Summary: Atkinson (2000) — Measuring Gentrification and Displacement in Greater London
Full Reference
Atkinson, R. (2000) ‘Measuring Gentrification and Displacement in Greater London’, Urban Studies, 37(1), pp. 149–165. https://www.jstor.org/stable/10.2307/26195250.
1
2
3
4
5
6
7
8
9
10
11
@article{twiggemolecey2014exploring,
title={Exploring Resident Experiences of Indirect Displacement in a Neighbourhood Undergoing Gentrification: The Case of Saint-Henri in Montr{\'e}al},
author={Twigge-Molecey, Amy},
journal={Canadian Journal of Urban Research},
volume={23},
number={1},
pages={1--22},
year={2014},
publisher={Institute of Urban Studies, University of Winnipeg},
url={https://www.jstor.org/stable/10.2307/26195250}
}
Questions
What is the article about?
The article investigates the connection between gentrification and displacement in Greater London during the 1981–1991 period. It uses census migration data and the ONS Longitudinal Study (LS) to measure whether professionalisation (used as a proxy for gentrification) is associated with the out-migration of working-class, elderly, unskilled, and economically inactive groups.
What kind of contribution is the paper trying to make?
The contribution is primarily empirical, but with significant methodological innovation.
The paper attempts to measure displacement quantitatively in a context where most prior research was either North American or qualitative.
It develops a method using census data and proxy indicators to identify gentrified areas and then traces the migration flows of vulnerable groups out of those areas.
What gaps does it identify/tackle?
The paper identifies several interconnected gaps:
While gentrification in London has been extensively studied (Hamnett and Williams, 1979; Munt, 1987; Butler, 1996), very little research has been conducted in Britain to measure the extent of displacement.
There is a broader methodological gap: it is extremely difficult to track displacees.
Although London has been the focus of work on polarisation and occupational restructuring (Hamnett, 1976; Hall and Ogden, 1992; Harloe, 1992), there has been little work connecting these macro-level economic processes to displacement at the neighbourhood level.
The author highlights that it is “exceedingly hard to distinguish between gentrification as a form of neighbourhood replacement or displacement” (p. 151). This distinction should be treated very carefully in relevant studies, including in the master’s dissertation.
The paper also situates itself against literature groups including: North American displacement research (Hartman, 1979; Marcuse, 1986; Sumka, 1979; DeGiovanni, 1984), London gentrification studies (Glass, 1964; Hamnett and Williams, 1979; Bridge, 1993; Butler, 1996), and work on polarisation and occupational restructuring (Hamnett, 1976; Lyons, 1996).
What are the main research arguments?
The central argument is that gentrification in London during the 1980s was associated with significant displacement of vulnerable groups, and that this displacement has been rendered invisible by a lack of measurement.
The paper argues that:
Professionalisation (as a proxy for gentrification) is negatively related to the presence of working-class, elderly, unskilled, and inactive groups.
Migration flows, rather than in-situ status changes, are the dominant mechanism behind neighbourhood compositional change.
Out-migration from gentrified areas is significantly greater than from London as a whole, suggesting displacement rather than normal residential mobility.
Displacement removes social problems spatially. It rearranges rather than ameliorates the causes of poverty.
How are they explored/presented?
The paper uses a multi-method approach:
Quantitative: Cross-sectional 1981 and 1991 census data are used alongside ONS Longitudinal Study data. Proxy indicators are constructed for both gentrification (professionalisation, degree holders, owner-occupiers) and displacement (working class, unskilled, elderly, private renters, inactive, lone parents). Four artificial borough-sized ‘G areas’ are built from wards experiencing above-average professionalisation, ranked from low to high gentrification. Exit flows from these areas are then compared with the rest of London.
Qualitative: Interviews with tenant’s rights project workers in Camden, Hammersmith, and Kensington provide accounts of displacement mechanisms including landlord harassment, rent increases, buy-outs, and the erosion of community networks and services.
Population weighting: Figures are weighted by population size changes over the period to ensure comparability.
How are data used to test them?
Data sources
- OPCS 1981 and 1991 Census data;
- ONS Longitudinal Study (a 1% sample of the population of England and Wales tracked across censuses); interviews with tenant’s rights workers;
- Metropolitan Police data on eviction offences; estate agent interviews.
Method logic
Define gentrification via proxy indicators of professionalisation;
- Construct four artificial borough-sized ‘G areas’ from wards exceeding the London-wide average rate of professionalisation;
Measure outflows from these areas for vulnerable groups;
- Compare with London-wide migration patterns to establish whether outflows are disproportionate.
Limitations and biases
The LS data are restricted to England and Wales, so flows to/from Scotland or abroad appear as exits, potentially inflating displacement estimates.
Census data only provides snapshots at 10-year intervals, so intervening moves cannot be captured. This means only ‘last-resident displacement’ is measured, and the total number of moves any individual made between 1981 and 1991 is unknown.
All figures should be considered underestimates because of the lack of detail and profiles on each household in the aggregate data.
The use of aggregate data to understand linkages between gentrification and displacement has been considered problematic because of a “lack of refinement or closeness to the nature of these processes” (p. 149). Such information does little to help us understand the qualitative impacts of migration on people who make location decisions out of choice versus constraint.
Professionalisation as a proxy for gentrification is imperfect: professions can be defined and narrated in various ways, whereas economic status has more general and well-acknowledged standards. The paper’s finding that 103,400 people in London shifted occupational status in situ raises questions about the stability of this proxy.
The construction of four artificial borough-sized ‘G areas’ from wards involves a substantial loss of spatial granularity. It is unclear why aggregated artificial areas were preferred over studying exact wards directly.
Two of the original seven displacement variables (households privately renting; ethnicity) had to be removed due to data limitations in the census migration data.
The voluntary/involuntary distinction cannot be resolved with aggregate data alone. As the author acknowledges, it is “easier to research the more tangible manifestation of gentrification, given that it is an observable end-state predicated on the removal or voluntary migration of previous residents” (p. 150).
What are the main findings?
- The gentrified areas covered approximately one-sixteenth of the London area, yet for many potential displacee groups, out-migration from these areas was almost equal to or exceeded that of London as a whole. For instance, 23,200 elderly people left gentrified areas compared to 20,000 elderly moves across all of London.
Migration was significantly greater than internal status change over the period, suggesting population replacement rather than in-situ socioeconomic improvement.
In the qualitative data, elderly residents were disproportionately represented among displacees. Displacement was achieved through rent increases, landlord harassment (mentioned in Marcuse, 1985 as well), violence, inducements, and the termination of assured shorthold tenancies.
Gentrification cumulatively eroded residents’ ability and desire to remain as social networks, shops, services, and public provision changed in ways unrelated to their lifestyles (argues in Twigge-Molecey, 2014 as “social” and “political” displacement).
Displacement removed social problems spatially rather than solving them: “problems are moved rather than solved.”
What is the contribution?
The paper provides the first systematic attempt to measure gentrification-induced displacement in Greater London using census and longitudinal data.
It demonstrates that displacement, though invisible in standard indicators, was a significant process in the 1980s.
Its methodological contribution lies in showing how proxy indicators and migration data can be combined to detect displacement signals even when direct tracking of displacees is impossible.
The qualitative strand adds depth by documenting the mechanisms and lived experiences behind the aggregate patterns.
Discuss!
On the definition of gentrification
The paper defines gentrification as “the rehabilitation of working-class and derelict housing and the consequent transformation of an area into a middle-class neighbourhood” (Smith and Williams, 1986). This definition is precisely about two distinct class groups. The annotation notes that one should also consider the economic factors underlying the process — “a capitalism is based precisely on its ability to displace the working class” (p. 149) — and the mechanism of targeting those with the “quietest political voice and with least equity.”
On displacement as relational
A crucial observation is that “lower SEGs may have similar abilities to displace those with lower resources than themselves and, in this sense, there is a degree of relativity to the process of displacement” (p. 153). This suggests displacement is not only a top-down phenomenon but operates at every level of the deprivation hierarchy — each group can displace those below it. This supports the cascade model developed in my master’s dissertation.
On the spatial pattern
The finding that gentrified growth was higher in outer than inner London, while the extent of gentrification was greater in inner London, suggests a spreading pattern that may be relevant to understanding how displacement cascades outward through the deprivation hierarchy.
On causality
The paper is cautious about causality, noting that “while causality cannot be definitive, it is probable that working-class household displacement was a reality” (p. 150).
So, my question here: is gentrification the cause of working-class displacement, or is displacement the final stage/observable outcome of the process? This question about the direction of the causal arrow remains relevant.
On profession versus economic status as an indicator
A fundamental question is why professionalisation is chosen as the primary indicator of gentrification rather than a more direct economic measure such as income or wealth.
Professions can be changed more easily than one’s underlying social or economic status: a job title may shift without any real change in material circumstances, and economic hardship can persist regardless of occupational category.
The paper’s own data reveals this instability — 103,400 people apparently shifted occupational status in situ over the decade. Moreover, because the analysis only counts movers who did not change occupation, anyone whose profession shifted during a move is excluded from the displacement estimate, systematically undercounting the process.
If a direct economic indicator (such as income level or deprivation score) had been used instead, it would arguably provide a more stable and meaningful measure of both the gentrification pressure and the vulnerability to displacement.
This is one reason why my master’s dissertation uses IMD deciles — a composite economic deprivation measure — rather than occupational proxies to classify neighbourhood position.
On social sieve and housing markets
The metaphor of housing costs acting as a “social sieve” is powerful and directly relevant to understanding how the deprivation hierarchy sorts who is displaced.
It connects to the framework with deprivation hierarchy: the hierarchy determines who is vulnerable to displacement pressures generated by polarisation.
What am I still confused about?
How can 103,400 people in London have shifted occupational status from working-class to professional within a single decade? Is this a function of how socioeconomic groups (SEGs) were defined at the time, or does it reflect genuine upward mobility? The scale of in-situ status change seems surprisingly large and could affect the interpretation of migration-based displacement estimates.
The paper mentions that “even in existing areas which have been gentrified, the continuation of the process may displace ever higher income groups” (p. 152). Should this read “lower income groups”? So confused about this point…
Why were four artificial borough-sized ‘G areas’ constructed rather than analysing the individual wards directly? The aggregation seems to lose spatial granularity without a clear methodological justification.
The relationship between professionalisation and gentrification is used as a proxy throughout, but the conclusion seems to stop at professionalisation without fully extending to gentrification. Can the displacement conclusion be directly attributed to gentrification, or only to professionalisation?
Would city size be a factor in the extent of gentrification and displacement? The author notes an “arguably global city focus” in the literature. So what is the causality between city size and the existence/intensity of these processes?
Based on the above, why was I encouraged to read this?
This paper is foundational for understanding the methodological challenges and empirical realities of measuring displacement in London.
It provides the template for a census-data-based approach to studying migration flows associated with neighbourhood change. The use of proxy indicators, the construction of comparison areas, and the comparison with London-wide baselines are all methodological moves that inform my master’s dissertation, even where the dissertation improves upon them.
The paper’s core argument — displacement is real but invisible in standard data, and that it rearranges rather than solves social problems — is the motivating concern behind my dissertation’s focus on where displaced populations go.
Connections to my master’s dissertation
Motivation
The paper’s key claim that “displacement removes social problems and rearranges rather than ameliorates the causes of poverty, environmental decay and the loss of neighbourhood vitality — problems are moved rather than solved” (p. 163) is the central motivation for the master’s dissertation.
If displacement merely relocates deprivation rather than addressing it, then understanding where displaced people go — and what happens to the deprivation hierarchy as a result — becomes a critical research question.
Methodological comparison
No proxy indicators for gentrification
Where Atkinson defines gentrification through professionalisation and then constructs artificial ‘G areas,’ the dissertation skips this step entirely. Instead, it focuses directly on migration flow patterns between hierarchical positions (IMD deciles) across the whole study area. The gentrification signal is embedded in the flow pattern rather than imposed as a prior category. This avoids the circularity of defining gentrification partly by the arrival of wealthier residents and then ‘discovering’ that wealthier residents are arriving.
Finer spatial resolution
The dissertation uses MSOAs rather than artificially constructed borough-sized areas, preserving far more geographic detail.
Fixed deprivation hierarchy
By fixing IMD deciles to the 2010 baseline, the dissertation avoids the problem of the classification shifting as displacement proceeds. Atkinson’s approach uses contemporaneous census snapshots, which means the gentrification classification and the displacement measurement are entangled in the same data.
Cascade flow index
Rather than measuring outflows from pre-classified areas, the dissertation develops a cascade flow index that captures the directional pattern of migration along the deprivation hierarchy — inflows from wealthier MSOAs and outflows to poorer MSOAs — for each neighbourhood. This measures the displacement-like dynamic directly.
Census OD migration data
The dissertation uses 2011 and 2021 census origin-destination migration data, compared against the 2010 IMD fixed decile classification. This is structurally similar to Atkinson’s use of census migration data but covers a more recent period (2011–2021 vs. 1981–1991).
Shared limitations
Several limitations identified in the annotations apply equally to my dissertation:
Voluntary/involuntary distinction: Neither study can determine whether individual moves were forced or voluntary. Census OD data records that someone moved from area A to area B, but is silent on why. However, systematic patterns — where out-migration from certain areas disproportionately flows toward more deprived destinations — provide structural evidence of displacement-like processes even without individual intentionality data.
Last-resident displacement only: Census snapshots at 10-year intervals capture only the final move, not intervening ones. The total number of moves any individual made between census points is unknown.
Aggregate data limitations: Using aggregate data lacks “refinement or closeness to the nature of these processes.” The qualitative impacts on individuals and households cannot be discerned.
Limitations specific to my dissertation
Two important limitations for the dissertation that should be explicitly acknowledged:
Intra-neighbourhood polarisation: The dissertation assigns a single deprivation decile to each MSOA, treating it as internally homogeneous. In reality, gentrifying MSOAs may contain both affluent newcomers and deprived long-term residents coexisting in the same area. The polarisation within neighbourhoods (Hamnett, 1976; Hall and Ogden, 1992) cannot be captured at the MSOA level. This is the standard trade-off of area-based analysis.
Micro-level investigation: The dissertation focuses on the structural patterns of flows between hierarchical positions, and cannot trace individual households or identify specific causal chains (rent increases, demolitions, family decisions). This would require longitudinal panel data or qualitative fieldwork, which are different research designs entirely. The dissertation’s contribution sits between the macro story of polarisation and the micro story of individual displacement experiences.
Key conceptual connections with this paper
Displacement as relational: Atkinson’s observation that lower socioeconomic groups can displace those below them supports the cascade model — displacement operates at every level of the hierarchy, not only at the top.
Social sieve: The metaphor of housing costs as a “social sieve” connects directly to the deprivation hierarchy’s role in sorting who bears the brunt of displacement. The hierarchy doesn’t cause the pressure but determines who is filtered through.
Feedback loop: Polarisation reshapes the deprivation hierarchy, which channels displacement, which deepens both the hierarchy and polarisation. The dissertation’s cascade flow index captures the displacement link in this cycle.