Fiscal policy instruments and the political economy of designing programs to reach the poorest
Ehtisham Ahmad October 2007
The views in this paper are personal and do not necessarily represent those of the IMF.
The issues Information needed to identify the
poorest Generally those unable to participate in the
labor market Marginal groups without family support Often excluded from formal programs:
pensions or unemployment insurance Information available to localities, but
not easily to the center/donors
The issues (continued) Local officials: may not have the
appropriate incentives to implement “central programs” “Capture by elites” Absence of information on “outcomes”
In addition, overlapping responsibilities: typically allow the local governments
to “pass the buck” to the center
Policy instruments: taxation Trend to centralize broad based
taxes (VAT, CIT); e.g., China and Mexico
Need for some own-source revenues at margin for accountability
Poorer areas may lack resources to meet poverty spending
Policy instruments: transfers Equalization transfers used in many
advanced countries (Australia, China) But may not ensure that needs of the
poorest are met in all cases Scope for targeted interventions?
Central government operations Could be costly and administratively demanding
Central financing and local operations Likely model: but possibility of leakages Political economy considerations and policy design
The policy agenda: examples from Mexico and China Can one use insights from
developments in method to design programs to support the poorest? Use of competition—both across regions as
well as over time Importance of defining program objectives;
monitoring mechanisms/ Contracts—enforceability? Moves towards performance budgeting?
Sequencing of measures to move in this direction
Some experiences
Local provision: China Wu bao (five guarantees)—minimum basic
guarantee for those without family support (largely rural)
Reasonably well targeted Based on community identification and support Variance based on community/local government resources Together with anti-poverty program, covered 6 percent of
the rural population in early 1990s. Variance in level of local support increases after
market based reforms Increasing difficulties in getting local governments
to implement central programs Grain silos (federal programs, financed by center,
administered by local governments)
Central provision:Mexican insights Mexican Progresa/Opportunidades
regarded as successful central anti-poverty program
Together with Seguro Popular to top up health care for uninsured Direct Federal administration Coverage: 70 percent of rural population; 25
percent of total population Disincentives to participate in the formal sector
Replication of Communidades Red Solidaridad (El Salvador); Tekoporã
(Paraguay) Also direct central provision Attempts to involve local governments not
successful Kudos to center; what’s in it for local politicians? Clear example of “overlapping responsibilities”
What could be done by the center to better utilize local information and implementation? Cost effectiveness is important
Hybrid: central finance, local provision Di bao; largely urban program in China, since
1999 local selection of recipients Variable eligibility criteria Central funding—0 in rich areas, 100 percent in Tibet
Protects center from open ended program, but possible exclusion of the poorest
Possibility of tighter central determination of criteria, without risk of open-ended costs?
Minimize incentives for local governments to misrepresent needs and divert resources
Method Draws on “competition” models
Used horizontally in the external aid literature Failure of conditionality in foreign aid Svensson J.
(1999) and (2003) Multi-tiered nested decision-making (Ahmad et
al, 2002) Intertemporal developments (simple 3-period
model—Ahmad and Martinez, 2004) Specific/conditional transfers to local governments Appropriate feedback mechanisms or penalties Inherent in multi-year budgets where future funds
depend on intermediate performance targets.
Main result Competition between recipient units
and freedom of allocation over time (multi-year budgeting) reduce diversion of transfers towards local goals
Requires: Clear definition of “outcomes” that can be
monitored Rather than just tracking of funds Multi-year budgets
The model Each year the central government
distributes: a transfer,2T, to regions A and B
A and B can, and prefer to, divert a share, maximum of F, of the transfer to their own goals
Center cannot monitor easily the use of funds/extent of diversion But can tell if there are no “grain silos” being
built Number of wu bao individuals/families housed
PoliticalA and B
No regional, no intertemporal competition
Regionalcompetition
Regional and intertemporal competition
1. First year in office
Each receives T, can divert maximum fraction F of T
Each receivesT, can divert maximum fraction F of T
Each receivesT, can divert maximum fraction F of T
2. Last year in office
Same Depending on use by A and B, transfer can be allocated totally to A or B. Winner receives 2 T
Depending on use by A and B, transfer or 4T is allocated to winner, or moved to period 3. Winner receives 4T
3. First year in office
Same as in 1. Same as in 1. Transfer can be 0, or 4T
Results Maximum diversion of transfer, F.
Diversion Less than F, but not zero
Still less diversion, with possibility of zero diversion. T totally spent for CG objective.
Conclusion: Options Direct central provision of the programs
Possible, but could be costly and demanding on administrative resources
Local provision, with own-financing, leads to accountability, but Variability of resources—poor regions worse off
Central financing of earmarked transfers, with local administration Likelihood of diversion of resources, but....
...there is an alternative Introduction of competition plays a role in
effective transfer design, Even in the presence of overlapping competencies Horizontal competition (using foreign aid literature
seen insufficient to remove incentives to divert) Multi-period budgets with future funds dependent
on demonstrable outcomes the most effective instruments
Can credible contracts be designed? (i.e., will funds be cut-off for non-performance?)
Conclusions Contracts should specify “outcomes”,
introducing relative performance criteria Can be developed over time, in moves
towards performance budgeting Preconditions: better monitoring of both non-
financial and “costing” elements More flexibility in cases where clarifying
responsibilities may be a lengthy process (e.g., both Mexico and China)