IDPR, 26 (3) 2004

Jennifer Davis

Scaling up urban upgrading

Where are the bottlenecks?

This paper discusses the challenges to expanding successful, small-scale urban upgrading initiatives in
developing countries, using empirical data from the Slum Networking Project (SNP) in Ahmedabad,
India. Despite demonstrated success in providing infrastructure and community development services in
27 slum communities, the SNP has had difficulty expanding its activities beyond this small proportion of
the city’s 2,431 low-income neighbourhoods. Our case analysis suggests that it is neither insufficient
funding nor a lack of political will – both of which are regularly identified as reasons why successful
upgrading programmes are not scaled up – that has impeded progress in the Ahmedabad case. Instead,
a mismatch between the bundle of services offered by the Municipal Corporation and the priorities of
slum households, along with the terms of the partnership between the corporation and the SNP’s NGO
partners, are among the principal explanations for the SNP’s slow rate of expansion.

A persistent challenge for cities in the developing world is devising effective, sustain-
able strategies for supplying low-income households with reliable infrastructure
services. Poor neighbourhoods often pose technical challenges for infrastructure
investment because of their density and location on marginal lands. Affordability
concerns, particularly given the current emphasis among development agencies on
greater cost recovery through user fees, are paramount in low-income areas. Many
sector professionals are also convinced that effective service delivery to poor
neighbourhoods requires the skills of social intermediation professionals who can
facilitate communication between households and service providers. In short, urban
service providers face some of their greatest challenges and resource needs in neigh-
bourhoods where revenue potential is thought to be the most limited (UNHSP, ).
Given these challenges, it is perhaps not surprising that, despite several decades of
efforts to improve conditions in poor urban areas, roughly a third of all urban
households are still considered to consist of slum-dwellers (UNHSP, ). Although
a large and growing best practice literature suggests that effective upgrading efforts do
exist, the majority of such programmes are time-bound efforts that benefit only a
small segment of those in need of improved services (Banes et al., ; Dua, ;

Jennifer Davis is Assistant Professor of Urban Studies and Planning in the Department of Urban Studies and Planning,
Massachusetts Institute of Technology,  Massachusetts Avenue, Cambridge, MA , USA; email: jd@mit.edu

Paper submitted February ; revised paper received July  and accepted August .
 The term ‘slum’ is employed in this article as an administrative classification (as in Indian municipalities),
referring to a wide variety of informal housing settlements that have unclear land tenure, limited infrastructure
and housing structures that are generally constructed with temporary or non-permanent materials.

IDPR26_3_04_Davis 305 1/11/04, 2:31 pm

 306 Jennifer Davis

Hasan, ; Kessides, ; Otiso, ). Current development endeavours such as
the Millennium Development Goals and the World Bank/UNCHS Cities Alliance
stress the need to ‘focus on scaling up successful approaches’ rather than devising new
approaches; effective strategies that ‘turn what works for , people into a successful
programme for ,, then  million, then  million’ are needed (Wolfensohn,
quoted in Davis and Iyer, ).
Frequent references to the challenges of scaling up exist in the urban upgrading
literature; the most common explanations for the failure to expand successful projects
are insufficient resources (primarily funding from government and/or international
agencies) and a lack of political will. Inadequate funding has long been at the centre
of scaling-up concerns for upgrading, including the sites-and-services projects of the
s and s (World Bank, ; Cohen, ). Such concerns persist today;
Olanrewaju (), for example, identifies uncertain funding as the principal reason
for the stalling of upgrading efforts in  low-income settlements in Lagos. Even
where funding is stretched to accomplish more, for example, through relaxed tech-
nical standards or the use of labour contributions, financial resources are still often
considered wanting. Reviewing upgrading programmes in sub-Saharan Africa,
Fekade () notes that in spite of ‘increasingly adaptive realistic standards’,
initiatives had limited impact because bi- and multilateral support organisations
‘failed to sustain funding’.
Resource constraints may also relate to human capacity (technical, administrative,
social intermediation, etc.), organisational or institutional capacity, supply chains,
financing, land availability, or other limitations that render a smaller, successful
programme unfeasible at a larger scale. The World Bank’s sites-and-services projects,
for example, fell short of benefiting their millions of target beneficiaries in part
because insufficient land was available for the self-help construction programmes
(Werlin, ). Gough () documents deficiencies in building materials and labour
as hampering progress in self-help housing construction in Pereira, Colombia.
Resistance to scaling up – the unwillingness of key stakeholders to support the
expansion or ‘mainstreaming’ of an effective upgrading initiative – is often implicated
in after-the-fact analyses of slum upgrading initiatives. Often, this resistance is related
to improving the security of tenure for poor households. What is termed ‘land un-
availability’ may in reality be the unwillingness of some stakeholders to undertake
reforms that would improve access by households. As Farvacque and McAuslan (,
) note, ‘politicians, senior public servants, traditional rulers, existing landowners’
and ‘[a]rmies of lower and middle civil servants, have a substantial stake in the
 Following the World Bank (), ‘slum upgrading’ is defined in a limited way as the systematic improvement of
the physical environment in slums. Slum upgrading generally includes improving and/or installing infrastructure
such as water supply, sanitation and storm drainage networks, as well as waste collection systems, access roads
and footpaths, lighting, public telephones, etc.

IDPR26_3_04_Davis 306 1/11/04, 2:31 pm

 Scaling up urban upgrading 307

confused, multi-layered and irrational systems’ of land management in developing

countries. In addition, elected officials may discourage beneficiary participation in
upgrading programmes in an effort to retain power among their constituents.
This paper discusses the challenges that have obstructed the progress of an
upgrading programme in western India – the Slum Networking Project (SNP) of
Ahmedabad – which notably do not include a lack of financial resources or
significant resistance from local government officials. Instead, a mismatch between
the bundle of services offered by the project and the priorities and preferences of
slum households, along with the terms of the partnership between the project and its
NGO partners, are among the principal explanations for the SNP’s slow rate of
expansion to date. The following section presents information about the SNP and its
objectives. The failure of the SNP to scale up in Ahmedabad, despite its well-
publicised success, is then analysed. The final section summarises the study findings
and conclusions.

Study site and methodology

The SNP in Ahmedabad is an urban upgrading initiative that has generally been
considered a success story, both locally and in international development circles. The
implementing agency of the SNP, the Ahmedabad Municipal Corporation (AMC), is
well acquainted with the challenge of providing infrastructure services to the poor.
Once called ‘the Manchester of the East’, Ahmedabad’s economy was built on textile
production until the industry declined rapidly in the s. During this same period,
the AMC almost doubled its service area (from . square kilometres to .
square kilometres) to include an eastern periphery that was largely unplanned and
lacked physical infrastructure. During this economic downturn and the city’s rapid
expansion, the proportion of residents living in low-income slums and chaals
increased dramatically. By , a total of , slums and , chaals had been
documented in Ahmedabad, comprising over  per cent of the city’s . million
residents (ASAG, ).
Since its beginnings in , the SNP has extended a bundle of services to roughly
, low-income residents in Ahmedabad’s slums. Residents have organised into
community-based organisations (CBOs), made cash contributions towards the costs
of the project, and worked with AMC and NGO staff to upgrade their communities
from slums into ‘colonies’. The SNP (also called ‘Parivartan’) pairs technical staff
from the AMC with NGOs and private corporations to offer a set of infrastructural,
 The project emerged from similar experiences in the neighbouring city of Indore, but with significant changes to
several project elements. See Verma () for details of the project’s history.
 In Ahmedabad – as in most Indian cities – the term chaal is used to denote a particular form of multi-storey or row
housing tenement constructed by firms for their workers, usually with permanent or semi-permanent materials.

IDPR26_3_04_Davis 307 1/11/04, 2:31 pm

 308 Jennifer Davis

Table 1 Cost sharing in the Slum Networking Project, Ahmedabad Municipal Corporation

Per-household costs* AMC NGO Household

On-site infrastructure costs: $199 $111 $44 $44

Community development activities: $22† $15 $7 –
Networking to main infrastructure: $100 $100 – –
Total cost: $321 $226 $51 $44
(70%) (16%) (14%)

Notes: * All figures are in US dollars; † Many NGO staff feel that their costs for community development activities are
actually higher than those stipulated in the project agreement.

environmental and social services to slum households at affordable prices. The

services offered by the SNP include private water supply and sewerage connections,
private toilets, paving of streets and alleys, storm-water drainage, community rubbish
bins and street lighting, and neighbourhood tree planting. The project’s costs are met
using subsidies from federal and state programmes, as well as from the AMC itself,
along with contributions from households and NGOs (see Table ). Three NGOs are
currently involved in the project: the Self-Employed Women’s Association (SEWA),
the Mahila Housing SEWA Trust (MHT), and Saath.
Between June  and January , we used a variety of primary data collection
techniques to obtain information about the SNP. A team of Massachusetts Institute
of Technology (MIT) faculty and graduate students completed more than  semi-
structured interviews with staff at the AMC and its partner NGOs, as well as with
local politicians, academics and development professionals familiar with infrastructure
service provision in Ahmedabad’s slums. In addition, we partnered a local NGO that
was not an SNP partner to develop and administer a household survey and a series of
focus groups in selected slum neighbourhoods throughout the city. The MIT team
trained approximately two dozen facilitators and enumerators from Ahmedabad
who, in turn, carried out the data collection activities with household members.
We completed  surveys with households in eight slums that have participated in
the Parivartan programme, and  surveys with households in nine slums that have
not enrolled in the programme. We also completed a rapid appraisal of conditions in
 Whereas the SNP was initially envisioned as a partnership between the public, private and NGO sectors, in
reality only one private company has been involved in the initiative, and in only one slum community.
 Saath is not an acronym; it is a word that from Hindi or Gujarati can be translated as ‘together’, ‘cooperation’ or
‘support’.
 In each survey cohort, households were selected to form a sample that was representative with respect to the
proportion of respondents residing in each of Ahmedabad’s five administrative zones. No other stratification was
employed in our sampling strategy.

IDPR26_3_04_Davis 308 1/11/04, 2:31 pm

 Scaling up urban upgrading 309

 other non-Parivartan slum communities. Focus group meetings were also held in
each community selected for inclusion in the household survey in order to gather
background information on the settlement, including the history of infrastructure
development, as well as qualitative information about the community’s interactions
with AMC staff. Our surveys and focus group discussion guides were developed and
pre-tested in consultation with the partner NGO staff and CBO activists; all data
collection was carried out using local languages. Respondents in Parivartan
communities were asked about the quality of their infrastructure services prior to and
following participation in the programme; the performance of AMC staff; and socio-
economic and demographic information about their household members. Respond-
ents in non-Parivartan communities were asked about the performance of existing
infrastructure services and the perceptions of AMC staff; their interest in enrolling in
the SNP; and the socio-economic and demographic characteristics of their households.
The results of our data collection activities in Parivartan slums suggest that,
among the communities where it has been implemented, the SNP has delivered
success in many forms. For example, before the upgrading activities more than  per
cent of households in Parivartan communities relied on shared public taps or bore-
wells that delivered an estimated  litres per capita per day. At the time of our survey,
 per cent of respondents used individual or shared piped water connections, and
estimated that their water use had increased by approximately  per cent. More than
 per cent of respondents in Parivartan neighbourhoods also said that, prior to their
enrolment in the project, their households had access to no formal sanitation facility,
and instead used open areas and drains to relieve themselves. Forty-five per cent of
respondents said that their principal sanitation facility before the upgrading project
was used by more than  people. By contrast, at the time of our survey  per cent
of respondents used private or shared household toilet facilities with sewer
connections. Of these, more than  per cent gave the cleanliness, privacy and quality
of plumbing in their facilities the highest rating of ‘good’.
Parivartan households also expressed a high level of satisfaction with the services
provided through the programme, compared both with their recollections of public
service provision before the project and with non-Parivartan households. More than
three-quarters of respondents in Parivartan communities reported being ‘very
satisfied’ with their current water supply and sanitation services, as compared with 
per cent who said they were also very satisfied before the project and  per cent of
respondents in non-Parivartan slums (Fig. ). The project has also improved the repu-
tation of AMC staff among residents. Among Parivartan respondents, more than 
per cent said they felt that AMC staff are ‘helpful’, ‘polite’ and ‘honest’, significantly
higher than the  per cent of residents in non-Parivartan communities.
 Verma (), however, challenges the notion that the slum networking approach has been successful in the
neighbouring city of Indore, where it won a ‘best practice’ citation from the UN Centre for Human Settlements.

IDPR26_3_04_Davis 309 1/11/04, 2:31 pm

 310 Jennifer Davis

Figure1 Satisfaction with water and sanitation services among Parivartan, and non-Parivartan
households in Ahmedabad

Despite the early success of the SNP, the progress of upgrading Ahmedabad’s
slums has stalled in recent years. Between its inception in  and our visit in ,
the SNP had either completed work or is active in approximately  of Ahmedabad’s
, slums and chaals. This accomplishment is laudable, yet is far short of the target
initially set by the AMC, which was to reach all of Ahmedabad’s , slum
households by the year . The following sections seek to explain this disappointing
progress in upgrading through the SNP.

Resource bottlenecks
Whereas insufficient resources are frequently cited as the reason that slum upgrading
programmes are ended or stall before they reach a substantial proportion of the
urban poor, the SNP has not wanted for financial resources. In the mid-s, the
AMC initiated a set of financial reforms, including simplifying the assessment of both
octroi and property tax, as well as improving tax collection efficiencies. In ,
Ahmedabad also became the first Indian municipality to float a municipal bond
without a state guarantee. At the close of the – fiscal year, the AMC recorded a
surplus of  million rupees (US$. million), an amount that has grown steadily
since  (AMC, ). Not only has the AMC had ample resources to support the
SNP, it has increased its budget for the SNP each year since its inception. Indeed, the
SNP has not been able to use more than  per cent of its allocated budget from the
AMC in any given year (Table ).
With respect to human capital, the project does suffer from under-resourcing
among its engineers. The Slum Networking Cell, the AMC department charged with
 Octroi tax is levied on goods moving in and out of a local government’s jurisdiction. Local governments in India
rely heavily on octroi tax for revenues, although recent efforts to phase out the tax have emerged at the national
level. In Ahmedabad, approximately  per cent of the AMC’s revenues are derived from octroi.

IDPR26_3_04_Davis 310 1/11/04, 2:31 pm

 Scaling up urban upgrading 311

Table 2 Annual budget allocation from AMC and funds utilisation

Year Allocated budget % increase Expenditure % utilised

1996–97 $44,400 – $1,600 4

1997–98 $240,200 441 $36,600 15
1998–99 $1,024,900 327 $283,800 28
1999–00 $1,155,400 13 $254,600 22
2000–01 $1,679,000 45 not available not available

Source: SNP. Figures are approximate.

administering the programme, has no senior engineers able to draft designs for the
often complex infrastructure installations that are required in participating slums.
Instead, the AMC has contracted a private engineering firm to prepare design
documents for the programme; SNP engineers oversee construction in the slums
according to these designs. From the perspective of resource constraints, however,
there is no shortage of highly skilled private-sector talent to fill this deficit of design
skills within the AMC.
The SNP is also fortunate to have three NGO partners with qualified staff, good
track records and a high level of credibility in Ahmedabad. Saath is a young
organisation originally focused on health programmes but with a growing interest in
urban infrastructure. Leaders of Saath are keenly aware that their performance in the
SNP will establish its reputation in the field of slum upgrading, with important
implications for its funding potential. The MHT is part of the well-known SEWA
network of organisations and, therefore, benefits from the legitimacy that SEWA has
earned from years of organising poor women workers in India (Dhatta, ;
Crowell, ; Rose, ). SEWA and the MHT have sophisticated leaderships, as
well as large and well-trained staffs. In particular, SEWA’s experience with community
savings programmes has made it the financial intermediary between participating
slums and the AMC.

Resistance from key stakeholders

There are many actors in Ahmedabad’s political and bureaucratic institutions who
might have cause to obstruct the progress of the SNP. For example, the urban poor
provide an enormous potential voting block, and the SNP could be viewed as
undermining the influence of elected officials in Ahmedabad. Given the extensive
publicity that the SNP has received, as well as its potential for attracting additional
donor funding to the city, it is perhaps not surprising that we documented very little
overt resistance to the programme from the city’s elected officials. Staff of the partner

IDPR26_3_04_Davis 311 1/11/04, 2:31 pm

 312 Jennifer Davis

NGOs were fairly uniform in their reports that, at worst, elected officials ignore the
programme; more often, they appear at ‘ribbon cutting’ events after months of
preliminary work by residents, community organisations and NGOs. Some observers
note that the relative lack of politicisation of the SNP may be related to the fact that
the programme was initiated during a period of administrator’s rule in Ahmedabad,
during which the elected arm of the AMC was suspended.
At the same time, the majority of Ahmedabad’s  councillors have funds
allocated to them from numerous state and federal government welfare and poverty
alleviation programmes. These funds are earmarked specifically for infrastructure
upgrading in slums and chaals. The value of political grants for infrastructure invest-
ments disbursed in Ahmedabad in the – fiscal year (US$,,) was .
times that of the SNP’s budget and . times the SNP’s expenditures. Unlike the
SNP, most elected officials require little or no contribution of labour or money from
households when infrastructure investments are provided. As such, the potential exists
for such politicians to undermine the scaling up of Parivartan by offering alternative
sources of infrastructure finance to slum communities.

Municipal civil servants

Like many South Asian bureaucracies, the AMC is a solidly hierarchical organisation
with strong norms of duty and deference. Middle-ranking staff are keen to push the
SNP forwards, in large part because a former municipal commissioner expressed a
personal interest in its success. At the same time, this senior tier of the AMC is also
motivated by the possibility of securing external grants and loans for urban up-
grading. The AMC leadership has thus compelled SNP staff to provide infrastructure
services in as many settlements as possible, as quickly as possible.
Somewhat more surprising is the enthusiasm with which lower-level staff at the
SNP have responded to this mandate. In the colony of Maachipir, for example,
community members reported that SNP engineers worked around the clock to
implement the project in their community. During a malfunctioning of one of the
colony’s drainage lines, SNP staff worked until the early hours of the morning to
correct the technical faults. Maachipir residents report that they have never witnessed
such commitment from AMC staff in the many years that they have been engaged
with ward and zone office engineers. Given the fact that, as in most public service
agencies in South Asia, few material incentives are available to senior AMC officials
to encourage good performance by employees, one wonders what it is that motivates
project engineers in the SNP. We have documented elsewhere (Davis et al., ) the
organisational arrangements and incentive structures that motivate SNP staff. This
effort has resulted in a markedly better reputation for SNP staff among slum-dwellers
compared with their colleagues in other AMC departments (Table ).

IDPR26_3_04_Davis 312 1/11/04, 2:31 pm

 Scaling up urban upgrading 313

Table 3 Attitudes towards AMC staff among Parivartan and non-Parivartan communities

Parivartan Non-Parivartan
(n =177) (n =147)

Percentage of respondents Before

who found AMC staff to be: project Currently Currently
Helpful 36% 95% 25%
Polite 39% 95% 32%
Honest 49% 91% 30%

Whereas there appears to be no lack of commitment to scaling up Parivartan

among the SNP, there is considerable resistance to the programme from technical
staff in the AMC’s ward offices. Once a slum community is networked with infra-
structure through the SNP, it becomes part of the operations and maintenance
responsibility of one of the AMC’s  ward offices. Indeed, the philosophy of Pari-
vartan is ‘to integrate the slums into the mainstream of the city’ (AMC, ). Staff at
the ward offices, however, enjoy a poor reputation for customer responsiveness, and
they have been particularly reluctant to provide post-construction support to Parivartan
communities. In the newly upgraded community Ramraman Chaal, for example, a
drainage line installed by the SNP subsequently became blocked. Despite registering
four official complaints over a -day period, residents received no assistance from
their ward office. Eventually, frustrated residents collected  rupees (approximately
US$.) from each affected household and hired a private contractor to clear the line.
For their part, ward office staff justify their reluctance to service infrastructure in
Parivartan communities by citing improper designs and poor construction by SNP
engineers. ‘We have had no part in installing these things’ one ward office engineer
explained. ‘It is not our fault that they are not working properly.’ Tensions between
the SNP cell and ward offices have escalated over operations and maintenance, with
the cell occasionally stepping in to address service problems even after their formal
obligations to Parivartan communities have been fulfilled. Despite substantial
implications for the long-term sustainability of infrastructure installed through the
SNP, it is not clear whether the resistance of ward staff has had much impact on the
project’s scaling up.
 In our survey of  slum households that had not participated in the SNP,  per cent of households said they had
not made a complaint to the AMC about their water supply or sanitation services within the previous six months.
Of these,  per cent did, in fact, have legitimate operation and maintenance problems, but did not contact the
AMC because they felt that no action would be taken by the ward officers. Among the households who did make
service complaints to the AMC, more than one-third reported that their problems were never resolved.
 Verma () notes that demand for a similar upgrading project in nearby Indore fell off in part as a result of
households’ exposure to the poor maintenance of infrastructure installed by the project in neighbouring slums.

IDPR26_3_04_Davis 313 1/11/04, 2:31 pm

 314 Jennifer Davis

NGO partners
The partnership between the AMC and the three NGOs participating in Parivartan
is unique in its allocation of significant power and control to the NGOs, as well as in its
production of mutual trust and respect between government and civic professionals.
Nevertheless, at the time of our visit, senior leaders in the NGOs were obstructing the
scaling-up process by refusing to work in a larger number of slum neighbourhoods
than is possible – even with available staff resources – for their organisations. This
resistance from the NGO partners has emerged from disagreement about the types of
community services that should be offered in the programme, and what level of
funding should be provided to support organisations.
SNP architects originally estimated that approximately US$ per household
would be required for the community development activities associated with the SNP.
Of this amount, NGOs are required to contribute  per cent while the AMC makes
up the balance. Over time, however, the expectations upon the NGO partners have
increased while the resources allocated to them have remained constant. Along with
community mobilisation, education, training and income-generating activities, NGO
staff have taken on virtually full-time positions monitoring construction work and
liaising with SNP staff and contractors. Over the past two years, each of the three
NGO partners has had to raise external funding to finance its continued participation
in the programme. One senior NGO member estimated that per-household com-
munity development costs for the slums in which she has worked are closer to US$
than the US$ allotted.
The discrepancy between budgeted and actual community development expendi-
tures stems in large part from the fact that the SNP has provided very little guidance
regarding what activities the partner NGOs are to undertake. SEWA, the MHT and
Saath each have distinct approaches to working with slum communities, comprising
different types of activities and different resource requirements. The AMC leadership
has allowed partner NGOs to pursue whatever approach they feel to be most effective
for successful participation, rather than imposing a standardised set of activities and
demanding a uniform approach. At the same time, regardless of the approach
pursued by an NGO, the budgetary support from the programme for community
development activities remains constant.
Project support for community development work is also constant from slum to
slum, despite the considerable variation in demand for such intervention from
different communities. For smaller, homogeneous communities with a history of
collective action, an NGO may need only to educate residents about the benefits and
obligations of the project and assist households in establishing savings accounts for
their financial contributions. In other settlements, where internal conflicts, previous
disappointments with infrastructure projects, extreme poverty or other barriers pose

IDPR26_3_04_Davis 314 1/11/04, 2:31 pm

 Scaling up urban upgrading 315

substantial challenges to successful community organising, an NGO might need

considerably more time and resources to help residents prepare for and implement
the SNP successfully. NGO representatives have suggested that funding for
community development be determined on a case-by-case basis, which would not
only relieve the financial burden that the organisations currently bear, but would
ostensibly give greater control to communities over the kinds of development
activities they would like to have.
Frustration with the funding of community development activities in the SNP has
made NGO staff reluctant to engage in heavy promotion of the SNP to candidate
communities. ‘We are holding back’ one representative admitted. ‘We want the
programme to expand, but we are more concerned about doing a good job … We
would rather work in fewer slums with a good outcome than in more slums with
problems.’ Indeed, the AMC has been open to increasing the number of civic
partners in the programme, but little interest has emerged thus far on the part of the
NGOs (Dutta, ). Representatives from several NGOs that considered, but
eventually decided against, participating in the SNP said that their decisions were
driven principally by reports of the insufficient resources allocated to the NGO
partners. ‘We are not SEWA … We don’t have that kind of money’, the head of one
such organisation explained. ‘I think the project is very good, but the AMC must see
that this kind of [community organising] work is not easy or fast. It takes good people
and money for a really successful experience.’

State government
Although the focus of our analysis was municipal-level institutions in Ahmedabad, it
is important to note that the SNP’s progress is influenced substantially by state insti-
tutions in Gujarat. Like the AMC, the government of Gujarat has a vested interest in
drawing increased donor support for slum upgrading activities in Ahmedabad. At the
same time, a principal obstacle to taking the SNP to scale is the state government’s
unwillingness to cooperate with the AMC’s innovative approach to land tenure issues.
In order to encourage community participation in the SNP, the AMC offers a virtual
-year tenure guarantee, which can be extremely attractive to households located on
holdings with disputed or uncertain titles. The guarantee does not confer an exchange-
able deed to the slum resident; instead, the AMC commits itself to not initiating any
action that would require the community’s relocation for a period of  years.
This tenure provision is not necessarily legally binding; in fact, it is unclear

 Of course, this approach would itself increase the cost of NGO participation, as some mechanism of debating
and coming to a consensus on the types of activities and the level of funding provided to each enrolling
community would have to be devised.

IDPR26_3_04_Davis 315 1/11/04, 2:31 pm

 316 Jennifer Davis

whether the AMC can legally provide such a guarantee for land that is owned by
either another branch of government or a private landholder. To date, the policy has
not been challenged, nor has the AMC’s guarantee to any community been breached.
At the same time, SNP staff have been instructed to review the tenure status for each
slum that applies to the project and reject those with a risk of tenure-related
complications.  If a slum is located on AMC land not reserved for a particular use
(e.g., a highway or park construction), or is located on private land not involved in
pending litigation, SNP staff are permitted to enrol the community in the pro-
gramme. If the community is on state government land, the SNP must seek approval
from these branches of government. None of these approval requests has been
granted in the past, and several AMC staff said that it has been indicated that none
should be expected. Thus, SNP staff consistently reject applications from slums
located on state-owned land.
In sum, whereas the SNP’s tenure provision makes the programme more
attractive to slum-dwellers, it has also made the process of approving communities’
applications more difficult. As of November , only  of the  slums that had
submitted applications to the project were approved. The remaining slums have been
either rejected or indefinitely delayed, at times because of their distance from main
water and sewer lines in the city, but more often because of the land ownership status
of the communities. In a random sample of  slum communities in Ahmedabad,
we found that  ( per cent) were located on reserved AMC, state or federal land,
and would thus be considered (at least initially) ineligible for participation in the SNP.

Untested implementation conditions

To the extent that the characteristics of residents or communities, institutional
arrangements and/or policy frameworks for pilot initiatives are not representative of
those extant in the target population, the scaling up of upgrading programmes will be
impeded. For example, projects are often launched in communities in which the likeli-
hood of success is high, either because of the particular characteristics of residents
(e.g., education levels, wealth or a history of collective action), or because of support
from influential stakeholders. When extended to new areas/communities, the pro-
gramme encounters difficulties because of the unique characteristics of the areas or
persons served (e.g., technical challenges, social conflicts). Revisiting the issue of land
availability, an upgrading pilot project might be successfully implemented in the small
proportion of a city’s slum neighbourhoods that have secure tenure; it cannot

 Slums in areas for which a land use plan has yet to be finalised or developed will also be barred from participating
in the project. A slum can also be refused permission to participate on technical grounds, e.g., if the settlement is
located on land that cannot be accessed with minor improvements to existing mainstream infrastructure.

IDPR26_3_04_Davis 316 1/11/04, 2:31 pm

 Scaling up urban upgrading 317

reasonably be expected to scale up, however, if insecure tenure and a reluctance to

invest characterise the majority of the city’s poor neighbourhoods.
The establishment of ad hoc organisational arrangements or the granting of
‘policy holidays’ are institutional manifestations of this same scaling-up bottleneck. In
an effort either to ease implementation (e.g., bypassing cumbersome bureaucratic
procedures) or to experiment with innovative upgrading strategies (e.g., testing new
technical designs or standards), upgrading projects often receive special consider-
ations that preclude a ‘real world’ test of these strategies. The widely acclaimed
Kampung Improvement Project in Indonesia, for example, was administered through
project units that existed outside the traditional institutional framework for urban
service delivery. The skills and capacity developed in the units were thus lost once
external support for the project was withdrawn, and upgrading efforts largely ceased
(Operations Evaluation Department, ).
A final component of untested implementation conditions concerns the nature of
household preferences. More than  years ago, Turner () highlighted the fre-
quent differences between planners’ perceived and slum-dwellers’ actual preferences
for housing; since that time, the consultation with households for planning and
implementing upgrading programmes has increased considerably. Nevertheless, if
upgrading programmes are designed using information about the priorities of a
subset of non-representative households, scaling up is likely to be impeded. Practi-
tioners often argue that choosing socially cohesive neighbourhoods with high levels of
articulated demand for a pilot phase is important in order ‘to conscientiously achieve
rapid success, to maintain a certain momentum and to encourage those involved’
(UNCHS, ). Yet implementation manuals and project budgets for scaling up that
are based on experiences from these relatively easy pilot sites are likely to provide
inadequate guidance and resources for the challenges that lie ahead.
At first glance, it does not appear that the communities that have participated in
the SNP in Ahmedabad would be considered easy locations for an urban develop-
ment programme. Surveys conducted before the commencement of upgrading
activities in several slums show them to be socio-economically comparable to the
city’s poor communities in general. Data from our own surveys conducted in
Parivartan and non-Parivartan communities support this view (Table ). Parivartan
communities range in size from  to , residents; some had strong pre-existing
CBOs, while others had very little organising experience. The homogeneity of
religious beliefs and ethnicity also varies across communities. SNP and NGO staff
have thus implemented Parivartan in a range of settings that, in most respects, mirror
the diversity of Ahmedabad’s slums.
The important exception to this observation is that, prior to participating in the
SNP, a small number (less than  per cent) of Parivartan households had access to any
of the eight services offered by the programme. Senior AMC officials we interviewed

IDPR26_3_04_Davis 317 1/11/04, 2:31 pm

 318 Jennifer Davis

Table 4 Profile of Parivartan and non-Parivartan communities

Parivartan Non-Parivartan
Characteristic (n =147) (n =177)

% female 64 64
Median age 36 40
% with no formal education 41 45
% with primary education only 33 35
Median household size 4.9 5.1
% homeowners* 96 66
Median length of household’s residence in community (years) 18 18
Median market value of home† 58,400 rupees 50,000 rupees
(~US$1,300) (~US$1,100)
Median monthly rental value of home† 390 rupees 315 rupees
(~US$8.60) (~US$7)
Median household monthly income 1,710 rupees 1,960 rupees
(~US$43) (~US$38)
Median daily water use, winter 16 (pre-Parivartan)
22
(litres per capita per day) 24 (post-Parivartan)

Notes: *An equivalent percentage (63%) of Parivartan respondents said they were not homeowners before the
SNP, suggesting that the AMC’s 10-year tenure guarantee is viewed as de facto land tenure; † Estimated by
respondent.

believe that conditions are similar in the majority of Ahmedabad slums, an assump-
tion which underlies the AMC’s requirement that communities accept the entire
bundle of services in order to enrol in the SNP. Our research suggests, however, that
the number of informal settlements in Ahmedabad that have no basic infrastructure
services is far below the , cited by the AMC. For example, of these informal
settlements approximately , are row tenements or chaals, none of which has
submitted an application to the SNP. Field staff that we interviewed explained that
the private-sector sponsors of these housing developments have, in most cases, already
installed basic water supply and sanitation services for residents at no cost to them.
Many of the city’s , slums have also obtained water and sanitation services,
using both formal and informal means. In the city’s West Zone, only about  per cent
of slum communities ( out of ) are entirely without services. The remaining
settlements have obtained basic infrastructure services through some combination of
political sponsorship, the informal market and the AMC’s ward offices. In visits to 
randomly selected slums throughout the city, we found that  per cent already had
individual water and sanitation services, and  per cent had water, sanitation and
road paving. Among respondents in our survey, only  per cent of households in
Parivartan communities had individual (household) water and sanitation connections

IDPR26_3_04_Davis 318 1/11/04, 2:31 pm

 Scaling up urban upgrading 319

prior to participating in the SNP; among respondents in non-Parivartan com-

munities,  per cent had individual services. The daily water use per capita in these
non-Parivartan communities ( litres per capita per day) is quite similar to that of the
colonies whose water services have been improved by the SNP ( litres per capita per
day).
AMC officials nevertheless share the conviction that the bundling of services is
advantageous to both the AMC and slum residents. As one staff member explained,
‘this package is much better than what [households] will get from the zone offices [of
the AMC]… It’s also cheaper than what they would have to pay for someone to
connect them illegally.’ To SNP staff, the only feasible explanation for a slum house-
hold’s lack of interest in a particular service is that the household already possesses
that service. As such, the project rules include a set of credits that are applied to
reduce the financial contribution required of any household that has already
obtained one or more of the eight services. However, this reduction only applies to
services that have been obtained legally by the resident. SNP staff subsequently see
little need to adjust the suite of services offered, nor to create opportunities for
residents to select subsets of the services according to their needs and preferences.
Findings from our surveys and focus group discussions suggest that offering only
individual-level, bundled services is one important explanation for the SNP’s slow rate
of expansion in the city. For example, in our survey of Parivartan households we
asked respondents to list all the services provided in their community through the
SNP. Without being prompted,  per cent,  per cent,  per cent and  per cent of
residents surveyed were able to name water supply, sanitation, drainage and road
paving services, respectively. Considerably fewer residents (less than  per cent), how-
ever, could recall any of the six other services provided in the project. Respondents
were also asked to identify which single service they felt was the most important to
their household. More than  per cent of respondents cited water supply, sanitation,
drainage or road pavement, while fewer than  per cent of respondents chose any of
the remaining services. When reminded of these project elements, many respondents
commented that they had no interest in these services at all; as one slum resident
noted, ‘the AMC doesn’t require rich households to buy trees with their water supply’.
NGO staff also reported that they were regularly asked by community members to
lobby the SNP to exempt them from having to purchase the less popular upgrading
services.
Similar findings emerge from slum communities that have not been part of the
SNP. In a series of focus group discussions, we asked which of the SNP services they
would be willing to invest in for their community. Water supply, sewerage and road
 For example, if a household already has a legal water supply connection, it can deduct US$ from its US$
contribution to the project. Similarly, an existing (legal) sanitation system will reduce a household’s obligation by
US$.

IDPR26_3_04_Davis 319 1/11/04, 2:31 pm

 320 Jennifer Davis

paving were virtually always chosen, whereas there was little demand for tree planting,
rubbish removal, streetlights or other upgrading services. When asked whether they
would be interested in participating in the SNP, two-thirds of the groups concluded
that their community could more easily obtain the services they were interested in
from other sources (e.g., local politicians or charitable organisations), while the
remaining one-third felt that they would be willing to enrol for the full suite of services
from the SNP.

Summary and conclusions

Whereas the literature on urban upgrading in developing countries suggests that
resource constraints and political resistance are common reasons for the failure of
successful initiatives to scale up, somewhat different conclusions emerge regarding the
SNP in Ahmedabad. Clearly, a lack of political will within the state of Gujarat
regarding regularising slum communities has had a substantial impact on the SNP’s
expansion. However, even if the SNP continues to reject applications from slums on
state, federal or reserved lands, there are still more than half a million slum-dwellers
residing in communities eligible for the programme.
Nor were shortages of financial or human-capital resources identified as principal
explanations for the failure to take the programme to scale. Technical support for
engineering designs is readily available in the private sector, and the AMC has been
very successful in raising and earmarking funds for the project. Much more important
were the systematic differences between the infrastructure endowments of pilot SNP
communities and the majority of the target population, combined with the inflexi-
bility of the AMC’s policy of providing only bundled, individual-level services.
Given the plurality of options that slum-dwellers have for obtaining access to the
services they most desire, the SNP will need to think more strategically about how to
make its offer attractive to communities.
For example, whereas the argument for bundling together water, sewerage and
storm-water drainage services in the city’s dense urban slums is clear, there is no
technical or financial rationale for requiring the provision of all upgrading services in
all SNP communities. Revising this policy to allow greater flexibility in service
selection would probably increase the attractiveness of the project to communities.
This strategy would also accelerate the rate of the AMC’s physical construction

 This finding stands in contrast to a body of literature that rejects ‘narrowly conceived and sectorally planned’
upgrading projects (Moser, ). Staged upgrading may proceed one sector at a time, but need not occur
without comprehensive planning.
 It is notable that, when asked where this set of services originated, the majority of AMC staff said that it had
simply been adopted from the project implemented in Indore (a few said that it had emerged from a discussion
between senior AMC staff and NGO representatives).

IDPR26_3_04_Davis 320 1/11/04, 2:31 pm

 Scaling up urban upgrading 321

accomplishments, thus improving its standing with development agencies interested

in supporting urban upgrading initiatives.
It is also important to recognise that, so long as the city’s elected ward councillors
are willing to use their own budgets to provide slum-dwellers with services at little or
no cost, enrolment in the SNP will probably suffer. The practice of councillors invest-
ing in low- or no-cost services to slums was well established long before the advent of
the SNP; however, the AMC did not consider it important to launch a policy
marketing campaign among councillors in order to promote consistent cost-sharing
policies for infrastructure projects. This decision may need to be revisited. It is
encouraging to note that similar strategies have met with success in other public
service projects in South Asia (Davis et al., ). Finally, it is important to note that,
while some critical bottlenecks to expanding the SNP can be resolved only with policy
reform at the state level (e.g. tenure policy), others are fully within the purview of the
AMC and its partner organisations. Even if broad economic and policy reforms are
not forthcoming in the short term, substantial gains in the number of slum-dwellers
served by the SNP can be made through the same effort and innovation that have
characterised the programme to date.

References
 (  ) (), ‘The Slum Networking Project’ (unpub-
lished mimeo), Ahmedabad, AMC.
 (  ) (), Statistical Report on the City of Ahmedabad,
Ahmedabad, AMC.
 (   ) (), ‘A survey of slums and chaals in Ahmedabad,
India’ (unpublished report).
, ., , . and , . (), ‘Towards a national community infrastructure
upgrading program for Ghana’, Urban Notes, , –.
  (), ‘Upgrading urban communities: a resource for practitioners’ (web-
based toolkit available at http://web.mit.edu/urbanupgrading/index.html).
, . . (), ‘The challenge of replicability: towards a new paradigm for urban shelter
in developing countries’, Regional Development Dialogue, , –.
, . (), The SEWA Movement and Rural Development: The Banaskantha and Kutch
Experience, New Delhi, Sage.
, . and , . (), ‘Scaling up sustainable rural water supply services’ (discussion
paper), Washington, DC, Water and Sanitation Program.
, ., , ., , ., , ., , ., , . and , . (), In Search of
Good Governance: Experiments from South Asia’s Water and Sanitation Sector, New Delhi, The Water
and Sanitation Program.
, . (), ‘From development to empowerment: the Self-Employed Women’s
Association in India’, International Journal of Politics, Culture and Society, , –.
, . (), ‘Towards shelter for all: the Bombay experiments’, Urban Management in Asia:

IDPR26_3_04_Davis 321 1/11/04, 2:31 pm

 322 Jennifer Davis

Issues and Opportunities, New Delhi, National Institute of Urban Affairs, –.
, . (), ‘Partnerships in urban development: a review of Ahmedabad’s experience’,
Environment and Urbanization, , –.
, . and , . (), Reforming Urban Land Policies and Institutions in Developing
Countries, Washington, DC, World Bank Urban Land Management Program.
, . (), ‘Deficits of formal urban land management and informal responses under
rapid urban growth: an international perspective’, Habitat International, , –.
, . (), ‘Self-help housing in Colombia: alternatives for the production and
distribution of building materials’, Habitat International, , –.
, . (), ‘Orangi Pilot Project, Karachi: a low-cost sewer system by low-income
Pakistanis’ in B. Turner (ed.), Building Community: A Third World Case Book, London, Building
Community Books, –.
, . (), ‘World Bank experience with the provision of infrastructure services for the
urban poor: preliminary identification and review of best practices’ (Operational Review
TWU-OR), Washington, DC, World Bank Transportation, Water & Urban Development
Department.
, . (), ‘Urban social policy and poverty reduction’, Environment and Urbanization, ,
–.
, . (), ‘Urban infrastructure: a critique of urban renewal process in Ijora
Badia, Lagos’, Habitat International, , –.
   (), Urban Projects: Indonesia, Washington, DC, The
World Bank.
, . (), ‘State, voluntary and private sector partnerships for slum upgrading and basic
service delivery in Nairobi City, Kenya’, Cities, , –.
, . (), Where Women Are Leaders: The SEWA Movement in India, London, Zed.
, . (), ‘Housing priorities, settlement patterns, and urban development in
modernizing countries’, Journal of the American Institute of Planners, , –.
 (     ) (), ‘Hai El Salam project: an
upgrading and sites-and-services project in Ismalia, Egypt’ (Report No. HS//E),
Nairobi, UNCHS.
 (    ) (), The Challenge of Slums: Global
Report on Human Settlements, London, Earthscan.
, . . (), ‘Indore’s Habitat Improvement Project: success or failure?’, Habitat
International, , –.
, . (), ‘The slum upgrading myth’, Urban Studies, , –.
  (), ‘Urbanization sector working paper’, Washington, DC, World Bank.
  (), ‘What is urban upgrading?’, http://www.worldbank.org/urban/upgrading/
what.html

Acknowledgements
This article is one output of a study executed by the UNDP-World Bank Water and Sanitation
Program, South Asia Regional Office, and funded by the Government of the Netherlands,

IDPR26_3_04_Davis 322 1/11/04, 2:31 pm

 Scaling up urban upgrading 323

Netherlands Partnership Programme. I am grateful to Karen Polenske and Judith Tendler for
helpful comments, to Taimur Samad for fieldwork assistance, and to Genevieve Connors for
desk research. I am also indebted to the staff of the Slum Networking Project (Ahmedabad
Municipal Corporation), and of the NGOs Saath and SEWA, for the generous sharing of
their time and information.

IDPR26_3_04_Davis 323 1/11/04, 2:31 pm

IDPR26_3_04_Davis 324 1/11/04, 2:31 pm