Centre for Research on Inequality,
Human Security and Ethnicity
University of Oxford
Policy briefing
www.crise.ox.ac.uk
March 2009 CRISE Policy Briefing No. 2
Inequality and fiscal policy
Properly designed, fiscal policies can be effective tools for redressing social and economic
inequality. Frances Stewart and Rajesh Venugopal examine how.
Fiscal policy—the level and composition of a government’s taxation and expenditure—can
be used to address different kinds of
inequality within a country: inequal-
ity between individuals or households
(vertical inequality) and inequality
between groups (horizontal inequal-
ity). Taxation is particularly relevant for
addressing economic inequalities, and
expenditure for social inequalities.
The first step in devising appropri-
ate fiscal policies is to gain an in-depth
understanding of the status and dynam-
ics of group identity and horizontal
inequalities along various dimensions
through a scoping study. The study
should examine whether these groups
are ranked hierarchically, how they are
regionally distributed, and whether they
are specialised in particular activities.
Tax policy
Tax policy in developing countries
typically suffers from problems of level
and composition: there is often weak
revenue mobilisation and the system
frequently fails to address inequality.
The tax system can contribute to re-
ducing inequalities directly by increas-
ing its progressiveness, and indirectly
by raising additional revenue to finance
expenditure devoted to reducing in-
equalities.
Increasing the progressiveness of
the tax structure
This method works best where groups
are hierarchical in terms of income such
that horizontal inequalities (HIs) and
vertical inequalities (VIs) overlap.
Direct taxes on income, profits and capi-
tal gains are usually the most progres-
sive part of any developed country tax
system. In addition, property taxes are
also typically progressive because poor-
er people (groups) have less property
than richer ones. A simple approach to
improving progressivity is to exempt
all property below a certain value. The
start-up costs of meeting data require-
ments in the form of a cadastre may be
onerous but have long-lasting benefits,
including by improving the security of
property rights. For developing coun-
tries, on average, land taxes account for
0.5% of GDP and around 18% of local
government revenue, but Indonesia has
a much more extensive tax with higher
rates which generate two-thirds of local
governments’ own revenue (Bird and
Slack 2006; Kelly 2004).
Indirect taxes A value-added tax which
has exemptions for basic necessities (in-
cluding food) can be progressive. This
is the situation, for example, in Kenya:
according to estimates for 1997, VAT
was progressive, allowing for exemp-
tions and zero rating (Wanjala 2007).
Almost half developing countries that
adopt VAT exempt food (ibid). VAT
could also incorporate higher rates u
Three principles of tax design
n Identity should not be used as the
basis for tax liability as this goes
against basic principles of tax equity.
Rather, horizontal inequality should
be addressed in an indirect way, e.g.
by using geographical, income level
or occupational criteria as a basis for
taxation.
n Redistribution is not the only goal
of tax policy: other objectives should
not be neglected. Where possible,
taxes should be designed so as to ful-
fil each of the objectives; where not,
any tradeoff needs to be considered
carefully.
n There is no single, overarching
policy framework: countries vary
greatly and each case has to be exam-
ined individually.
Inequality and fiscal policy
Policy Brief ing No. 2
Inequality and fiscal policy
on luxuries, although multiple rates
cause administrative complexity. An
alternative is to levy particular excise
taxes on luxuries. Other indirect taxes
that are likely to improve VI and prob-
ably HI would be heavy motor vehicle
duties and taxes on airlines and flights.
Fuel taxes can be differentiated so
as to tax the sorts of fuels that richer
people use and not poor people’s fuels.
Similarly, food subsidies can improve
distribution when specifically directed
at staple foods.
Tailoring taxes to group behaviour
Privileged versus deprived groups can
often be identified through their loca-
tion, their productive activities and their
consumption behaviour and taxes can
then be designed so as to reduce group
inequality:
Location Decentralised administrations
can devise revenue-sharing formulae to
improve the position of poorer districts
and ipso facto of poorer groups. This
was attempted in Indonesia’s post-2001
‘Big Bang’ decentralisation. Govern-
ments can also use property taxes to
differentiate between wealthier and
poorer locations.
Occupation Many HIs arise from differ-
ential group concentrations in special-
ised economic activities, such as export
versus food crop producers; merchants
versus farmers; or formal versus in-
formal sector activities. This makes it
possible to address HIs by differential
taxation of exports, wholesale trade, or
the formal sector.
Consumption It is possible to target
taxes towards forms of consump-
tion that benefit wealthier groups. For
example, taxes on alcohol would not
affect Muslims, who form the relatively
deprived groups in many societies.
User fees that target services little
used by poorer groups, such as higher
education or heavy use of water (with
appropriate exemptions), may yield
beneficial results.
Altering the regional balance of taxa-
tion Regionally based HIs can be due
to economic geography, as well as
political and historical factors, often
including colonial policy. For unitary
states, the main redistributionary con-
tribution of taxes is to address VI, while
raising revenues to finance HI reduc-
tion through allocating expenditures
disproportionately to poorer regions. A
more federal or decentralised state can
be directly responsive to regional HIs,
by adjusting the regional distribution
of revenue and expenditure to benefit
poorer regions. In Nigeria, for example,
the 1989 formula for the allocation of
central revenues to the states was 30%
equally distributed among the states,
40% according to population, 15% (in-
versely) for social development, and 5%
for local tax effort.
Raising revenue
In addition to increasing the progressiv-
ity of the tax system to reduce inequal-
ity, a major contribution of the tax sys-
tem is in raising revenues to undertake
HI-reducing expenditures. Where total
revenues are low, raising tax revenues
is an important part of policy towards
HI. Most low-income countries have a
low tax base (average 10.4% of GDP),
and need to focus on increasing rev-
enues. Key areas to be addressed are
low tax rates, weak administration and
low compliance.
Low tax rates By comparing rates and
revenue raised from different types
of taxation in relation to GDP in the
country with the average in a compara-
ble set of countries, particularly low tax
rates can be identified where rises in
rates would be warranted.
Administration and compliance Weak-
nesses in administration can be
overcome by reforms and technical
assistance. Low compliance is partly a
matter of the nature of the tax admin-
istration: effective tax administration
requires well-trained officials and low
levels of corruption. Compliance is also
affected by the prevalent culture: if it
is believed that few comply then low
levels of compliance are likely. Donors
and central governments can introduce
incentives for greater compliance by
making matching contributions. Durand
and Thorp’s (1998) study of the suc-
cessful case of Peru, in which reforms
of the administration greatly increased
revenue, identifies three necessary
conditions for success: political support
from on top; ongoing bureaucratic re-
generation; and general improvements
in the relationship of state and society.
Expenditure policy
The impact of government expenditure
on HIs can be analysed in relation to
Selected developing countries 2005
Government revenue
(excluding grants) as a
percentage of GDP
Source: WDI, World Bank *2002
*
*
Inequality and fiscal policy Inequality and fiscal policy
Policy Brief ing No. 2
three distinct types of expenditure:
(a) infrastructure investment (the
government accounts for over three-
quarters of investment in poor coun-
tries); (b) contracts and employment
(the public sector typically accounts
for half of formal sector employment);
and (c) service provision (the provision
of health and education, important in
themselves, and they contribute to
improved incomes).
Evidence suggests that there is
often inequality in the distribution in
each of the types across racial, reli-
gious, ethnic or geographical groups.
It is essential to ensure appropriate
group participation in each of the three
types of government expenditure.
Most attention is generally paid to the
distributional implications of the third
type (service provision), whereas the
first two are often more important in
their impact on HIs and, in particular,
the visible and felt impact. At a mini-
mum this implies anti-discrimination
legislation and monitoring, but in some
contexts it may also be appropriate to
introduce positive discrimination. Such
policies towards the allocation of both
government contracts and govern-
ment employment have been adopted
in quite a large number of countries,
including Canada, Malaysia, South
Africa, Northern Ireland and the US.
They include:
n Anti-discrimination employment
legislation.
n Guidelines for all public decision
makers.
n Setting targets and quotas.
n Careful monitoring of progress.
n Competitive and transparent bid-
ding procedures for contracts.
n Incentives and penalties, including
allocating contracts only to firms where
fair employment practices are being
followed.
n Providing set asides for contracts
for businesses controlled by particular
groups, or preferences for such groups.
The third category of expenditure,
government-provided goods and serv-
ices, can be categorised in the follow-
ing way: (a) goods with non-allocable
benefits (e.g. defence expenditure or
debt servicing); (b) goods with alloca-
ble benefits across groups but not by
region (e.g. expenditure on national
television which benefits only those
with TV sets; or patronage of a particu-
lar religion); (c) goods with allocable
benefits across regions but not intra-
regionally (e.g. expenditure to limit
environmental damage); (d) goods
with allocable inter-regional and intra-
regional benefits—this includes much
government expenditure, e.g. expendi-
ture on social services and economic
infrastructure; and (e) transfer pay-
ments, which include direct payments
to communities or households or
individuals administered according
to particular criteria. These can be al-
located among individuals and groups
according to their design.
Each of these categories of spend-
ing affects HIs to a different degree, and
can hence be used to remedy them in
distinct ways. Broadly speaking, the last
two categories—goods and services the
benefits of which are directly allocable,
and transfer payments—are the most
useful for addressing VIs and can also
be designed to address HIs.
To summarise, with respect to gov-
ernment expenditure, improvements in
HI can be achieved via reallocation in
the production, running and consump-
tion of publicly funded services:
n By a policy towards participation in
government contracts for the construc-
tion of public facilities, and in employ-
ment in the public sector.
n By a focus on pro-poor public
goods and those that favour particular
groups, e.g. subsidising radio not tel-
evision, bus transport not air travel; and
supporting the language and religion of
poor groups.
n By locating a higher proportion of
public goods in regions where poor
groups are concentrated; and within
regions in specific locations where
poorer groups are concentrated. For
example, in Lima, the white elite live in
one area and indigenous migrants in
another. Schools, transport, sanitation,
and housing located in the slum areas
of Lima would benefit indigenous peo-
ple, not the white population. Any such
improvements in HI would, of course,
also improve VI.
n By appropriately designed transfer
payments which can improve both VI
and HI significantly. u
n Government expenditure is often progressive, benefiting the poor more than
the rich, at least in proportion to their original incomes. It follows that in so far as
taxes raised are neutral vis-à-vis income distribution, even if not progressive, higher
revenue and expenditure will improve income distribution.
n It is politically much easier to achieve redirection of expenditure towards particu-
lar groups if this is through extra expenditure, rather than if it involves taking some
benefits away from existing groups.
n Raising taxes also plays a political role which is especially important in post-
conflict countries: those who pay taxes have a perception of having a stake in the
government, particularly representative government. As Ross (2004) has argued,
turning around a one-time rallying cry, there is ‘no representation without taxation’.
His evidence demonstrates that the greater the proportion of government expendi-
ture financed by taxation the more representative the forms of government, suggest-
ing that raising revenue would contribute to reduced political HIs.
Why is raising government revenue important for
redistribution?
Inequality and fiscal policy
Policy Brief ing No. 2
Data
A first requirement in designing fiscal
policies to reduce HIs is to identify the
nature of the HIs in the country. Data on
HIs can be difficult to find as household
surveys often do not include questions
on ethnicity or religion, sometimes
because of their political sensitivity.
The first need in a country is to conduct
an inventory of available data, and of
urgent data needs. Working with re-
cipient countries to develop appropriate
data collection tools is then important.
In the short term, it is often possible
to get sufficient data using shortcut
methods. One possibility is to take
some other characteristic as a proxy for
ethno-cultural difference, for example
data by region or by the language spo-
ken, which are often more readily avail-
able than socio-cultural data. The de-
gree to which regional data are useful,
however, depends on how far identity
groups are geographically segregated.
As a rule of thumb, region is a useful
proxy for measurement and policy if
more than half the members of the
deprived group, and less than half of
the privileged group, are concentrated
in the targeted region. In many African
countries, ethnic and religious groups
are regionally concentrated so regional
inequality may be a suitable proxy for
ethnic or religious inequality and in
some cases region itself defines group
identities. Another proxy for ethnicity is
the use of a language variable, which is
sometimes available where ethnic vari-
ables are not—as in Indonesian surveys
in the New Order period.
Imaginative use of existing data
generally contributes considerably.
Possible data sources include:
Census data, where ethnic or language
data are often included, and sometimes
religion.
Centre for Research on Inequality, Human Security and Ethnicity (CRISE)
Oxford Department of International Development (Queen Elizabeth House)
University of Oxford
3 Mansfield Road
Oxford OX1 3TB, UK
Tel +44 1865 281810, Fax +44 1865 281801
The Demographic and Health Surveys
(DHS), many of which include ethnic
identification, as well as information on
access to social services and ownership
of domestic assets.
Living Standards Measurement Surveys
(LSMS) which sometimes include eth-
nic variables.
Regional data from household surveys,
the census and sometimes public ex-
penditure accounts.
Specific sectoral data (e.g. from schools
and hospitals) which often contain eth-
nic and regional information.
Where there remain major gaps,
small household surveys can be rapidly
conducted to supplement existing data,
including a limited number of basic
variables (e.g. asset ownership; nature
of employment; incomes; educational
attainments; and anthropometric data).
The role of aid-givers
A first requirement is to promote
awareness of the need for such policies
and their nature. Opportunities for this
arise where aid donors are involved in
general policy discussions, for example
in association with public expenditure
reviews, or the Poverty Reduction Strat-
egy Papers, or economic policy reform
and adjustment more generally. In each
case, donors should include discus-
sions about the need to review HIs
and to use fiscal policy to correct them
where they are high.
Donors can provide technical as-
sistance in data analysis and support
for supplementary surveys as judged
necessary.
Donors can assist governments in
policies to raise revenues, including
supporting improvements in tax ad-
ministration, and in compliance, and in
identifying which tax rates are relatively
low and might be increased. This policy briefing is based on CRISE
Working Paper 65: The Implications of
Horizontal and Vertical Inequalities for
Tax and Expenditure Policies by Frances
Stewart, Graham Brown and Alex Cobham.
Available to download at www.crise.ox.ac.
uk/pubs/workingpaper65.pdf
Donors can also assist in identify-
ing types of expenditure that would
reduce inequalities, drawing on avail-
able data. Where governments are
unwilling to correct sharp HIs, donors
may themselves contribute to reducing
HIs through the distribution of aid; they
may work with NGOs to promote ex-
penditures reducing HIs; and they may
try to change government policy via
policy dialogue and conditionality. n
References
Bird, Richard, and Enid Slack. 2006. The role of
the property tax in financing rural local gov-
ernments in developing countries. Internation-
al Tax Program Paper 0608. International Tax
Program, Institute for International Business,
Joseph L. Rotman School of Management,
University of Toronto.
Durand, Francisco and Rosemary Thorp. 1998.
Reforming the state: a study of the Peruvian
tax reform. Oxford Development Studies 26 (2):
133–151.
Kelly, Roy. 2004. Property taxation in Indonesia.
In Richard Bird and Enid Slack (eds), Interna-
tional Handbook of Land and Property Taxa-
tion. Northampton, MA: Edward Elgar.
Ross, M. 2004. Does taxation lead to represen-
tation? British Journal of Political Science 34:
229–249.
Wanjala, Bernadette. 2007. Design and per-
formance of Kenya’s tax system: an inequal-
ity perspective. In Society for International
Development (SID), Readings on Inequality in
Kenya: Sectoral Dynamics and Perspectives.
Nairobi: SID.
www.crise.ox.ac.uk
Read more…