The Distributional Effects of Tax‐benefit Policies under New Labour: A Decomposition Approach*
| DOI | http://doi.org/10.1111/j.1468-0084.2011.00684.x |
| Published date | 01 December 2012 |
| Date | 01 December 2012 |
856
©Blackwell Publishing Ltd and the Department of Economics, University of Oxford 2012. Published by Blackwell Publishing Ltd,
9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 74, 6 (2012) 0305-9049
doi: 10.1111/j.1468-0084.2011.00684.x
The Distributional Effects of Tax-benefit Policies
under New Labour: A Decomposition Approach*
Olivier Bargain
DEFI, Aix-Marseille Université, Chˆateau La Farge - Route des Milles 13290 Les Milles,
France (e-mail: olivier.bargain@univmed.fr)
Abstract
I revisit the distributional effects of tax-benefit policy reforms under New Labour using
counterfactual microsimulations embedded in a Shapley decomposition of time change
in inequality and poverty indices. This makes it possible to quantify the relative effect
of policy changes compared to all other changes, and to check the sensitivity of this pol-
icy effect to the use of (i) income vs. price indexation, and (ii) base vs. end period data.
Inequality and poverty depth would have increased, and the sharp fall in child poverty
would not have occurred, had the reforms of income support and tax credits not been
implemented.
I. Introduction
For analysts and policy makers, it is crucial to know whether actual tax-benefit reforms
have achieved their objectives in terms of redistribution. Ausual way to identify the impact
of policy changes on income distribution is to decompose inequality indices by income
components, possibly isolating the role of taxes or social transfers, and to repeat the assess-
ment at different points in time (before and after important reforms). Yet, the decomposi-
tion by income types, as introduced and axiomatized by Shorrocks (1982), has well-known
limitations (see Shorrocks, 1999, for an overview). In particular, the contribution assigned
to a specific factor is not always interpretable in an intuitively meaningful way (see Chant-
reuil and Trannoy, 1997). Moreover, constraints are often placed on the types of poverty
and inequality indices which can be used, some indices requiring the introduction of a
vaguely defined ‘interaction’ term to maintain the decomposition identity. Equally crucial
for policy analyses, measuring the contribution of taxes and transfers to overall inequal-
ity/poverty at different points in time does not allow disentangling the pure effect of policy
changes from their interaction with the underlying population. For instance, social assis-
tance schemes may appear more redistributive because of their increased generosity or
ÅI am grateful to TimCallan for useful comments. I am indebted to all past and current members of the EUROMOD
consortium as well as to those involved in the development of the model. Simulations performed in this study rely
on the Family Expenditure Survey (FES) made available by the UK Office for National Statistics through the Data
Archive. Material from the FES is Crown Copyright and is used by permission.
JEL Classification numbers: H23, H53, I32.
The distributional effects of tax-benefit policy changes 857
because of automatic increase in welfare payments as unemployment rises. This approach,
sometimes referred to as the ‘actual payments’ method, is used by Jenkins (1995) and
Goodman, Johnson and Webb (1997) to analyse inequality trends in the United Kingdom
in the 1970s and 1980s. Their finding that the tax-benefit system of the late 1980s was
not less redistributive than that of the late 1970s is partly due to the fact that they do not
account for changes in the underlying market income distribution.
Alternatively, it is possible to use tax-benefit microsimulations to construct counter-
factual situations and to disentangle the pure effect of a policy change from changes in the
environment in which the policy operates, particularly changes in market income inequal-
ity (see Atkinson, 2005, for a general statement). However, measures may be sensitive to
the choice of indexation factor used in the ‘no reform’ counterfactual scenario. Results
may also depend on the underlying population used to evaluate policy change, either the
base-period or the final-period data. This issue has been investigated in the literature on
tax progressivity (e.g. Dardanoni and Lambert, 2002; Lambert and Thoresen, 2009), but
has received little attention in actual policy evaluations.1
The present study revisits the effect of tax-benefit policy changes under the first New
Labour government (1997–2001) in the United Kingdom. This era has received a lot of
attention in the literature given the important redistributive policies put in place.2In fact,
these few years are also characterized by an economic upturn accompanied by a large
increase in market income inequality. In this context, it is important to disentangle the
pure role of policy reforms from all other changes and notably changes in market income
inequality.To do so, I produce a series of counterfactual simulations using data for the base
and end periods and relying on tax-benefit calculations for the base- and end-period fiscal
systems. These simulations are embedded in a formal framework whereby the change in
poverty/inequality indices is decomposed into three components: (i) the effect of changes
in tax-benefit policy; (ii) the effect of adjusting tax-benefit monetary parameters according
to market income growth; and (iii) all the changes not directly linked to tax-benefit policies.
This decomposition approach offers several advantages. First, it allows quantifying the
pure policy effects relatively to the other effects (e.g. changes in the population, in market
income inequality, in unemployment rates, etc.). Assessing the relative size of these two
effects is generally not done in the policy literature and in particular in the evaluation of the
UK reforms (for instance in Adam and Browne, 2010). Secondly, I check the sensitivity of
the policy effect measure to the underlying population, either the base- or end-period data.
I am not aware of evaluations that actually check the role of the reference year, apart from
Jenkins and van Kerm (2005) in a slightly different context.3Thirdly, symmetry arguments
suggest that the two alternative policy effects, that is, measured on end- or base-period
data, should be averaged, as suggested by Shorrocks (1999) and Kolenikov and Shorrocks
1An exception is the study of Clark and Leicester (2004) who carefully investigate the distributional effect of
policy changes over the 1980s and 1990s in the United Kingdom and provide an extensive sensitivity analysis.
Their approach is not embedded in a formal decomposition framework as that suggested in the present article. The
decomposition helps to clarify many of the issues regarding the assessment of the pure policy effect, as explained
below.
2See Sutherland (2001), Brewer et al. (2004), Hills, Sefton and Stewart (2009) and, specifically on child poverty,
Sutherland and Piachaud (2001), Dickens and Ellwood (2003) and Brewer et al. (2003).
3For instance, Adam and Wakefield (2005) analyse the distributional impact of tax-benefit reforms implemented
between 1997 and 2005 in the United Kingdom as assessed on end-period data only.
©Blackwell Publishing Ltd and the Department of Economics, University of Oxford 2012
To continue reading
Request your trial