Declining Enrollments and Socio-Economic Divide in Urban Schools: The Fragile Link Between School Utilization,Housing,and Educational Opportunity

Abstract

Due to recent demographic shifts in urban neighborhoods, many public school boards have experienced fluctuations in their enrollment levels. This reconfiguration of school populations is particularly problematic for schools with declining enrollment, as low facility utilization is a commonly used rationale for school closures. This study utilizes census, enrollment, and school achievement data from a large urban school district to explore the academic and socioeconomic characteristics of schools experiencing under-enrollment and under-utilization. The analysis revealed that schools with low utilization rates serve a higher number of marginalized students, are located in areas with higher housing instability, and experience lower achievement rates. We note that underutilization is symptomatic of neighborhood decline, and when combined with the potential for school closures, it reveals compounding challenges for marginalized communities. We argue that greater integration between education and urban development policies is essential to help communities with underutilized schools enhance and balance their educational opportunities.

Keywords

school reform social justice urban education

Introduction: School Enrollments, Utilization, and Educational Opportunity

In school systems where attendance is geographically determined, the socioeconomic composition of schools reflects the demographic characteristics of the surrounding neighborhoods. As local communities prosper or decline, schools may reflect these changes through transformations in the characteristics of the student body, including an increase or decrease in enrollments. Disinvested neighborhoods with deteriorating social infrastructure may deter families from moving in or staying, unless the cost of living elsewhere is so high that no other option is available (Gingrich & Ansell, 2014). This socioeconomic sorting creates segregation patterns whereby mid and high-income families, who have the resources to move, seek residence in more prosperous neighborhoods, while low-income families, who do not have a choice, remain in the declining neighborhoods. Naturally, the decline in population at the neighborhood level is reflected in the decline of population at the school level. Also, if this decline is accompanied by a concentration of poverty, then schools in these neighborhoods would reflect the increase in poverty levels (Reardon et al., 2024). In this paper, we further this insight by focusing on the relationship between school enrollment levels and the distribution of educational opportunities in school districts.

Fluctuating enrollments within a school district can lead to school overcrowding or underutilization. Overcrowded schools may struggle to provide a comfortable learning environment due to space limitations that result in congestion in classrooms and common areas. School boards address overcrowding by deploying portable classrooms, changing catchment areas to redistribute students within the district, or changing school hours and schedules (Riveros, 2023; Riveros & Zhou, 2025). On the other side of the enrollment spectrum, underutilization may limit the quantity and quality of programs for schools with declining student populations. When school funding is tied to enrollment levels, school boards may need to redirect a share of their budget to maintaining unused spaces and transportation. Also, when funding is contingent on student numbers, underutilized schools may be unable to offer additional programming, which could limit learning opportunities for students. Further, since enrollment numbers are a significant criterion in the decision to close a school (Collins et al., 2023; Ontario Ministry of Education, 2018), low student counts put underutilized schools at risk of closure. The literature on school closures has shown the devastating consequences of this practice for minoritized communities (Burdick-Will & Keels, 2013). Thus, enrollment disparities present significant challenges for public school boards seeking to provide high-quality and equitable access to education for their students.

For many years, large urban centers across North America have experienced an increase in out-migration toward smaller peripheral communities (Karp et al., 2022; Hartt, 2021; Weaver et al., 2016). The high cost of living in big cities, decreased housing affordability, and the growth in remote work have drawn households to smaller communities often located at the periphery of large urban centers (McQuillan, 2024). As a result, some urban school boards have experienced a decline in enrollment. This investigation focuses on the largest school district in Canada, the Toronto District School Board (TDSB). The TDSB serves ~239,000 students in 579 schools distributed throughout the city of Toronto (Toronto District School Board, 2025). Between the school years 2019–2020 and 2022–2023, the Toronto District School Board experienced a 6% decrease in enrollment (Toronto District School Board, 2024), which reflects the population decline trend across the city (McQuillan, 2024).

Changes in school enrollment levels may reflect changes in the distribution of educational opportunities (Riveros & Zhou, 2025). That is, the location, size, and composition of the school, which are often informed by policy, could aggravate segregation or promote social integration (Richards, 2014). At the neighborhood level, schools could promote social and economic well-being by attracting families, businesses, and social services. Conversely, schools could be used to maintain the socioeconomic status of affluent families by limiting the enrollment of less affluent students through direct or indirect exclusionary mechanisms (Tieken & Auldridge-Reveles, 2019). Direct mechanisms include exclusive programming, such as gifted or specialized programs (Grissom et al., 2019); indirect mechanisms include limiting access through location; for example, when attendance is determined by proximity, schools in wealthy neighborhoods can reject low-income students by virtue of their attendance area, intensifying school socioeconomic segregation (Yoon, 2024). Since these processes are both social and spatial, the adoption of vocabularies and sensitivities that capture these features becomes urgent and necessary. Unfortunately, the education literature has paid scant attention to questions of school capacity, size, and composition, underestimating their relevance for equalizing educational opportunity. This paucity has likely been driven by a simplistic understanding of school facilities as just physical spaces and the exclusive concern of managers and bureaucrats. This oversimplification has prevented the advancement of a robust and academically informed discussion about the implications of social-educational spaces for equity, learning, and well-being.

The following questions guided this study: (1) What are the socioeconomic characteristics of schools experiencing low utilization (i.e. under-enrollment) in the TDSB?, (2) What neighborhood-level housing and demographic factors are linked to school underutilization, and what are the broader implications for marginalized communities in the district?, and (3) How is school underutilization in the TDSB associated with student achievement across grades and subjects? In this paper, utilization is defined as the extent to which a school building is occupied by students, due to enrollment, relative to the available instructional space (NCES, 2003). For example, 80% utilization means that 80% of the spaces available for instruction in the building, such as classrooms, are currently used. We adopted a quantitative approach and used data from the Ontario Ministry of Education, the Toronto District School Board, Statistics Canada, and Ontario Health. The methods section will discuss specific details about the sources, characteristics of the data, and analytical approach.

The Spatial Politics of Schooling

Schools have a fundamental role in shaping the social dynamics of neighborhoods and cities. They are sites of social formation, reproduction, and contestation (Riveros & Nyereyemhuka, 2023). For some communities, a school can be a hub, a source of pride and identity; yet for others, it can be a place of struggle, isolation, and exclusion. Policies, social practices, and discourses shape the school’s nature and operation. As social institutions, schools do not exist in a vacuum; they are products and producers of social relations. By portraying schools as relational sites, we can better understand how they interact with other social processes and institutions. Enrollment levels, for instance, cannot be explained as a mere function of the school’s characteristics. Several factors influence the number of students attending a school, including housing policy, neighborhood demographics, and school board policies, which determine who lives in a neighborhood and, therefore, affect who attends the local school. Relatedly, the confluence of several social, political, and spatial factors, including exclusionary urban development, speculative housing markets, economic displacement, and migration (Gingrich & Ansell, 2014), may shape the composition of neighborhoods and, by extension, the social characteristics of schools.

The notable scholarship that has examined how socioeconomic disparities shape schooling illuminates the relational nature of education processes. By relational, we mean that the very existence and formation of a school, and a school system, as a social and material reality, are predicated on a complex arrangement of social, political, and economic forces. This characterization translates to a spatial understanding of schools whereby educational spaces are products and producers of social relations. Educational spaces, thus, are not mere material backgrounds; they are political sites deeply entangled with social forces and the practices of social actors.

This relational depiction of space draws from Harvey’s (2006) tripartite framework, in which space is understood as simultaneously absolute, relative, and relational. Absolute space is an extensive totality, a container that exists independent of material or social arrangements. Absolute space is conceptualized and expressed through the language of geometry. In relative space, “the standpoint of the observer plays a critical role” (Harvey, 2006, p. 273). This is space understood in terms of different epistemological, disciplinary, ideological, or political frameworks. For example, urban mobility may be represented differently by car drivers, commuters, bicycle riders, and pedestrians, respectively. Finally, relational space is constituted by the interaction of social, economic, political, and material processes. It depends on these processes for its existence. In this sense, space does not exist independently of the processes that define it. In Harvey’s words, “an event or a thing at a point in space cannot be understood by appeal to what exists only at that point. It depends upon everything else going on around it” (p. 274). The tripartite model of space offers a sophisticated theorization of the relationship between schools and their surrounding socio-economic and political processes. It suggests that schools should not be understood and investigated as isolated institutions. Their formation, operation, and change are intimately connected to other social institutions and forces. Thus, when considering changes in the school’s population, including school demographics and enrollment levels, the analysis must recognize the influence of several social, economic, and political factors that coalesce to produce the school as a social site.

This spatial approach to the study of schooling reasserts the political nature of schools. By deploying spatial vocabularies and sensibilities, this research demonstrates how space is implicated in exclusions, privileges, scarcity, hierarchies, and differences in education. Often, questions of school planning, enrollment, capacity, and utilization are dismissed by educational researchers as mere bureaucratic and technical issues. However, the repositioning of these questions from a socio-spatial perspective reveals the connections between education reform, urban restructuring, demographic changes, and the material conditions of schooling.

Declining Enrollments in Socio-Spatial Perspective

Enrollment data is essential in funding, planning, and the overall organization of school systems. Changes in the number of students attending a school or district can signal demographic and residential trends. These changes offer critical insights into the demand for educational services in a neighborhood or city. When analyzed in relation to other factors, such as race, gender, housing, school funding, or income, researchers can obtain meaningful insights about the composition and characteristics of schools. Jackson (2021), for instance, argued that the traditional accounts that attribute improvements in educational opportunity to the expansion of enrollments in educational systems fail to recognize the role of institutional goals and social reproduction in educational supply. While some forms of expansion may be driven by compensatory mechanisms, such as the aim to increase the proportional representation of underserved students, other expansions may be driven by financial goals, such as the increase in allocations for fee-paying students or the introduction of specialized programs to increase revenues (Poole et al., 2021). For schools with declining enrollments, this means that a reduction in demand must be understood in the larger context of the factors that may influence school utilization, such as school assignment policies, funding, neighborhood change, housing, labor, and demographic trends.

Urban Transformation and Schooling

The link between housing and schooling invites further examination of the nexus between school reform and urban restructuring. In many urban centers, residential location significantly shapes educational opportunities in terms of access to and availability of educational resources. Lipman (2013), for instance, argued that contemporary shifts toward market-driven social housing policies and the increased marketization of education have intensified socioeconomic disparities at the school level. This shift reflects the role of schooling as a mechanism for social reproduction. For many families, proximity to schools is a key consideration when choosing a residence. However, in a speculative housing market, residential choices are mediated by the affordability and availability of housing options. As housing costs increase, a sorting process emerges whereby families gravitate toward the neighborhoods they can afford (Gingrich & Ansell, 2014). In addition to socioeconomic disparities, racial discrimination has been cited as another factor in residential sorting (Bischoff, 2008; Frankenberg, 2013; Owens, 2020). For instance, through redlining, financial institutions can limit access to credit and mortgages for people living in areas with high concentrations of racialized populations, blocking homeownership and preventing wealth accumulation within racialized communities. Also, through steering, some real estate agents have directed racialized homebuyers toward less desirable neighborhoods (Ross, 2011), furthering segregation.

This situation, coupled with the dismantling of public housing and a decrease in government-funded housing assistance, leaves marginalized families in a precarious position regarding their residential options (Holme et al., 2020). In school districts where school attendance is mainly determined by geographic location, neighborhood sorting has a profound impact on the socioeconomic composition of the local schools. Research on the housing-schools nexus reveals that measures of achievement, such as standardized test scores, correlate with housing prices, household income, levels of parental education, employment status, and single parenthood (Chetty et al., 2016; Reardon et al., 2024). This association reflects not only the essential link between housing and schools but also the importance of housing policy in promoting equitable educational opportunity.

Urban transformation has been cited as a key influence on school utilization. Processes of gentrification, suburbanization, and urban decline lead to changes in the socioeconomic composition of neighborhoods, which in turn affect the levels of enrollment in local schools. Pearman (2020), for instance, found that public schools in gentrifying neighborhoods experience declines in their enrollment levels. This reduction is more pronounced when the gentrifiers are predominantly White. Candipan (2020) examined the decrease in demand for local schools in gentrifying neighborhoods, finding that gentrifier parents tend to avoid local public schools when choice schools are available to them. The availability of schools of choice, such as private and charter schools, has an impact not only on the socioeconomic composition of local public schools but also on their enrollment levels. More recently, Dee and Murphy (2021) noted that after the COVID-19 pandemic, public schools in some jurisdictions experienced a decline in enrollment; however, charter, virtual, and vocational schools increased their enrollment. These authors noted that these declines were higher in smaller districts and those that serve higher percentages of White and less affluent students.

School Choice

Some school systems have adopted school choice policies aiming to break the coupling between residential location and school location. In addition to the conventional market-driven rationale that suggests competition between education providers will increase quality, these policy initiatives assume that families will be willing to travel longer distances to attend schools that match their educational priorities, thereby relieving local schools of enrollment pressures. However, research in this area has consistently demonstrated that this policy strategy tends to exacerbate racial and socioeconomic segregation. Affluent families, who are capable and willing to bear the cost of transportation to access high-performing schools, mobilize their resources to take the most desirable educational opportunities in the district, leaving the less desirable options to less affluent families (Yoon & Lubienski, 2017). This shift of educational opportunities in education markets was studied by Denice and Gross (2016), who found that low-income and racialized families have fewer opportunities to access high-performing schools, especially when these schools are in White and high-income neighborhoods. Adding to these findings, Pogodzinski et al. (2021) noted that even with district-sponsored transportation, low-income students continue to face challenges in attending school regularly due to longer distances, unreliable transit, and a lack of school-based supports. Further, Billingham and Hunt’s (2016) study revealed that when given a choice, White parents tend to avoid Black majority schools, which suggests that, at least in the United States, race plays an important role in informing the geographies of school choice. Finally, Bischoff and Tach (2020) found that an increase in school choice options is associated with an increase in racial imbalances between attendance area-bound public schools and their neighborhoods. These studies demonstrate that choice policies influence the racial and socioeconomic composition of the schools, which ultimately affects the schools’ enrollment levels.

School Closures

The literature on school closures highlights declining enrollment as a typical criterion in a district’s decision to close a school (Tieken & Auldridge-Reveles, 2019). When school funding is tied to enrollment levels, the costs of maintaining and operating underused facilities create a financial incentive for closures (Tieken & Auldridge-Reveles, 2019). Research in this area has noted that closures are often concentrated in neighborhoods with high percentages of racialized and low-income populations (Ewing, 2018; Lee & Lubienski, 2016). These studies have documented the adverse consequences of closures, including the fragmentation of local communities (Collins et al., 2023), an increase in travel-to-school time for affected students (Nerenberg, 2021), and a negative impact on student learning due to the stress associated with change (Brummet, 2014). Therefore, when low utilization is a key criterion for closure, a decrease in enrollment levels could signal the beginning of the end for many schools.

For many communities, losing a school represents losing a community hub, a site for social and political participation (Collins et al., 2023; Ewing, 2018). Some strategies have been proposed to ameliorate and counter the negative effects of school closures. Examples include repurposing unused buildings or parts of them to house community activities, groups, and events such as afterschool programs, daycares, adult education, sports, and recreational activities (Clegg & Williams, 2019). Evidently, declining enrollments are a clear indication of a reduction in the demand for schooling. Yet, this reduction in educational demand should not be understood as a reduction in the need for social infrastructure. More innovative, inclusive, and participatory alternatives to closing a school are needed. As noted in the literature reviewed above, the demographic and socioeconomic composition of a school is influenced by several out-of-school factors that inform the socioeconomic characteristics of the neighborhood, including migration patterns, housing, urban development, and employment. Therefore, strategies to address enrollment decline, and thus to prevent school closures, must include revitalizing neighborhoods through economic and social opportunities for residents. Vibrant neighborhoods attract families, which in turn bolsters enrollment in local schools.

Overcrowding

Research on overcrowding has explored the causes, responses, and effects of schools operating at overcapacity. Demographic shifts related to increased birth rates, urban growth, domestic and international migration, and premature school closures have been cited as factors that lead to a surge in enrollment levels in schools (Ready et al., 2004). School boards respond to these pressures by reshaping attendance areas, flexibilizing enrollment policies, modifying schedules, and adding capacity with portable classrooms (Riveros, 2023; Riveros & Zhou, 2025). Studies on overcrowding have found negative effects on achievement, particularly for low-income students (Burnett, 1996). These findings have been confirmed by another strand in the literature that investigates the impact of reducing school sizes on academic achievement. Kuziemko (2006) found that smaller schools tend to have improved academic achievement and that “the effect of changes in enrollment does not appear the year after a shock, but 2 and 3 years later, suggesting that the longer students spend in larger schools, the greater the decline in their achievement indicators” (p. 73). Relatedly, Jones et al. (2008) noted that as schools increase in size, attendance rates decrease, which has adverse impacts on achievement. Overcrowding has been a lasting concern in many education systems. Reduced capital investment in schooling infrastructure has created pressures for schools that see enrollments increase due to demographic shifts (Filardo, 2016; Riveros, 2023). The following section situates the question of school enrollment in the context of education reform in Ontario.

Enrollment and School Reform in Ontario

In the late 1990s, Ontario experienced a widespread restructuring of its educational system. Most of the changes relied on austerity measures aimed at limiting government spending on social programs (Basu, 2013). These reforms introduced a new funding model for school boards that, among other things, required school boards to maximize the use of their facilities to avoid the underutilization of schools (Basu, 2007). The perception that enrollment was declining prompted the Ontario Ministry of Education to create new guidelines to determine the optimal utilization of schools (Basu, 2013; Collins et al., 2023). The assumption was that a more efficient use of school space would bring significant savings, which is in line with the politics of austerity that have prevailed in the province for many decades. To enforce this message, the Ontario Ministry of Education warned school boards that their funding would be compromised if the utilization targets were not met (Basu, 2007). As a result, 745 schools were closed between 1998 and 2005 (Basu, 2013), and 402 additional schools were closed between 2011 and 2021 (Collins et al., 2023). The main rationale for these closures was underutilization, under the argument that a more efficient allocation of instructional spaces would prevent wasteful spending on empty classrooms. It should be noted, however, that research from Basu (2007, 2013) and Collins et al. (2023) found that many of these closures impacted minoritized communities. These authors pointed out that the loss of public amenities, like public schools, exacerbates the economic decline of the neighborhoods, furthering socioeconomic marginalization for vulnerable groups.

Declining enrollment has been a concern in Ontario since the late 1970s (Chamberlin, 1980). Low birth rates, an aging population, and migration from rural to urban areas have brought a reduction in the number of students for many school boards (Statistics Canada, 2007). In 2008, the Ontario Ministry of Education created the Declining Enrolment Working Group (Ontario Ministry of Education, 2009) and tasked it with providing recommendations to address the impacts of dwindling enrollments. In their report, the working group confirmed the negative enrollment trends and suggested adjustments to the funding formula to offer additional support to school boards experiencing reductions in school population. Notably, the report stated that enrollment fluctuations were not the same across the province. While rural and small municipalities were projected to lose students, some mid-size cities, especially around Ontario’s largest city, Toronto, were expected to increase their enrollment numbers. These projections were confirmed by Riveros (2023), who found that between the years 2010 and 2020 most large school boards, including those in cities around Toronto, experienced growth and had to rely on portable classrooms to accommodate their increased school populations (Riveros & Zhou, 2025).

It should be noted that the Declining Enrolment Working Group report focused on province-wide and between-school boards measures. They found that, on average, the province experienced a decline in student population, and this situation was evident in many school boards. However, the Declining Enrolment Working Group did not investigate within-school board fluctuations; that is, while a school board may appear to be experiencing average decreases, it is possible for enrollment declines to be concentrated in certain areas of the district while other areas may experience increases. Conversely, it is possible for some areas within a school district to experience a decrease while, on average, the district experiences an increase in student population. The error was to assume that a pattern in the aggregated data, such as an average decline in enrollment, applies to all individuals or sub-groups within the sample, an example of the ecological fallacy (Jargowsky, 2004). Ignoring these subtle but relevant differences has implications for the conceptualization, diagnosis, and responses to the issue. A province-wide or district-wide perspective obscures social processes that shape school composition at the local level. As noted in the literature reviewed in previous sections, variations in school enrollment patterns could be understood as an interaction between the socioeconomic dynamics of the neighborhood, housing, and education policies. Strategies designed to address enrollment decline must recognize the interactions between in-school and out-of-school factors.

Data and Methods

Dataset Creation

The data for the analysis were obtained from several sources. The first source was the Ontario Ministry of Education’s School Information and Student Demographics datasets (Ontario Ministry of Education, 2017, 2021). These variables are listed in Table 1. For this study, we did not include grade 9 math achievement results due to measurement inconsistencies. Before 2021, there were two math streams in grade 9, applied math and academic math, each stream received a separate test, and the results were reported separately. After 2021, both streams were combined, and only one test was conducted. This created a discontinuity in the measurement of grade 9 math. Thus, to avoid data reliability issues, we decided not to include this measure in the models.

Table 1.

Descriptive Statistics Dataset 1.

Variable	2017					2021					Change rate (%)
Variable	Median	Mean	S.D.	Min.	Max.	Median	Mean	S.D.	Min.	Max.	Change rate (%)
% Gifted	0.00	2.24	5.91	0.00	40.00	0.00	2.09	5.77	0.00	50.00	−6.30
% Low-income	25.00	26.85	11.79	0.00	63.00	26.00	26.78	11.44	3.00	63.00	−0.29
% New to Canada from non-Eng. speaking country	10.00	9.44	8.73	0.00	90.00	10.00	8.79	7.64	0.00	85.00	−6.84
% Parents no diploma	10.00	9.21	7.35	0.00	35.00	10.00	8.97	6.94	0.00	32.00	−2.66
% Special education	14.00	16.51	14.11	0.00	100.00	11.00	15.68	14.76	0.00	100.00	−5.03
% Grade 3 math^a	62.00	60.72	18.68	6.00	100.00	60.00	59.52	19.36	5.00	100.00	−1.97
% Grade 3 reading^a	77.00	75.02	14.86	22.00	100.00	75.00	73.83	16.25	16.00	100.00	−1.59
% Grade 3 writing^a	76.00	74.23	15.48	21.00	100.00	69.00	67.88	18.07	16.00	100.00	−8.55
% Grade 6 math^a	50.00	50.64	20.56	0.00	100.00	49.00	49.38	20.72	0.00	100.00	−2.49
% Grade 6 reading^a	84.00	81.31	12.19	34.00	100.00	88.00	84.98	11.58	37.00	100.00	4.51
% Grade 6 writing^a	82.50	80.81	12.58	33.00	100.00	88.00	85.22	12.12	37.00	100.00	5.47
% Grade 10 OSSLT^a	80.00	78.49	14.02	13.00	100.00	82.00	79.39	13.84	20.00	100.00	1.16
Utilization rate (%)	84.17	85.81	29.84	0.72	249.82	80.65	81.28	24.81	2.38	192.05	−5.28

Percentage of students achieving the provincial standard.

The second source included utilization rate (UR) data for each school in the district for the years 2016–2017, 2017–2018, 2018–2019, 2019–2020, and 2021–2022. The UR measures the school’s under/over utilization by dividing the number of students by the number of instructional spaces and multiplying the result by 100. Most literature accepts 75%–90% as an ideal utilization rate (NCES, 2003), so in theory, a school above 90% UR is overcrowded, and below 75% is underutilized. We used the Jenks Natural Breaks Method (De Smith et al., 2018) to sort the schools into three utilization categories: high, medium, and low. This method reorganizes the dataset values into classes by grouping similar values together, maximizing the average differences between these classes. The Jenks method allowed us to sort schools with similar UR levels into three distinct groups: low, medium, and high UR. We organized the data by education panel (elementary/secondary; Appendix A). Finally, the data were merged using the school’s identification number as the joining criterion.

The third source included Canadian Census data at the dissemination area (DA) level for the city of Toronto for the 2016 and 2021 cycles (Statistics Canada, 2016, 2021). With an average population of 400–700 people, dissemination areas are the smallest geographical areas for which census data is aggregated (Statistics Canada, 2016, 2021). These variables are listed in Table 2. All count data were converted to percentages by dividing the count by the total population or total households for that variable in the DA and multiplying the result by 100.

Table 2.

Descriptive Statistics Dataset 2.

Variable	2016					2021					Change rate (%)
Variable	Median	Mean	S.D.	Min.	Max.	Median	Mean	S.D.	Min.	Max.	Change rate (%)
UR (%)	80.81	81.53	29.03	0.00	207.55	78.15	76.70	26.23	0.00	177.38	−5.93
Household and dwelling marginalization	0.52	0.51	0.64	−1.40	2.92	0.42	0.42	0.65	−1.45	2.80	−17.35
Material resource marginalization	0.51	0.61	1.00	−1.23	4.49	0.52	0.58	0.99	−1.15	3.83	−4.58
Median dwelling value	552,101	59,9092	27,9353	9124	1,921,423	836,899	912,624	386,810	71,255	2,735,112	52
% Living in subsidized housing	7.22	10.40	10.85	0.00	69.38	7.12	10.13	10.09	0.00	57.46	−2.66
% Renters	43.76	43.67	18.59	5.44	98.45	44.95	44.81	17.27	7.91	97.25	2.60
% Living in unsuitable housing	13.04	14.30	8.35	1.48	55.97	13.46	14.72	8.78	1.49	59.22	2.98
% Living in single detached houses	27.38	30.71	19.86	0.00	94.33	26.73	30.19	19.37	0.07	89.90	−1.70
% Living in apartment bldg. over five storeys	30.55	33.03	22.87	0.00	99.79	30.91	34.79	22.49	0.00	99.77	5.32
% Moved-in 5 years ago	36.52	37.60	7.18	20.82	64.84	32.06	33.14	6.69	19.59	62.22	−11.86

The fourth source included two cycles of the Ontario Marginalization Index (Matheson & van Ingen, 2016; Matheson et al., 2023). The index was created by Ontario Health and includes four dimensions: household and dwellings, material resources, age and labor force, and racialized and newcomer population. Each dimension is a separate index with a score for each DA. The index used 42 Census-based indicators. Through a series of iterative factor analyses, variables with low factor loading were removed. Eighteen indicators remained to create the four dimensions of marginalization. This study focuses on the households and dwellings dimension and the material resources dimension (Table 2), given their relevance for housing and socioeconomic status. The household and dwelling dimension includes the following census variables expressed as percentages within each DA: population living alone, population older than 16 years old, population who are divorced/widowed, apartment buildings, dwellings not owned, population who moved within the past 5 years, and average number of persons per dwelling. This dimension aims to reflect the degree of “family and neighborhood stability and cohesiveness” (Matheson et al., 2023, p. 6). The material resources dimension includes the following census variables expressed as percentages within each DA: Population aged 25–64 without a high school diploma, lone-parent families, income from government transfers for population 15 years and older, unemployed aged 15 years and older, population considered low income (Statistics Canada, 2022), and population living in dwelling in need of major repair. This dimension aims to represent levels of poverty and access to basic material needs.

The fifth source is the Postal Code Conversion File (PCCF). The study used the 2016 and 2021 versions of the Postal Code Conversion File (PCCF), produced by Statistics Canada and Canada Post (Statistics Canada, 2024). The PCCF links six-character postal codes and census geographic areas such as dissemination areas and census tracts. The PCCF also provides latitude and longitude coordinates for mapping applications. The sixth source is the students’ postal code list from the TDSB in 2016–2017 and 2021–2022. In addition to the students’ postal code for each school, the dataset also includes the student count in each postal code for each school.

Since the analysis is undertaken at the school level, the census data had to be linked to each school. We followed a three-step process to assign DA-level data to each school in the sample. The first step was to link DA-level data to each postal code. The PCCF links the DA to the student’s postal code. Since each postal code may correspond to a single or multiple DAs, the DA data for each postal code is calculated by averaging the variable data from all DAs within or intersect with the particular postal code. The second step was to calculate a set of weights that were generated for the postal code associated with each school. For each school in the sample, we divided the number of students from a specific postal code by the total number of students at the school to determine the weight of that postal code. Third, each student’s postal code was linked to their school through the school’s unique ID. For each school, we calculated a weighted average of census variables by multiplying each postal code’s data by the proportion of students from that postal code attending the school. We then added these products to obtain the school-level variable.

To answer the research questions, we organized the data into two datasets. The first dataset included utilization rates, school demographics, and achievement data at the school level between 2017–2018 and 2020–2021 (Table 1). This dataset was used to answer question one: “What are the socioeconomic characteristics of schools experiencing low utilization (i.e. under-enrollment) in the TDSB?” and question three: “How is school underutilization in the district associated with student achievement across different grades and subjects?.” The second dataset included school utilization rates, housing-related census variables, and two dimensions of the Ontario marginalization index, all aggregated at the school level (Table 2). As noted above, this dataset corresponds to two cycles of the Canadian census, 2016 and 2021 and was used to answer research question number two: “What neighborhood-level housing and demographic factors are linked to school underutilization, and what are the broader implications for marginalized communities in the district?” Also, this dataset was used for the analysis of variance (Appendix C) and the visualization of marginalization levels in the city (Figure 1).

Figure 1.

Changes in the concentration of marginalization TDSB 2016–2021.

Analytical Strategy

We began the analysis by measuring the concentration of marginalization in the school district. We used the Global Moran’s I Index (Appendix B) to test for the spatial autocorrelation of two dimensions of the Ontario Marginalization Index: Household and Dwelling, and Material Resources in 2016 and 2021. The purpose of the test was to determine whether these dimensions were clustered, dispersed, or randomly distributed within the study area (Mitchell & Griffin, 2021) and whether the degree of clustering increased or decreased between 2016 and 2021. Spatial clustering of the data results in a positive Moran’s I Index, such as when high values are clustered near other high values, and low values cluster near other low values. Conversely, the Moran’s I Index will be negative when high values are near low values. After finding that the Global Moran’s I was significant, we conducted an Anselin local Moran’s I analysis to measure where the clustering occurred within the study area. A positive index (I) value means that a feature has similar high or low neighboring features. A negative index (I) value means that a feature has dissimilar neighboring features (Appendix B). The next step was a one-way analysis of variance (ANOVA) test aiming to determine whether there is a statistically significant difference between levels of utilization rate (high, medium, low) in relation to two dimensions of the Ontario marginalization index: dwelling and household marginalization, and material resource marginalization (Appendix C). Next, we conducted a series of Tukey HSD tests to identify the differences between the groups tested in the ANOVA.

Finally, we conducted three sets of multivariate regression analyses using the Ordinary Least Squares (OLS) method (Appendix D). The first set of models, addressing research question one, used UR as the response variable and the demographic variables from the Ministry datasets as explanatory variables. The second set of models, addressing research question two, also used UR as the response variable, with housing variables as explanatory variables. The third set of models, addressing question three, treated UR and demographic data as explanatory variables and examined their associations with the achievement variables. Each relationship was modeled separately.

Findings

Socioeconomic Profile of Underutilized Schools

The regression models revealed that schools with lower utilization tend to serve higher proportions of students from low-income households, students receiving special education services, and students whose parents do not have a high-school diploma (Table 3). These findings confirm results from previous literature on the demographic characteristics of schools experiencing declining enrollment (Green et al., 2022; Pearman, 2020). Also, as noted above, downward trends in public school enrollment are associated with overall neighborhood decline. When neighborhoods experience socioeconomic decline, households that can afford to relocate move to more prosperous areas. Low-income families, who cannot afford to move, are often left behind, which increases the local concentration of poverty. Since public schools tend to serve their adjacent communities, it follows that poverty would concentrate in these schools, too. Relatedly, previous research (Weber et al., 2025) has demonstrated that the rates of students receiving special education services tend to increase as household income decreases, suggesting that schools experiencing a reduction in enrollment would have higher rates of students receiving special education services.

Table 3.

Regression Results – The Association Between School Utilization and Demographic Variables and in Elementary and Secondary Schools.

Predictor	Elementary	Secondary	All
Intercept	104.46**** (1.23)	139.10**** (5.04)	107.54**** (1.15)
% New to Canada from non-English-speaking country	0.33**** (0.06)	−0.61**** (0.12)
% Low income	−0.39**** (0.05)	−0.98**** (0.25)	−0.31**** (0.05)
% Parents no diploma	−0.40**** (0.08)	−0.68*** (0.25)	−0.53**** (0.07)
% Gifted	−0.18** (0.08)
% Special education	−0.47**** (0.04)	−0.68**** (0.07)	−0.53**** (0.32)
R ²	0.14	0.35	0.17

Note. Variables were selected through stepwise regression; non-significant predictors were removed from the final models.

p < .1. **p < .05. ***p < .01. ****p < .001 (standard error).

The findings reported in Table 3 support these conclusions by showing that, keeping all other variables constant, a 10 percentage point increase in special education students is associated with a 4.7 percentage point decrease in utilization for elementary schools, a 6.8 percentage point decrease for secondary schools, and a 5.3 percentage point decrease for elementary and secondary schools combined. Similar reductions were observed for elementary schools with higher numbers of low-income students; namely, a 10 percentage point increase in low-income students is associated with a 3.9 percentage point decrease in utilization. The downward trend is more dramatic in secondary schools, where a 10 percentage point increase in low-income students is associated with a 9.8 percentage point decrease in utilization. For the model that combines both elementary and secondary schools, the decrease in utilization is 3.1 percentage points, all else equal. Similarly, reductions in utilization are associated with an increase in the number of students whose parents have no high school diploma or certificate. For instance, for every 10 percentage point increase in students in this group, elementary schools reduce their utilization by four percentage points and secondary schools by 6.8 percentage points. These findings suggest that schools with low utilization tend to serve higher numbers of marginalized students.

It should be noted that the relationship between the percentage of students new to Canada and utilization rate differs by school level (Table 3). In elementary schools, and maintaining everything else equal, a one percentage point increase in newcomer students is associated with a 0.33 percentage point increase in utilization. In contrast, secondary schools show the opposite pattern, with a decrease of 0.61 percentage points in utilization for every percentage point increase in newcomer students. These findings suggest that newcomer families often arrive with younger children, resulting in an increase in demand for elementary schools. Following this logic, it could be argued that secondary schools may serve lower rates of newcomers, which could explain the opposite trend in enrollment for this population. In the next section, we examine the clustering of marginalization in more detail using the Ontario Marginalization Index.

Overall Increase in the Clustering of Marginalization

Historically, large urban centers like Toronto present various forms of socio-spatial clustering. Residential sorting, which contributes to clustering, results in the concentration of poverty in some neighborhoods as the limited availability of affordable housing and economic opportunities pushes low-income families toward declining neighborhoods (Walks, 2016). Figure 1 illustrates some of these clusters with a small increase in the concentration of marginalization over time. The Moran’s I results (Appendix B) show a statistically significant clustering of schools with higher levels of household and dwelling marginalization in the city’s southern-central areas and primarily in the inner-city north of downtown (Figure 1). Between 2016 and 2021, the concentration of household and dwelling marginalization increased by six percentage points (Appendix B). Schools in these areas tend to serve higher numbers of students living in dense neighborhoods with higher rates of residential mobility. Families in these areas tend to live in more crowded or unsuitable dwellings, which, as shown in Table 4, model 1, is associated with an average reduction in utilization.

Table 4.

Regression Results – Housing and School Utilization.

Predictor	Elementary				Secondary
Predictor	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Intercept	89.73**** (1.60)	91.07**** (2.12)	68.03**** (3.00)	53.93**** (6.24)	80.81**** (12.34)	44.76**** (5.47)	128.58**** (8.81)	112.21**** (17.92)
% Living in unsuitable housing	−0.19* (0.10)
% Living in subsidized housing	−0.56**** (0.08)	−0.54**** (0.09)		−0.38**** (0.09)	−1.21**** (0.34)		−1.13**** (0.33)
% Renters		−0.10** (0.05)			−0.35* (0.20)		−0.39** (0.19)
% Living in single detached house			0.34**** (0.05)	0.31**** (0.05)		0.93**** (0.18)
% Living in apartment bldg. over five storeys			0.09* (0.05)
Median dwelling value					2.995e-05**** (8E-06)
% Moved-in within last 5 years				0.61**** (0.13)				−1.16** (0.50)
Low-median value dwelling (dummy)							−32.16**** (7.04)
Medium-median value dwelling (dummy)							−27.34**** (6.61)
R ²	0.07	0.07	0.05	0.11	0.30	0.14	0.33	0.03

Note. Variables were selected through stepwise regression; non-significant predictors were removed from the final models.

p < .1. **p < .05. ***p < .01. ****p < .001 (standard error).

The map in Figure 1 shows a significant concentration of material resource marginalization in the schools located in the northwest and east areas of the city. Between 2016 and 2021, the concentration of material resource marginalization in these schools increased by eight percentage points (Appendix B), suggesting that schools in these areas serve higher numbers of students living in low-income households, with inadequate access to basic material needs. It should be noted that these patterns take place in more peripheral or suburban neighborhoods, supporting previous literature on the growing suburbanization of poverty (Allen et al., 2024; Mordechay & Terbeck, 2023). The discussion in subsequent sections substantiates these findings by revealing a significant association between socioeconomic disadvantage and declining enrollment.

An examination of the levels of marginalization against school utilization levels reveals that schools with low utilization are more likely to have higher levels of marginalization. As noted above, we identified the schools with low utilization by using the Jenks natural breaks algorithm (De Smith et al., 2018). The Jenks procedure maximizes the differences between classes, thus grouping similar values together. This allowed us to find the actual distribution of utilization rates in the school board (Appendix A). Based on this distribution, we counted the number of schools in each category (low, medium, high) and found that between 2016 and 2021, the overall number of schools with low utilization increased by 17%, while schools with medium utilization increased by 10%. In contrast, schools with high utilization decreased by 32%, which suggests a downward trend in utilization across the board (Appendix E).

Using the low, medium, and high utilization breaks, we conducted one-way ANOVA tests to compare the levels of marginalization (material resources, and household and dwelling) between the different school utilization levels. For material resources marginalization, the ANOVA test revealed statistically significant differences between utilization levels. A post hoc Tukey’s HSD test for multiple comparisons found that low-utilization schools have significantly higher levels of material resource marginalization than medium and high-utilization schools (Appendix C). In other words, schools with low utilization rates have higher percentages of students living in resource-marginalized households. The ANOVA test that compared the levels of household and dwelling marginalization in different utilization ranges was not significant.

Housing Characteristics and School Underutilization

Along with socio-economic factors, housing factors have a statistically significant association with low utilization. Housing characteristics are important to school utilization as residential affordability, availability, and suitability could attract or deter households from moving or staying in a neighborhood, affecting the enrollment levels in the local schools. Furthermore, as we show below, as enrollment levels vary, so do residential characteristics, which suggests that the low utilization of schools is associated with a concentration of residential marginalization.

Housing conditions are important factors in understanding educational inequities; for instance, Holme (2022) found that affordability, availability, and suitability of a child’s residence have a significant influence on the child’s educational outcomes and experiences. The regression models summarized in Table 4 reveal significant relationships between school utilization and housing, tenure, type, and conditions. In relation to tenure, the schools that serve large numbers of students living in households that rent their residences tend to have lower utilization rates (Table 4, models 2, 4, and 6). Similar declines can be observed in schools that serve high rates of students in subsidized housing (Table 4, models 1, 2, 5, and 7), with a more pronounced decline in secondary schools.

Low utilization in schools serving high rates of renters could be associated with increased transiency and housing instability in the neighborhoods served by these schools. While we did not include a home ownership variable in the models, it stands to reason that home ownership, as opposed to renting, creates more residential stability, which has a positive impact on school utilization. This insight is supported by the results of model 8, according to which utilization levels decrease by 1.16 percentage points for every percentage point increase in the proportion of people who moved into the school’s attendance area during the last 5 years. It should be noted, however, that this trend reverses in elementary schools; that is, an increase in movers in the previous 5 years is associated with an average increase of 0.61 percentage points in utilization. We believe these differences reflect size differences between elementary and secondary schools. Elementary schools have smaller catchment areas and lower capacity than secondary schools, which makes them prone to higher utilization. Further, all models that include variables related to residential mobility, such as the percentage of renters and the percentage living in subsidized housing, show a decline in utilization. This supports the idea that schools with low utilization tend to serve students with high levels of transiency.

In addition to tenure, housing values have a positive association with utilization in secondary schools (Table 4, model 5), suggesting that more desirable neighborhoods where housing values are higher are experiencing increased demand and, thus, increased school utilization. Relatedly, schools located in neighborhoods with a high number of single-detached houses tend to have increased utilization levels (Table 4, models 3 and 6), as well as those serving students living in high-rise buildings (model 3). Neighborhoods with a larger number of unsuitable dwellings, namely, residences without enough bedrooms given the household size, tend to be served by schools with lower utilization levels (Table 4, model 1). This suggests that neighborhoods with small residential units, where families would have to live in overcrowded conditions, are becoming less attractive, affecting the enrollment levels of local schools. However, low-income households may gravitate to these neighborhoods as housing may be more affordable in these areas. This insight is supported by the results shown in Table 3, which suggest that schools with low utilization tend to serve higher numbers of low-income students.

As noted above, utilization increases alongside housing values in areas served by secondary schools (see Table 4, model 5). To expand this finding, we examined the extent of utilization declines in areas with low housing values by categorizing median housing values into three groups (dummy variables): low median dwelling value, medium median dwelling value, and high median dwelling value. The groups were identified through the Jenks natural breaks procedure explained above. The regression model that incorporates these dummy variables (Table 4, model 7) shows that schools serving students in lower-priced housing areas (based on median value) tend to have a 32 percentage points lower utilization rate compared to schools in higher-priced housing areas (high median-value dwellings). For example, based on the intercept of model seven, schools in high housing value areas tend to have an average utilization rate of 128.58%. Thus, keeping all else equal, schools in the lowest housing price areas are expected to have an average utilization of 96.42% net of other influences. These results show the pronounced difference in utilization between schools serving neighborhoods with high and low median housing values. While the results of model seven suggest high utilization levels relative to the variables included in the model, it is important to note that the intercept depends on the model’s configuration. Other models in Table 4 show a lower intercept when other variables are included.

Overall, these findings demonstrate that housing quality, tenure, income levels, and residential stability are essential factors affecting school utilization. The results show that students living in unstable and unsuitable housing are more likely to attend schools with declining utilization. Since low utilization is a key criterion in school closure decisions, students vulnerable to housing insecurity could be facing the closure of their schools.

Achievement and School Utilization

We used ordinary least squares regression models (Appendix D) to explore the association between achievement and enrollment levels (Table 5). The data corresponds to annual provincial standardized achievement tests administered by the Education Quality and Accountability Office of Ontario (EQAO) to students in grades 3 6, 9, and 10. We used 5 years of data (2017–2018–2021–2022). The resulting dataset included each school’s annual achievement rates as separate observations. This increased the statistical precision and robustness of the models by capturing yearly fluctuations. To explore the differences in achievement in relation to utilization levels, we categorized the schools into three groups using the Jenks procedure outlined above. The groups were included as dummy variables in the models using the medium utilization rate as the reference group.

Table 5.

Regression Results – Achievement and School Utilization.

	Grade 3				Grade 6				Grades 3 and 6
Predictor	Math	Reading	Writing	Average	Math	Reading	Writing	Average	Math	Reading	Writing	Overall
Intercept	84.7	92.5	87.2	88.1	73.3	100.6	98.9	91.0	79.1	96.5	92.4	89.3
Low utilization	−0.47 (0.72)	−0.78 (0.65)	−1.14 (0.73)	−0.87 (0.62)	−0.57 (0.94)	−0.73 (0.58)	−1.88*** (0.59)	−1.06* (0.62)	−0.71 (0.78)	−1.08* (0.59)	−1.96*** (0.64)	−1.36** (0.60)
High utilization	3.95**** (0.84)	3.07**** (0.76)	3.63**** (0.86)	3.44**** (0.73)	2.78*** (1.00)	0.30 (0.62)	0.81 (0.63)	1.30** (0.66)	3.58**** (0.79)	1.79*** (0.60)	2.50**** (0.65)	2.45**** (0.61)
R ²	0.41	0.28	0.22	0.36	0.42	0.40	0.33	0.45	0.48	0.41	0.33	0.47

Note. Demographic variables not included due to space limitations. The models for secondary schools were not statistically significant.

p < .1. **p < .05. ***p < .01. ****p < .001 (standard error).

The results show that, relative to medium and high-utilization schools, low utilization rates in elementary schools are negatively associated with some achievement measures. The results for secondary schools were not statistically significant; therefore, we focus exclusively on the elementary models. Table 5 shows that there are negative and significant associations between utilization and grade 6 writing, the average of all three subjects in grade 6, the average of grades 3 and 6 in reading, the average of grades 3 and 6 in writing, and the average overall achievement for all grades and subjects. Although the other relationships between low utilization and achievement are not statistically significant, it is noteworthy that all coefficients consistently show a negative direction, suggesting a general tendency toward lower achievement in low-utilization schools.

For example, the overall model that combines all subjects for grades 3 and 6 shows that holding all other factors constant, 1.36 percentage points fewer students in low-utilization schools achieve the provincial standards on all subjects combined. That is, if, on average, 89.3% of grade 3 and 6 students in medium-utilization schools achieve the provincial standard in all subjects combined, then 87.9% of students will achieve the standard in low-utilization schools (89.3 − 1.36 = 87.9). The difference becomes more pronounced when comparing low and high-utilization schools. Keeping all other variables constant, for the overall model, low-utilization schools have a predicted achievement rate, that is, 3.81 percentage points lower than high-utilization schools (2.45 − (−1.36) = 3.81). Given a predicted rate of 91.75% for high-utilization schools, low-utilization schools are expected to have ~87.94% of students achieving the provincial standard. These results provide additional insight into the dynamics of enrollment decline.

Underutilization often results in a lack of enriched and varied programs, affecting the students’ educational experience and overall performance. When school funding is tied to enrollment levels, undersubscribed schools will struggle to offer additional academic support, such as counselors, supplementary instructional resources, and specialized instruction, compounding the challenges for the communities served by these schools.

Limitations

A key methodological feature of this study is the assignment of census variables to schools via students’ postal codes. This allowed us to allocate data to each school more accurately than using fixed catchment areas. By knowing how many students from a postal code attend a given school, we could assign weights to each postal code, which gave us a more accurate representation of the school’s socioeconomic characteristics. A catchment area map does not allow for such weighting because we would not be able to ascertain the areas that contribute the most or the least students to the school; thus, we would be required to average the variables within the area’s boundaries. Our analysis was limited, however, by the type and amount of data available to us. The school board provided the postal codes from 2016 and from 2021, which allowed us to use two cycles of the Canadian census (2016 and 2021). While a larger timeframe would provide more detailed and nuanced insights, we believe the results of this study offer relevant considerations about the relationships between socioeconomic disparity and declining enrollment in urban schools.

Naturally, our findings are circumscribed to the specific social, political, and spatial contexts of the city of Toronto and the province of Ontario. Issues of school enrollment and utilization are not abstract or generic phenomena; rather, they are deeply embedded in the social and material conditions of schools and the neighborhoods they serve. Accordingly, we are cautious about generalizing our findings, recognizing that social policies, funding models, and demographic dynamics vary across jurisdictions. Nonetheless, the conclusions of this study invite further examination of the intersections between education policy, financing, planning practices, and housing in large urban centers. In this sense, our findings may resonate with other contexts facing enrollment decline and uneven spatial development. Finally, although a qualitative exploration of the lived experiences of the social actors involved in these processes lies beyond the scope of this study, we recognize that such an inquiry could complement the findings presented here.

Conclusions

Although school utilization is an understudied and underutilized metric in education policy analysis and research, this study demonstrates its potential for studies of educational disparities and educational opportunity. As migration from large urban centers to smaller communities persists, urban school boards face a decrease in enrollment, creating challenges for funding, programming, and the overall operation of their schools (Hartt, 2021; Karp et al., 2022). This study demonstrates that these challenges are more pronounced in low-utilization schools due to the concentration of different forms of marginalization in the communities they serve.

Since low teacher-student ratios tend to have a positive effect on instruction (Laitsch et al., 2021), it could be tempting to argue that low enrollment levels are beneficial. However, when funding is tied to enrollment, as in the case of Ontario, a low student count imposes limitations on the number of instructional resources and support that could be offered to students in these schools. Under this model, schools that maintain healthy enrollment levels could fund more enriched and diverse instruction and programs. One strategy to address the underutilization of schools is to repurpose unoccupied sections of the building for non-school-related activities, such as community centers, recreational facilities, or adult education (Ontario Ministry of Education, 2015). Arguably, these partnerships have the potential to offset some of the operational costs of underused schools; however, these strategies do not address the reduction in instructional capacity that results from a decrease in funding due to low enrollments. This invites a critical examination of per-pupil funding formulas and the consideration of alternative funding schemes.

The finding that schools that serve low-income and marginalized students have lower enrollment could be interpreted as a sign that poverty is declining in the district. Still, this conclusion must be weighed against two crucial considerations. First, the overall decline in school enrollment suggests a decline in the city’s population, particularly families with school-aged children. Perhaps a decrease in poverty means that low-income families are moving out of the city, looking for more affordable locations (McQuillan, 2024). This suggestion is supported by a growing body of literature that has called attention to the increasing suburbanization of poverty (Allen et al., 2024; Mordechay & Terbeck, 2023). Conversely, the trends identified in this study may suggest that wealthier households are returning to urban cores, pushing housing prices up and lower-income families out. Second, this study found that material resource marginalization is becoming more concentrated in some areas of the city (Figure 1 and Appendix C). This suggests that while poverty may have decreased due to an outflow of low-income households towards suburban communities, many low-income households remain in some neighborhoods.

This study reveals that schools in areas of high marginalization tend to have low utilization, which makes them candidates for closure. As noted above, closures have been widely criticized for their devastating consequences for local communities (Collins et al., 2023; Tieken & Auldridge-Reveles, 2019). Another area that could be further examined is the impact of catchment areas on marginalization and the potential of altering catchment areas to balance enrollments. The question is whether adjusting each school’s catchment area could balance the share of students among all the schools in the district. Notable literature has examined the impact of attendance areas on school segregation, suggesting that even within the same school district, the student’s residence determines access to varied levels of school quality (Asson, 2024; Richards, 2014). By noting that a concentration in marginalization often accompanies a decrease in utilization, this study offers additional insights for policymakers interested in rebalancing and redistributing educational opportunities in their districts.

Our focus on low utilization and declining enrollments offers a new perspective on educational opportunity in urban centers. The findings underscore the nexus between enrollment and poverty, contributing to analytical frameworks investigating school and socioeconomic segregation. Further, the significant connections between school utilization levels and socioeconomic status suggest the existence of trends that could be further investigated. We recommend that, in addition to in-school interventions to support these students academically, more integrated and intersectoral policy interventions should be devised to promote housing stability, affordability, and availability at the municipal level, including neighborhood improvements that could bring about social and economic revitalization.

Footnotes

Appendices

Acknowledgements

We thank the Planning Division, Strategy and Planning Department, Toronto District School Board, for sharing the necessary data that enabled us to conduct this study.

ORCID iDs

Augusto Riveros

Chenyang Zhou

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The merged and anonymized datasets that support the findings of this study are available from the corresponding author upon reasonable request.

Author Biographies

Augusto Riveros is an Associate Professor in the Faculty of Education at the University of Western Ontario, Canada. His research focuses on the application of spatial methods and theories to education policy analysis.

Chenyang Zhou is a PhD candidate in the Faculty of Education at the University of Western Ontario, Canada. Her research explores how urban demographic shifts relate to school enrollment, composition, and outcomes.

References

Allen

Farber

Tiznado-Aitken

(2024). Pathways to suburban poverty in nine Canadian metropolitan areas. Cities, 147, 104848.

Asson

(2024). Rezoning educational opportunities: The relationship between school attendance zone boundary changes and access to schools of varying quality. Educational Policy, 38(7), 1563–1607.

Basu

(2007). Negotiating acts of citizenship in an era of neoliberal reform: The game of school closures. International Journal of Urban and Regional Research, 31(1), 109–127.

Basu

(2013). Multiplying spaces of subalterity in education: From ideological realms to strategizing outcomes. The Canadian Geographer, 57(3), 260–270.

Billingham

C. M.

Hunt

M. O.

(2016). School racial composition and parental choice: New evidence on the preferences of White parents in the United States. Sociology of Education, 89(2), 99–117.

Bischoff

(2008). School district fragmentation and racial residential segregation: How do boundaries matter?. Urban Affairs Review, 44(2), 182–217.

Bischoff

Tach

(2020). School choice, neighborhood change, and racial imbalance between public elementary schools and surrounding neighborhoods. Sociological Science, 1(7), 75–99.

Brummet

(2014). The effect of school closings on student achievement. Journal of Public Economics, 119, 108–124.

Burdick-Will

Keels

T. S.

(2013). Closing and opening schools: The association between neighborhood characteristics and the location of new educational opportunities in a large urban district. Journal of Urban Affairs, 35(1), 59–80.

10.

Burnett

(1996). Enrollment growth and overcrowding in New York City schools. IUME Briefs.

11.

Candipan

(2020). Choosing schools in changing places: Examining school enrollment in gentrifying neighborhoods. Sociology of Education, 93(3), 215–237.

12.

Chamberlin

R. J.

(1980). Too few apples: The challenge of declining enrolments and teacher redundancy in Canada. Canadian Education Association.

13.

Chetty

Hendren

Katz

L. F.

(2016). The effects of exposure to better neighborhoods on children: New evidence from the moving to opportunity experiment. American Economic Review, 106(4), 855–902.

14.

Clegg

Williams

(2019). Mixed-use developments In Taylor

Wright

(Eds.), Urban schools: Designing for high density (pp. 43–56). RIBA Publishing.

15.

Collins

P. A.

Barber

Masuda

J. R.

Snow

(2023). Socio-spatial dimensions of school closures and neighbourhood change in Ontario: An environmental injustice?. Wellbeing, Space and Society, 4, 100138.

16.

Dee

T. S.

Murphy

(2021). Patterns in the pandemic decline of public school enrollment. Educational Researcher, 50(8), 566–569.

17.

Denice

Gross

(2016). Choice, preferences, and constraints: Evidence from public school applications in Denver. Sociology of Education, 89(4), 300–320.

18.

De Smith

M. J.

Goodchild

M. F.

Longley

P. A

. (2018). Geospatial analysis: A comprehensive guide to principles, techniques and software tools (6th ed.). The Winchelsea Press.

19.

Ewing

E. L.

(2018). Ghosts in the schoolyard: Racism and school closings on Chicago’s south side. University of Chicago Press.

20.

Filardo

(2016). State of our schools: America’s K–12 facilities. 21st Century School Fund, National Council on School Facilities.

21.

Frankenberg

(2013) The role of residential segregation in contemporary school segregation, Education and Urban Society, 45, 548–570.

22.

Gingrich

Ansell

(2014). Sorting for schools: Housing, education and inequality. Socio-Economic Review, 12(2), 329–351.

23.

Green

T. L.

Germain

Castro

A. J.

Latham Sikes

Sanchez

Horne

(2022). Gentrifying neighborhoods, gentrifying schools? An emerging typology of school changes in a gentrifying urban school district. Urban Education, 57(1), 3–31.

24.

Grissom

Redding

Bleiberg

(2019). Money over merit? Socioeconomic gaps in receipt of gifted service. Harvard Educational Review, 89(3), 337–369.

25.

Hartt

(2021). Quietly shrinking cities: Canadian urban population loss in an age of growth. UBC Press.

26.

Harvey

(2006). Space as a keyword. In Castree

Gregory

(Eds.), David Harvey: A Critical Reader (pp. 270–294). Blackwell.

27.

Holme

J. J.

(2022). Growing up as rents rise: How housing affordability impacts children. Review of Educational Research, 92(6), 953–995.

28.

Holme

J. J.

Frankenberg

Sanchez

Taylor

De La Garza

Kennedy

(2020). Subsidized housing and school segregation: Examining the relationship between federally subsidized affordable housing and racial and economic isolation in schools. Education Policy Analysis Archives, 28(169), n169.

29.

Jackson

(2021). Expansion, enrollment, and inequality of educational opportunity. Sociological Methods & Research, 50(3), 1215–1242.

30.

Jargowsky

P. A.

(2004). The ecological fallacy. In Kempf-Leonard

(Ed.), The encyclopedia of social measurement (Vol. 1, pp. 715–722). Academic Press.

31.

Jones

J. T.

Toma

E. F.

Zimmer

R. W.

(2008). School attendance and district and school size. Economics of Education Review, 27(2), 140–148.

32.

Karp

D. N.

Bagchi-Sen

Rogerson

(2022). Not all shrinking places are similar: The variegated nature of population decline in the United States. Applied Geography, 138, 102581.

33.

Kuziemko

(2006). Using shocks to school enrollment to estimate the effect of school size on student achievement. Economics of Education Review, 25(1), 63–75.

34.

Laitsch

Nguyen

Younghusband

C. H.

(2021). Class size and teacher work: Research provided to the BCTF in their struggle to negotiate teacher working conditions. Canadian Journal of Educational Administration and Policy, 196, 83–101.

35.

Lee

Lubienski

(2016). The impact of school closures on equity of access in Chicago. Education and Urban Society, 49(1), 53–80.

36.

Lipman

(2013). The New Political Economy of Urban Education: Neoliberalism, Race, and the Right to the City. Routledge

37.

Matheson

F. I.

Moloney

van Ingen

(2023). 2021 Ontario Marginalization Index: User guide. Public Health Ontario. https://www.publichealthontario.ca/en/Data-and-Analysis/Health-Equity/Ontario-Marginalization-Index

38.

Matheson

F. I.

van Ingen

(2016). Ontario Marginalization Index. Public Health Ontario. https://www.publichealthontario.ca/en/Data-and-Analysis/Health-Equity/Ontario-Marginalization-Index

39.

McQuillan

(2024). Leaving the big city: New patterns of migration in Canada. The School of Public Policy Publications, 17(1), 22–35.

40.

Mitchell

Griffin

(2021). The ESRI guide to GIS analysis, volume 2: Spatial measurements and statistics. Environmental Systems Research Institute.

41.

Mordechay

Terbeck

F. J.

(2023). Moving out and apart: Race, poverty, and the suburbanization of public school segregation. American Journal of Education, 129(2), 205–235.

42.

NCES. (2003). Facilities information management: A guide for state and local education agencies. National Center for Education Statistics.

43.

Nerenberg

K. M.

(2021). Decoupled or doubled? Evaluating geospatial dispersion of student-resident displacement by school closure. Social Science Research, 95, 102519.

44.

Ontario Ministry of Education. (2009). Planning and possibilities: The report of the declining enrolment working group. Queen’s Printer for Ontario.

45.

Ontario Ministry of Education. (2015). Community planning and partnerships guideline. Ontario Ministry of Education.

46.

Ontario Ministry of Education. (2017). School information and student demographics 2017-2018. https://data.ontario.ca/dataset/school-information-and-student-demographics

47.

Ontario Ministry of Education. (2018). Draft pupil accommodation review guideline. Ontario Ministry of Education.

48.

Ontario Ministry of Education. (2021). School information and student demographics 2021-2022. https://data.ontario.ca/dataset/school-information-and-student-demographics

49.

Owens

(2020). Unequal opportunity: School and neighborhood segregation in the USA. Race and Social Problems, 12(1), 29–24.

50.

Pearman

F. A.

(2020). Gentrification, geography, and the declining enrollment of neighborhood schools. Urban Education, 55(2), 183–215.

51.

Pogodzinski

Lenhoff

S. W.

Cook

Singer

(2021). School transit and accessing public school in Detroit. Education and Urban Society, 54(6), 695–713.

52.

Poole

Sen

Fallon

(2021). Neoliberalism and public education finance policy in Canada: Reframing educational leadership as entrepreneurship. Routledge.

53.

Ready

D. D.

Lee

V. E.

Welner

K. G.

(2004). Educational equity and school structure: School size, overcrowding, and schools-within-schools. Teachers College Record, 106(10), 1989–2014.

54.

Reardon

S. F.

Weathers

E. S.

Fahle

E. M.

Jang

Kalogrides

(2024). Is separate still unequal? New evidence on school segregation and racial academic achievement gaps. American Sociological Review, 89(6), 971–1010.

55.

Richards

M. P.

(2014). The gerrymandering of school attendance zones and the segregation of public schools: A geospatial analysis. American Educational Research Journal, 51(6), 1119–1157. https://doi.org/10.3102/0002831214553652

56.

Riveros

(2023). The overreliance on portable classrooms in Ontario schools: New evidence for the study of austerity and disinvestment in Canadian public education. Canadian Journal of Education, 46(3), 570–596.

57.

Riveros

Nyereyemhuka

(2023). Conceptualizing space in educational administration and leadership research: Towards a spatial justice perspective. International Journal of Leadership in Education, 26(4), 666–679.

58.

Riveros

Zhou

(2025). Portable classrooms: Socioeconomic context and implications for academic achievement. Canadian Journal of Educational Administration and Policy, (206), 54–71.

59.

Ross

S. L.

(2011). Understanding racial segregation: What is known about the effect of housing discrimination?. In Newburger

H. B.

Birch

E. L.

Wachter

S. M.

(Eds.), Neighborhood and life chances: How place matters in modern America (pp. 288–302). University of Pennsylvania Press.

60.

Statistics Canada. (2007). Education indicators in Canada: Report of the pan-Canadian education indicators program 2007. Catalogue no. 81-582-XIE. Statistics Canada. http://www.statcan.gc.ca/pub/81-582-x/81-582-x2007001-eng.pdf

61.

Statistics Canada. (2016). Profile of census dissemination areas: Toronto. Statistics Canada. https://datacentre.chass.utoronto.ca/census/

62.

Statistics Canada. (2021). Profile of census dissemination areas: Toronto. Statistics Canada. https://www12.statcan.gc.ca/census-recensement/2021/dp-pd/index-eng.cfm

63.

Statistics Canada. (2022). Low-income cut-offs, after tax (LICO-AT). In Dictionary, census of population 2021. Statistics Canada https://www12.statcan.gc.ca/census-recensement/2021/ref/dict/az/Definition-eng.cfm?ID=fam019

64.

Statistics Canada. (2024). Postal code conversion file: Reference guide. Statistics Canada. https://doi.org/10.5683/SP3/MR8S4U

65.

Tieken

M. C.

Auldridge-Reveles

T. R.

(2019). Rethinking the school closure research: School closure as spatial injustice. Review of Educational Research, 89(6), 917–953. https://doi.org/10.3102/0034654319877151

66.

Toronto District School Board. (2024). Financial facts & expenditure trends. Toronto District School Board. https://www.tdsb.on.ca/Portals/0/docs/24-001-Financial-Facts_AODA%20(1).pdf

67.

Toronto District School Board. (2025, February 19). About us. Toronto District School Board. https://www.tdsb.on.ca/About-Us

68.

Walks

(2016). Homeownership, asset-based welfare and the neighbourhood segregation of wealth. Housing Studies, 31(7), 755–784.

69.

Weaver

Bagchi-Sen

Knight

Frazier

A. E.

(2016). Shrinking cities: Understanding urban decline in the United States. Routledge.

70.

Weber

R. N.

Waitoller

F. R.

Drucker

J. M.

(2025). Disposable spaces: How special education enrollment affects school closures. Urban Education, 60(1), 154–184.

71.

Yoon

E.-S.

(2024). An invitation to Bourdieusian space analysis: Applying Pierre Bourdieu’s theory to geospatial research in education. In Stahl

Ayling

Brock

(Eds.), The Bloomsbury handbook of Bourdieu in educational research (pp. 34–46). Bloomsbury.

72.

Yoon

E. S.

Lubienski

(2017). How do marginalized families engage school choice in inequitable urban landscapes? A critical geographic approach. Education Policy Analysis Archives, 25, 42.