- Email: [email protected]
Whatever happened to our land use plan?
Sock-Econ Plan Sri, Vol. 13. pp. 215.783 0 Pergamon Pros Ltd.. 1979 Printed in Great Britain
WHATEVER HAPPENED TO OUR LAND USE PLAN? A METHOD
FOR ANALYSING TRANSFORMATIONS PLANNED LAND USESt RACHELLE
Center
for
IN
ALTERMAN
Urban and Regional Studies, Technion Institute for Research and Development, Architecture and Town Planning, Technion Institute of Technology, Haifa, Israel
(Received
18
and
Faculty
of
August 1978)
Abstract-This paper reports on a study of the degree to which an urban plan has been implemented over an 11-yr period and the transformations that have occurred in planned land uses. The data have been drawn from a case-study of a statutory plan for a medium-size town near Haifa, Israel. Smallest Space Analysis (a multi-variate mapping technique) is applied to the analysis of the propensity for planned land-uses to be changed into other land uses, in the hope of identifying the underlying policy of the decision-makers in approving changes. The results indicate that the substitution of land-uses tends to reflect the traditional concepts of compatability regarding contiguity in location,
1.
INTRODUCTION
With the passage to time, new forces emerge in urban environments which lead to changes in the plans in force. This paper describes the degree to which the urban plan in the case study has been implemented over the years, in the hope of learning something both about the trends of change and about the decision-making policy. The analysis is undertaken with the aid of statistical tools, particularly Smallest Space Analysis-a multivariate “mapping” technique. Local plans in Israel as required by the PIarming and Building Law, l%5,$ are intended to be the major regulative tool for directing and moulding urban development. The Law makes a distinction between outline and detailed plans. An outline plan is akin to a traditional British development plan,! usually specifying permitted land uses in particular zones and covering the :This research has been supported by a Ford Foundation Grant. We acknowledge with thanks the help of Dr. Israel Adler of the Hebrew University and the Center for Applied Social Research who has advised us on the SSA method and its possible applications, fThe Law was the offspring of the Town Planning Ordinance of 1936. Although some significant changes were made in the new law, especially with respect to national land-use planning, for the present purposes the two laws could be regarded as more or less the same. For a description of planning law in Israel, see Gouldman [ I], Strong [Z] and Katin and Virshubsky [3]. §For a description of the British Development plan see for example Reynolds(41 and Keeble ([S], pp. 37-45). Through the 1936 Ordinance Israel has imported the basic concepts and procedures of British town planning. The new Israeli-Law of I%5 still bears marked resemblance to the 1947 British Town and Country Planning Act (Clarke[6]). For a comparison with more recent legal changes see Telling ([7], S2ff). For a statement of the recent approach to urban planning see Solesbury ([8]. pp. 122-135): and Sharpe [9]. fOne hectare = 10 dunams = approx. 2.5 acres. ‘Because detailed plans are, by their very purpose, likely to provide more detailed specifications than the outline plan, the registration of accordances and deviations could not be done mechanicallv It was necessary to construct a set of assumptions on the permitted range within which a difference would be construed as detailing rather than as a deviation. 275
entire municipal area. Detailed plans, on the other hand, designate land uses and site layout more specifically and reflect ongoing decisions. They usually stem from initiatory rather than requlative planning. Normally, in order for a building permit to be issued, the detailed plan submitted earlier must accord with the outline plan-or else an amendment must be approved. The decisionmaking is carried out in a two-tier system, whereby the decisions of the local commission must be approved by the district commission. This legally-instituted procedure has served as a source of data for the purpose of measuring the degree to which the outline plan in the case-study has been effectuated over the years. The case-study is of the Krayot area near Haifa with a population of 70 thousand, a conurbation composed of three municipalities which have long been earmarked for amalgamation and which are therefore designated as a single local planning district with its own local planning commission. The outline (master) plan which was approved in I%4 is a relatively conventional land-use plan. It also specifies, in addition to the
main
road
network
and
the
generally
permitted
uses
in the various areas, the maximum permitted floorarea ratios (regarding residential use) and (often) the permitted maximum building heights. The method of measuring degree of effectuation of the outline plan deserves separate discussion. 2.THE MEASURE OF DEGREE OF EFFECTUATION
The operational measure of effectuation is obtained by means of a comparison of the land use and development specifications of the outline plan with all the detailed or amendment plans submitted in the period 1%4-74. This comparison is carried out by the division of the area into a uniform grid of cells (one hectare each),( and the registration of accordances and deviations by land areas for each category of land-use specifications.” In order to allow for small fluctuations reflecting site conditions, any “reshufling” within a cell was ignored, a deviation being registered only if it falls outside the cell boundaries. This yielded a matrix of “transitions” from the outline plan to the given detailed plan by land use categories. The
RACHELLE ALTERMAN
276
diagonal registers accordances. To get an over-all measure of degree to which a given detailed plan implements the outline plan, the land-areas registering an accordance were cumulated and divided by the total land area.
decision-makers seemed to have attributed to the roads demarcated by the outline plan. Finally, the three categories indicating ecomonic activity (INDUS. LT. INDUS and COMM.) have intermediate accordance with COMM. the lower of the three.9
3. TRANSITION MATRIX OF LAND USES The advantage of the method whereby the data is collected is that it pertains not only to each detailed plan but, because it has preserved the original area-sizes, it may also (through aggregation) cover all areas of the outline plan that have come up for detailed planning. Table 1 shows the transition in planned land uses from the outline plan, to the aggregate of detailed plans (196474). The units are percentages of total area in dunamst for each land use category-i.e. the measure here is unweighted, every type of change being regarded as equal to any other type of change. Table 1 is termed LUSE. Before turning to the analysis of the patterns emerging from the matrix as a whole, it may be worthwhile to dwell on some of the highlights in the data, noting the trends that have occurred in the urban system.
3.3 The pattern of transitions in residential land use If we aggregate all the residential land uses, it may be observed that the overwhelming proportion of areas designated as residential (of one intensity or another), have remained in that use category-as high as 93%. Residential land-use as a whole is also the major recipient of transformations from non-residential categories, as shown in Table 2. In all the categories excepting ROADS and COMM., residential land use accepts comparatively the largest portion of the deviations. In the case of roads and commercial uses, it is the second-largest recipient. It is thus apparent that, contrary to the recommendations of the plan, there is an increasing allocation of the land uses for residential purposes in the Krayot. These findings seem to indicate that the Krayot-which had been planned as a self-sufficient town with its own industrial area and services-is increasingly becoming “suburbanized”, being drawn under the influence of Haifa. The outline plan apparently did not predict these trends and is now finding it difficult to contain the pressures for change. A major recipient of transition has been RES.-Hi; that is, a large portion of the total transitions that have occurred through the years, have been due to conversion to higher-density residential use. RES.-Hi has absorbed as much as 24.6% of the area that had undergone any change. This is not surprising if we note, on the one hand, that housing in Israel is a lucrative investment and, on the other hand, that there has been increasing ac-
3.1 Degree of implementation The over-all degree of accordance in the aggregated LUSE matrix-ACCORD-is 66% of the area (if we exclude the FUTURE category, the number declines somewhat to 63%). It may thus be said that during the 1l-year period covered by the present study, about twothirds of the Krayot area that has fallen under some detailed plan, has followed the outline plan. 3.2 Accordance by land-use categories The pattern of degree of accordance by land use may be of interest, and is as follows (extracted from Table 1):
OPEN SPACE
ROADS
RES.$ -50%
RES. -75%
RES. -87%
RES. -100%
RES. Hi.
PUB. BLDGS.
COMM.
LT. INDUS.
INDUS.
57%
91
36
74
78
64
93
52
60
74
65
Among the various intensities of residential use we find a general trend where the larger the floor-area ratio permitted, the greater is the degree of accordance with the outline plan. The only exception is RES-100%, and that may be due partially to problems of interpreting the intentions of the plan regarding number of floors permitted once land assembly has been undertaken. Among the non-residential uses we find that the two public uses-OPEN SPACE and PUB. BLDGS-have had the greatest encroachment (despite public land ownership), being transformed into uses with higher direct economic returns. The category of ROADS has very high accordance, indicating the importance the
*I dunam = 1000 sq. meters = 0.25 acre (approx.). tThe per cent figures in the RES categories signify the permitted
maximal floor-area ratios in per cent terms,i.e. (total floor space/lot size) X 100. $In the case of INDUS, we regard transformation to LT. INDUS as a permitted lower intensity, especially because the planning administrators do not seem to make meaningful distinctions between the two categories, often placing the latter in areas designated for the former without specifying the substantive difference.
ceptance in the last decade of high-rise apartment blocks (which constitute 30% of the RES-Hi category). Generally, the higher the intensity of the original residential category, the more often is it changed to a higher intensity yet; this is especially notable regarding RES85% and RES.-100%. Among the non-residential uses, PUB. BLDGS. has changed to RES.-Hi most often, with all the implications this substitution is likely to have had for the amount and quality of public services. Another public use category, OPEN SPACE, has shown few transformations to RES.-Hi but rather has tended to change to residential use of middle density or to light industry and public buildings. 4. THE PATTERN OF CHANGES IN LANDUSE MARGINAL TOTALS
4.1 The coeficient
of dissimilarity
Although our measure of effectuation pertains to localized land uses where both the origin and the destination of each cell in the grid are known, it may be of interest to see to what extent the supply of planned land-use areas has increased within the urban region as a whole. This can be achieved by looking at the marginal totals of Table 1. A measure has been proposed for
Out1
6.5
0.2
RES. -87%
INDUS.
I
: are
same
ratio.
-
*FAR = floor
area
: : : :
PUB.BLDGS. COHH. LT. INDUS. FUTURE
Accordance
:
RES.Hi
: : :
23.9
1317
up
75%
to
the
15.6
856
15.8
outline
2.8
0.2
0.4
6.4
are
ratio,
FAR, 3-4 floors. FAR, calculated
for
as
1.6
90
floors.
large
2-3
below:
best
12.9
projects
*
7.3
402
land
use
_lL__!
Accordance
-
transition
IO
m
m 712
632
to
lesser
but
permitted
dunams
5504
452
456 130.5j
60
134
77
124
=
hectare
use
1
100.0
8.2
11.5
8.3
1.1
2.4
1.4
2.3
33.3
13.0
715 1835
7.2
4.2
231 395
7.1
TOTAL PCT
393
TOTAL AREA dunams:?)
a:<
15.0
1.8
( in
INDUS.
174.11
3.3
0.1
22.5
5.9
LT. INDUS.
Permits higher density residential, with no density or height limitations. Public buildings, such as school, community centers, synagogues. Commercial areas. Light Industrial; not operationally distinct from INDUS. To be planned in the future; no designation at present.
87%
m
18.7
le2
0.9
1.3
COW!.
defined
plans,
3.6
200
3.3
152.21
6.5
50% floor-area FAR: 3-4 floors.
definition
in
2.9
161
as above, with 100% Residential, up to with no height limit.
with
further
appearing
needing
those
16.8
926
Residential, as above,
4.1
Those
:
4.4
221
RES. -50% RES.-75% RES.-100% RES. -87%
The categories interpreted.
6.8
Notes
376
AREA
TOTAL PCT
FUTURE
0.2
7.0
LT. INDUS. 1.1
8.3
5.0
5.0
118.2)
COW. 1.8
0.8
0.2
2.5
PUB. BLDG
plans as % of total area in each land use category
3$3@ le7 163.71 36.3
4.2
0.2
RES. Hi
plan to detailed
RES. -100%
24.6
243
m
15.3
11.:
17.0
RES. -07%
L13.01162.31
Fcc
ls4.1)
&-._?r;_x;
24.7
5.3
135.71
2.0
RES. -50%
RES. -75%
land uses from outline
4.5
/xJ
ROADS
in planned
PUB. BLDGS.
RES. Hi
-100%
RES .
5.6
OPEN SPACE
LUSE-transitions
ROADS
plan
plans
I. Matrix
RES. -50% RES. -75%
OPEN SPACE
ine
Detailed
Table
RACHELLEALTERMAN
Fig. I. The Krayout outline plan, 1964-5. Table 2. Transitionsfrom land uses of outline plan to residential use (% area in each I.u. category) LAND in
USE OUTLINE
RESID.
in
OPEN
ROADS
RESID.
1.7
93.0
SPACE
27.8
PUB. BLDGS.
COHH.
24.6
13.3
LT. NDUS,
INDUS.
FUTURE
I
la.9
30.4
34.3
DETAILED PLAN
comparing two percentage distributions, called the Coefficient of Dissimi1arity.t Applied to the present problem, it would read: DISSIM
=
2 lLUSEi i LusEI’J =26,
when
i =
j
i.j = I
where LUSEi and LUSEj are the total area in land use categories i (of the outline plan) and j (of the detailed plans) respectively. Cumulation of the absolute per cent differences in the marginal distributions yields DISSIM = 26.$ The converse is (100 - 26) = 74, which is higher than ACCORD. This means that, if we disregard location of the land use in question but consider only its existence anywhere in the urban area as implementation of the outline plan, then the degree of effectuation aggregated over time is quite high, higher yet (as may be expected) than if we require the land use to locate in the specific place designated. 4.2 Description of the changes in marginal totals The coefficient of dissimilarity shows the tendency for detailed planning to maintain the approximate total area assigned in the outline plan for each land use, providing tExplainedby Matras([9al,pp. M-157) who describes its use in demography. $Here the category FUTURE has been excluded in order to maintain the symmetry-but this makes little difference.
it somewhere else in the urban area. However, there are variations among land use types. A comparison of the marginal totals in Table 1 shows that there has been very little decline in the total area of planned OPEN SPACE. That is, while there has definitely been encroachment on the areas specifically designated as open space in the outline plan (with an accordance of only 57%), this does not register in the land use totals, since other areas have been provided instead. Such additions do not necessarily fulfill the same open-space function: for example, a stadium and an agricultural-school farm are part of the additional area, without having appeared in the outline plan; while open space in areas conceived originally as “green paths” from the seashore inward, have declined. By using the location-specific measurement in the subsequent analysis we have adhered to the plan’s terms-of-reference and have avoided the evaluative question of deciding whether the function envisioned for open space is being fulfilled elsewhere despite the transformations. The area planned for ROADS has remained strikingly the same. There has been a general adherence to the routes of main roads demarcated in the outline plan, with very few additions (and only some small changes in route). The two categories often regarded as “amenities”PUB. BLDGS. and COMM.-have both increased, possibly reflecting the rise in the standard of living and the concomitant change in accepted norms for planning public facilities. At the same time, both these categories
Whatever
happened
have a relatively low degree of location-specific
ACCORD, indicating that some of the area initially in these two categories has yielded to development pressures and has been transformed into other uses. The two types of industrial use show opposite trends. While LT. INDUS. has increased significantly, INDUS. has declined. Because the distinction between the two is not made clear in the plan, it is perhaps more meaningful to add the two together, noting that there has been almost no change in the joint area. The discussion will now return to the full transition matrix. 5. SMALLEST
279
The SSA-I version is applicable to square, symmetric matrices-i.e. the two halves split by the diagonal are the same. As inputs it may take correlations, distances, percentage distributions, or raw frequencies. The rank order comparison among the entries is undertaken over the matrix as a whole. A second version, SSA-II, is applicable to asymmetric matrices i.e. the two halves of the matrix are not the same (although it must be square in this case too). That is, where Pr(alb) # Pr(bla), the monotonicity conditions is:! distance (a, b) < distance (6, c) whenever Pr(aJb) > Pr(blc).
SPACE ANALYSIS OF LAND USE TRANSITIONS
5.1 Introduction to SSA-I and SSA-ZZ The description up to this point has pointed to some of the relationships, without attempting to comprehend the full matrix all at once. That would have been a difficult task to accomplish unaided. But on the basis of the full transition matrix data, we would have liked to obtain answers to questions such as: which land-use categories have a strong “affinity” (in terms of likelihood of transition) for other particular land uses? Do they tend to cluster into groups? Is there some underlying structure? The various versions of Smallest Space Analysis proposed by Louis Guttman, are a set of techniques that may help to provide answers to these questions.? The techniques attempt to fit the data into a Euclidean space of a minimum of dimensions (hopefully two or three so that they can be visually comprehended), while yet maintaining as best as possible the monotonicity condition which, in the case of the SSA-I version is:$ distance (a, 6) corr. (c, d).
tSee for example the exposition in Milton BloombaumllO]: many and varied examples of the application of SSA-I and II are available. E.g.. Mortimer[l I], Laumann and Guttman[lZ] (for SSA-II), Elizur [ 131 and Guttman et al. [ 141. When searching for a suitable method for the present purposes. we scanned literature on social mobility, having found that the nature of the problem has many aspects similar to our own: frequent presentation in the form of a transition matrix: distinction between mobile (deviations) and immobile (accordances): and an inherent rank-order among the occupational or educational categories. This area has also been a subject of many attempts at methodological innovations, from which we hoped to benefit. We found there the use of the Coefficient of Dissimilarity and SSA, both of which have also been applied in the present work (see Blau and Duncan)[lS]. $This criterion applies to data of “similarities”. For data based on “dissimilarities” the signs of inequality are reversed. $For a mathematical presentation of the method where algorithms are derived, see Guttman[l6] and Lingoes and Guttman [ 171. Vn other words, we had to make the assumption that changes from x to y and from y to .r are in some way comparable. Although this assumption is not always reasonable, we feel the method would still be useful, since it provides some answers about the underlying relationship between land uses, a relationship that is not dependent on direction of transition. “The decision-rules are fluid, depending on the extent to which an added dimension substantially improves the solution (see Bloombaum[lO]). Coefficients as high as 0.21 are sometimes still considered acceptable by researchers associated with Guttman (e.g. 1973 book by E. Katz-“Israel’s Leisure Culture”, in Hebrew).
to our land use plan?
The rank-order comparison whereby the distances are calculated is undertaken not over the whole matrix, but rather within each row or column separately. The algorithm attempts to map the best arrangement of points in the minimum number of dimensions, so as to maintain the rank-order of each row as best as possible. In both methods, only off-diagonal elements enter the computation. Both SSA-I and SSA-II will be applied, serving different purposes suited to their differing attributes. SSA-I will be used to describe the structure of the land-use transition matrix; SSA-II will be applied to derive weights for the various transformations. 5.2 SSA-Z analysis of the matrix LUSE The SSA-I technique is suitable for describing the LUSE transition matrix in raw form since it compares each cell entry with all other cells, applying the rankorder criterion over the whole matrix. Because the technique applies to symmetrical matrices, it has been necessary to “fold” the raw transition matrix over the diagonal, summing every pair of opposite cells into a single cell.7 The matrix used here was not the row percentages described in Table 1, but rather the raw numbers matrix (or, equivalently, the percentages of the grand total). The SSA-I solution is reproduced in Fig. 2 for the 2-D space. The measure of goodness of fit, the Coefficient.of Alienation, is 0.19, which is satisfactory.” For the 3-D solution, it is 0.11, considered to be a good fit. It so happens that the 3-D solution maintains the basic structure of the 2-D case, only adding further refinement to it, to be described below. The 3-D solution is reproduced in Fig. 3. From the 2-D solution, we may learn something about the underlying policy of trends of land use transitions. A division down the center of the space may be observed, separating the land use categories along the RESIDENTIAL-NON-RESIDENTIAL dimension (excepting LT. INDUS. which falls in the residential “region”). The mapping also shows a tendency for low-density residential uses to cluster separately, away from the higherdensity residential uses. OPEN SPACE and PUB. BLDGS. are located close to each other, as are INUDS. and COMM. This means that, on the one hand, residential uses generally show “affinity” for each other in terms of likelihood of substitution more than for non-residential uses, and, on the other hand, that there are differences within the two major land-use types, with clustering by density or intensity of use. This general split is maintained in the 3-D solution, but an added dimension appears: the “affinity” of INDUS.,
280
RACHELLE
1
NON-RESIDENTIAL
Ii -___-__---___A$
I _5m.o
-+___-_____---__c
-500 0
0.0 Key:
I -OPEN Z*RES. 3.RES. 4*RES.
/ I
RESIDENTIAL
SPACE - 50% - 75X - 07%
Fig. 2. SSA-I two-space
500.0
5. RES. -100% 6. RES. Hi 7.PUB. BLDGS 8. COMMERCIAL
solution to distribution use transitions.
9.LT. INOUS, O~INDUSTRIAL A. ROADS
of planned
land-
LT. INDUS. and COMM. for higher-density residential (RES.-100% and RESHi), and the affinity of OPEN SPACE and PUB. BLDGS. for low-density residential (RES.-75% and RES.-50%). The ROADS category tends toward separation (because of the scarcity of any transitions) in both the 2-D and the 3-D solutions. It is interesting that the pattern emerging from the SSA analysis of planned land-use transitions strongly reminds one of the traditional notions of land-use planning regarding proximity in location, even though the problem at hand pertains to substitution of one land use for another, rather than to location side-by-side. These
tFor an example of how some of these concepts have found expression in land use planning and urban design, see Keeble[S], pp. 214; 222-224. In his book Siegan[22] attempts to show how, in a city like Houston that has never instituted zoning, the land-use pattern resembles that of zoned cities to a remarkable degree. He argues that this similarity is due to the fact that the traditional notions attributed to zoning are in reality based on economic considerations in location that are made by the individual developers or buyers regardless of whether zoning directs them to do so or not (especially in view of the frequent amendments and variations to the zoning scheme which, he argues, are the attempts to fit the zoning scheme to the exigencies of economic considerations). Whether or not this argument is correct in all respects, the data presented by him comparing zoned and non-zoned cities provide us with clues to understanding the notions underlying land-use planning. However, his argument is best applicable to privately-owned land where development decisions are made on a private basis. Uheories stressing the economic basis of zoning (Siegan[ZZ]; Ross[23]) draw their argument from a context where most development is undertaken by private initiative, on private land, for the sake of furthering private economic objectives: the Israeli scene is in some points almost opposite: much development is publicly initiated on public land, although economic considerations are usually also very strong. Do our findings about the land-use concepts reflect common economic and environmental values? Or do our findings reflect merely the preconceived notions imported from abroad by the professionals or bureaucrats?
The answer
would
require
further
research.
ALTERMAN
notions have been based on assumed “compatability” between uses, having their origins in conceptions about nuisances and economic effects. On the North American planning scene, these concepts have through several decades received legal expression in court decisions about the legitimacy of planning or zoning and have been Delafons[ll]; extensively documented (see, Heyman[l9]; Mandelker[20]; Adler[Zl]; Siegan[22]). On the basis of these concepts, traditional planning has tended, on the one hand, to separate residential from non-residential use, and, on the other hand, to further isolate the various densities of residential use from each other, allowing certain non-residential uses to be located in proximity only to higher density r~esidential uses.? It would appear that the decision-makers and planners have perceived deviation from the outline plan in terms of accepted concepts about desirable and undesirable contiguity of land-uses. They have tended to permit deviations more often and in greater amounts where the distance, so to speak, from one land use to the other in terms of desirability of contiguity in location has been smaller. Likely this tendency reflects the decision-makers’ unarticulated feeling that if the outline plan has designated the original land use as it has, it had done so on the basis of these very concepts of “good” planning, and that they ought to be adhered to as best possible. Although none of the amendments or deviations constituted bona fide attempts at comprehensive re-planning, it seems that the original plan still had some trace of influence on the decisions-even though they were decisions to deivate from it. If our assumptions are correct, one is struck by the similarity in land-use concepts between North America, Britain and Israel, despite the differences among these countries in landownership, public housing, etc.* 6. SSA.11 APPLICATION TO THE PROBLEM OF WEIGHTING LAND USE TRANSITIONS
6.1 The dilemma of assigning weights To this point, it has been assumed that a unit of change in planned land-use from X to Y has the same weight as a unit of change from Y to 2; and that transition from X to Y has the same weight as transition from Y to X. In other words, the deviation has been measured by the sheer size of the area involved, the land units each receiving a weight of 1. The measure presented above for the degree of the plan’s effectuation has this underlying assumption. However, needless to say, not all changes are equivalent, whether in terms of the underlying motivations for the transition, or in terms of their urban impact. For example, deviation from planned open space to industry may not be of the same weight (by whatever criteria) as deviation from open space to residential use. And, furthermore, a change from planned open space to industry is not equivalent to a change from industry to open space. Intuitively, one or two ways of weighting immediately come to mind. One possibility is to arrange the land-uses by assumed intensity of activity. However, this may entail a high degree of arbitrariness. For intensity of activity varies from one situation to the other and is not necessarily an “attribute” of the land-use type as such-excepting perhaps for a few clear relationships, such as between residential densities, which can be thus ranked by definition. Another possibility is to assign weights, or rank-order by the degree
Whatever happened to our land use plan?
Fig. 3. SSA-I
three dimensional
solution
to distribution
of “amenity” attributable to each land use category. However the term “amenity” is often used by planners in a loose fashion even when speaking about a specific site and specific conditions: not to speak of the vagueness of the term when used to pertain to a general relationship between land-uses. Unfortunately, the solution to the problem of assigning weights is quite elusive. For one thing, whatever criteria we choose (others that come to mind are traffic generation, land values, noise levels, etc.), may depend on so many factors that the problem becomes almost unmanageable. These include the supply of the planned and built-up land-use at each point in time; the unique site conditions of every location where a change occurs (a change in one locality may not be equivalent to the same type of change in another locality); the social values, habits, likes and dislikes of various groups of residents and decision-makers: the surrounding land-uses for each instance of change; and more. For another thing, even if we were to choose some sophisticated criteria and overcome the information gap with a great investment of effort, we would still be open to the criticism of having measured a complex phenomenon by one, or two, or three criteria, that, in the final analysis, are subjectively selected by the researcher. To avoid the role of judge of others’ decisions (even if it be a judged armed with information), we decided to focus on the aggregate of decisions about land-use changes made by the appropriate local planning bodies, and to attempt to draw out the set of weights that may underlie these decisions, without subjecting them to any further evaluation. After all, the decision-makers were acquainted with local conditions at each point in time,
of planned
281
land-use
transitions
better than a researcher who has to reconstruct the situation. One may assume that by looking at the macro scale of the decisions (i.e. all detailed plans together), whatever regularity might exist is likely to emerge from the analysis. The derived set of weights would then be projected back to each individual case of decision-making. The rationale underlying this procedure is analogous to the comparison of some local pattern (such as dwelling density) with some average, macro datum which serves as a standard (such as the equivalent national average). 6.2 The SSA-II solution to land-use transitions The SSA-II method is more suited than SSAI to the problem of assigning weights because rather than comparing the rank-order of transformations over the whole matrix, it compares the rank-order within every row. That is, the solution tries to find the optimal arrangement of the points in minimum dimensions, putting together the distributions for each land-use. When looking at the relationship of any given point to the others, one is in fact looking from the “point of view” of each land-use. It is important to adopt this point of view because some land-uses require large areas, while others need only small bits of area; and a transition is not a priori of less importance just because it pertains to a small area. Another advantage of SSA-II for this problem is that it operates not on the “folded” semi-matrix, but on the entire asymmetrical LUSE matrix, attempting to find the pattern that takes into account changes from X to Y and from Y to X separately. The solution with 11 land-use categories (excluding FUTURE) yields an acceptable goodness-of-fit for the
282
RACHELLE
ALTERMAN
I
I
Fig. 4. SSA-II two-space solution of matrix LUSE (with example of ordinal weighting,from
“point of view” of OPEN
SPACE). 2-D
solution (Coefficient of Alienation of 0.14). The 2-D mapping is reproduced in Fig. 4. The weights were obtained thus: each land-use category serving as the starting point in turn, the shortest linear distances were drawn between it and 11 other points: these distances were ranked ordinally, in each case separately, the shortest receiving a score of 1 and the longest a score of 5, intermediate distances being ranked as possible increments of 0.5 (see example in Fig. 4).t This procedure produced a set of rank-ordered scores for each land-use category of the outline plan in combination with each detailed-plan category. The weights served as multipliers of the land-area percentage in each cell of matrix LUSE for every detailed plan. A new variable, called SSAW, was constructed so as to be equivalent to the reciprocal of ACCORD. 6.3 Sensitivity to weighting The sensitivity to the new weights has been checked by comparing the degree of ACCORD (unweighted) with SSAW (weighted) for all cases of detailed plans submitted for approval. The correlation between SSAW and ACCORD is as high as 0.87. The similarity in the results leads to the conclusion that the macro-level of the decisions (which underlies the SSAW measure) and each micro case, generally do not differ too much. Thus the observations made in the unweighted analysis still hold. Nevertheless, the theoretical problem of weighting does exist, and weights can make a difference. Further analysis of other possible methods is beyond the present work, but is undoubtedly desirable.
also distinguished between the degree of deviation with respect to the various land uses, and have found differing patterns. The most striking of these have been the high degree of adherence to the original land allocation for roads which contrasts with the low degree of adherence to land allocation for open space and public areas. Also significant was the preponderance of changes to higherdensity residential use which reflected the trend in the urban system during the IO-yr period of the plan’s effectuation-trends that the guiding plan was apparently unable to contain. From the Smallest Space Analysis we found a remarkable similarity between the “affinity” of one land-use for the other in terms of tendency for transformation, and the accepted concepts of compatability between land-uses regarding desirable or undesirable contiguity in location. This pattern seems to indicate that despite the apparent absence of any conscious awareness of the planning import of a given amendment decision, there is a tendency to adhere to the accepted notions of land-use planning as closely as possible, on the assumption that deviation from the plan would somehow be minimized. REFERENCES
Legal Aspects of Town PIanning in Israel.
I. M. D. Gouldmann,
Hebrew University of Jerusalem, Institute for Research & Comparative Law, “Merkaz” Press 2. Ann Louise
Finland.
Legislative (1%6).
Planned Urban Environments: Sweden, the Netherlands, France. John Hopkins
Strong,
Israel,
Press, Baltimore (1971). 3. Ernest Katin and Mordechai and administration in Israel.
Virshubski, Environmental law Tel Aviv University Studies in
Law I. 197-238 (1975). CONCLUSION
From the analysis we learned that the guiding urban plan has been adhered to with respect to about 60% of the total urban area planned. This is a significant degree of implementation and yet leaves much to be desired. We
+Because of the monotonicity condition. the SSA should be interpreted from an ordinal, rather than an interval point-of-view.
4. Josephine P. Reynolds, The changing objectives of the drawn plan. In F. J. McCulloch et al., Land Use in an Urban Environment, Chap. 6, pp. 151-184. Liverpool University Press (1%5). 5. Lewis Keeble, Principles & Practice of Town & Community Planning. The Estates Gazette Ltd., London (1%9). 6. John J. Clarke, Law Housing and Town Planning, 5th Edn. Pitman, London (1949). 7. A. E. Telling. Planning Law and Procedure, 5th Edn. Butterworth. London (1977).
Whatever happened to our land use plan? 8. William Solesbury, Policy in Urban Planning: Structure Plans, Programs and Local Plans. Pergamon Press, Oxford (1974). 9. L. J. Sharpe, Innovation and change in British land use planning. In Planning Politics and Public Policy: The British, French and Italian Expereince (Ed. by Jack Hayward and Michael Watson), pp. 316357. Cambridge University Press (1975). 9a. Judah Matras, Populations and Societies. Prentice Hall, Englewood Cliffs, New Jersev (re. Statistical methods) (1970).
IO. Milton Bloombaum, Doing smallest space analysis. J. Conflict Resolution 14(3). 409-416 (1970). I I. Jeytan T. Mortimer, Pattern of intergenerational occupational mobility: a smallest space analysis. Am. J. Sociology 79(5), 1278-1299(1974). 12. Edward 0. Laumann and Louis Guttman, The relative associational contiguity of occupations in an urban setting. Am. Sociotog. Reo. 31(2) (April), 169-178. (re. SSA application) (1%6). 13. Dov Elizur, Adapting to Innovation: a Facet Analysis of the Case of the Computer. Jerusalem Academic Press, Jerusalem (re. SSA method) (1970). 14. Ruth Guttman, Louis Guttman and N. A. Rozenzweig, Crossethnic variation in dental, sensory and perceptual traits: a non-metric multivariate derivation of distances for ethnic groups and traits. Am. J. Physical Anthropology 27(3), 259276 (re. SSA application) (1%7).
283
IS. Peter M. Blau and Otis Dudley Duncan, The American Occupational Structure. Wiley, New York (re. Statistical Methods) (1%7). 16. Louis Guttman, A general nonmetric technique for finding the smallest co-ordinate space for a configuration of points. Psychometrica 33(4), 469-506 (1968). 17. James C. Lingoes and Louis Guttman, Nonmetric factor analysis: a rank-reducing alternative to linear factor analysis. hfultivatiate BehavioraiRes. 2(4) (Oct.), 485-505 (1967). 18. John Delafons, Land Use Controls in the United States, M.I.T. Press, Cambridge, Mass. (l%9). 19. I. Michael Heyman, Innovative land regulation and comprehensive planning. The New Zoning: Legal, Administrative and Economic Concepts and Techniques. (Ed. by Norman Marcus and Marlyn W. Groves), pp. 23-37. Praeger, New York (1970). 20. Daniel R. Mandelker, The basic philosophy of zoning: incentive or restraint?. (See Ref. [lq). 21. Gerald M. Adler, Land Planning by Administrative Regulation: the Policies of the Ontario Municipal Board. University of Toronto Press, Toronto (1971). 22. Bernard H. Siegan, Land Use Without Zoning. Lexington Books, Toronto (1972). 23. John M. Ross, Land use controls in metropolitan areas: the failure of zoning and a proposed alternative. Southern Calif. Law Rev. 45, 335-364 (1972).
WHATEVER HAPPENED TO OUR LAND USE PLAN? A METHOD
FOR ANALYSING TRANSFORMATIONS PLANNED LAND USESt RACHELLE
Center
for
IN
ALTERMAN
Urban and Regional Studies, Technion Institute for Research and Development, Architecture and Town Planning, Technion Institute of Technology, Haifa, Israel
(Received
18
and
Faculty
of
August 1978)
Abstract-This paper reports on a study of the degree to which an urban plan has been implemented over an 11-yr period and the transformations that have occurred in planned land uses. The data have been drawn from a case-study of a statutory plan for a medium-size town near Haifa, Israel. Smallest Space Analysis (a multi-variate mapping technique) is applied to the analysis of the propensity for planned land-uses to be changed into other land uses, in the hope of identifying the underlying policy of the decision-makers in approving changes. The results indicate that the substitution of land-uses tends to reflect the traditional concepts of compatability regarding contiguity in location,
1.
INTRODUCTION
With the passage to time, new forces emerge in urban environments which lead to changes in the plans in force. This paper describes the degree to which the urban plan in the case study has been implemented over the years, in the hope of learning something both about the trends of change and about the decision-making policy. The analysis is undertaken with the aid of statistical tools, particularly Smallest Space Analysis-a multivariate “mapping” technique. Local plans in Israel as required by the PIarming and Building Law, l%5,$ are intended to be the major regulative tool for directing and moulding urban development. The Law makes a distinction between outline and detailed plans. An outline plan is akin to a traditional British development plan,! usually specifying permitted land uses in particular zones and covering the :This research has been supported by a Ford Foundation Grant. We acknowledge with thanks the help of Dr. Israel Adler of the Hebrew University and the Center for Applied Social Research who has advised us on the SSA method and its possible applications, fThe Law was the offspring of the Town Planning Ordinance of 1936. Although some significant changes were made in the new law, especially with respect to national land-use planning, for the present purposes the two laws could be regarded as more or less the same. For a description of planning law in Israel, see Gouldman [ I], Strong [Z] and Katin and Virshubsky [3]. §For a description of the British Development plan see for example Reynolds(41 and Keeble ([S], pp. 37-45). Through the 1936 Ordinance Israel has imported the basic concepts and procedures of British town planning. The new Israeli-Law of I%5 still bears marked resemblance to the 1947 British Town and Country Planning Act (Clarke[6]). For a comparison with more recent legal changes see Telling ([7], S2ff). For a statement of the recent approach to urban planning see Solesbury ([8]. pp. 122-135): and Sharpe [9]. fOne hectare = 10 dunams = approx. 2.5 acres. ‘Because detailed plans are, by their very purpose, likely to provide more detailed specifications than the outline plan, the registration of accordances and deviations could not be done mechanicallv It was necessary to construct a set of assumptions on the permitted range within which a difference would be construed as detailing rather than as a deviation. 275
entire municipal area. Detailed plans, on the other hand, designate land uses and site layout more specifically and reflect ongoing decisions. They usually stem from initiatory rather than requlative planning. Normally, in order for a building permit to be issued, the detailed plan submitted earlier must accord with the outline plan-or else an amendment must be approved. The decisionmaking is carried out in a two-tier system, whereby the decisions of the local commission must be approved by the district commission. This legally-instituted procedure has served as a source of data for the purpose of measuring the degree to which the outline plan in the case-study has been effectuated over the years. The case-study is of the Krayot area near Haifa with a population of 70 thousand, a conurbation composed of three municipalities which have long been earmarked for amalgamation and which are therefore designated as a single local planning district with its own local planning commission. The outline (master) plan which was approved in I%4 is a relatively conventional land-use plan. It also specifies, in addition to the
main
road
network
and
the
generally
permitted
uses
in the various areas, the maximum permitted floorarea ratios (regarding residential use) and (often) the permitted maximum building heights. The method of measuring degree of effectuation of the outline plan deserves separate discussion. 2.THE MEASURE OF DEGREE OF EFFECTUATION
The operational measure of effectuation is obtained by means of a comparison of the land use and development specifications of the outline plan with all the detailed or amendment plans submitted in the period 1%4-74. This comparison is carried out by the division of the area into a uniform grid of cells (one hectare each),( and the registration of accordances and deviations by land areas for each category of land-use specifications.” In order to allow for small fluctuations reflecting site conditions, any “reshufling” within a cell was ignored, a deviation being registered only if it falls outside the cell boundaries. This yielded a matrix of “transitions” from the outline plan to the given detailed plan by land use categories. The
RACHELLE ALTERMAN
276
diagonal registers accordances. To get an over-all measure of degree to which a given detailed plan implements the outline plan, the land-areas registering an accordance were cumulated and divided by the total land area.
decision-makers seemed to have attributed to the roads demarcated by the outline plan. Finally, the three categories indicating ecomonic activity (INDUS. LT. INDUS and COMM.) have intermediate accordance with COMM. the lower of the three.9
3. TRANSITION MATRIX OF LAND USES The advantage of the method whereby the data is collected is that it pertains not only to each detailed plan but, because it has preserved the original area-sizes, it may also (through aggregation) cover all areas of the outline plan that have come up for detailed planning. Table 1 shows the transition in planned land uses from the outline plan, to the aggregate of detailed plans (196474). The units are percentages of total area in dunamst for each land use category-i.e. the measure here is unweighted, every type of change being regarded as equal to any other type of change. Table 1 is termed LUSE. Before turning to the analysis of the patterns emerging from the matrix as a whole, it may be worthwhile to dwell on some of the highlights in the data, noting the trends that have occurred in the urban system.
3.3 The pattern of transitions in residential land use If we aggregate all the residential land uses, it may be observed that the overwhelming proportion of areas designated as residential (of one intensity or another), have remained in that use category-as high as 93%. Residential land-use as a whole is also the major recipient of transformations from non-residential categories, as shown in Table 2. In all the categories excepting ROADS and COMM., residential land use accepts comparatively the largest portion of the deviations. In the case of roads and commercial uses, it is the second-largest recipient. It is thus apparent that, contrary to the recommendations of the plan, there is an increasing allocation of the land uses for residential purposes in the Krayot. These findings seem to indicate that the Krayot-which had been planned as a self-sufficient town with its own industrial area and services-is increasingly becoming “suburbanized”, being drawn under the influence of Haifa. The outline plan apparently did not predict these trends and is now finding it difficult to contain the pressures for change. A major recipient of transition has been RES.-Hi; that is, a large portion of the total transitions that have occurred through the years, have been due to conversion to higher-density residential use. RES.-Hi has absorbed as much as 24.6% of the area that had undergone any change. This is not surprising if we note, on the one hand, that housing in Israel is a lucrative investment and, on the other hand, that there has been increasing ac-
3.1 Degree of implementation The over-all degree of accordance in the aggregated LUSE matrix-ACCORD-is 66% of the area (if we exclude the FUTURE category, the number declines somewhat to 63%). It may thus be said that during the 1l-year period covered by the present study, about twothirds of the Krayot area that has fallen under some detailed plan, has followed the outline plan. 3.2 Accordance by land-use categories The pattern of degree of accordance by land use may be of interest, and is as follows (extracted from Table 1):
OPEN SPACE
ROADS
RES.$ -50%
RES. -75%
RES. -87%
RES. -100%
RES. Hi.
PUB. BLDGS.
COMM.
LT. INDUS.
INDUS.
57%
91
36
74
78
64
93
52
60
74
65
Among the various intensities of residential use we find a general trend where the larger the floor-area ratio permitted, the greater is the degree of accordance with the outline plan. The only exception is RES-100%, and that may be due partially to problems of interpreting the intentions of the plan regarding number of floors permitted once land assembly has been undertaken. Among the non-residential uses we find that the two public uses-OPEN SPACE and PUB. BLDGS-have had the greatest encroachment (despite public land ownership), being transformed into uses with higher direct economic returns. The category of ROADS has very high accordance, indicating the importance the
*I dunam = 1000 sq. meters = 0.25 acre (approx.). tThe per cent figures in the RES categories signify the permitted
maximal floor-area ratios in per cent terms,i.e. (total floor space/lot size) X 100. $In the case of INDUS, we regard transformation to LT. INDUS as a permitted lower intensity, especially because the planning administrators do not seem to make meaningful distinctions between the two categories, often placing the latter in areas designated for the former without specifying the substantive difference.
ceptance in the last decade of high-rise apartment blocks (which constitute 30% of the RES-Hi category). Generally, the higher the intensity of the original residential category, the more often is it changed to a higher intensity yet; this is especially notable regarding RES85% and RES.-100%. Among the non-residential uses, PUB. BLDGS. has changed to RES.-Hi most often, with all the implications this substitution is likely to have had for the amount and quality of public services. Another public use category, OPEN SPACE, has shown few transformations to RES.-Hi but rather has tended to change to residential use of middle density or to light industry and public buildings. 4. THE PATTERN OF CHANGES IN LANDUSE MARGINAL TOTALS
4.1 The coeficient
of dissimilarity
Although our measure of effectuation pertains to localized land uses where both the origin and the destination of each cell in the grid are known, it may be of interest to see to what extent the supply of planned land-use areas has increased within the urban region as a whole. This can be achieved by looking at the marginal totals of Table 1. A measure has been proposed for
Out1
6.5
0.2
RES. -87%
INDUS.
I
: are
same
ratio.
-
*FAR = floor
area
: : : :
PUB.BLDGS. COHH. LT. INDUS. FUTURE
Accordance
:
RES.Hi
: : :
23.9
1317
up
75%
to
the
15.6
856
15.8
outline
2.8
0.2
0.4
6.4
are
ratio,
FAR, 3-4 floors. FAR, calculated
for
as
1.6
90
floors.
large
2-3
below:
best
12.9
projects
*
7.3
402
land
use
_lL__!
Accordance
-
transition
IO
m
m 712
632
to
lesser
but
permitted
dunams
5504
452
456 130.5j
60
134
77
124
=
hectare
use
1
100.0
8.2
11.5
8.3
1.1
2.4
1.4
2.3
33.3
13.0
715 1835
7.2
4.2
231 395
7.1
TOTAL PCT
393
TOTAL AREA dunams:?)
a:<
15.0
1.8
( in
INDUS.
174.11
3.3
0.1
22.5
5.9
LT. INDUS.
Permits higher density residential, with no density or height limitations. Public buildings, such as school, community centers, synagogues. Commercial areas. Light Industrial; not operationally distinct from INDUS. To be planned in the future; no designation at present.
87%
m
18.7
le2
0.9
1.3
COW!.
defined
plans,
3.6
200
3.3
152.21
6.5
50% floor-area FAR: 3-4 floors.
definition
in
2.9
161
as above, with 100% Residential, up to with no height limit.
with
further
appearing
needing
those
16.8
926
Residential, as above,
4.1
Those
:
4.4
221
RES. -50% RES.-75% RES.-100% RES. -87%
The categories interpreted.
6.8
Notes
376
AREA
TOTAL PCT
FUTURE
0.2
7.0
LT. INDUS. 1.1
8.3
5.0
5.0
118.2)
COW. 1.8
0.8
0.2
2.5
PUB. BLDG
plans as % of total area in each land use category
3$3@ le7 163.71 36.3
4.2
0.2
RES. Hi
plan to detailed
RES. -100%
24.6
243
m
15.3
11.:
17.0
RES. -07%
L13.01162.31
Fcc
ls4.1)
&-._?r;_x;
24.7
5.3
135.71
2.0
RES. -50%
RES. -75%
land uses from outline
4.5
/xJ
ROADS
in planned
PUB. BLDGS.
RES. Hi
-100%
RES .
5.6
OPEN SPACE
LUSE-transitions
ROADS
plan
plans
I. Matrix
RES. -50% RES. -75%
OPEN SPACE
ine
Detailed
Table
RACHELLEALTERMAN
Fig. I. The Krayout outline plan, 1964-5. Table 2. Transitionsfrom land uses of outline plan to residential use (% area in each I.u. category) LAND in
USE OUTLINE
RESID.
in
OPEN
ROADS
RESID.
1.7
93.0
SPACE
27.8
PUB. BLDGS.
COHH.
24.6
13.3
LT. NDUS,
INDUS.
FUTURE
I
la.9
30.4
34.3
DETAILED PLAN
comparing two percentage distributions, called the Coefficient of Dissimi1arity.t Applied to the present problem, it would read: DISSIM
=
2 lLUSEi i LusEI’J =26,
when
i =
j
i.j = I
where LUSEi and LUSEj are the total area in land use categories i (of the outline plan) and j (of the detailed plans) respectively. Cumulation of the absolute per cent differences in the marginal distributions yields DISSIM = 26.$ The converse is (100 - 26) = 74, which is higher than ACCORD. This means that, if we disregard location of the land use in question but consider only its existence anywhere in the urban area as implementation of the outline plan, then the degree of effectuation aggregated over time is quite high, higher yet (as may be expected) than if we require the land use to locate in the specific place designated. 4.2 Description of the changes in marginal totals The coefficient of dissimilarity shows the tendency for detailed planning to maintain the approximate total area assigned in the outline plan for each land use, providing tExplainedby Matras([9al,pp. M-157) who describes its use in demography. $Here the category FUTURE has been excluded in order to maintain the symmetry-but this makes little difference.
it somewhere else in the urban area. However, there are variations among land use types. A comparison of the marginal totals in Table 1 shows that there has been very little decline in the total area of planned OPEN SPACE. That is, while there has definitely been encroachment on the areas specifically designated as open space in the outline plan (with an accordance of only 57%), this does not register in the land use totals, since other areas have been provided instead. Such additions do not necessarily fulfill the same open-space function: for example, a stadium and an agricultural-school farm are part of the additional area, without having appeared in the outline plan; while open space in areas conceived originally as “green paths” from the seashore inward, have declined. By using the location-specific measurement in the subsequent analysis we have adhered to the plan’s terms-of-reference and have avoided the evaluative question of deciding whether the function envisioned for open space is being fulfilled elsewhere despite the transformations. The area planned for ROADS has remained strikingly the same. There has been a general adherence to the routes of main roads demarcated in the outline plan, with very few additions (and only some small changes in route). The two categories often regarded as “amenities”PUB. BLDGS. and COMM.-have both increased, possibly reflecting the rise in the standard of living and the concomitant change in accepted norms for planning public facilities. At the same time, both these categories
Whatever
happened
have a relatively low degree of location-specific
ACCORD, indicating that some of the area initially in these two categories has yielded to development pressures and has been transformed into other uses. The two types of industrial use show opposite trends. While LT. INDUS. has increased significantly, INDUS. has declined. Because the distinction between the two is not made clear in the plan, it is perhaps more meaningful to add the two together, noting that there has been almost no change in the joint area. The discussion will now return to the full transition matrix. 5. SMALLEST
279
The SSA-I version is applicable to square, symmetric matrices-i.e. the two halves split by the diagonal are the same. As inputs it may take correlations, distances, percentage distributions, or raw frequencies. The rank order comparison among the entries is undertaken over the matrix as a whole. A second version, SSA-II, is applicable to asymmetric matrices i.e. the two halves of the matrix are not the same (although it must be square in this case too). That is, where Pr(alb) # Pr(bla), the monotonicity conditions is:! distance (a, b) < distance (6, c) whenever Pr(aJb) > Pr(blc).
SPACE ANALYSIS OF LAND USE TRANSITIONS
5.1 Introduction to SSA-I and SSA-ZZ The description up to this point has pointed to some of the relationships, without attempting to comprehend the full matrix all at once. That would have been a difficult task to accomplish unaided. But on the basis of the full transition matrix data, we would have liked to obtain answers to questions such as: which land-use categories have a strong “affinity” (in terms of likelihood of transition) for other particular land uses? Do they tend to cluster into groups? Is there some underlying structure? The various versions of Smallest Space Analysis proposed by Louis Guttman, are a set of techniques that may help to provide answers to these questions.? The techniques attempt to fit the data into a Euclidean space of a minimum of dimensions (hopefully two or three so that they can be visually comprehended), while yet maintaining as best as possible the monotonicity condition which, in the case of the SSA-I version is:$ distance (a, 6)
tSee for example the exposition in Milton BloombaumllO]: many and varied examples of the application of SSA-I and II are available. E.g.. Mortimer[l I], Laumann and Guttman[lZ] (for SSA-II), Elizur [ 131 and Guttman et al. [ 141. When searching for a suitable method for the present purposes. we scanned literature on social mobility, having found that the nature of the problem has many aspects similar to our own: frequent presentation in the form of a transition matrix: distinction between mobile (deviations) and immobile (accordances): and an inherent rank-order among the occupational or educational categories. This area has also been a subject of many attempts at methodological innovations, from which we hoped to benefit. We found there the use of the Coefficient of Dissimilarity and SSA, both of which have also been applied in the present work (see Blau and Duncan)[lS]. $This criterion applies to data of “similarities”. For data based on “dissimilarities” the signs of inequality are reversed. $For a mathematical presentation of the method where algorithms are derived, see Guttman[l6] and Lingoes and Guttman [ 171. Vn other words, we had to make the assumption that changes from x to y and from y to .r are in some way comparable. Although this assumption is not always reasonable, we feel the method would still be useful, since it provides some answers about the underlying relationship between land uses, a relationship that is not dependent on direction of transition. “The decision-rules are fluid, depending on the extent to which an added dimension substantially improves the solution (see Bloombaum[lO]). Coefficients as high as 0.21 are sometimes still considered acceptable by researchers associated with Guttman (e.g. 1973 book by E. Katz-“Israel’s Leisure Culture”, in Hebrew).
to our land use plan?
The rank-order comparison whereby the distances are calculated is undertaken not over the whole matrix, but rather within each row or column separately. The algorithm attempts to map the best arrangement of points in the minimum number of dimensions, so as to maintain the rank-order of each row as best as possible. In both methods, only off-diagonal elements enter the computation. Both SSA-I and SSA-II will be applied, serving different purposes suited to their differing attributes. SSA-I will be used to describe the structure of the land-use transition matrix; SSA-II will be applied to derive weights for the various transformations. 5.2 SSA-Z analysis of the matrix LUSE The SSA-I technique is suitable for describing the LUSE transition matrix in raw form since it compares each cell entry with all other cells, applying the rankorder criterion over the whole matrix. Because the technique applies to symmetrical matrices, it has been necessary to “fold” the raw transition matrix over the diagonal, summing every pair of opposite cells into a single cell.7 The matrix used here was not the row percentages described in Table 1, but rather the raw numbers matrix (or, equivalently, the percentages of the grand total). The SSA-I solution is reproduced in Fig. 2 for the 2-D space. The measure of goodness of fit, the Coefficient.of Alienation, is 0.19, which is satisfactory.” For the 3-D solution, it is 0.11, considered to be a good fit. It so happens that the 3-D solution maintains the basic structure of the 2-D case, only adding further refinement to it, to be described below. The 3-D solution is reproduced in Fig. 3. From the 2-D solution, we may learn something about the underlying policy of trends of land use transitions. A division down the center of the space may be observed, separating the land use categories along the RESIDENTIAL-NON-RESIDENTIAL dimension (excepting LT. INDUS. which falls in the residential “region”). The mapping also shows a tendency for low-density residential uses to cluster separately, away from the higherdensity residential uses. OPEN SPACE and PUB. BLDGS. are located close to each other, as are INUDS. and COMM. This means that, on the one hand, residential uses generally show “affinity” for each other in terms of likelihood of substitution more than for non-residential uses, and, on the other hand, that there are differences within the two major land-use types, with clustering by density or intensity of use. This general split is maintained in the 3-D solution, but an added dimension appears: the “affinity” of INDUS.,
280
RACHELLE
1
NON-RESIDENTIAL
Ii -___-__---___A$
I _5m.o
-+___-_____---__c
-500 0
0.0 Key:
I -OPEN Z*RES. 3.RES. 4*RES.
/ I
RESIDENTIAL
SPACE - 50% - 75X - 07%
Fig. 2. SSA-I two-space
500.0
5. RES. -100% 6. RES. Hi 7.PUB. BLDGS 8. COMMERCIAL
solution to distribution use transitions.
9.LT. INOUS, O~INDUSTRIAL A. ROADS
of planned
land-
LT. INDUS. and COMM. for higher-density residential (RES.-100% and RESHi), and the affinity of OPEN SPACE and PUB. BLDGS. for low-density residential (RES.-75% and RES.-50%). The ROADS category tends toward separation (because of the scarcity of any transitions) in both the 2-D and the 3-D solutions. It is interesting that the pattern emerging from the SSA analysis of planned land-use transitions strongly reminds one of the traditional notions of land-use planning regarding proximity in location, even though the problem at hand pertains to substitution of one land use for another, rather than to location side-by-side. These
tFor an example of how some of these concepts have found expression in land use planning and urban design, see Keeble[S], pp. 214; 222-224. In his book Siegan[22] attempts to show how, in a city like Houston that has never instituted zoning, the land-use pattern resembles that of zoned cities to a remarkable degree. He argues that this similarity is due to the fact that the traditional notions attributed to zoning are in reality based on economic considerations in location that are made by the individual developers or buyers regardless of whether zoning directs them to do so or not (especially in view of the frequent amendments and variations to the zoning scheme which, he argues, are the attempts to fit the zoning scheme to the exigencies of economic considerations). Whether or not this argument is correct in all respects, the data presented by him comparing zoned and non-zoned cities provide us with clues to understanding the notions underlying land-use planning. However, his argument is best applicable to privately-owned land where development decisions are made on a private basis. Uheories stressing the economic basis of zoning (Siegan[ZZ]; Ross[23]) draw their argument from a context where most development is undertaken by private initiative, on private land, for the sake of furthering private economic objectives: the Israeli scene is in some points almost opposite: much development is publicly initiated on public land, although economic considerations are usually also very strong. Do our findings about the land-use concepts reflect common economic and environmental values? Or do our findings reflect merely the preconceived notions imported from abroad by the professionals or bureaucrats?
The answer
would
require
further
research.
ALTERMAN
notions have been based on assumed “compatability” between uses, having their origins in conceptions about nuisances and economic effects. On the North American planning scene, these concepts have through several decades received legal expression in court decisions about the legitimacy of planning or zoning and have been Delafons[ll]; extensively documented (see, Heyman[l9]; Mandelker[20]; Adler[Zl]; Siegan[22]). On the basis of these concepts, traditional planning has tended, on the one hand, to separate residential from non-residential use, and, on the other hand, to further isolate the various densities of residential use from each other, allowing certain non-residential uses to be located in proximity only to higher density r~esidential uses.? It would appear that the decision-makers and planners have perceived deviation from the outline plan in terms of accepted concepts about desirable and undesirable contiguity of land-uses. They have tended to permit deviations more often and in greater amounts where the distance, so to speak, from one land use to the other in terms of desirability of contiguity in location has been smaller. Likely this tendency reflects the decision-makers’ unarticulated feeling that if the outline plan has designated the original land use as it has, it had done so on the basis of these very concepts of “good” planning, and that they ought to be adhered to as best possible. Although none of the amendments or deviations constituted bona fide attempts at comprehensive re-planning, it seems that the original plan still had some trace of influence on the decisions-even though they were decisions to deivate from it. If our assumptions are correct, one is struck by the similarity in land-use concepts between North America, Britain and Israel, despite the differences among these countries in landownership, public housing, etc.* 6. SSA.11 APPLICATION TO THE PROBLEM OF WEIGHTING LAND USE TRANSITIONS
6.1 The dilemma of assigning weights To this point, it has been assumed that a unit of change in planned land-use from X to Y has the same weight as a unit of change from Y to 2; and that transition from X to Y has the same weight as transition from Y to X. In other words, the deviation has been measured by the sheer size of the area involved, the land units each receiving a weight of 1. The measure presented above for the degree of the plan’s effectuation has this underlying assumption. However, needless to say, not all changes are equivalent, whether in terms of the underlying motivations for the transition, or in terms of their urban impact. For example, deviation from planned open space to industry may not be of the same weight (by whatever criteria) as deviation from open space to residential use. And, furthermore, a change from planned open space to industry is not equivalent to a change from industry to open space. Intuitively, one or two ways of weighting immediately come to mind. One possibility is to arrange the land-uses by assumed intensity of activity. However, this may entail a high degree of arbitrariness. For intensity of activity varies from one situation to the other and is not necessarily an “attribute” of the land-use type as such-excepting perhaps for a few clear relationships, such as between residential densities, which can be thus ranked by definition. Another possibility is to assign weights, or rank-order by the degree
Whatever happened to our land use plan?
Fig. 3. SSA-I
three dimensional
solution
to distribution
of “amenity” attributable to each land use category. However the term “amenity” is often used by planners in a loose fashion even when speaking about a specific site and specific conditions: not to speak of the vagueness of the term when used to pertain to a general relationship between land-uses. Unfortunately, the solution to the problem of assigning weights is quite elusive. For one thing, whatever criteria we choose (others that come to mind are traffic generation, land values, noise levels, etc.), may depend on so many factors that the problem becomes almost unmanageable. These include the supply of the planned and built-up land-use at each point in time; the unique site conditions of every location where a change occurs (a change in one locality may not be equivalent to the same type of change in another locality); the social values, habits, likes and dislikes of various groups of residents and decision-makers: the surrounding land-uses for each instance of change; and more. For another thing, even if we were to choose some sophisticated criteria and overcome the information gap with a great investment of effort, we would still be open to the criticism of having measured a complex phenomenon by one, or two, or three criteria, that, in the final analysis, are subjectively selected by the researcher. To avoid the role of judge of others’ decisions (even if it be a judged armed with information), we decided to focus on the aggregate of decisions about land-use changes made by the appropriate local planning bodies, and to attempt to draw out the set of weights that may underlie these decisions, without subjecting them to any further evaluation. After all, the decision-makers were acquainted with local conditions at each point in time,
of planned
281
land-use
transitions
better than a researcher who has to reconstruct the situation. One may assume that by looking at the macro scale of the decisions (i.e. all detailed plans together), whatever regularity might exist is likely to emerge from the analysis. The derived set of weights would then be projected back to each individual case of decision-making. The rationale underlying this procedure is analogous to the comparison of some local pattern (such as dwelling density) with some average, macro datum which serves as a standard (such as the equivalent national average). 6.2 The SSA-II solution to land-use transitions The SSA-II method is more suited than SSAI to the problem of assigning weights because rather than comparing the rank-order of transformations over the whole matrix, it compares the rank-order within every row. That is, the solution tries to find the optimal arrangement of the points in minimum dimensions, putting together the distributions for each land-use. When looking at the relationship of any given point to the others, one is in fact looking from the “point of view” of each land-use. It is important to adopt this point of view because some land-uses require large areas, while others need only small bits of area; and a transition is not a priori of less importance just because it pertains to a small area. Another advantage of SSA-II for this problem is that it operates not on the “folded” semi-matrix, but on the entire asymmetrical LUSE matrix, attempting to find the pattern that takes into account changes from X to Y and from Y to X separately. The solution with 11 land-use categories (excluding FUTURE) yields an acceptable goodness-of-fit for the
282
RACHELLE
ALTERMAN
I
I
Fig. 4. SSA-II two-space solution of matrix LUSE (with example of ordinal weighting,from
“point of view” of OPEN
SPACE). 2-D
solution (Coefficient of Alienation of 0.14). The 2-D mapping is reproduced in Fig. 4. The weights were obtained thus: each land-use category serving as the starting point in turn, the shortest linear distances were drawn between it and 11 other points: these distances were ranked ordinally, in each case separately, the shortest receiving a score of 1 and the longest a score of 5, intermediate distances being ranked as possible increments of 0.5 (see example in Fig. 4).t This procedure produced a set of rank-ordered scores for each land-use category of the outline plan in combination with each detailed-plan category. The weights served as multipliers of the land-area percentage in each cell of matrix LUSE for every detailed plan. A new variable, called SSAW, was constructed so as to be equivalent to the reciprocal of ACCORD. 6.3 Sensitivity to weighting The sensitivity to the new weights has been checked by comparing the degree of ACCORD (unweighted) with SSAW (weighted) for all cases of detailed plans submitted for approval. The correlation between SSAW and ACCORD is as high as 0.87. The similarity in the results leads to the conclusion that the macro-level of the decisions (which underlies the SSAW measure) and each micro case, generally do not differ too much. Thus the observations made in the unweighted analysis still hold. Nevertheless, the theoretical problem of weighting does exist, and weights can make a difference. Further analysis of other possible methods is beyond the present work, but is undoubtedly desirable.
also distinguished between the degree of deviation with respect to the various land uses, and have found differing patterns. The most striking of these have been the high degree of adherence to the original land allocation for roads which contrasts with the low degree of adherence to land allocation for open space and public areas. Also significant was the preponderance of changes to higherdensity residential use which reflected the trend in the urban system during the IO-yr period of the plan’s effectuation-trends that the guiding plan was apparently unable to contain. From the Smallest Space Analysis we found a remarkable similarity between the “affinity” of one land-use for the other in terms of tendency for transformation, and the accepted concepts of compatability between land-uses regarding desirable or undesirable contiguity in location. This pattern seems to indicate that despite the apparent absence of any conscious awareness of the planning import of a given amendment decision, there is a tendency to adhere to the accepted notions of land-use planning as closely as possible, on the assumption that deviation from the plan would somehow be minimized. REFERENCES
Legal Aspects of Town PIanning in Israel.
I. M. D. Gouldmann,
Hebrew University of Jerusalem, Institute for Research & Comparative Law, “Merkaz” Press 2. Ann Louise
Finland.
Legislative (1%6).
Planned Urban Environments: Sweden, the Netherlands, France. John Hopkins
Strong,
Israel,
Press, Baltimore (1971). 3. Ernest Katin and Mordechai and administration in Israel.
Virshubski, Environmental law Tel Aviv University Studies in
Law I. 197-238 (1975). CONCLUSION
From the analysis we learned that the guiding urban plan has been adhered to with respect to about 60% of the total urban area planned. This is a significant degree of implementation and yet leaves much to be desired. We
+Because of the monotonicity condition. the SSA should be interpreted from an ordinal, rather than an interval point-of-view.
4. Josephine P. Reynolds, The changing objectives of the drawn plan. In F. J. McCulloch et al., Land Use in an Urban Environment, Chap. 6, pp. 151-184. Liverpool University Press (1%5). 5. Lewis Keeble, Principles & Practice of Town & Community Planning. The Estates Gazette Ltd., London (1%9). 6. John J. Clarke, Law Housing and Town Planning, 5th Edn. Pitman, London (1949). 7. A. E. Telling. Planning Law and Procedure, 5th Edn. Butterworth. London (1977).
Whatever happened to our land use plan? 8. William Solesbury, Policy in Urban Planning: Structure Plans, Programs and Local Plans. Pergamon Press, Oxford (1974). 9. L. J. Sharpe, Innovation and change in British land use planning. In Planning Politics and Public Policy: The British, French and Italian Expereince (Ed. by Jack Hayward and Michael Watson), pp. 316357. Cambridge University Press (1975). 9a. Judah Matras, Populations and Societies. Prentice Hall, Englewood Cliffs, New Jersev (re. Statistical methods) (1970).
IO. Milton Bloombaum, Doing smallest space analysis. J. Conflict Resolution 14(3). 409-416 (1970). I I. Jeytan T. Mortimer, Pattern of intergenerational occupational mobility: a smallest space analysis. Am. J. Sociology 79(5), 1278-1299(1974). 12. Edward 0. Laumann and Louis Guttman, The relative associational contiguity of occupations in an urban setting. Am. Sociotog. Reo. 31(2) (April), 169-178. (re. SSA application) (1%6). 13. Dov Elizur, Adapting to Innovation: a Facet Analysis of the Case of the Computer. Jerusalem Academic Press, Jerusalem (re. SSA method) (1970). 14. Ruth Guttman, Louis Guttman and N. A. Rozenzweig, Crossethnic variation in dental, sensory and perceptual traits: a non-metric multivariate derivation of distances for ethnic groups and traits. Am. J. Physical Anthropology 27(3), 259276 (re. SSA application) (1%7).
283
IS. Peter M. Blau and Otis Dudley Duncan, The American Occupational Structure. Wiley, New York (re. Statistical Methods) (1%7). 16. Louis Guttman, A general nonmetric technique for finding the smallest co-ordinate space for a configuration of points. Psychometrica 33(4), 469-506 (1968). 17. James C. Lingoes and Louis Guttman, Nonmetric factor analysis: a rank-reducing alternative to linear factor analysis. hfultivatiate BehavioraiRes. 2(4) (Oct.), 485-505 (1967). 18. John Delafons, Land Use Controls in the United States, M.I.T. Press, Cambridge, Mass. (l%9). 19. I. Michael Heyman, Innovative land regulation and comprehensive planning. The New Zoning: Legal, Administrative and Economic Concepts and Techniques. (Ed. by Norman Marcus and Marlyn W. Groves), pp. 23-37. Praeger, New York (1970). 20. Daniel R. Mandelker, The basic philosophy of zoning: incentive or restraint?. (See Ref. [lq). 21. Gerald M. Adler, Land Planning by Administrative Regulation: the Policies of the Ontario Municipal Board. University of Toronto Press, Toronto (1971). 22. Bernard H. Siegan, Land Use Without Zoning. Lexington Books, Toronto (1972). 23. John M. Ross, Land use controls in metropolitan areas: the failure of zoning and a proposed alternative. Southern Calif. Law Rev. 45, 335-364 (1972).