The Political Economy of Devolved Provision: Measuring the Pro Poor Impact of Provision Targeting an

1
The Political Economy of Devolved Provision
Measuring the Pro Poor Impact of Provision
Targeting and Participation?Case Evidence from
Punjab Unions

• Ali Cheema and Shandana Khan Mohmand
• Lahore University of Management Sciences
• June 2nd, 2005

2
Plan

• Motivation and Main Questions
• Institutional Setting
• Accountability and Government Responsiveness
• Empirical Research Questions
• Sampling Strategy
• Features of Case Villages
• Empirical Strategy, Identification and Results
• Accountability and Participation
• Conclusion and Insights for Policy

3
Motivation

• Does decentralization increase accountability of
  governments to local citizens
• In particular non-elite and marginalised groups

4
Motivation

• It is a lamentable fact that Indias ancient and
  self-governing village communities
  havedisappeared, under the too centralised
  administration of British Ruleno representatives
  of the village population help the
  administrationand an alien Government lacks that
  popular basis, that touch with the people.
• R.C.Dutt (1903)
• What is the village community but a sink of
  localism, a den of ignorance, narrow-mindedness
  and communalism. I amsurprised that those who
  condemncommunalism should come forward as
  champions of the village community.
• Ambedkar (1948)
• The smaller the society, the fewer the distinct
  parties and interests composing itthe more
  frequently will a majority be found of the same
  party and the smaller the number of individuals
  composing a majority, and the smaller the compass
  within which they are placed, the more easily
  will they concert and execute their plans of
  oppression.
• Madison (Federalist Papers No. 10)

5
Motivation

• Accountability inferred by
• Responsiveness of government to poor and
  marginalized?
• Paper analyzes accountability changes by
  examining post reform provision targeting
  exhibits systematic pro-poor bias
• Extent to which reforms have catalyzed
  participation of marginalized in new citizen
  bodies/local collectivities
• Far reaching local government reform in Pakistan
  in 2001
• Allows us to analyze impact on the equity of
  service provision and pro-poor participation
• Infer if accountability improving (similar
  studies for India Bardhan and Mookherjee 2003,
  Besley et. al. 2003)
• Focus on Union level
• Only directly elected tier
• Improved accountability at this level important
  gauge for accountability changes in the system
• Analysis allows identification of interventions
  to strengthen union level accountability

6
Part 1. Government Responsiveness

• Q1. Has the magnitude of provision to previously
  underprovided villages increased Post LG reforms
  (2001)?
• Q2. Has equality of provision between villages,
  social groups and classes increased Post LG
  reforms (2001)?

7
Part 2. Participation

• Q1. Do poor/marginalized groups have equal
  access to the new citizen bodies?
• Q2. Is there gender equality in access to the
  new citizen bodies?
• Q3. Do poor/marginalized groups have equal
  access to union level demand aggregation
  mechanisms?

8
Institutional Setting - LG Reforms 2001

• Direct elections of councilors, nazims and naib
  nazims at union level
• Union lowest tier of three-tier LG
• Elected union level representatives integrated
  into higher tiers of LG
• Union nazims/naib nazims - district and tehsil
  councilors, respectively
• Union councilors, nazims and naib nazims
  electoral constituency for election of
• District, tehsil nazims/naib nazims
• Councilors elected on reserved seats at higher
  tiers

9
Institutional Setting - LG Reforms 2001

• Consequences/Expectations
• Increased electoral competition for local
  influentials at union level
• Union nazims/naib nazims
• Pivotal players in selection of development
  schemes
• Especially for services like soling/sanitation
  that are poorly documented
• Important role in creating awareness of new
  community bodies
• Increased mandates at union level

10
Institutional Setting - LG Reforms 2001

• 3. Autonomy of union nazims/naib nazims as
  agenda setters increased
• Role as councilors gives them a voice over scheme
  selection in case of services like
  soling/sanitation
• Development funds for union nazims/naib nazims
• Promises access to development funds for many
  previously underprovided unions
• Unions nazims/naib nazims virtual monopoly over
  local level information
• District/tehsil does not have capacity to
  observe and verify the use of development
  funds at the local level for services like
  soling/sanitation
• Discretion with union representatives regarding
  selection and implementation of development
  schemes

11
Part 1. Accountability and Government
Responsiveness
12
Empirical Research Questions

• Three district case studies ask
• Has the gap in provision between the union
  nazims own village and other villages increased
  or decreased post-reform.
• Besley et. al.s (2003) cross country results
  show
• Targeting bias towards the LG heads village in
  multi-village local governments
• Has majority biraderi gained in provision
• Is targeting geared towards the majority biraderi

13
Empirical Research Questions

• Ask whether voting for nazim increases HH chances
  of obtaining provision
• At minimal level tells if local level democracy
  electorally responsive
• Whether belonging to the village influentials
  political faction gives a household a higher
  chance of obtaining drain provision
• Field work reveals electoral contests organized
  around patron-client factions or dharas
• Analyse effect of HH belonging to a socially
  powerful village level dhara (political faction)
• Elite bias of post-reform targeting inferred by
• Analysing the sociological composition of each
  dhara

14
3 District Case Studies

• Choosing districts/tehsils
• 2 dimensions
• Small peasant v. large landlord
• Variation in exposure to urbanization
• Results similar
• Difference in political dynamics
• Present results of small peasant case with high
  exposure to urbanization
• Expect higher proclivity for change

15
Outcome Variable

• Objective to estimate post-reform targeting bias
  between villages/social groups
• Needed data on government provided HH level good
• Use pucca drains (sanitation) at household (HH)
  level as measure
• Pre-survey visits revealed drain
• Targeted towards different villages and
  residential clusters
• At times provided not at lane but HH level
• Caveat - Data measuring quantity of provision not
  access to a functional drain system
• Sanitation an important budgetary and need item
• Post-reform effect
• Measure change in provision at HH level using
  survey and physical mapping
• Drawing a pre-LG election baseline at HH level
• Measuring current provision at HH level
• Measure targeting bias between social
  groups/classes
• Using survey data on HH characteristics

16
Sampling Strategy

• Household Surveys
• Mapping
• Villages mapped by professionals
• Create biraderi census
• Map pre/post reform provision of soling and
  sanitation
• Sampling
• From this biradiri census data a random
  stratified sample of 20 of the population is
  drawn.
• 22-30 HH in each village surveyed
• District sample 364-500 HH
• Total HH sample 1000
• 2/3rd male and 1/3rd female respondents

17
Sampling Strategy
18
Case Study Sampling Strategy

• Rural focus
• Union characteristics
• Small peasant and proximate to urban areas
• Weak political linkages with higher tier
• Little pre-reform funding
• No money from MPAs/MNAs post-reform
• No CBO/NGO provision/activity
• District and Tehsil nazim proactive and equitable
  disbursement of funds

19
Case Study Sampling Strategy

• Unions chosen (UC1 and UC 2)
• Received equal per capita development spending on
  sanitation post-reform
• Control for revenue effect
• Union nazims won by small margin, 2-5 of votes
  cast
• Expect these executives to be electorally
  responsive, especially to majorities
• Expectation
• Elite capture lower and impact of devolution
  higher
• Because areas expected to have seen
  socio-economic change
• Because absence of large landlordism

20
Sampling Strategy

• Each union 2 villages chosen
• Nazim village (NV1 and NV2)
• Non-Nazim Village (NNV1 and NNV2)
• Low or no representation in union council
• Consequence of union being a multi-member
  electoral ward
• No politician resident
• Weak political linkages with councilors and
  higher tier politicians
• Allows focus on union nazim/naib nazim decision
  making
• A majority or near majority (47 62) HH where
  head voted for nazim/naib nazim
• Expect nazim/naib nazim to be electorally
  responsive

21
Village level Sample Statistics and Voting
Behaviour
22
Distribution of Councillors across Case Villages
Note Data on sample villages is given in bold.
Number of councillors given in parenthesis. In
case UC 1 one seat is vacant because of the
resignation of the elected councillor.
23
Features of Nazim/Non Nazim Villages

• Socio-economic features similar
• Expect in NNVs
• Village influential from minority landholding
  biraderi
• Majority biraderi small peasant landholders and
  labourers
• Majority biraderi much poorer condition of
  dwellings
• Large population (21) of village servants or
  Muslim Sheikh biraderi
• Abject living conditions
• NVs more equitable
• Village influential (nazim) from majority
  biraderi which is comprised of
• Small peasant landholders and labourers

24
Features of Nazim/Non Nazim Villages

• Politically
• NV
• majority biraderi and majority of non agrarian
  biraderis voted for nazim
• NNV
• Majority biraderi and lower caste voted for nazim
• Village influential and his biraderi electorally
  opposed nazim
• Given voting affiliations, we expect non-elite
  bias in post-reform targeting in both villages
• Also expect targeting to favour majority biraderi

25
Socio-Economic and Political Features of Case
Villages
26
Socio-Economic and Political Features of Case
Villages (contd.)
Note (1) Standard deviation in parenthesis. (2)
Dominant biraderi is the village influentials
biraderi (3) NA implies that in the nazim
villages the dominant and majority biraderi is
one and the same.
27
Socio-Economic Features of Dominant and Majority
Biraderis
28
Socio-Economic Features of Dominant and Majority
Biraderis
29
Does it Pay to Reside in the Nazims Village?
30
Does it Pay to Reside in the Nazims Village?
Notes The dependent variable is a dummy variable
which equals one if the household was provided a
pucca drain by government outside the house.
Household controls included are agricultural
ownership, literacy and household size. We also
include the interaction terms of all these
variables with Post (i.e. post 2001). We ran
separate regressions, which included these
controls as well as occupational dummies and
their interaction terms but this did not change
the results. Robust standard errors in brackets.
Standard errors have been clustered at the
village level. significant at 10,
significant at 5, significant at 1.
31
Does it Pay to Reside in the Nazims Village?

• Results
• Substantial increase in overall provision in
  previously underprovided villages (positive)
• Post reform increase in spatial inequality
  favours NVs (negative)
• Increased mandates interacting with a political
  process to create incentives for nazims to target
  own village
• Remainder of paper attempts to prize open black
  box called union level politics

32
Does Majority Biradari Gain?
33
Majority Biradari and Provision Targeting
Notes The dependent variable is a dummy variable
which equals one if the household was provided a
pucca drain by government outside the house.
Household controls included are agricultural
ownership, literacy and household size. We also
include the interaction terms of all these
variables with Post (i.e. post 2001). We ran
separate regressions, which included these
controls as well as occupational dummies and
their interaction terms but this did not change
the results. Robust standard errors in brackets.
Standard errors have been clustered at the
village level. significant at 10,
significant at 5, significant at 1.
34
Majority Biradari and Targeting

• Results
• Highly significant increase in post reform
  provision to majority biradari HH in NVs
• Nazim electorally responsive to majority biraderi
  in his village
• Which were small landholders
• So..non-elite groups benefiting as well

35
Majority Biradari and Targeting

• Results
• Suggests majority biradari HH in NNV get a small
  post-reform increase
• In NNVs minority dominant biradaries benefiting
• Means non-elite groups losing out in NNVs
• Intriguing result given small margin of victory
  for nazims
• In spite of this, discounting a potential vote
  bank in NNVs
• Voting result consistent in other districts as
  well
• What explains result?

36
(No Transcript)
37
Does it Pay to Vote for the Union Nazim?
Notes The dependent variable is a dummy variable
which equals one if the household was provided a
pucca drain by government outside the house.
Household controls included are agricultural
ownership, literacy and household size. We also
include the interaction terms of all these
variables with Post (i.e. post 2001). We ran
separate regressions, which included these
controls as well as occupational dummies and
their interaction terms but this did not change
the results. Robust standard errors in brackets.
Standard errors have been clustered at the
village level. significant at 10,
significant at 5, significant at 1.
38
Patron-Client Factions and the Inequality of
Provision

• Key respondent surveys underscored importance of
  village level factions in bargaining over
  provision
• Sociology of provision explained as a bargaining
  game between the dhara of village influential,
  union nazim/naib nazim
• Individual voters do not matter
• What mattered was whether HH part of a dhara
  (patron-client faction) that was socially
  powerful
• Survey data shows over 85 of HHs part of dharas
  and vote with these factions
• Village influential prioritises provision towards
  own dhara members because of embedded
  alliances based on
• Biraderi, influentials transact intermediation
  for votes, historically recognized as socially
  powerful
• Political party affiliation not important
• Non member voters lose out
• Because alliances not embedded
• Do not have a socially powerful influential to
  negotiate on their behalf

39
Patron-Client Factions and the Inequality of
Provision

• Decided to test hypothesis
• Being a member of a village influentials dhara
  gives a HH higher likelihood of getting post
  reform provision
• Can empirically test hypothesis
• Rich survey data on dhara formation and
  membership
• Identify a HH as a member of the village
  influentials dhara if
• Respondents report being a member of a dhara
• Reported that dhara was organized directly by the
  village influential
• Consistency in voting patterns

40
Patron-Client Factions and the Inequality of
Provision
41
Patron-Client Factions and the Inequality of
Provision
42
Patron-Client Factions and the Inequality of
Provision
Notes The dependent variable is a dummy variable
which equals one if the household was provided a
pucca drain by government outside the house.
Household controls included are agricultural
ownership, literacy and household size. We also
include the interaction terms of all these
variables with Post (i.e. post 2001). We ran
separate regressions, which included these
controls as well as occupational dummies and
their interaction terms but this did not change
the results. Robust standard errors in brackets.
Standard errors have been clustered at the
village level. significant at 10,
significant at 5, significant at 1.
43
Patron-Client Factions and the Inequality of
Provision

• Results
• Being a member of a village influentials dhara
  in NVs and NNVs has resulted in
• Substantive increase in post reform provision
• Relative to non-members of dhara
• Benefits to NNV influentials faction not
  conditional upon electoral alliance
• NNVs village influentials dhara electorally
  opposed the nazim

44
Patron-Client Factions and the Inequality of
Provision

• Results
• Competitive democracy at union level may not be
  working as well as assumed
• Even in rural areas exposed to urbanization

45
Patron-Client Factions and the Elite Biased
Provision

• Examining social composition of village
  influentials dharas gives us a handle on elite
  bias
• NV influentials dhara encompassing of majority
  biraderi and non-agricultural biraderis
• Elite bias low
• NNV influentials dhara exclude majority biraderi
  and non-agrarian castes
• Dharas dominated by large landholding dominant
  biraderi
• Elite bias in provision high
• Interesting question why cant non-elite groups
  form own dharas to give political competition?
• Dharas exist but factionalized and unstable

46
Patron-Client Factions and the Elite Biased
Provision
Features of Village Influential Dharas (Vote
blocks)
Note (1) Majority biraderi is the occupational
caste that is in a population majority in the
village. (2) Dominant biraderi is the
occupational caste of the village influential. In
nazim villages the dominant biraderi is the same
as the majority biraderi because the nazim is
from the latter. (3)NA means that the dhara has
no members in this particular village
47
Patron-Client Factions and the Elite Biased
Provision
Features of Village Influential Dharas (Vote
blocks)
Note (1) Majority biraderi is the occupational
caste that is in a population majority in the
village. (2) Dominant biraderi is the
occupational caste of the village influential. In
nazim villages the dominant biraderi is the same
as the majority biraderi because the nazim is
from the latter. (3)NA means that the dhara has
no members in this particular village
48
Accountability and Participation
49
Summary

• Location bias in provision
• Provision not conditional on voting for the
  winning nazim Or
• Being from the majority demographic group
• Even where elected nazims margin of victory
  small
• Provision conditional upon being a member of the
  village influentials dhara

50
Interpreting the Results

• Informal political networks trump planned
  provision
• Are the impediments we have identified structural
  or transitory
• Too early to judge
• Need data from another electoral cycle
• Next election results interesting
• Can ask whether participation in new citizen
  bodies/planning processes opening spaces for
  poor/ marginalized?
• Do we see organic change at this level

51
Interpreting the Results

• Key Questions
• Do women/marginalized groups have equal access to
  new participatory bodies?
• Do poor and marginalised have access to formal
  planning and scheme selection processes?

52
Equality of Access

• Access measured in terms of awareness of SMCs
  and CCBs
• Formalised on the ground
• VNCs not empowered
• Awareness of bodies poor on whole except SMCs
  CCBs somewhat
• Significant correlates of awareness
• Women (-ive)
• Dominant Biraderi (ive)
• Av. HH adult primary school attainment (ive)
• Last worrying as HH adult primary school
  attainment low
• Reveals problems with equal access to
  information hypothesis!

53
Equality of Access

• SMCs
• Women respondents 23-34 less like to be aware!
• Members of dominant biraderi 14-16 more likely
  tol be aware!
• 1 increase in the average HH adult primary
  school attainment results in awareness increasing
  by 23-27

• CCBS
• Women respondents 12 less like to be aware!
• Members of dominant biraderi 12 more likely tol
  be aware!
• 1 increase in the average HH adult primary
  school attainment results in awareness increasing
  by 11-34

54
Equality of Access in Planning

• No formal mechanisms for aggregating village
  community needs
• 60-90 respondents report highly infrequent
  meetings with union councillors/nazims
• 90 and more respondents say meetings if any
  informal

55
Equality of Access to Union Representatives

• Far from equal access to these informal meetings
  for poor/marginalised
• Results show
• Location bias
• Respondents in NNV 8 less likely to report an
  infrequent meeting
• Dominant Biraderi bias
• Members of dominant biraderi 8-10 more likely to
  report an infrequent meeting
• Structural constraints to participation
  persisting!

56
Conclusions Findings
57
Government Responsiveness

• Argue Pakistan LG reforms (2001) have increased
  mandates and agenda setting power of union
  nazims/naib nazims
• This has
• Substantially increased provision to
  underprovided areas
• Social inequality in provision result mixed
• NV elite bias declining
• NNV elite bias increasing
• Most interesting finding
• Informal village level patron-client factions
  determine winners and losers of post-reform
  provision
• Specific pattern of dhara politics more important
  determinant of provision than individual voting
• Majority biradaries dont gain in all villages
  despite voting for winning candidate and being
  electorally important
• Given this, elite bias becomes a function of the
  social composition of dharas

58
Government Responsiveness

• Continued importance of village level
  patron-client politics interesting
• Case union not domain of larger landed interests
• Exposed to urbanization and industrialization
• Future research question
• What explains persistence of exclusionary
  dhara-based politics in these areas

59
Participation

• Average levels of awareness of new citizen bodies
• Low
• Biased in favour of dominant biraderi members
• Biased against women
• Schooling matters for participation
• More disturbing absence of inclusive union
  level demand aggregation mechanisms
• Demand aggregation domain of informal meetings
• Informal meetings have inbuilt location and
  dominant biraderi bias

60
Caveats

• Note of caution, small village and service level
  sample
• Not easy to generalize
• Idea that conditions we identify represent a
  reality that is prevalent in many (though not
  all) parts
• Reality produces outcomes, which need further
  systemic interventions to overcome

61
Policy I

• How functional a unit is the union to further
  local democracy?
• Nature of union as multi-member ward means
  villages will be underrepresented in council
• Appears electoral politics at union level
  trumped by bilateral dhara-based bargaining
• Village communities and HH dont have direct
  input (monitoring) in the selection of schemes
• Need to redesign mechanisms to give equitable
  voice and representation to all villages and
  social groups in
• Governance
• Planning and scheme selection
• How essential are village councils
• Incentivise equity from above
• Equity incentives in fund allocation

62
Policy II

• Design mechanisms of new bodies do not promote
  inclusion
• Awareness cannot be assumed on part of women and
  marginalised
• Is it time to consider affirmative action beyond
  councillor reservation?
• Is it time to make planning processes more
  participatory?