Attitudes to Income Distribution
is a substantial literature on the way in which concepts such as
“equity” can be given meaning in an economic context. These are
typically expressed in the form of income inequality, poverty or social
welfare. However, there is a fundamental problem in that the constructs
developed in the economics literature and elsewhere may not accord well
with the way lay people view issues of social equity. This raises
questions that may impact upon policy design and evaluation: for example
the sort of consensus approach to social equity that underlies the
concept of "fairness" in the design of taxes and social
security systems, or the notion of effectiveness of anti anti-poverty
programmes at local or national levels.
1 Intellectual Origins
are two principal strands to the work of the current workshop group (see
section 5 for those involved).
Social welfare and the economics of income distribution
researchers involved in the proposed workshop have all been active in
aspects of this research (see for example Amiel and Cowell 1999, Cowell
2000, Schokkaert 1992). However much empirical work remains to be done
on the structure of personal and social determinants of distributional
Use of experimental techniques in economics
methodology is now well established in a number of economic
applications. It is used to analyse behavioural choices where an
econometric approach to market outcomes is inappropriate because of lack
of data, measurement difficulties or thinness of samples. Although there
is a huge literature on experimental bargaining and the problem of
experimental design in that context, similar work on normative economics
is still rare.
However related techniques have been applied to
normative issues. A prime example is the use of the questionnaire
approach (Amiel and Cowell 1999, Gaertner and Jungeilges 1999,
Schokkaert and Devooght 1998). This literature has provided important
insights on (a) whether or not standard assumptions made in the
formulation of distributional orderings are appropriate, and (b) the
factors accounting for systematic differences between certain groups of
persons in their attitudes on issues such as inequality, poverty and
1.2 Types of Experiment
types of experimental approach have become established procedures in
economics, both of which are relevant to the current proposal:
methods involving active participation by subjects have been developed
for a variety of media. The role of laboratory experiment is usually
performed by interactive computer methods: individual choices from
menu-driven programs are logged, and the responses used as a database
from which to infer the structure of people's preferences and judgements.
Although computerised approaches to the analysis of
distributional attitudes are rare, there are alternative methods,
including participation in group decision-making. Notable among these is
the pioneering experimental development by Kroll and Davidovitz (1999)
that investigates attitudes to inequality and risk simultaneously.
complementary approach to the participatory experiments outlined above
is the sort of questionnaire experiment discussed in Amiel (1999). The
general form of these experiments has been as follows
are asked numerical and/or verbal questions about alternative
distributions in carefully specified hypothetical environments.
high level of participation is ensured by making use of student
respondents during part of their lecture/class programme.
controls for consistency of responses are incorporated in the question
studies focused upon whether responses conform to standard axioms of
economic analysis and whether the background of respondents is relevant
to the type of response elicited. Results from paper questionnaire
studies show that while studying economics and being male increases the
likelihood of responses being consistent with economic orthodoxy, other
background variables such as income, political attitudes and age are
1.3 Computer-Based Methods
are several reasons why computer-based investigations may yield greater
opportunities to the researcher than a paper-based questionnaire:
of question format. Computer technology can easily provide respondents
with a variety of formats of presenting the same information – such as
numerical problems, in words, graphics, for example. This enables the
researcher to control for biases that may arise from concentrating on a
single-format approach as in a paper questionnaire.
flexibility. Computer technology also allows the researcher to enrich
the content of the experiment: for example respondents can be presented
with a variety of specific instances drawn at random, which would be
very difficult in a questionnaire study.
It is important to learn whether responses to questionnaire
investigation are conditioned by the medium itself – whether for
example the procedure of filling in responses in order on a piece of
paper subtly affects the way in which the respondents express their
Limitations of the laboratory approach
traditional laboratory used in experimental economics typically consists
of a suite of computer equipment on a single site running specially
written programs. Apart from the substantial set-up and maintenance
costs of the computers and specialised software, this approach incurs
other financial penalties such as the recruitment of participants and
the monitoring of lengthy experimental sessions. Even where a laboratory
is already set up and well maintained the techniques are too inflexible
and restrictive for the purposes of developing techniques for
investigating distributional judgements rather than, for example, market
are several advantages of replacing the traditional laboratory by an
and familiar technology. Respondents are usually familiar with the
“feel” of the Internet from work and leisure experience; the
interface with respondents does not have to be specially written, as it
is managed by a standard browser.
participation, By contrast to a fixed, conventional laboratory
accommodating perhaps only a dozen participants at one time, the
marginal cost of increasing the sample size is comparatively low.
with oral interview: subjects can be allowed to regress, clarify or
expand on responses; logging these extensions and modifications is
easily be done on the web server.
Internet experiment can be done in a classroom environment so as to
supervise conditions of participation and can be subject to the same
controls as laboratory or questionnaire experiments.
Relationship to questionnaire studies
involvement in experiments. Subjects may respond differently according
to whether they see themselves as directly involved in the world of the
questionnaire experiment or in a position of Olympian detachment.
· Process of evaluation. The basis by which people reach judgements that conform to or deviate from economic orthodoxy can be illuminated by keeping track of a respondent’s progress through an interactive questionnaire that permits regressions and changes of mind.