London, as viewed in the data, may be seen as a discretely identified area even though

this is not the case. Any migratory changes may occur within that area but also from

outside London into it and vice versa. The problem of this ‘real’ movement and that

created by apparent moves, but really through some form of occupational mobility or

otherwise, are confounding factors that make inferences guarded and also came to be

important concerns in the construction of the multiple regression models. The later

use of the Longitudinal Study was intended as an antidote to this measurement

inaccuracy. Inferences from the data at this level should only be utilised at that level

and not at smaller spatial distributions which would make such inferences less able to

cope with the contextual features of such areas.


The first thing that one notes about the data is the basic rates of change for the

variables as a whole in the three areas defined. Mean ward level changes give an

indication of the average percentage point rates of change for the spatial areas.

The most dramatic initial changes were the increases in unemployment

(likely to be subject to many problems of measurement error) and owner occupation.

These changes would appear to fit the period from which the data was taken in which

council house sales, house sales in general and structural declines in various industries

were taking place. The large increases in unemployment are matched in each area by

decreases in working class occupations although one cannot assume that one accounts

for the other the similarity of the figures would suggest a relationship born out by a

correlation of-0.41 showing that roughly 17% of unemployment is explained by the

decreases in working class occupations for Greater London as a whole (with p less

than 1 in 10,000).

It was interesting to note a fairly clear correspondence between rises in each of the

gentrification categories for each area, becoming more pronounced in inner London,

and general decreases in the displacee variables for London as a whole, again

becoming stronger declines for the inner areas. This seemed to be taking place with

the exception of associated rises for unemployment and for lone parents (later shown

to be positively correlated with increases in professionals). It is plausible to suggest



that lone parents have become more widely distributed across socio-economic groups

which would account for these effects.

Clearly change was more pronounced in inner London. Inner London drew larger

increases in the gentrifier variables and greater reductions in the displacee variables.

This is significant even though a direct link between the two cannot be made, it shows

that the division of the data into these areal subsets does have an effect on both the

relationship between gentrification and displacement and the extent of the observed

changes in the variables. This could be for a number of reasons. The decrease in the

size of the population of inner London over the years makes the increases in the

gentrifier variables all the more significant. While working class groups sharply

dropped in inner London there was also a large mean ward increase in the number of

unemployed which might be accounted for by the large scale structural decrease in

working class employers in the centre of the city.


An analysis of the data re-aggregated into quartiles using professionals as the

grouping variable showed these effects more strongly, see table 5.2 below, but one

can lose any idea of the location of such effects (see appendix A, part two for the full

details of the quartiles). These abstract areas also formed the basis of the LS research

which utilised four areas on the basis of increasing levels of gentrification, defined as

an increase in the number of professionals above the city-wide mean.


‘inner London is defined as the following boroughs; Camden, Hackney, Hammersmith and Fulham,

Haringey, Islington, Kensington and Chelsea, Lambeth, Lewisham, Newham, Southwark, Tower

Hamlets, Wandsworth and the City of Westminster (Same as the OPCS’s classification of inner



One can see the linkages between the displacee variables and the

gentrification variables in the wards when divided into quartiles. The ordering

variable is increases in the percentage points of professionals by ward for Greater

London. The data, ordered in this way, provides a demonstration of an inverse

relationship between the gentrification and displacee variables, except for the lone

parents and unemployed which always appeared to increase where gentrification

increased. There appears to be strong correspondence between increases in all

gentrification variables by quartile, as there is for working class, unskilled, ethnic,

renting and elderly; all of which decrease by ascending quartile.


Each displacee variable’s mean ward percentage point change decreases as the number

of professionals increase (apart from lone parents and unemployed). However, if one

looks at the range for working class it is possible to see that although the arithmetic

mean’s decrease over the quartiles remains relatively stable so that one cannot view

this reduction as being due to the increases in professionals. For renting it would

appear that the second quartile was the most significant ‘location’ for what might be

seen as displacement. The dramatic decreases in working class could be for a number of reasons. First, fewer people may be employed in these occupations. Second, such occupations may have

migrated from the capital to areas with something to offer for such groups. Third, part

of the impact of increasing numbers of professionals may be a displacement of

working class. In reality it is likely that a combination of these factors is responsible.

The quartile analysis was too abstract to be conclusive since the mean ward decreases

masked relatively small differences between each quartile for the displacee variables.

Strictly speaking one could not say that increases in the number of professionals was

the key factor in reducing the frequency of the displacee variables. However, it should

be noted that this is only a crude measurement and one which had little intuition

toward contextual changes. That all of the gentrification variables showed marked

increases and that five out of the seven displacee variables showed decreases can be

considered a starting point in the evidence for a process of replacement and



  1. b) The location of gentrification and displacement in London

Whereas the later correlation and regression models are based upon the use of all of the

data extracted for London as a whole (see later), the analysis of the data in its

geographical distribution utilised an idea of ‘abnormal’ change more strongly. Inner and

Outer London provide artificial boundaries with which the data can be divided whereas

use of the arithmetic mean (with the additional criteria indicated in the methodology)

indicates a division of the total area into a new social geography based upon a notion of

the abnormal growth of particular characteristics. This cut off point provides a well-used

method for establishing the proxy occurrence of gentrification as an event signified by

numeric change over a discrete time period.

Gentrification definitions based upon owner occupation and educated workforce were

left out since professionals were the most accurate variable in the measurement of

gentrification activity and in terms of the variables extraction from the census. This

phase of the analysis used the relationship between gentrification (measured as the

increase above mean, of professionals and managers) and the various displacement

variables. In taking a mean growth for London this left 130 wards which could be

considered to have gentrified to varying degrees, from just above the mean to extensive

gentrification. A criteria of elimination was applied for all wards which appeared to

have gentrified but whose increase in the number of professionals had come about

because of relative decreases in the other groups in that ward – in other words

gentrification could not be seen to be occurring in wards where there were actually

fewer professionals in 1991 than 1981 (A list of these wards and their respective

changes is provided in appendix B).

In simple numbers there were more ‘gentrified’ wards in outer (79) than inner (51)

London, which is an interesting result in itself showing that, in general, the wards with

more than average growth were expanding more in the outer than inner areas of London.


However, if one takes these numbers of gentrified wards and express them as a

percentage of the total number of wards in each area one finds that similar ward areas of

inner and outer London had been gentrified (17.1% and 17.2% respectively). This shows

that the concentration and proportion of gentrification in inner London is actually no

more than that of the outer area. On the other hand, if one looks at the number of

‘gentrified’ wards in a borough like Wandsworth one can see that fully 21 out of 22

wards had gentrified to some degree showing an uneven distribution of gentrification in

the larger areas that comprise London. The social and political centrifugal forces that

shape this distribution are considered further in chapter nine.

Key locations were also observed. In particular the riverside in the Docklands areas

appeared to be heavily gentrified, a result consistent with the other views of that area

(Hall and Ogden, 1992). More unexpected locations were also observed. South east

London revealed a number of wards in which gentrification had taken place and outer

London, as previously noted, had seen dispersed but widespread gentrification by

professionals and managers. Areas that had been further gentrified like Richmond,

Kingston and Blackheath highlighted places that would have been excluded from an

analysis which used a criteria of exclusion since their starting percentages were above

the mean. In other words, areas which were already gentrified or established middle

class areas could be gentrified even further.