Bert Hofman
and Kai Kaiser World Bank1
Paper Presented
at the Conference: CAN DECENTRALIZATION HELP
REBUILD INDONESIA?
A Conference
Sponsored by the International Studies Program,
Andrew Young School of Policy Studies,
Georgia State University
May 1-3 2002
Atlanta,
Georgia
Indonesia’s
new intergovernmental fiscal system devolves on aggregate enough resources
to cover the devolved expenditure responsibilities. But the intergovernmental
fiscal system is as of yet far from ideal. The distribution of
resources and tasks has caused budgetary problems in some of the regions,
especially at the provincial level. The regions’ high dependence
on central transfers could undermine local accountability. Inadequate
provisions for local taxes risks inappropriate taxation and unhealthy
tax exporting. And finally, the system has few means for central
government to finance national priorities at the local level.
Law 25 of 1999
meant fundamental reforms of Indonesia’s intergovernmental fiscal
relations. The reforms strongly increased the regional government’s
share of government resources, moved the transfer system from one dominated
by earmarked grants to one largely relying on general grants supplemented
by revenue sharing, and—with the reforms introduced by law 34/2000—gave
broad taxing authorities to local government.
Before the
2001 decentralization, most resources were transferred from central
to regional governments through earmarked grants. The largest
of these was the SDO (Subsidi Daerah Autonom or subsidy for autonomous
region) grant which covered all civil service salaries and recurrent
expenditures for the regions. In addition, INPRES (Instruksi President)
grants aimed to finance development spending in the regions. The
INPRES grants started as a block grant for development spending in the
1980s, but gradually evolved into an array of specific grants for purposes
ranging from re-greening18
In the new
system, central regional transfers remain the dominant means of financing,
but the earmarking is gone. The bulk of regional government spending
is financed by transfers from the center (see Table 1:DAU Dominates).
Well over 90 percent of regional revenues come from the Balancing Fund
(dana perimbangan) which includes a general grant (the Dana Alokasi
Umum or DAU), shared taxes, natural resource revenues (SDA, sumber daya
alam), and a special allocation grant channel (DAK, dana alokasi khusus)
Local governments have limited own revenues (PAD, pendapatan asli daerah),
which constitutes less than 7 percent of total revenues. Starting
2002, the center is also making additional special autonomy transfer
arrangements with two provinces.
Dana Alokasi
Umum. The Dana Alokasi Umum (DAU) or general grant is the mainstay of
the new intergovernmental fiscal system. The DAU adds up to some
65 percent of regional revenues, and to a little over 70 percent of
the Balancing Fund. The DAU is by law a minimum of 25 percent
of central government revenues after tax sharing.19
For 2001 and 2002 this minimum allocation has been maintained by Government
and Parliament. However, although the law and the regulations
suggest that the 25 percent is the share of actual revenues after revenue
sharing, for FY2001 the budgeted amount was disbursed. In
all, this cost the region some Rp. 9 Trillion in revenues, or 15 percent
of the total DAU for that year.20
Because the
“hold harmless” element was interpreted to be a minimum DAU allocation
rather than a guaranteed amount, this element took almost 80 percent
of the total DAU. In 2002, the minimum DAU was reduced to 50 percent
of the total amount, but rather than relating it to past SDO and INPRES,
is became a minimum amount per region, plus an amount related to the
actual wage bill of 2001. But “hold harmless” obtained a new meaning:
Parliament objected against the proposed distribution of the DAU, because
the richer regions stood to lose compared to the 2001 distribution.21
The formula
part of the allocation relies on the notion of expenditure needs and
own fiscal capacity. The share in the DAU pool for a region depends
on the difference between its fiscal needs and its fiscal capacity.
For 2001 there concepts were interpreted different from 2002, in part
due to practical reasons, in part due to more analysis done for 2002.
In 2001, at the time the formula had to be presented to the Regional
Autonomy Advisory Council, the data on shared revenues were not yet
available, and it was decided to ignore them. For 2002 they were included,
but natural resource revenue shares only for 75 percent. As indicators
for expenditure need the formula includes (i) population; (ii) poverty
rate; (iii) land area; and(iv) the construction price index as an indicator
of “geographical circumstances.” The formula must include these
variables, as they arementioned in the elucidation of Law 25/99.
In the 2001 formula each of the variables was included with equal weight,
whereas in the 2002 formula, population and area both received higher
weights than the others.
Contingency.
The DAU allocation was supplemented by a “contingency fund” to absorb
any mismatches between devolved expenditure responsibilities and revenues.
Of a budgeted amount of Rp. 6T. in 2001, some Rp. 3 T was disbursed.
The first tranche of Rp. 1.1 T. related to genuine mismatches caused
by decentralization. A process of application, review and allocation
set out in a Presidential decree was followed for this tranche.
The second tranche, however, became necessary because of the centrally
mandated salary increase which pushed up the regional wage bill by some
15-30 percent.
Shared revenues.
The 2001 decentralization greatly increased the importance of shared
revenues. The most important factor was the inclusion of oil and
gas revenues and personal income tax in the taxes to be shared.
The former were included to accommodate long-standing dissatisfaction
of natural resource rich regions which felt that “Jakarta” took
their resources, and they did not get anything in return. True
or not, with the implementation of Law 25/1999, they now get a significant
share of those revenues (see Table 2: Revenue shares). In addition,
the personal income tax was included for sharing through Law 17 of 2000.22
For each of these shared taxes, the province gets a minor part, whereas
the bulk of revenues goes to the local governments.
The sharing
formulae for most of the shared revenues contain an additional element
of equalization. For oil and gas, mining, and forestry, the local
governments of regions neighboring the producing region receive a share
as well. For fisheries, property tax and land transfer tax, a
small percentage of the revenues is shared by all local governments
in Indonesia. Whereas the underlying motivation may well be one
of equalization, with the initiation of a formula-based DAU, these complex
sharing mechanisms may well be redundant—whatever a region gets from
those shared taxes is counted as own fiscal capacity, and reduces the
allocation of the DAU.
Own Revenues.
Law 34/2000 greatly expands the scope for local government revenues.
The law amended law 18 of 1997, which intended to stop the then-prevailing
local government practice of issuing a plethora of local government
taxes, many with little revenue potential, and high costs to the taxpayer
and the economy. Law 18/1999 therefore restricted regional taxes
to a closed list, and made any additional taxes conditional upon approval
of the Ministry of Finance.
Law 34 reverses
the burden of proof. The law still gives a list of regional taxes,
but regional governments can add taxes through regional regulations
approved by the regional government council, as long as it abides by
the principles mentioned in the law. These principles are sound
(Box), but supervising them has turned out to be problematic—not least
because the law has tight deadlines for central government to meet if
it wants to cancel a local tax. An added complication is the way
supervision is structured: Law 22/1999 gives the Minister of Home Affairs
the authority to cancel regional regulations, including those on regional
taxes. Up until now, the Minister has been hesitant to invoke
these powers, not least because regional governments have the right
to appeal his decision to the Supreme Court. As a result, there
has been little to check regional government’s creativity in taxation,
and although the damage still remains limited, 84 out of the more than
1,000 regulations on local taxes have been found to be in conflict with
the law. Among them are taxes on the “import” of goats into
kabupaten Bogor, and an advertisement tax on Coca-Cola bottles in Lampung
province. Meanwhile, the Minister of Home Affairs has as of now formally
cancelled only one regional tax.
Do the Regions
Get Enough?
A key question
for the new intergovernmental fiscal system is whether the regions on
aggregate receive enough resources. This question can be
considered in three ways: (i) do the regions receive enough resources
to cover the expenditures needed for the tasks they are expected to
perform? (ii) do the regions receive an amount compatible with what
government as a whole can afford? and (iii) do the regions receive enough
to cover the spending obligations they inherited from the central government
in the course of decentralization.
Method (i),
sometimes labeled the costed minimum standards approach, has practical
and theoretical issues. For starters, as was argued previously,
Law 22/1999 does not clearly define the functions of the regional governments,
as the functions are defined negatively: local government does everything
that the center and the provinces does not do. And for the obligatory
functions of local government, which are defined in the law, it does
not clearly define what part of the function local government performs,
not what minimum standards of services should be delivered. Even
if these issues could be overcome, the in formation to cost out the
functions is lacking at present. Moreover, determining what the
functions cost at present may not be very telling for what the functions
should cost if efficiently delivered. But apart from all these
practical objections, there is a more fundamental objection against
this method: unless carefully managed, minimum standards are but a wish
list of spending developed independently of what government as a whole
can afford.
Method (ii)
the affordability approach faces several issues as well.
The method requires the Indonesian government to make choices for the
nation as a whole as to what it wants to spend its scarce resources
on. If the priorities so determined are tasks of regional government,
then more resources would need to be devolved—be it through grants,
revenue sharing, or devolution of more tax bases to regional governments.23
Although the method is to be preferred over the costed minimum standard,
there are numerous practical impediments, not least the lack of information
in the current budget and accounting system which does not allow a link
between policy goals and spending.
Central government
could devolve more resources if it wanted to do so. Currently,
a significant part of its spending is devoted to tasks that could be
considered local government tasks. Taking the 2002 budget as a
guide, the development budget still contains as much as 10-20 trillion
or \[1 percent] of GDP in spending which could be further devolved to
the regions, together with a corresponding increase in revenues.
Note that implementation of these projects is already largely done at
the sub-national level. However, since the financing is done from
the central budget, there is no local scrutiny over the spending.
On the recurrent budget, the wage bill probably offers further scope
for savings, as not all civil servants that ought to have been decentralized
actually were. Moreover, the Government has already decided to
phase out the fuel subsidies over time, and this will further free up
resources that could be made available to the regions. And finally,
the government is determined to increase the tax ratio to GDP over time.
One quarter of that increase will already be automatically transferred
to the regions through the DAU, but more could be made available to
the regions.
Whether increased
resources should be made available remains to be seen. First,
several areas of central government’s own responsibility that have
been chronically underfunded, most notably Operations and Maintenance.
Second, the central government is aiming for a zero budget deficit by
fiscal year 04, and achieving this goal is likely to absorb much of
the savings and additional revenues mobilize. And third, local
governments may not be ready to absorb additional spending at this time,
as they have just almost doubled their levels of spending, and their
local planning, budgeting and financial management systems may already
be stretched, and accountability at the local level is still weak.
(iii) Do the
devolved resources cover the devolved responsibilities? On aggregate,
more than enough revenues seem to have been devolved to match the transferred
revenue responsibilities. This holds even if we take account of
the July 2001 wage increase, and correcting the region’s own development
spending for inflation. In total, the regions received “surplus”
revenues of some Rp.21 Tr. in 2001, or 1.5 percentage point of GDP (Figure
3.__: More than enough).24
One could,
therefore, argue that decentralization “cost” the center this very
same amount: if Government could have perfectly targeted the devolved
resources, it could have transferred Rp. 21 Trillion less than it actually
did. Lewis (2001, p.330) estimates an even higher surplus of Rp.27.5,
but this was before the wage increase and the subsequent disbursement
of the contingency fund. Lewis also estimates separately the surpluses
of the provincial level and the local level. His judgment is that
whereas at the provincial level the extra revenues more or less just
cover the extra expenditures, local governments received most of the
surplus. This findings jives also with the disbursements from
the contingency fund, of which a disproportional amount was disbursed
to the provinces.
Further evidence
for the finding that more than enough resources were transferred can
be found in development spending. Budgeted regional development
spending made a significant jump in 2001, from an (annualized) Rp. 14
Tr. in FY2000 to a planned Rp. 26 Tr. in FY2001 (Table 3: Consolidated
Development Spending).
The regions
may have been forced to cut back slightly on these plans after the July
wage increase, but there is every reason to believe that development
spending in the regions rose significantly. This is good news
for government development spending as a whole: because of the increase
in regional development spending, the drop of central development spending
as a percent of GDP did not lead to a n overall decline: for both 2000
and 2001, this was budgeted to be about 5.1 percent of GDP.
Thus, the regions
as a whole do not seem strapped for funds. The increase in development
spending in the regions also suggests that there is more than sufficient
funds to cover the (recurrent cost of) functions transferred.
Therefore, the often-heard argument that the regions spend mostly on
“bureaucrats” and have too few resources for “services to the
people,” therefore seems to be a red herring. In fact, on aggregate,
wages make up little over 50 percent of regional spending (Table 4:
Summary Regional Expenditures). Moreover, it is a misconception
that only development spending can be considered as service delivery:
as the civil service numbers in chapter 2 suggest, some 70 percent of
the wage bill is paid to teachers and health workers, civil servants
that provide direct services to the people.
In conclusion,
on aggregate more than sufficient revenues were devolved to cover the
additional expenditure responsibilities of the regions. Little
can be said about whether this was enough to cover expenditure levels
sufficiently large to cover some minimum standard of services in the
regions.
How equal
is the new intergovernmental fiscal system?
Sufficient
resources for the regions on aggregate disguise large variations among
the regions in fiscal capacity. Even after redistribution through
the DAU, in FY2001 the richest local government had more than fifty
times as much revenues per capita as the poorest one (Table A.2).
The poorest region has only 20 percent of the per capita revenues as
the average. And the variation among the regions as measured by
the Gini coefficient for the per capita revenues for the regions is
some 0.39.
Some of this
variation can be explained by the small size of the units of local government
in Indonesia.25
But even if
one aggregates revenues at the provincial level, the variation in revenue
capacity remains large: the richest province has about ten times
as much revenues per capita as the poorest one, and the poorest one
has only some 40 percent of the revenue capacity of the average one.
For comparison, in the US, the poorest state as about 65 percent of
the revenues of the average state, and in Germany, any state falling
below 95 percent of average gets subsidized. In Russia, the variation
in the \[56] oblasts is more in line with that of Indonesia: the richest
of the 89 regions has revenues per capita some 40 times higher than
the poorest, which is still considerably less than that among Indonesian
local governments, although larger than among the provinces.26
In Brazil, the richest state has 2.3 times the revenues per capita of
the poorest state.27 In China, expenditures per capita
in the richest province was some 17 times that of the poorest one, but
excluding the city provinces of Shanghai, Beijing, and Tianjin, the
disparity fell to 5.5 to 1.28
Why are inequalities
in fiscal capacity so high among Indonesia’s regions? After
all, the Law on Fiscal Balance promises a system that would equalize
fiscal capacity among the regions taking into account the regions’
own fiscal capacity and fiscal needs. Two causes of inequality
stand out: (i) The large variation in own fiscal capacity among Indonesia’s
regions, and (ii) the imperfections in the equalization mechanism of
the DAU.
Disparities
in own fiscal capacity.
Own fiscal
capacity among the regions varies widely. Much of the variation
is due to revenues from natural resource sharing. The few regions
that do receive substantial amount, but not all. The distribution
of the personal income tax share is highly unequal as well, with some
regions receiving no revenues at all from this source. Relative
to these two sources of revenues, variation in own taxes (PAD) per capita
is relatively small.
Equalizing
properties of the DAU29
The DAU allocations
reduce disparities in fiscal capacity among the regions. Whether
it does so by enough is hard to tell, in part because the concept of
equalization itself is left vague in the law, the regulations, and even
in the political debate surrounding the DAU allocation. Law 25/1999
requires the DAU to be allocated such that it reduces the disparity
between expenditure needs and “economic potential.” The allocation
is to be done by formula taking objective factors of needs into account
and own fiscal capacity. The elucidation of the law mentions explicitly
the factors population, area, poverty, and geographical condition—which
was later interpreted as cost differences. There are several reasons
for this, but the key one was that the allocation of the DAU was restricted
by the need to give regions enough finance to pay for the devolved government
apparatus, and the resulting “hold harmless” clause was misinterpreted.
The hold harmless
clause. In the run-up to 2001, the authorities had to solve the
intractable issue of matching the new expenditure assignments with the
new intergovernmental fiscal system. Expenditures “landed”
in those regions where the government apparatus that was to be decentralized
was located. To avoid that the bulk of the money, the DAU, would
land somewhere else, and thereby risking that civil servants went unpaid
and services break down, it was decided that the DAU allocation should
hold regions harmless compared to what they received before in SDO and
INPRES and what they had to spend extra on the devolved government apparatus.
This became known as the “base amount,” equal to 130 percent of
SDO and 110 percent of INPRES of (annualized) FY2000 amount. A
true hold harmless clause would have ensured that regions would not
fall below that amount. Instead, it became a minimum—absorbing
some 80 percent of the total DAU.
The DAU allocation
for 2001 therefore became strongly correlated with past distribution
of grants. This favored the resource rich regions, because the
old INPRES system was implicitly compensating these regions for the
revenues generated by them. On top of the base amount, each region
received an amount based on a formula that included fiscal needs and
own fiscal capacity. But because information on revenue sharing
was not yet known at the time of formulating the DAU, natural
resource revenue shares were not counted as own fiscal capacity.
The “hold
harmless” took on a different meaning in 2002. MOF had proposed a
more equalizing DAU allocation for 2002—among others by now taking
natural resource revenues into account, and the Regional Autonomy Advisory
Board had approved the proposal. But the regions that stood to lose
lobbied hard with Parliament, and even though according to Law 25/99
Parliament had no formal say in the distribution of the DAU (as opposed
to the aggregate amount), it insisted that each region would get at
least as much as DAU as in 2001.
The bottom
line of all this is that the DAU allocations are less equalizing than
one would expect based on the Law. Regression analysis in Hofman, Kajatmiko,
Kaiser (partially reproduced in Annex 3) shows that the DAU is positively
correlated with own fiscal capacity, and shows a strong relationship
with the wage bill. Nevertheless, the DAU still equalizes in the
sense that variation in revenue per capita as measured by Gini coefficient
or coefficient of variation is reduced by the DAU (Table 5). The
reason for this is that regions with high own fiscal capacity do receive
less DAU as a proportion of that own fiscal capacity even though in
absolute amounts they may receive more that regions with low own fiscal
capacity.
Fiscal Dependency
Historically
Indonesia has had one of the most centralized tax systems in the world
(Ma, Jun 1997). The recent fiscal decentralization actually increased
regional fiscal dependence, as measures by the share of own revenues
(PAD) in total revenues.International evidence suggests that this high
degree of dependence is inversely associated with governance outcomes
(de Melo, Luiz and Matias Barenstrein 2001), and fiscal dependence should
therefore be a concern for Indonesia.
Law 34/200
on regional taxes should have addressed the issue of fiscal dependency.
However, the approach taken in that law led to another set of problems.
Law 34/2000 allows for regions to issue their own tax regulations, as
long as they abide by certain (sound) principles. This is far
more liberal than Law 18/1997 which only allowed a limited number of
taxes specified in the law, with high hurdles on additional taxes. At
the same time, Law 34/2000 did not devolve a tax most suited for regional
governments: the land and real estate tax.30 Arguably the
approach of Law 18/1997 may be better, even though it did perhaps not
encompass an appropriate tax base for the regions “Nuisance” or
“predatory” taxation have received some attention during the first
year of decentralization. Many of those taxes are technically illegal,
and improved central supervision is clearly one important remedy to
this problem,. But a more fundamental solution to the “revenue
hunger” of the regions will probably include enhanced local tax bases
and marginal levels of revenue discretion. The trouble with this
solution is that those taxes that may be most lucrative from a revenue
basis, are also be less desirable from an equalization of fiscal capacity
perspective (e.g., natural-resources or property/service based taxes
which will disproportionately benefit urban areas).31
Directions
for Reforms
Indonesia’s
intergovernmental fiscal system can be much improved. The broad
orientation of reforms is to have the relatively rich regions “fend
for themselves” with own tax base, shared taxes and commercial borrowing.
The poorer regions are to get support through DAU, DAK, and access to
well-managed central lending and on-lending facilities to enable them
to provide similar quality services at similar local tax rates throughout
Indonesia.
Improving the
intergovernmental fiscal system in this direction requires, among others:
- Moving to a more
equalizing DAU by phasing out the transitional elements in the allocation.
- Restricting local
taxes to a closed list over which the regions have tax rate autonomy—possibly
within centrally set limits.
- This list of taxes
should include the property taxes, and could include a local surcharge
in the personal income tax and payroll taxes and selective business
taxes. Expanding motor vehicle use or fuel taxation is a further
option.
- Deciding on a transparent
and consistent treatment of natural resource revenues in revenue sharing
and in the equalization formula.
- Introducing a selective
system of specific grants—combined with an (on) lending window—to
promote the financing of national priorities at local level. A
larger DAK could be financed from a gradual reduction in the center’s
own development spending on regional functions.
References
1
The findings, interpretations and conclusions expressed in this paper
are entirely those of the authors. They do not represent the views of
the World Bank, its Executive Directors, or the countries they represent.
This paper draws on the forthcoming World Bank Regional Public Expenditure
Review for Indonesia, and on Hofman, Kaiser and Kajatmiko (2001). The
authors wish to express thanks to Jorge Martinez, Roy Bahl, Richard
Bird, Roy Kelly, Dana Weist, Blane Lewis, Bernd May, and Machfud Sidik
for the many helpful discussions on the topic of the paper. We thank
Fitria Fitriani for excellent research assistance.
18
See Annex Table XXX for detail, and Silver, Christopher, Iwan J. Azis,
and Larry Schroeder. 2001. Intergovernmental Transfers and Decentralization
in Indonesia. Bulletin for Indonesian Economic Studies 37 (3):345-62
19
Law 25 is not specific on whether the 25 percent is before or after
revenue sharing. PP104 has taken this interpretation, which has apparently
been accepted by Parliament and the regions.
20
Revenues in the approved budget for 2001 were Rp.263T and revenue sharing
Rp. 20 T, which yields a DAU of Rp. 60T. Actual revenues as per
preliminary outcome data suggests revenues of Rp. 299T. Assuming
the same amount of revenue sharing, this would result in a DAU of Rp.
69T
21
This Parliamentary involvement in the distribution seems against Law
25/1999 which specifies that the Regional Autonomy Advisory Council
proposes the distribution of the DAU to the President, who approves
it by Presidential decree.
22
The sharing of the personal income tax on a derivation basis was decided
at the last moment, and inspired by a conversation the Minister ff Finance
had on a trip to disseminate the decentralization laws. Art .31C
of Law 17/2000 describes the sharing. However, the article is unclear
whether the sharing of the personal income tax is based on the residence
principle or the place of work.
23
South Africa is operating such a system. The constitution prescribes
functions for the provinces and municipalities, and orders the government
to give each level of government its “equitable share” of the national
revenues. The equitable share is determined in the context of
budget preparation, which in the case of South Africa is ased on a medium
term expenditure framework. If, say, more priority is put on health
case, the equitable share of the province—which is responsible for
it, will; get a larger share.
24
One caveat here is the assumption that all development projects implemented
by the Kanvils and Kandeps continue to be financed from the central
budget. This seems to have been the case but the center is now
trying to find ways do devolve this financing responsibility.
25
This point was made by Richard Bird in a comment at the Bali conference
on Decentralization in East Asia, January 10-11 2002.
26
Martinez-Vazquaesz, Jorge and Jameson Boex (1998): Fiscal Decentralization
in the Russian Federation: Main Trends and Issues. Report prepared
for the World Bank/EDI, December.
27
Issues in Brazilian Federalism, Draft World Bank Report, May 8, 2001
28
World Bank, 2001 China: Provincial Expenditures Review, Report No. 22591-CHA),
November, Washington DC.
29
For more detail on this issue see Lewis (2001) and Hofman, Kajatmiko,
and Kaiser (2002)
30
The thinking in the ministry of Finance on this has already shifted
in the direction of devolvingthis tax.
31
Zhuravskaya (2000) argues that the intergovernmental fiscal system prevailing
in the Russian federation in the ninetees provided no incentives to
increase the local tax base or provide public goods. Shared revenue
allocations effectively penalized regions that raised their own revenues.
When regions effectively have no opportunity to increase local revenues,
they will also have little incentives to increase the local tax base
and will over-regulate local business.
