant challenges in the conceptualization, identification, and measurement of social boundaries.
More systematic work is needed for the development of standardized boundary characterization metrics.
How stable in time is a detected boundary? Are there different types of boundary effects? How do boundaries differ from neighbourhood effects?

Take-away message and future work

A place based understanding of economic activities at the neighborhood level is essential for the development of effective urban policies.

The study aims to provide a principled methodological framework to detect and analyze social boundaries.

This research opens up new directions for further study, focusing on areas of transition between relatively homogeneous zones.

Many thanks!

Comments and critics are welcome: awonge01@bbk.ac.uk