Devolution, Accountability, and Service Delivery in Pakistan
Devolution, Accountability, and Service Delivery in Pakistan
Devolution, Accountability, and Service Delivery in Pakistan
130
Zahid Hasnain
131
Local
policymakers
Local Policy-makers
Poor
Poor people
People
Source: Ahmad, et al. (2004).
Providers
Providers
Zahid Hasnain
132
Holding Policy-makers Accountable
Important recent works in this literature include Persson and Tabellini (2000), Keefer (2002), Keefer
and Khemani (2003), Besley, et al. (2004).
133
The key question is whether local governments are better able to resolve these
political market failures than provincial or national governments. On the one hand closer
proximity might ameliorate some of these information problems that lead to political
market failure. On the other hand, and as per the classic argument against
decentralisation, articulated first by James Madison in the Federalist Papers, local
governments may be more prone to lite capture as there will be fewer distinct political
groupings in smaller jurisdictions, less media coverage compared to national elections,
and less costly for lites to purchase the votes of less-informed voters.
Holding Providers Accountable
In order to be able to act on the demands imposed on them by citizens, policymakers must be able to hold the bureaucracy accountable for actually delivering these
services. In general, mechanisms for monitoring and disciplining front-line providers
such as teachers are weak in developing countries. Decentralisation can further
exacerbate these problems, as administrative devolution usually lags behind political and
fiscal decentralisation, and local policy-makers have limited authority over the staff
assigned to them.
The Inter-governmental Framework
Local governments incentives for improving service delivery are significantly
influenced by the allocation of expenditure responsibilities between the local and higher
tiers of government; the assignment of taxes across these levels; the design of the intergovernmental grant system; and the budgeting and monitoring of fiscal flows between
different tiers of government.2 The literature posits that expenditure and tax assignments
should be based on considerations of economies of scale, spillover benefits, tax efficiency
and equity issues. In general these considerations imply that local governments are
unable to meet most of their resources through taxes, and must rely on intergovernmental grants, which creates problems of soft-budget constraints and may weaken
service delivery incentives. An effective fiscal transfer system should be predictable and
formula-driven, consisting of a good balance between unconditional grants, to provide
local governments expenditure discretion, and conditional grants, to enable the centre to
directly specify some accountability for service delivery.
Effective expenditure assignments are also important from the perspective of
political market failures. As noted earlier, to be motivated to deliver public goods,
politicians must be able to effectively claim credit from voters for providing such goods
to them. With multiple tiers of government, voters might have little knowledge about
which tier of local government is responsible for which functions. Even though the
responsibilities of the different tiers of government are clearly defined, it is likely that
opportunistic politicians will exploit voters relative lack of knowledge by avoiding
responsibility for their own lack of efforts or claiming credit for others actions in
providing public goods. If voters are likely to be grateful to all the tiers of government
when public goods increase, even if only one government contributed, then the incentive
to provide that public good is diminished.
2
Zahid Hasnain
134
Triesman (2002).
135
Zahid Hasnain
136
Whom
Who households
the households
would contact,
contact, 2004
2004(percentage)
(percentage)
100%
MPA/MNA
MPA/MNA
90%
30
80%
Community
Community
members
Members
25
70%
20
60%
50%
15
40%
Union nazim
or
Union
Nazim
or
councillor
Councillor
10
30%
5
20%
10%
Punjab
Punjab
Sindh
Sindh
Khyber NWFP
Pakhtunkhwa Balochistan
Balochistan
2002
2004
Pakistan
Pakistan
0%
1
The survey also revealed that most of the reasons for contact with union
councillors cited by male and female respondents alike, related to personal issues, such as
financial support, the issuance of identity cards, a police problem, or some form of
dispute. Those who contacted for reasons of service delivery did so regarding mainly
about water, roads, and electricity. Less than 2 percent of the respondents approached
local officials for education and health matters. Clearly, this suggests that increased
proximity and greater accessibility of policy-makers does not necessarily imply that
pressures for improving social service delivery increase; it may imply that while elected
officials may be less able to shirk their responsibilities to their respective constituencies,
they may also be consumed by satisfying the personal demands of voters, or demands for
improving physical infrastructure.
Electoral Accountability
As discussed, an argument against local government accountability is that local
government elections may not be as competitive as national elections, due either to
relatively lower citizen interest and media coverage, or lite capture.
The
competitiveness of local government elections can be examined along three dimensions:
(a) voter turnout; (b) the level of contestation (percent of seats filled by contest between 2
or more candidates); and (c) level of partisan politics, or the number of independent
candidates who win relative to candidates who are affiliated with political parties. The
first two indicators are almost self-evidentthe higher the voter turnout and the greater
the level of contest for individual seats, the more representative a candidate. With
regards to the third indicatorpartisan affiliationit can be argued that intensity of
election campaigns is likely to be higher when political parties are involved as opposed to
when independent candidates are contesting, primarily because candidates affiliated with
political parties have greater resources at their disposal for their campaign. Partisan
politics is also likely to impact the first two indicators and lead to a higher voter turnout
and greater contestation.
As Figure 3, left panel shows, voter turnout in union council elections in Pakistan
has been higher than national electionsfor example, their was a 48 percent turnout in
the 2005 local government elections, as compared to a 42 percent turnout in the 2002
national assembly electionsindicating that citizens are if anything more interested in
137
local issues than national ones. Turnout varied considerably across the provinces, from a
high of 54 percent in Punjab to a low of 31 percent in Balochistan, but was in all
provinces higher than the turnout in the national elections.
Fig. 3. The Competitiveness of Local Government Elections
Voter turnout in national and local elections
60
2001
2005
50
40
35
2002
1993
1997
30
40
25
30
20
15
20
10
5
10
0
Punjab
0
National Assembly elections
Sindh
Khyber Pakhtunkhwa
NWFP
2001
Balochistan
2005
The level of contestation, with one notable exception, has also been
considerable. In Punjab, Sindh, and Khyber Pakhtunkhwa in only 6 percent, 12
percent, and 11 percent respectively of the seats in the 2005 elections were
candidates returned unopposed (Figure 3, right panel). By contrast, in the
Balochistan local government elections of 2001 almost 40 percent of the union
council seats were uncontested, a number far higher than in any election at the
national or provincial level, and reflected the fact that in many cases the elections
were managed up front by tribal elders with the aim of ensuring adequate
representation of most of the sub-tribes and clans that were politically important in
the district. The 2005 elections appear to have been more competitive, with the
number of uncontested seats in the province declining to 17 percent.
Political parties are by law not allowed to participate in local government
elections, and candidates have to compete on an individual capacity. However, in reality
the vast majority of candidates in the local elections belonged to one or another of the
established political parties, and the resources of these parties were, again with the
notable exception of some districts in Balochistan, utilised in the campaigns [World Bank
(2004)]. The nature of local partisan politics appears to be quite similar to that at the
provincial levelin the districts of Punjab for example, traditional patron-client or
factional politics based on biraderis (patrilineal networks) were the main basis for
political mobilisation [Cheema and Mohmand (2005)].
The indirect election of the district and tehsil nazim implies that the relative
competitiveness of union council elections does not automatically translate into the
electoral accountability of these key local policy-makers. The re-election of an
incumbent district or tehsil nazim is dependent not on voters but on a few hundred union
councilors and therefore, downward accountability is dependent on the degree to which
union councillors act as agents of the public, and the extent to which nazims have an
incentive to listen to councillors.
One uniform way that district nazims have responded across districts to this
dependence on union councillors is by allocating some portion of the district development
Zahid Hasnain
138
budget to each individual union nazim to be used at his or her discretion. This is the
equivalent of the MPA and MNA funds that have been constituted at the provincial and
national levels respectively. However, the district nazims relatively greater independent
political standingthey are usually from established political familiesalso implies that
the future electoral fortunes of union councilors are as much linked to their endorsement
by the nazim. There are many examples of district nazims campaigning on behalf of
union councilors in order to secure his own future re-election from a pliant electoral
college [World Bank (2004)].
An interesting feature of local government elections is the relatively high degree of
incumbency disadvantage of district nazims. As Figure 4 shows, only 24 percent of the
incumbent district nazims were re-elected in the 2005 across Pakistan, as compared to 38
percent of members of provincial assemblies in 1997.4 Across the provinces, only in
Punjab were district nazims and MPAs equally disadvantaged, with 35 percent getting reelected. The incumbency disadvantage of nazims was particularly acute in Khyber
Pakhtunkhwa and Balochistan, with only 13 percent and 11 percent of nazims getting reelected respectively.
Fig. 4. Incumbency Disadvantage in Local Elections
Incumbent winning percentage
60
50
40
30
20
10
0
Punjab
PA elections 1990
Sindh
Khyber NWFP
Pakhtunkhwa
PA elections 1993
Baloch.
PA elections 1997
Pakistan
LG elections 2005
The 2002 provincial assemblies were not considered due to the complications caused by the
intervening period of military rule.
139
Table 1
Correlates with the Election of a District Nazim
Prob. of
Nazim
Getting Reelected
Citizen Satisfaction with:
Basic Health Facilities
0.13
(1.88)
Schools
6.70
(4.59)
Roads
3.29
(2.58)
Drinking Water Supply
2.67
(2.79)
Perceived Improvement (Over Past 12 Months) in:
Basic Health Facilities
Schools
Roads
Drinking Water Supply
Prob. of
Nazim
Getting Reelected
1.01
(0.62)
0.01
(0.97)
0.54
(1.08)
0.702
(1.78)
Sources: PSLM 2004/05 and data from the Election Commission of Pakistan.
Note: Logit regression; standard errors in parenthesis.
Electoral manipulation is also a factor that cannot be ruled out. The provinces with high incumbency
disadvantageSindh, Khyber Pakhtunkhwa, and Balochistanalso had a majority of incumbent district nazims
that were in political opposition to the ruling provincial coalition, and anecdotal evidence does suggest that in
part their failure to get re-elected was due to the provincial manipulation of the votes of their electoral colleges.
140
Zahid Hasnain
141
amount in the ratio of 58:42, the latter of which is transferred to the local governments
into their account (Account 4). This ratio is determined on the basis of expenditure
baselines of the province and local governments, taking into account the obligatory
expenditures of the province.
The provincial allocable amount is transferred to the local government broadly
under two separate block grants, one for recurrent expenditures, which constitutes
approximately 88 percent of the allocable amount, and the remainder for development
expenditures, with no de jure discretion to local governments to re-allocate across these
two windows.6 The development grantRs 12 billion in 2006-07is distributed
horizontally across the districts on the basis of a formula that gives equal weights to
population and backwardness.
As Figure 5 shows, due to overall improved resource availability, transfers to local
governments in Punjab have been increasing steadily at an annual rate of approximately
15 percent over the past five years, with growth rising to 18 percent and 22 percent in
2005-06 and 2006-07. In 2006-07, approximately Rs 100 billion was transferred to local
governments, with districts receiving Rs 84 billion, TMAs Rs 12.7 billion, and Unions Rs
4 billion. Given that salary expenditures have increased at a lower rate, districts have
been able to utilise some of the savings from the recurrent grants to fund development
activities, despite the official earmarking of these transfers.
Fig. 5. PFC Transfers in Punjab
PFC transfers to local governments (Rs. million)
120,000
100,000
80,000
60,000
40,000
20,000
0
2002/03
2003/04
2004/05
Districts
TMAs
2005/06
2006/07
Unions
This fiscal space is unique to the districts of Punjab; however even here it is
overshadowed by the provincial presence in the districts, particularly in the social sectors.
Transfers from the Provincial Retained Amount: The province makes a number of
additional transfers to local governments from the provincial retained amount and outside
6
To be precise, the 2.5 percent of the General Sales Tax is added to the Provincial Allocable and the
total amount is transferred under four grants: a General Purpose Grant to meet current expenditures; an
Equalisation Grant to meet any shortfalls from the baseline expenditures after distribution of the General
Purpose Grant; a Development Grant to meet development needs; and a Tied Grant to meet the social sector
priorities of the provincial government.
Zahid Hasnain
142
the framework of the PFC. The most significant at the district level are the tied or
conditional grants for education and health under the provincial Education Sector Reform
Programme (ESRP) and Health Sector Reform Programme (HSRP) respectively, which
are earmarked for the improvement of school and health facility infrastructure. These
tied grants are distributed among the districts on a formula based on both need and
performance. Additional funds for the social infrastructure improvement are also
transferred to local accounts on a more ad hoc basis under the Chief Ministers
Accelerated Programme for Social Development (CMAPSD). The allocations in the
provincial public sector development programme for ESRP, HSRP, and CMAPSD
(education and health components) in 2006-07 were Rs 5 billion, Rs 1.3 billion, and Rs
5.5 billion respectively.
An additional significant source of local development work is through federal
vertical programmes such as the Khushal Pakistan Programme, the federal Education
Sector Reforms (ESR), and the Presidents Programme for the Improvement of
Watercourses. These funds are usually tied to certain pre-defined interventions, and are
administered by the concerned DCO or EDO, usually through Personal Ledger Accounts
held outside the provincial treasury, again with little or no involvement of the local
political leadership.
In 2006-07, these and other non-PFC development transfers amounted to Rs 14.5
billion or larger than the local governments allocation under the block development
grant.7 The analysis of district expenditure priorities has to take into consideration this
large provincial presence, particularly in the sectors that have been de jure devolved to
local governments.
Local governments are also involved in executing provincial development schemes
through the provincial Account 1. Specifically, the concerned department will, through
the respective EDOs, implement these projects. There is no involvement of the district
political leadership in the execution of these schemes.
On the recurrent side, the province also transfers resources for the recruitment
of government employees against vacant posts, and to fund salary increases
mandated by the federal government that cannot be financed under the provincial
allocable amount.
District Development Priorities in Punjab
Four stylised facts emerge from an analysis of provincial and district development
expenditures in Punjab. First, physical infrastructure, in particular roads, is by far the
highest priority of the district governments; second, these infrastructure schemes are
small and largely neighbourhood-specific; third, district policy-makers appear to attach a
lower priority to operations and maintenance than their provincial counterparts; and
finally provincial interventions in education and health appear to have provided
additional incentives for districts to prioritise the physical infrastructure sectors.8
7
Punjab (2007).
This paper focuses on budget allocations as opposed to actual expenditures. The reason is that only
the budget data allows for a separation of what development schemes in the district ADB are being financed by
the district (i.e., from the block development grant, own source revenues, and any savings from the recurrent
grant) and what are being financed by the province.
8
143
Health
Health
Rs 447m,
Rs.447m,
2%2%
Others
Others
Rs 4721m, 21%
Rs.4721m,
21%
Roads
Roads
Rs.7274m,
Rs 7274m,32%
32%
Rs 22,640
Rs.
22,640Million
Million
CCBs
CCBs
Rs Rs.
5987m,
26%26%
5987m,
Electrification
Electrification
Rs
6%6%
Rs.1289m,
1289m,
Water&&sanitation
Sanitation
Water
Rs.
Rs1769m,
1769m,8%
8%
The other defining feature of the district ADP is the small size of the typical
scheme. As Figure 7 left panel shows, there were, excluding block allocations, over
17,000 schemes in the development portfolio of these districts in Punjab. Of these
9
Given that only Rs 12 billion was transferred by the province under the Development Grant in the
PFC, it is likely that actual utilisation of the district ADPs would be much less.
10
As per the LGO, a CCB is to be a non-elected voluntary organisation, consisting of at least 25
members, which can be established for a variety of purposes, including initiating and improving development
projects, establishing cooperatives, forming monitoring bodies over police and other service providers and
reinforcing the capacity of monitoring committees at the behest of the concerned council. At least 25 percent of
the total development budget of each tier of local government (district, tehsil, and union) must be earmarked for
projects identified by CCBs, and each CCB has to make a cash contribution of 20 percent in order to tap into
these funds for a specific project.
11
As per the LGO, water and sanitation is the responsibility of tehsils; however district governments
are also spending in the sector.
Zahid Hasnain
144
4685 were roads schemes, 6248 water and sanitation schemes, 1586 electrification
schemes, 880 education, and 201 health schemes. The size of a typical road scheme
was roughly Rs 1.5 million, and of a water and sanitation scheme only Rs 0.3 million
(Figure 7, right panel). These were typically neighbourhood-specific schemes
construction or rehabilitation of a small road or drains (what is referred to as
soling).
Fig. 7. District ADPs in Punjab, 2006-07: Number and Average Size of Schemes
Average allocation per scheme (Rs.)
1827
2,174,010
2,250,000
1976
Union schemes
2,000,000
1,750,000
1586
Electrification
1,500,000
1,481,089
1,287,978
1,250,000
6248
1,000,000
Roads
4685
Education
733,861
750,000
500,000
880
280,508
250,000
Health
201
0
Education
17,403
Health
Roads
Water &
sanitation
Rural
electrification
It should be noted that due to a number of block allocations the above is not a
complete picture of sectoral district development spending. For example, most
districts provide a block allocation for schemes to be identified by the union
councils, and anecdotal evidence suggests that these are also mostly small
infrastructure projects the details of which are unknown. Similarly, the details of
CCB schemes, where formed, are also not available. Therefore the total number of
schemes is likely to be larger, and the focus on physical infrastructure greater, than
what is portrayed in these figures.
How do these allocations compare to that of the provincial government in the
post-devolution period? Figure 8, provides the average sectoral composition of the
provincial ADP for 2001-02 to 2006-07. Roads and large-scale infrastructure
projects are also a priority for provincial policy-makers, but education also features
as a priority sector. Specifically on average 27 percent of the ADB was allocated to
roads, 16 percent to education, 12 percent to irrigation, and 5 percent to health. The
provincial government has therefore prioritised social sectors much more than the
district government.
The average provincial schemes were considerably larger than their district
counterparts. In total, there were about 2400 schemes in the provincial ADP of 2006-07,
with 623 roads, 286 water and sanitation, 174 health, and 149 education schemes. These
translated into average scheme sizes of for example Rs 22 million for roads and Rs 13
million for water and sanitation.
145
Block alloc.
9%
Education
16%
Health
5%
Others
24%
Roads
27%
Irrigation
12%
W&S
7%
48% annual
growth
250
160
200
140
13% growth
120
16% growth
150
100
2% growth
80
12% growth
100
60
40
50
20
0
2002/03
2003/04
Education
2004/05
Health
2005/06
Community services
2002/03
2003/04
Education
Health
2004/05
Community services
146
Zahid Hasnain
This relative neglect of non-salary expenditures, combined with the large size of
the consolidated district ADPs relative to the block development grant transfer in the
PFC, suggests that districts have reallocated funds from the non-salary PFC grant to fund
development expenditures.
The greater focus on physical infrastructure investments, the emphasis on small,
localised schemes, and the relative lack of prioritisation of operations and maintenance
suggests that incentives for providing visible, targeted benefits are even higher for local
policy-makers as compared to their provincial counterparts. Water and sanitation and
village electrification in particular exemplify this targeting. As Cheema and Mohmand
(2006) found in villages they surveyed in Punjab, sanitation and drain provision was
being targeted to specific households even within a lane in a given village. Similarly,
village electrification in many cases is actually the provision of electricity connections to
particular households in a village.
Given the considerable direct contact between citizens and their local political
representatives elaborated on in Section IV, and the relatively greater demands of citizens
for improvements in roads, water, and electricity as compared to education and health,
this emphasis on physical infrastructure could simply be reflective of the relatively
greater responsiveness of local governments. It could also be an outcome of the local
government political structure. As also discussed in Section IV, district nazims have
responded to their electoral dependency on union councillors by allocating some portion
of their development funds to each union nazim. This parcelling of the development
budget into small portions would automatically result in a bias towards small-scale
infrastructure schemes targeted to the village and neighbourhood that are the
constituencies of these councillors.
These local priorities are also significantly influenced by the actions of the province,
particularly in the sectors that are, as per the LGO, local responsibilities. As discussed, over
the past four years, the Punjab government has launched large-scale reform programmes in
the education and health sectors. The Punjab Education Sector Reform Programme is a
high-profile programme of the government, with a number of interventions, including the
provision of missing facilities for primary schools, provision of free textbooks to public
primary and middle schools, and the provision of stipends to female middle and secondary
students in 15 low literacy districts. The funding for the provision of missing facilities is
transferred to local governments as tied grants. A total of Rs 5 billion were allocated for
this component in 2006-07, and distributed across districts on the basis of need and
performance. The mechanics for scheme identification and execution are as follows: the
districts provide the provincial government with the total number of missing facilities, and
the prioritisation within this list is done by the concerned MPAs of the particular district.
Therefore, while there is some role for district governments in identifying the schools that
will be provided these facilities, the decision about which schools will be targeted in a given
year primarily rests with the provincial political leadership. In the initial years of the
programme, these schemes were implemented by the works and services departments of the
concerned district. Starting in 2006-07, implementation has been contracted out to the
National Logistics Corporation (NLC), an agency outside of the Punjab public
administration. The districts therefore no longer have any role in the execution of these
schemes, and serve mainly as a conduit for funds.
147
The Health Sector Reform Programme similarly provides tied grants for missing
facilities, albeit on a smaller scale, with the schemes again identified by the concerned
MPAs. As in education, the implementation of these schemes as also been contracted out
to the NLC. A total of Rs 1.3 billion was allocated for this programme in 2006-07 in the
provincial PSDP, of which Rs 0.9 billion was for the provision of missing facilities in
BHUs and RHCs.
In addition to these tied grants, the CMAPSD is another major intervention in the
provincial PSDP in the devolved education and health sectors. Of the total allocation of
approximately Rs 5.5 billion for the education and health components of this programme,
Rs 2.6 billion is for investments in schools, and Rs 0.8 million for infrastructure
improvements in local hospitals, BHUs, and dispensaries, with the remainder targeted at
the higher education sector (a provincial responsibility). While these are not referred to
as tied grants, as in the ESRP and HSRP individual schemes in this programme are
identified by the provincial political leadership, but with the districts involved in the
execution of these schemes through their works and services departments. The only
substantive difference between these programmes is that CMAPSD does not have a
formula-based allocation mechanism to the districts and is instead distributed in a more
ad hoc manner.
In total therefore, the provincial ADP allocated Rs 7.6 billion and Rs 1.7 billion for
primary and secondary school and health infrastructure, far greater than the Rs 1.1 and Rs
0.4 billion that the districts are spending in these devolved sectors. This predominance of
provincial priorities at the local level is evident in Fig. 10, which displays per capita
Fig. 10. Per Capita Allocations in the Districts ( 2006-07)
Per capita education development allocations (Rs.)
450
400
100
350
80
300
250
60
200
40
150
100
20
50
Province
Sialkot
T.T.Singh
T.T.Singh
R.Y.Khan
Okara
Rajanpur
Narowal
Mianwali
Lahore
M.B.Din
Khushab
Jhang
Jhelum
Gujrat
Hafizabad
District
Province
R.Y.Khan
Rajanpur
Nankana Sahib
Multan
Mianwali
M.B.Din
Lahore
Khushab
Khanewal
Kasur
Jhelum
Jhang
Hafizabad
Gujranwala
Chakwal
Faisalabad
Bhakkar
Vehari
T.T.Singh
Sialkot
Sargodha
Sheikhupura
R.Y.Khan
Rawalpindi
Rajanpur
Okara
Narowal
Pakpattan
Mianwali
Nankana Sahib
Multan
Layyah
M.B.Din
0
Lahore
50
0
Kasur
100
100
Khushab
150
200
Khanewal
200
300
Jhang
250
400
Jhelum
300
500
Hafizabad
350
600
Gujrat
400
700
Faisalabad
450
800
Gujranwala
900
Bhakkar
Province
Chakwal
Tied grants
Sheikhupura
Bahawalnagar
Chakwal
District
1000
District
Faisalabad
Bhakkar
T.T.Singh
Sialkot
R.Y.Khan
Okara
Province
Bahawalnagar
Tied grants
Rajanpur
Narowal
Mianwali
Lahore
Khushab
M.B.Din
Jhang
Jhelum
Gujrat
Hafizabad
Chakwal
Faisalabad
Bhakkar
Bahawalnagar
District
Bahawalnagar
Zahid Hasnain
148
allocations in education, health, roads, and water and sanitation by the district and
provincial governments in districts on which data was available. Tied grants, the
CMAPSD, and provincial schemes executed from Account 1 form by far the largest
source of development expenditures in education and health in the districts for which data
was available (Figure 10, top panels). On average these are five to six times the district
allocations.
The provincial presence in the districts in the infrastructure sectors is also large,
but not as predominant (Figure 10, bottom panels). In roads the province and districts are
allocating on average roughly equivalent amounts, while in water and sanitation district
allocations form a larger proportion.
It is not unreasonable to presume that these provincial interventions have an
impact on the decision-making of local policy-makers. Credit claiming for a particular
service becomes difficult if multiple tiers of government are providing the same service.
This distortion can then create incentives for politicians to focus on targeted benefits or
on more visible interventions.
Figure 11 provides some clues as to why districts are focusing on small-scale
infrastructure. Since both the education and health tied grants are for the provision of
missing facilities, there is very little distinction between these and district schemes, as
indicated by the similar average size of these schemes. By contrast, the district is able to
distinguish itself much more from the province in roads and water and sanitation by
focusing on smaller schemes. The average district road scheme is one-tenth the size of a
provincial scheme, and the average water and sanitation scheme one-twentieth the size.
The reason it can do so is that physical infrastructure is intrinsically more
heterogeneous than social infrastructure roads and drains come in all sizes whereas
primary schools and basic health units are all quite similar. This heterogeneity allows for
specialisation, enabling district policy-makers to focus on and to take credit for the
small, neighbourhood schemes and provincial policy-makers to take credit for the larger
ones.
Fig. 11. Average Size of District and Provincial Schemes
Average size of development schemes executed in specific
districts (Rs.Thousand)
16,000
14,839
14,000
Province
12,000
10,000
Province
8,000
5,786
Tied grants
6,000
Tied grants
4,000
3,102
2,174
2,000
1,288
1,481
930
281
0
Education
Health
District
Roads
Tied grants or province
149
The above data also suggests that many districts have responded to the provincial
initiatives in education and health by essentially ceding all responsibility for these sectors
to the province. For example, districts, such as Bhawalnagar, Mianwali, Narowal,
Rajanpur, and Toba Tek Singh allocated either zero or negligible development funds to
health in their ADPs in 2006-07. Similarly, Bhakkar, Bhawalnagar, Jhelum, and
Mianwali had no education schemes in their ADPs.
This replacement of district social sector funding by the province is also evident in
the sectoral trends in the provincial and district ADPs (Figure 12). Provincial real
development allocations for roads and education have increased significantly in the postdevolution period. Road allocations have risen fourteen-fold in real terms between 200102 and 2006-07, and education allocations six-fold (Figure 12, left panel). By contrast,
consolidated district education ADP allocations (based on available data from 13 Punjab
districts) have declined by approximately 50 percent in real terms from 2003-04 to 200607. District health allocations have also declined in real terms over this period, while
provincial health allocations have roughly doubled from 2001-02. Districts prioritisation
of physical infrastructure is again evident, with roads allocations increasing by 70 percent
and rural electrification by 30 percent.
Fig. 12. Trends in PunjabReal Setoral, District, and Provincial ADP Allocations
Growth in provincial real development allocations
1500
1400
1300
180
Roads
160
1200
140
1100
1000
Electrification
120
900
800
Education
700
100
80
600
Health
500
60
400
Health
300
Education
40
200
W&S
100
0
2001/02
2002/03
2003/04
2004/05
2005/06
2006/07
20
0
2003/04
2004/05
2005/06
2006/07
The fact that district allocations in education and heath have declined over time
suggests that relative lack of emphasis on social infrastructure is not simply due to lower
citizen demand (one would not expect these to change in the short run), or the structural
constraints imposed by the local government electoral system (which are constant). It
reinforces the finding that the general unforeseen impact of the provincial tied grants is
an even greater incentive for local politicians to emphasise physical infrastructure, a
prioritisation that is understandable given the problems of credit-claiming in social
infrastructure.
VI. CONCLUSION
This paper explored the linkage between devolution, accountability, and service
delivery in Pakistan by first examining the degree of accessibility of local policy-makers
and level of competition in local elections, and second the expenditure priorities of local
150
Zahid Hasnain
governments, and the extent to which they were focused on patronage as opposed to the
provision of public goods.
The available evidence suggests that the direct accessibility of local policy-makers
to citizens is considerably greater than that of provincial and national policy-makers,
allowing the public many more channels to communicate their demands to the
government. It is also the case that union council elections in general, barring some
notable regional variations, are as competitive as provincial and national elections. The
high incumbency disadvantage of district nazims also suggests that lite capture is not a
significant problem at the district level. However, the indirect elections of the district and
tehsil leadership imply that there are some question marks around the electoral
accountability of these key policy-makers.
This greater accessibility has not translated into incentives for improved social
service delivery. Local government priorities are heavily tilted towards the physical
infrastructure sectors, and within physical infrastructure to small schemes that can be
targeted to specific localities and even households. The fact that roads and water and
sanitation schemes are small and neighbourhood specific indicates that local policymakers are focusing on areas of spending for which they can get maximum visibility.
This conclusion is also evident in the emphasis on development expenditures at the
expense of operations and maintenance expenditure. This in part may be an outcome of
their relatively greater accountabilityas discussed, most citizens contact councillors for
personal favours or for physical infrastructure. It may also be a product of the indirect
election system, as district nazims are dependent on the support of union councillors, and
the latters constituency is the village and neighbourhood. It is also, as elaborated at
length, a response to the provinces focus on the social sectors.
If the provincial government has taken on the agenda for education and health, and
the district is focusing on physical infrastructure, then is that necessarily an inefficient
outcome? As discussed in Section III, administrative devolution remains incomplete in
Pakistan, and local governments have little effective authority over the provincial staff
assigned to them. Therefore, given that personnel management is such an important
feature of effective delivery of education and health service, this de facto re-centralisation
may indeed produce beneficial outcomes.
The counterargument to this concerns sustainability. Given that it will be the
districts responsibility to maintain these social infrastructure schemes, what incentives
would they have to keep them going if they have had little say in their identification and
construction? One of the major criticisms of the Social Action Programme of the 1990s,
under which the government and donors invested approximately $9 billion in the social
sectors with very little to show in the way of outcomes, was precisely that the programme
was too centralised and that the infrastructure investments did not take into account local
conditions. So there is a risk that the same mistake is being repeated.
The main policy implication of these findings is the need for coherence in the key
accountability relationships delineated in Figure 1. If administrative devolution is
unlikely to take place, and this appears to be the case in Pakistan, then recentralisation of
education and health may indeed be a sensible strategy, but then this should be complete
recentralisation, and the province should also be responsible for the recurrent budget in
these sectors. Given the citizen pressures for provision of physical infrastructure, this
151
recentralisation could be part of a bargain and agreement between the province and local
governments that (intra-district) roads and water and sanitation services would then be an
area the province would not interfere in and where local governments would have
considerable expenditure autonomy.
REFERENCES
Ahmad, Junaid, Shantayanan Devarajan, Stutu Khemani, and Shekhar Shah (2004)
Decentralisation and Service Delivery. World Bank. (Mimeographed).
Bardhan, Pranab and Dilip Mookherjee (2004) Pro-poor Targeting and Accountability of
Local Governments in West Bengal. Institute for Economic Development, Boston
University. (Working Paper).
Besley, Timothy, Rohini Pande, Lupin Rahman, Vijayendra Rao (2004) The Politics of
Public Goods Provision: Evidence from Indian Local Governments. Journal of the
European Economic Association Papers and Proceedings 2: (2-3), 416426.
Cheema, Ali and Shandana Khan Mohmand (2006) Bringing Electoral Politics to the
Doorstep: Who Gains who Loses? Lahore University of Management Sciences.
(Mimeographed).
Devolution Trust for Community Empowerment (DTCE) and CIET (2005) Social Audit
of Governance and Delivery of Public Services: National Report. 200-05. Islamabad:
DTCE.
Faguet, Jean-Paul (2001) Does Decentralisation Increase Government Responsiveness to
Local Needs? Decentralisation and Public Investment in Boliia. Centre for Economic
Performance, London School of Economics. (Working Paper No. 999).
Keefer, Philip (2002) Clientelism, Credibility, and Democracy. Development Research
Group. World Bank. (Mimeographed).
Keefer, Philip (2004) What Does Political Economy Tell Us about Economic
Developmentand Vice Versa? Washington, DC. (World Bank Policy Research
Working Paper No. 3250).
Keefer, Philip and Stuti Khemani (2003) Democracy, Public Expenditures, and the Poor.
Washington, DC. (World Bank Policy Research Working Paper No. 3164).
Pakistan, Government of (Various Issues) Accountant General of Punjab. 2002-030607. Monthly Civil Accounts.
Pakistan, Government of (2005) Pakistan Social and Living Standard Measurement
Survey (PSLM), 2004-05. Islamabad: Federal Bureau of Statistics.
Persson, Torsten and Guido Tabellini (2000) Political Economics. Cambridge, MA: MIT
Press.
Punjab, Government of (2007) White Paper on the Budget, 2007-08. Lahore: Finance
Department.
Punjab, Government of (Various Issues) Annual Development Programme. Planning and
Development Department. 2001-0206-07. Lahore, Government of Punjab.
Triesman, Daniel (2002) Decentralisation and the Quality of Government. University of
California, Los Angeles. (Mimeographed).
World Bank (2001) Indonesia Poverty Report. Washington, DC: The World Bank.
152
Zahid Hasnain
World Bank (2003) Making Services Work for Poor People: World Development Report
2004. Washington, DC.: The World Bank.
World Bank (2004) Devolution in Pakistan: An Assessment and Recommendations for
Action. Washington, DC.